DESC:FORM 2
THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENTS RULES, 2005
COMPLETE SPECIFICATION
(See section 10, rule 13)
1. TITLE OF THE INVENTION:
IOT-BASED CATTLE HEALTH AND HEAT MONITORING SYSTEM
2. APPLICANTS:
(a) Name (b) Nationality (c) Address
Areete Business Solutions Pvt Ltd An Indian Company G404, Sylvan heights, Sanewadi, Aundh, Pune 411007, Maharashtra, India
3. PREAMBLE TO THE DESCRIPTION
PROVISIONAL
The following specification describes the invention. COMPLETE
The following specification particularly describes the invention and the manner in which it is to be performed.

FIELD OF THE INVENTION
The present invention relates to monitoring the condition of cattle, and more particularly the invention relates to an IoT-based system for monitoring cattle health and heat conditions.
BACKGROUND OF THE INVENTION
India is home to a population of over 350 million cattle, and these livestock play a significant role in the country's economy. In fact, they contribute approximately 4.5% to India's GDP and provide a source of livelihood for around 300 million families residing in small and medium households. Looking ahead, the United Nations Development Programme (UNDP) predicts that the global population could reach 9.5 billion by 2050, resulting in a substantial surge in the demand for animal products worldwide. This surge is expected to be around 70%, including a proportional increase in the demand for dairy products within India.
Being primarily an agricultural nation, India heavily relies on livestock for its food production, with more than six daily consumption items directly sourced from these animals. The Asian Development Bank highlighted in a report from May 2004 that enhanced management practices related to livestock offer an effective means to alleviate poverty.
In summary, India's sizable cattle population, along with other livestock, is instrumental in contributing to the country's economy, supporting millions of families, and fulfilling the dietary requirements of its populace. Furthermore, recognizing the projected increase in global demand for animal products, including dairy, India's reliance on livestock for food production is set to play a crucial role in meeting these demands. Implementing improved livestock management practices can thus serve as a viable strategy for poverty reduction in the country.
Livestock health monitoring often involves several key challenges that require attention and proactive measures. These challenges include the identification of missed heat cycles, enhancing cattle yield, ensuring milk quality, preventing vaccination failures, implementing proper feed management, and early detection of diseases. Here are some additional points to consider:
• Timely detection of heat cycles is crucial for effective insemination practices, and any missed alerts can hinder the process.
• Providing adequate care and timely veterinarian visits are essential after insemination to support successful outcomes.
• Early vaccination plays a vital role in preventing diseases at their early stages and maintaining overall herd health.
• Calves require extra care and attention compared to adult cattles due to their vulnerability and specific needs.
• Improper feeding practices can negatively impact the taste and odor of milk, emphasizing the importance of correct feed management.
• Sick or diseased cattle often exhibit changes in behavior, such as increased resting time, which may go unnoticed if not carefully monitored.
• Failure to perform artificial insemination (AI) in a timely manner can lead to unsuccessful attempts and financial losses associated with the investment in AI.
• The cost of maintaining dairy cattle can increase, and farmers may face losses due to the rising expenses associated with dairy farming.
By addressing these challenges through proactive measures and effective livestock management, farmers can improve the overall health, productivity, and profitability of their cattle operations.
In India, the availability of comprehensive solutions for efficient cattle health management is currently limited, with only a few large-scale farmers having adopted complete herd management systems, including milking parlors, which lean towards farm automation. However, these automation solutions primarily focus on enhancing farm productivity and lack health-related alerts for cattle. Moreover, the high cost, complexity, and infrastructure requirements associated with these solutions make them unaffordable for small and medium-scale farmers. As a result, the management of cattle health and heat cycles predominantly relies on traditional knowledge and experience, leading to the potential oversight of crucial information by farmers due to their limited understanding.
Furthermore, the existing monitoring techniques utilized by farmers do not provide adequate and precise health data for the cattle. This limitation gives rise to various issues, including the manual and often missed task of checking cattle temperature. Fluctuations and prolonged high temperatures in cattle can significantly impact milk yield and production.
To address these challenges, there is a need for accessible and user-friendly technology solutions that cater to the specific requirements of small and medium-scale farmers. These solutions should provide accurate health data and alerts, empowering farmers to make informed decisions regarding their cattle's well-being and productivity. By bridging the gap between traditional knowledge and modern technology, such advancements can greatly benefit the livestock industry in India and ensure optimal cattle health management.
Accordingly, there is a need for an Internet of Things (I?T) based system for cattle health monitoring along with behavior dynamics and tracking to make the farmer’s life easier and automate the monitoring systems, alerts, and care.
OBJECTS OF THE INVENTION
An object of the present invention is to monitor various conditions related to cattle’s health and heat parameters using an IoT-based system.
Another object of the present invention is to keep an accurate record of cattle’s health and develop the Cattle genome.
Yet, another object of the present invention is to enhance labor efficiency by offering guided inputs for managing the cattle herd.
Yet, another object of the present invention is to reduce the lab?ur ??st and medication for sick cattles.
Yet, another object of the present invention is to improve cattle longevity.
Yet, another object of the present invention is to improve cattle‘s comfort.
Yet, another object of the present invention is to improve the conception rate.
SUMMARY OF THE INVENTION
An IOT-based cattle health and heat monitoring system comprising, a plurality of collar units, each worn by a cattle, a server in communication with each of the plurality of collar units via the communication network, and a atleast one portable computing device communicatively coupled to the server.
Each of the collar unit includes a plurality of sensors configured to observe and measure a plurality of health parameters related to temperature conditions and movements of the cattle in contact, a controller communicatively coupled to the plurality of sensors, that is configured to receive process the sesnor output signals and convert the signals into data packets suitable for further calculations, a power supply unit electrically connected to the plurality of sensors, the controller and the communication unit, and a housing enclosing the plurality of sensors, the controller and the communication unit therein, wherein the housing is equipped with a strap/belt and locking mechanism to secure the housing onto the body of the cattle.
Each of the plurality of the collar units is configured to measure health parameters of the cattle in contact by way of the plurality of sensors, preprocess the sensor output signals by the controller therein and route to the communication network by means of a communication module. The server is configured with an artificial intelligence module includes a data preprocessing module, a feature extraction module, a training module, and an inference engine. The computing device configured with a dashboard/mobile application, that connects with the server via a communication interface and receive notifications from the server, regarding health and heat conditions of the cattle that is in contact with each of the plurality of collar units.
The AI module is configured for receiving the data packets from each of the collar unit processing the data packets individually representing real time sensor output values transformed as a combination of time domain and frequency domain data, predicting health, and heat from the temperature sensor data, and statistical parameters derived from the movement sensor data corresponding to the cattle.
BRIEF DESCRIPTION OF THE DRAWINGS
The implementation of various embodiments can be better understood with reference to the following drawings and description. The components in the figures are not necessarily to scale, the emphasis instead being placed upon illustrating the principles of the embodiments. Moreover, the figures, like reference numerals designate corresponding parts throughout the different views.
Reference will be made to embodiments of the invention, examples of which may be illustrated in the accompanying figures. These figures are intended to be illustrative, not limiting. Although the invention is generally described in the context of these embodiments, it should be understood that it is not intended to limit the scope of the invention to these particular embodiments.
The above and other objects, features, and advantages of the present invention will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings, in which:
Figure 1 illustrates a block representation of data architecture of the IoT-based cattle health and heat monitoring system, in accordance with the present invention;
Figure 2 illustrates a pictorial view of the IoT-based cattle health and heat monitoring system, in accordance with the present invention; and
Figure 3 illustrates a block representation of the collar unit in the IoT-based cattle health and heat monitoring system, in accordance with the present invention.
DETAILED DESCRIPTION OF THE INVENTION
The present invention introduces an IoT-based system for monitoring the health and heat conditions of cattle. This system enables continuous monitoring of various health parameters by utilizing collar units attached to or worn by each individual cattle. The collar units communicate with a computing device and a server through a communication network. Each collar unit is equipped with internet of things (IoT) devices to capture multiple parameters from the cattle. The measured data is transmitted to a server for further processing and analysis. Based on predictions generated by the server or server application, alerts are produced and conveyed to the user or farmer through the computing device via the communication network. These alerts provide valuable insights into the health conditions of the cattles under observation, enabling early detection of any potential issues. With timely detection, the user can administer appropriate remedies to improve the well-being of the cattle
In the following description, for the purpose of explanation, specific details are set forth in order to provide an understanding of the present invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced without these details.
The process described herein is explained using examples with specific details for better understanding. However, the disclosed embodiments can be worked on by a person skilled in the art without the use of these specific details.
Embodiments of the present invention include various steps, which will be described below. As used in the description herein and throughout the claims that follow, the meaning of "a, an," and "the" includes plural reference unless the context clearly dictates otherwise. Also, as used in the description herein, the meaning of "in" includes "in" and "on" unless the context clearly dictates otherwise.
Hereinafter, embodiments will be described in detail. For clarity of the description, known constructions and functions will be omitted. While embodiments of the present invention have been illustrated and described, it will be clear that the invention is not limited to these embodiments only. Numerous modifications, changes, variations, substitutions, and equivalents will be apparent to those skilled in the art, without departing from the spirit and scope of the invention, as described in the claim.
The present invention is illustrated with reference to the accompanying drawings, throughout which reference numbers indicate corresponding parts in the various figures.
Referring to Figures 1 to 3, an IoT-based cattle health and heat monitoring system (hereinafter “the system (100)”), in accordance with the present invention is shown. The system (100) comprises of a plurality of collar units (10), a communication module (18), a server (20), and a computing device (40).
The plurality of collar units (10) are wearable devices, each comprises a plurality of sensors communicatively coupled to a controller (16) and power supply unit. The collar unit (10) further comprises of a housing for accommodating the plurality of sensors, the controller (16), and the power supply unit.
In an embodiment of the present invention, each of the plurality of collar units (10) is configured to measure a plurality of cattle parameters including rumination (%), body temperature (F), standing/sitting ratio, activity level, heat cycle, lameness, GPS Location of the cattle, and the like. The collar unit (10) is designed as a wearable device that can be worn by a cattle and is held in place by a strap/belt provided with a locking mechanism.
In one of the exemplary embodiments of the present invention, the collar unit (11) is placed on the neck of one of the cattle at a predefined zone and is secured by means of the strap/belt with the locking mechanism. The locking mechanism prevents the collar unit (10) to be detached or slipping from the strap/belt. The strap/belt is a comfortable and durable cattle-friendly, yarn-based belt for holding the housing and other accessories secured therein. In one of the exemplary embodiments of the present invention, the strap/belt is a double-strand polycarbonate material, providing a higher tensile strength of of 1 ton.
In one of the exemplary embodiments of the present invention, the housing is made by a durable casing, preferably, ABS polycarbonate material, allowing for customization in different automotive colors. The housing is provided with a waterproof design.
In one of the exemplary embodiments of the present invention, the strap/belt that securely holds the collar unit (10) is further provided with a positioning weight, and numbering tags. The positioning weight restricts the movement of the strap/belt within the neck of the cattle in position and sets the beacon position as needed on the cattle’s neck. The numbering tag is attached with the cattle neck for the purpose of identification and further correlating with the collar unit (10).
The collar unit (10) customized to observe and measure the plurality of parameters related to the cattle accurately by means of the plurality of sensors therein. The output data from the plurality of sensors indicates the position, movements and heat characteristics of the cattle on 24/7 basis.
In one of the exemplary embodiments of the present invention, the plurality of sensors includes a temperature sensor (11), an accelerometer (12), a magnetometer (13), a gyroscope (14), and a GPS sensor (15) all electrically coupled to the controller (16). The plurality of sensors is powered by the power supply unit secured inside the housing. In a preferred embodiment, the power supply unit includes a battery unit.
In one of the exemplary embodiments of the present invention, the battery unit is a rechargeable/ non-rechargeable battery that can be secured inside the housing safely.
The plurality of sensors is configured to communicate with controller (16) and transfer the measured readings for processing. The processed data from the controller (16) is further sent to the server (20) as data packets via a communication network by way of the communication module (18).
The controller (16) that is configured to receive the plurality of sensor output signals is further connected to the communication module (18) via wired or wireless medium. In one of the exemplary embodiments of the present invention, the communication module (18) includes a communication device (18a) configured with the collar unit (10) capable of connecting the collar unit (10) direct to the communication network.
In one of the exemplary embodiments of the present invention, the communication module includes a network interface (18a) coupled to a gateway (18b) that connects the collar unit (10) with the communication network. The network interface (18a) is configured within the collar unit (10) and the gateway (18b) is secured external to the collar unit (10). In an embodiment, the network interface (18a) is designed to establish communication with the nearest gateway (18b) in proximity, while the gateway (18b) is configured to communicate with a set of adjacent collar units (10) within its range. In an embodiment, the network interface (18a) is a beacon. In an embodiment, the gateway (18b) is a BLE 5.0 Gateway with a range of 100 meters. Further, the gateway (18a) is powered by 220 Volts 50 Hz AC Power supply and is equipped with 2G/3G/4G SIM and Wi-Fi Connectivity. The communication module (18) is also provided with a Wi-Fi module that connects with a modem using 2G/4G connectivity.
In one of the exemplary embodiments of the present invention, the controller is a microcontroller configured with a firmware (17) that receives the plurality of sensor output signals process and convert them into digital data packets of predefined time frame suitable for transmission over the communication network and further calculations.
In alternative embodiments of the present invention, based on farm requirements the magnetometer (13), gyroscope (14) and GPS sensor (15) are replaced by devices with similar functionality. The temperature sensor (11) is configured to detect the temperature of the cattle using a physical contact-based temperature sensing probe. The accelerometer (12), magnetometer (13), and gyroscope (14) are configured to read the cattle movements and provide the measured readings in X, Y, and Z directions.
In an embodiment of the present invention, the temperature sensor (11) includes a device with high conductivity aluminium sensing probe. The device is designed in conical shape and is secured inside the housing of the collar unit (10) such that optimum contact with the skin surface of the cattle can be ensured. The sensing probe is connected to an active temperature sensor mounted on a printed circuit board in the collar device (10). In the embodiment, the air gap between the probe internal surface and the temperature sensor (11) is filled using a temperature conducting gel to ensure no air gaps. The actual temperature on the cattle is measured using thermometer for rectal temperature and device temperature probe for skin temperature for establishing the correlation during the initial stage of configuration. Thus, a correlation is established between device temperature measurements and actual body temperature through rectal and skin temperature measurements.
The server (20) is a server grade computing device connected to the communication network. The server (20) is configured with an AI module (30), an artificial intelligence-based application comprising an artificial intelligence unit (30), a data preprocessing module (1), a feature extraction module (2), a training module (3), and an inference engine (4).
The data preprocessing module (1) and the feature extraction module (2) are configured for processing the data received from the collar unit (10) via the communication network. After feature extraction, the training module (3) creates a plurality of AI-based analytics models with custom algorithms that are used to process and finalize the prediction data. The advanced data analytics models derive accurate and timely predictions related to cattle health from the plurality of sensor input readings. These analytic models are trained on the various data sets gives enhanced accuracy and reduced false/negative alerts. The AI module (30) applies a plurality of machine learning classifiers to predict the cattle health conditions based on the processed data, and the data is collected in a duration of the window on the cloud server (20), which is identified by an Annotator (Observer) trained by the data preprocessing module (1). This is further presented to the end-user (farmer) by means of a computing device (40) at the user end.
In one of the exemplary embodiments of the present invention, the server (20) is a cloud server application accessible via the communication network. The cloud server application is configured with an AI module (30), and artificial intelligence-based application comprising an artificial intelligence unit (30), a data preprocessing module (1), a feature extraction module (2), a training module (3), and an inference engine (4).
In one of the embodiments, the data preprocessing module (1) is a first submodule of the AI module (30), that is configured to clean and break the sensor data and rearranges the variables in the desired format.
In one of the embodiments, the feature extraction module (2) is a second submodule of the AI module, that is configured for preparing the summary of statistical features from the raw data received from the collar unit (10). The module is configured for calculating atleast 32 features from each of the parameters received.
In one of the embodiments, training module (3) is a third submodule of the AI module (30), that is configured for training an artificial neural network model based on the extracted features, with feature engineering, best model selection, hyperparameter tuning, and ensembling.
In one of the embodiments, the Inference Module (4) is a fourth submodule of the AI module (30), that is configured for deploying many different models for predicting the behaviors and combines the same at different layers for meaningful interpretation and subsequent actions.
In one of the exemplary embodiments of the present invention, the controller (16) in each of the plurality of collar unit (10) is also configured to execute sleep mode for a predefined time after regular time intervals. This customization is made for energy saving of the collar unit (10) and therefore achieves prolonged battery life. This forms the data collection frequency of the collar unit (10) to be customized to observe and measure the parameters related to temperature (11) and accelerometer (12) at a predefined time intervals. In the embodiment, each temperature reading is taken at every X time intervals, and the accelerometer (12) reading is taken once in Y time intervals for Z time duration. The values of X, Y, and Z are customizable by suitably coding the controller (16). In a preferred embodiment, the collar unit (10) is a robust all-weather, sealed IP67-based, BLE technology-based, water and dust-resistant device.
In an implementation of the embodiments, the signals captured by each of the plurality of sensors is preprocessed using the controller (16) and then routed to the server (20) directly by means of the communication module (18) as data packets. In one of the embodiments, the server (20) is a cloud server coupled to a plurality of collar units (10) via the communication network, wherein the data routing is facilitated using the communication module (18) configured within the housing which uses either Wi-Fi or BLE technology. Additionally, the data includes the square root of the accelerometer (12) reading, data related to the movement of cattle in horizontal direction, temperature conditions and the like. The data received by the cloud server (20) is further routed to the artificial intelligence (AI) module (30) for classification and prediction of the cattle health. The AI module (30) is configured for analyzing cattle behavior, temperature conditions, anomalies, heat periods, and illness from the sensor data by incorporating a plurality of methods and algorithms.
In an implementation of one of the exemplary embodiments of the present invention, the AI module (30) comprises of a plurality of machine learning classifiers including Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), AdaBoost (ADB), and K-Nearest Neighbors (KNN) AI-based analytics models with custom parameters for processing the data packets from the collar unit (11) and generating prediction results.
The computing device (40) includes a processing unit in communication with a network interface. The processing unit includes atleast one processor in communication with an internal memory stored with processor executable instructions and with a storage medium configured with a dashboard/mobile application. The dashboard/mobile application upon executed by the processing unit, gets connected with the server (20) via the communication interface. Thus, the computing device gets connectivity with each of the plurality of collar units (11) mounted on cattle’s body via the communication network and receives notifications regarding the cattle health from the server (20). The notifications received by the computing device (40) are further considered by the user/farmers to enhance productivity and to predict disease of the cattle for timely interventions.
The server (20) is configured to provide timely alerts on various health and heat conditions of the cattle that wears the collar unit (10). These alerts are provided through the dashboard/mobile application configured within the computing device (40). In an embodiment, the computing device (40) is a portable computing device with high-speed internet connectivity via a communication network.
In one of the exemplary embodiments of the present invention, the computing device (40) is a smart phone and the dashboard/mobile application is a mobile application.
In one of the exemplary embodiments of the present invention, the dashboard/mobile application is a multi-lingual application and providing instant alerts. This includes distinctive way of visualizing the processed information received from the server (20). The dashboard/mobile application employs multiple icons and digits, minimizing the use of excessive text. This unique approach makes it farmer/user friendly.
In an implementation of one of the exemplary embodiments of the present invention, the server is configured for anomaly detection, and cattle behavior recognition based on the individual data packets received from the collar unit (10) and raising alerts via the computing device (40).
In the embodiment, the anomalies detection involves calculating a long-term and short-term averages for various behavior indicators such as movements per hour, standing time, eating/ruminating time, and average body temperature. Further, alerts are raised if the current behavior significantly deviates from the generic behavior based on the averages. The heat-related alerts are generated based on activity levels, rumination, and standing behavior, wherein the heat is confirmed when these alerts match with historical or predicted values for similar parameters on a specific lagged period. The illness sensing includes health-related alerts which are generated based on activity levels, rumination, sitting behavior, and temperature. Illness is suspected when these alerts indicate significant deviations from normal behavior.
In the embodiment, the server (20) is configured to recognize the cattle behavior based on the data packets received from the collar unit (10). In the embodiment, the axial movements of the cattle are determined from the data from a 9-axis accelerometer (12) configured within the collar unit (10). The accelerometer (12) readings are first transformed into time and frequency domain data. Further, multiple features are extracted from each data packets, including mean, mode, median, standard deviation, and descriptive statistics. Further, physical annotations are made to label the readings with actual activity and position of the cattle. Thereafter machine leaning algorithms are applied on the labeled data by the AI module (30) to predict behavior indicators of a cattle such as sitting/standing, activity level, and rumination/chewing. The trained machine learning models are deployed to predict behavior in real-time based on the collar unit (10) readings.
In an implementation of one of the exemplary embodiments of the present invention, an operation of the system is explained with reference to the diagrams 1 to 3. The preferred embodiment the accelerometer (12) is an axial meter specifically a 9-axis accelerometer secured within the collar unit (20). The accelerometer (12) within the collar unit (20) is configured in such a way that it can measure various axial movements of the cattle in contact. The readings from the sensors such as accelerometer (12), magnetometer (13) and gyroscope (14) are communicated to the AI module (30) via the communication network as data packets in a rate of approximately 55 observations per minute. The data packets represent the observed values transformed as a combination of time domain data and frequency domain data. For predicting the activity of the cattle, the AI module (30) calculates number of FFT peaks from one data packet to represent movements/ min. The time duration is taken in minutes for those periods during which the collar unit (10) is not in sleep mode. Then aggregate the measured values over an hour to represent the number of movements per hour. This represents the basic unit of activity of the cattle under observation.
In the implementation of one of the exemplary embodiments of the present invention, the server (20) is configured to generate alerts on health and heat conditions of the cattle by using the AI module (30) based on determining the temperature conditions of the cattle from the temperature sensor data received from the collar unit (10); jaw movements of the cattle from the accelerometer (12), magnetometer (13), and gyroscope (14) data received from the collar unit (10); movements of the cattle such as sitting and standing from the accelerometer (12), magnetometer (13), and gyroscope (14) data received from the collar unit (10); and heat generation on the cattle from the accelerometer sensor (12) data received from the collar unit (10). Based on these predictions, the AI module (30) is configured to generate two sets of average values, including a long-term average representing the generic behavior of cattle and a short-term average representing the current behavior of cattle, and communicates these results to the computing device (40) and generates alerts upon the current behavior of the cattle significantly differs from the estimated generic behavior.
The jaw movements of the cattle are determined by a two-step process such as identifying whether the cattle are chewing/eating and whether the cattle are not chewing/eating. This gives idea on whether the cattle are ruminating or idle. To identify whether the cattle is chewing, the AI module (30) considers the axial sensor data and determine the coefficient of variation of resultant acceleration. Further, consider the features extracted from the accelerometer (12) data and determining the coefficient of variation of resultant acceleration and calculating a coefficient of variation of first difference of resultant acceleration for final prediction. In the second case, for identifying whether the cattle are not chewing/eating, The AI module (30) considers accelerometer (13), gyroscope (14) and magnetometer (15) data, derive variables therefrom and applying a Band pass filter (BPF) on the derived variables to extract the number of peaks for resultant values. The Band pass filter (BPF) is applied on the derived variables to extract the number of peaks for resultant accelerometer, gyroscope and magnetometer values. In the implementation, the variables are chosen such that the calculation and prediction procedures are device independent. The AI module (30) is configured with a simple supervised learning algorithm for deriving the decision tree model for each cattle breed such as Gir Cattle and Holstein-Friesian cattle to capture the general pattern and replicate it across all the variables.
The movements of the cattle such as sitting and standing are determined by a three-step process which involves, extracting 55 point time series from each data packet received from the collar unit (10) for 9 time series variables by the AI module (20); converting each of these 9 time series variables into frequency domain data using FFT and form 18 time series of values, by the AI module (20); extracting 16 statistical features for each of the 18 time series of values and calculating 288 features from each of the data packets, which is then reduced to a limited number by using principal component analysis, by the AI module (20); and applying a multi-predictor machine learning algorithm on the labelled data for predicting sitting/ standing of the cattle, by the AI module (20).

In a preferred embodiment, the AI module (30) is configured to extract the statistical parameters such as Mean, Mode, Median, Standard deviation Mean, Maximum, Minimum, Median Energy, Kurtosis, Skewness, Mean absolute deviation, Positive counts, Negative counts, Interquartile range, Standard deviation, Count above mean, Range, Peak count, and Median absolute deviation. Further, the extracted 288 features are reduced to a limited number preferably 20 by using principal component analysis.
Based on the said predictions, the AI module (30) generates two sets of average values. This includes a long-term average (of past 1 week) to represent the generic behavior of cattle and a short-term average to represent the current behavior of cattle. This includes past 6 hours in a preferred embodiment and past 24 hours in alternative embodiments, may be considered. The characteristics to be monitored are number of movements per hour, time duration which the cattle are standing in an hour in percentage, time duration which the cattle is eating in percentage, ruminating, or having no jaw movement in an hour and average body temperature for every hour.
In the implementation of one of the exemplary embodiments of the present invention, the conditions for detecting heat generation on the cattle is determined based on the accelerometer sensor (12). Normally, a cattle exhibits restlessness, lower feed, longer hours of standing and tendency of mounting close to other cattle when it is on heat. In a specific embodiment, the accelerometer (12) is a 9-axis accelerometer sensor that identifies the feeding, standing, and activity levels, as mentioned above.
The server (20) is configured to generate alerts when the current behavior of the cattle is significantly different from the generic behavior extracted previously. The alerts related to Heat include High Activity Alert, Low Rumination Alert and Long-Standing Alert. The Heat alerts are generated when any of the said alerts are generated and the intensity is proportional to the number of alerts. Heat is confirmed when of any or all these alerts matches with the actual records or predicted values for similar parameters on 21 +/- 3 days lagged period.
The server (20) is configured to generate alert related to Health issues. Normally, cattle are unwell, it may exhibit restlessness or very low activity level, drop in feed, longer hours of sitting and/or a fever. Based on these habits of cattle, the system is configured to produce alerts such as High Activity Alert, Low Activity Alert, Low Rumination Alert, Long Sitting Alert and High Temperature/ Fever Alert. The health alerts are generated when any of the said alerts are generated wherein the intensity of the alert is proportional to the number of alerts.
In an implementation of one of the embodiments of the present invention, a jaw movement of the cattle predicted by the AI module (30) by following the steps below.
1. Receiving the data packets corresponding to sensor output signals from the collar unit (20).
2. Breaking sensor data to extract a 55 point time series from each data packets for 9 variables (ax, ay, az, gx, gy, gz, mx, my, mz) derived from the sensor output signals and calculate the resultant of a, m and g. Here, the accelerometer measures the linear accelerations decomposed in three orthogonal axes – ax, ay, az. Gyroscope measures the rotational movement about three orthogonal axes – gx, gy, gz. Magnetometer measures the earth’s magnetic field decomposed into three orthogonal axes – mx, my, mz. All the features, used for the machine learning algorithms are derived out of these 9 raw variables, received from the sensor. The resultant of all the vectors are calculated as: a_res = sqrt(ax^2+ay^2+az^2), g_res = sqrt(gx^2+gy^2+gz^2), m_res = sqrt(mx^2+my^2+mz^2).
3. Generating a time series from the first difference of resultants, and calculate the variance of the resultants and first difference of resultants for each resultant respectively.
4. Training a first level model with the said two features for differentiating between eating/ non-eating characteristics of a cattle with a decision tree. To make the decision model, device independent, the system (100) employs a coefficient of variation (CV) instead of variance or SD.
5. For the non-eating scenario, the AI module (30) applies filter specifically a BPF to capture data having characteristics related to rhythmic movements of rumination in the processed sensor data. Further, a second level model created by AI module (30) uses 3 features, such as the number of peaks for resultants a, g and m transformed to frequency domain and applies the filter. Again, a decision tree created for the said characteristics and applies to differentiate between rumination and idle jaw movement.
The conjunction of the said two model is capable of accurately predicting the jaw movement pattern of the cattle, across different breeds.
In an implementation of one of the embodiments of the present invention, the activity level of the cattle predicted by the AI module (30) by following the steps below.
1. Receiving the data packets corresponding to sensor output signals from the collar unit (20)
2. Breaking sensor data to extract a 55 point time series from each sensor data packets for 3 variables (ax, ay, az) derived from the sensor output signals and calculate the resultant of the said 3 variables (ax,ay,az).
3. Converting the said resultant of the 3 variables from time domain to frequency domain by using Fast Fourier transform (FFT).
4. Calculating the number of peak values for the frequency domain data
5. Taking aggregate of the total number of peaks over the hour, to represent the movement per hour with sum as the aggregator function
In this is the method of movement measurement, next averaged over hours and days to monitor the current and general activity level of the animal.
In an implementation of one of the embodiments of the present invention, movements of the cattle such as sitting/ standing identifier is performed by the AI module (30) involves the following steps.
1. Receiving the data packets corresponding to sensor output signals from the collar unit (20).
2. Break sensor data and extracts 55-point time series from each packet for all the said 9 variables (ax, ay, az, gx, gy, gz, mx, my, mz) and convert each of these 9-time series into frequency domain data using Fast Fourier Transform (FFT). Thus 18 time series are created.
4. Further, a plurality of descriptive statistics is calculated for each of the said time series, including, Mean, Maximum, Minimum, Median Energy, Kurtosis, Skewness, Mean absolute deviation, Positive counts, Negative counts, Interquartile range, Standard deviation, Count above mean, Range, Peak count, and Median absolute deviation.
Thereby, 18*16=288 features are extracted by the AI module (30).
5. Thereafter, variable reduction is performed using Principal Component Analysis and derive at 20 final features.
6. Finally, a multipredictor Autogluon model is applied in these derived values for predicting sitting/standing position of the cattle.
In one of the exemplary embodiments of the present invention, the functionality of the temperature sensor and the probe contact is tested with blue contact test for ensuring skin contact on cattle. The temperature sensor is calibrated offline for measurement accuracy and sensitivity on a test bench and the results are found to be within +/-0.5 degree about normal body temperature of 102-degree Fahrenheit. Changes in body temperature due to illness like continuous increase of body temperature are captured using control chart rules/ long and short-term average rules to check for any trend and report as alerts.
The collar unit (10) further provides an additional functionality of tracking of cattle. The collar unit (10) attached to the cattle is configured to make connectivity with a nearest gateway (18b) in a prescribed range. In a specific embodiment, the connectivity range is set to 100 m. Hence a plurality of gateways (18b) installed within an area may communicate with the collar unit (10) worn by the cattle in the predefined distance and allows to track the position of the cattle.
In one of the exemplary embodiments, the gateway (18b) is designed as a battery-powered device providing uninterrupted service even during electricity failures. A preferred embodiment of the gateway offers data storage facility allowing us to back up data for extended periods.
In one of the exemplary embodiments, the gateway the system (100) provides scalable data storage, and a fixed storage facility that can store data from the collar unit (10) for over 15 hours.
In one of the exemplary embodiments, the dashboard/mobile application is configured with an augmented reality interface. This provides an enhanced visualization of the farm, providing a more immersive and informative experience.
ADVANTAGES OF THE INVENTION
1. Sensor Technology used in the collar unit (10) is accurate and provides high-quality data to be able to determine the exact behavioral pattern of the cattle. The collar unit is designed as a complete serviceable device ensuring easy maintenance and repair.
2. The present invention provides a solution for better management of cattle health and this is achieved by a care feature in our solution which provides timely support to the farmer. An easy data access provided with the system allows the users to easily download their data for further analysis and record-keeping.
3. The present invention uses advanced data analytics to derive accurate and timely predictions. These analytic models are trained continuously on the various data sets to keep enhancing the accuracy and reducing the false/negative alerts.
4. The present solution does not need any special skills or complex technical know-how to implement. The solution can be easily implemented by the farmer.
5. The system (100) further empowers the analytical skills of the users as the dashboard/mobile application provides accessibility to correlation graphs, compare herd versus individual animal data, and examine parameter comparisons at different times.
The most critical part of the implementation of the detection procedures is the sensitivity measurement during body temperature change. Another challenge was related to the efficiency of the device. As the device is not a 100% positive contact device, whenever the temperature probe loses skin contact with cattle skin, the device reads based on the dead weight mounted on the belt. The mathematical procedures implemented with the AI module (30) is capable of adjusting with such stray observations due to loss of skin contact and considers stable value over 1 hour for establishing the body temperature of the cattle.
,CLAIMS:We claim:
1. An IOT-based cattle health and heat monitoring system (100), the system (100) comprising:
a plurality of collar units (10), each of the collar unit (10) removably secured firmly onto the body of a cattle, the collar unit (10) includes
a plurality of sensors, the plurality of sensors is configured to observe and measure a plurality of health parameters related to temperature conditions and movements of the cattle in contact,
a controller (16) communicatively coupled to the plurality of sensors, the controller (16) configured with a firmware (17) that receives the plurality of sensor output signals, process and convert the signals into data packets suitable for further calculations,
a power supply unit electrically connected to the plurality of sensors, the controller (16) and the communication unit (18), and
a housing enclosing the plurality of sensors, the controller (16) and the communication unit therein, wherein the housing is equipped with a strap/belt and locking mechanism to secure the housing onto the body of the cattle,
wherein each of the plurality of the collar units (10) is configured to measure health parameters of the cattle in contact by way of the plurality of sensors, preprocess the sensor output signals by the controller (16) therein and route to the communication network by means of a communication module (18);
a server (20) in communication with each of the plurality of collar units (10) via the communication network, wherein the server is configured with an artificial intelligence module (30) includes a data preprocessing module (1), a feature extraction module (2), a training module (3), and an inference engine (4); and
atleast one portable computing device (40), the portable computing device (40) includes processing unit in communication with a storage medium configured with a dashboard/mobile application, wherein the dashboard/mobile application upon executed by the processing unit facilitates connectivity with the server (20) via a communication interface and receive notifications from the server (20), regarding health and heat conditions of the cattle that is in contact with each of the plurality of collar units (10);
wherein the AI module (30) is configured for receiving the data packets from each of the collar unit (10) processing the data packets individually representing real time sensor output values transformed as a combination of time domain and frequency domain data, predicting health, and heat from the temperature sensor data, and statistical parameters derived from the movement sensor data corresponding to the cattle.
2. The system (100) as claimed in claim 1, wherein the plurality of sensors includes, a temperature sensor (11) configured to measure temperature of the cattle, and movement sensors include accelerometer (12), magnetometer (13), and gyroscope (14) configured to measure the cattle movements.
3. The system (100) as claimed in claim 2, wherein the accelerometer (12), magnetometer (13), and gyroscope (14) are configured to read the cattle movements and provide the measured readings in X, Y, and Z directions suitable for the calculations performed by the AI module (30).
4. The system (100) as claimed in claim 2, wherein temperature sensor (11) is a conically shaped device designed with a high conductivity aluminum sensing probe and is securely positioned inside the housing of the collar unit (10) to ensure optimal physical contact with the skin surface of the cattle body.
5. The system (100) as claimed in claim 2, wherein the accelerometer (12) is an axial meter specifically a 9-axis accelerometer that is configured to measure various axial movements of the cattle in contact.
6. The system (100) as claimed in claim 1, wherein the collar unit (10) is designed as a wearable device for cattle and is held in place by the strap/belt strap/belt provided with a positioning weight, and numbering tags that securely holds the collar unit (10).
7. The system as claimed in claim 1, wherein the controller (16) is configured to execute sleep mode for a predefined time after regular time intervals and the data collection frequency for the collar unit (10) is customized to observe the parameters related to temperature and movements at a predefined time intervals.
8. The system (100) as claimed in claim 1, wherein the server (20) is a computing device connected to the communication network, that is configured with an AI module (30) having an artificial intelligence-based application comprising an artificial intelligence unit (30), a data preprocessing module (1), a feature extraction module (2), a training module (3), and an inference engine (4).
9. The system (100) as claimed in claim 1, wherein the server (20) is a cloud server application accessible via the communication network, wherein the cloud server application is configured with an AI module (30), and artificial intelligence-based application comprising an artificial intelligence unit (30), a data preprocessing module (1), a feature extraction module (2), a training module (3), and an inference engine (4).
10. The system (100) as claimed in claim 1, wherein the computing device (40) is a portable computing device with high-speed internet connectivity with a communication network.
11. The system (100) as claimed in claim 1, wherein the collar unit (10) communicates with the server via a communication module (18) configured therein.
12. The system (100) as claimed in claim 1, wherein each of the plurality of collar units (10) is configured to communicate with the server (20) via a communication module (18) having a network interface (18a) configured therein connected to a nearest gateway (18b).
13. The system (100) as claimed in claim 1, wherein the AI module (30) comprises of a plurality of machine learning classifiers including Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), AdaBoost (ADB), and K-Nearest Neighbors (KNN) AI-based analytics models with custom parameters are employed in the AI module (30).
14. The system (100) as claimed in claim 1, wherein the server (20) by means of AI module (30) is configured to predict and generate alerts on health and heat conditions of the cattle by,
determining the temperature conditions of each of the cattle from the temperature sensor (11) data received from the respective collar unit (10);
determining the cattle movements, jaw movements and heat generation on the cattle from the readings of movement sensors including accelerometer (12), magnetometer (13), and gyroscope (14) in the collar unit (10);
wherein based on the determined values, the AI module (30) is configured to generate two sets of average values, including a long-term average representing the generic behavior of the cattle and a short-term average representing the current behavior of the cattle, and communicates these results to the computing device (40) and generates alerts upon the current behavior of the cattle significantly differs from the estimated generic behavior.
15. The system (100) as claimed in claim 14, wherein the jaw movements of the cattle are determined by,
identifying whether the cattle is chewing/eating, by considering the features extracted from the accelerometer (12) data and determining the coefficient of variation of resultant acceleration and calculating a coefficient of variation of first difference of resultant acceleration for final prediction; and
identifying whether the cattle is not chewing/eating, indicating whether the cattle is ruminating or idle, by considering accelerometer (13), gyroscope (14) and magnetometer (15) data, derive variables therefrom and applying a Band pass filter (BPF) on the derived variables to extract the number of peaks for resultant values.
16. The system (100) as claimed in claim 14, wherein the movements of the cattle such as sitting and standing are determined by,
extracting 55-point time series from each data packets received from the collar unit (10) for 9 time series variables by the AI module (20);
converting each of these 9 time series variables into frequency domain data using FFT and form 18 time series of values, by the AI module (20);
extracting 16 statistical features for each of the 18-time series of values and calculating 288 features from each of the data packets, which is then reduced to a limited number by using principal component analysis, by the AI module (20); and
applying a multi-predictor machine learning algorithm on the labelled data for predicting sitting/ standing of the cattle, by the AI module (20).
17. The system (100) as claimed in claim 1, wherein the statistical parameters extracted by the AI module (30) include Mean, Mode, Median, Standard deviation Mean, Maximum, Minimum, Median Energy, Kurtosis, Skewness, Mean absolute deviation, Positive counts, Negative counts, Interquartile range, Standard deviation, Count above mean, Range, Peak count, and Median absolute deviation.
18. The system (100) as claimed in claim 14, wherein the alerts include Heat alert and health alert.
19. The system (100) as claimed in claim 18, wherein the heat alert include High Activity Alert, Low Rumination Alert and Long-Standing Alert.
20. The system (100) as claimed in claim 18, wherein the health alerts include High Activity Alert, Low Activity Alert, Low Rumination Alert, Long Sitting Alert and High Temperature/ Fever Alert based on number of movements observed per hour, time duration which the cattle is standing in an hour, time duration which the cattle is eating, ruminating, or having no jaw movement in an hour and average body temperature for every hour.
Dated this on 5th day of July, 2023

Ragitha K
Agent for Applicant
(IN/PA/2832)