Description:Complete Specification:
The following specification describes and ascertains the nature of this invention and the manner in which it is to be performed

Field of the invention
[0001] The present disclosure relates to the field emission predictions for an automobile. In particular, the present invention discloses method of predicting emissions in emission prediction system using artificial intelligence (AI) module.

Background of the invention
[0002] An unprecedented growth in traffic in the past decade has deteriorated the commuting environment to a great extent because of the air pollution that it causes. Studies have indicated that such emissions from the mobility sector have caused serious health problems such as respiratory issues, cardiovascular diseases, perinatal mortality etc. Transportation systems planning and operational models and the policies that are proposed to provide mobility solutions also now taking into account the environment related aspects. Emission standard and legislations like Bharat stage (BS) emission standards or likewise the European emission standards are laid down by the government to regulate the output of air pollutants from internal combustion engine and spark-ignition engine equipment, including motor vehicles. In India, recently BS-VI emission norms came into force from April 2020 that set the maximum permissible levels for pollutants that an automotive or a two-wheeler exhaust can emit. Similarly, Euro 7 emission standards are expected soon in European Union.
[0003] These increasingly stringent emission standards incur a need for stricter monitoring of emissions from the automobiles. Estimating the emissions from these automobiles in real world on road conditions has become a necessity in light of these stringent emission standards. These estimates have many diverse applications. Emissions include THC, CO, NOx, CO2, NO2 amongst others. With the advent of data science and data processing, systems are implemented using artificial intelligence (AI) modules. Today AI systems find their application in every filed where they process data to generate required output based on certain rules/intelligence acquired through training. While there are conventional emission predictors in smaller constrained cases, there is no framework, which can work agnostically and in an automated fashion using the benefits of data science and artificial intelligence.

[0004] Chinese Patent application CN114282680 AA titled “Machine learning algorithm based vehicle exhaust emissions prediction method and system” discloses a machine learning algorithm, the method comprising screening a variable parameter having a correlation to NOx parameters in a raw data set to form an initial data set, wherein the raw data set comprises road environment parameters, engine parameters and vehicle parameters, reducing the parameter dimensions of the initial data set by principal component analysis in a dimensionality reduction analysis, Constructing mutually uncorrelated principal component data sets, discretizing the principal component data sets by density clustering algorithms and dynamic temporal bending, and classifying the principal component data sets to obtain characteristic parameter data sets; constructing a vehicle NOx emission model using a convolutional neural network embedded in an error feedback unit using different categories of characteristic parameter data sets as training and testing samples for constructing an emission model. The emission prediction model provided by the present application is close to sensor measurements, enabling technical requirements of an aftertreatment system through the model, and replacing sensors to improve economics.

Brief description of the accompanying drawings
[0005] An embodiment of the invention is described with reference to the following accompanying drawings:
[0006] Figure 1 depicts an emission prediction system (10) in an automobile;
[0007] Figure 2 illustrates method steps (200) of training an AI module (1021) for the emission prediction system (10);
[0008] Figure 3 illustrates method steps (300) of predicting emissions in the emission prediction system (10).

Detailed description of the drawings
[0009] Figure 1 depicts an emission prediction system (10) in an automobile. The emission prediction system (10) comprises at least a processor (102) in communication with a plurality of automobile sensors. Emissions for the purposes of this invention refers to amount of total hydrocarbons (THC), carbon monoxide (CO), nitrogen oxides (NOx), carbon-dioxide (CO2), nitrogen dioxide (NO2) amongst others.

[0010] The processor (102) can either be a logic circuitry or a software programs that respond to signals received from various automobile sensors and processes them to get a meaningful result. The processor (102) may reside inside the automobile or may be implemented in cloud or a server further wherein the processing is further manifested through an edge computing device, such as a tablet, computer or a mobile phone. A hardware processor (102) may be implemented in the system as one or more firmware, an application specific integrated circuit (ASIC), and/or a field programmable gate array (FPGA), microcomputers, microcontrollers, digital signal processor (102)s, central processing units, state machines, logic circuitries, and/or any component that operates on signals based on operational instructions. A software processor (102) may reside inside a cloud or a server. The processor (102) in the present invention executes an AI module (1021).

[0011] An AI module (1021) with reference to this disclosure can be explained as a component which runs a model. A model can be defined as reference or an inference set of data, which is use different forms of correlation matrices. A person skilled in the art would be aware of the different types of AI models such as linear regression, naïve bayes classifier, random forest, gradient boosting trees, support vector machine, neural networks and the like. It must be understood that this disclosure is not specific to the type of model being executed in the AI module (1021) and can be applied to any AI module (1021) irrespective of the AI model being executed. A person skilled in the art will also appreciate that the AI module (1021) may be implemented as a set of software instructions, combination of software and hardware or any combination of the same. For example, a neural network mentioned herein after can be a software residing in the system or the cloud or embodied within an electronic chip.

[0012] The plurality of automobile sensors are electro-mechanical devices that monitor various engine parameters and vehicle parameters. The plurality of sensors (101) include but are not limited to air-fuel ratio meter, engine speed sensor, throttle position sensor, crank position sensor, cam position sensor, knock sensor, Manifold Absolute Pressure or MAP Sensor, Mass Air Flow or MAF Sensor, Oxygen or Lambda sensor, fuel pressure sensor, vehicle speed sensor and the like. The plurality of sensors (101) measure a plurality of parameters and send the information to the sensor. The plurality of parameters comprise at least one or more from the group of engine operating parameters and automobile operating parameters.

[0013] The emission prediction is characterized by the functionality of the processor (102). The processor (102) is configured to: generate one or more drive cycles by concatenating a plurality of microtrips; receive data comprising values of a plurality of parameters for the one or more drive cycles from the plurality of sensors (101); input the collected data to an artificial intelligence module; predict emission based on output of the AI module (1021). The processor (102) is configured to apply a correlation factor learnt by the AI module (1021) on the values of the first set of parameters to predict emission. The functionality and configuration of the processor (102) is explained through the method steps 200 and 300.

[0014] It should be understood at the outset that, although exemplary embodiments are illustrated in the figures and described below, the present disclosure should in no way be limited to the exemplary implementations and techniques illustrated in the drawings and described below.

[0015] Figure 2 illustrates method steps of training an AI module (1021) for the emission prediction system (10). The AI module (1021) and the emission prediction system (10) have been elucidated in accordance with figure 1. For clarity it reiterated that the system comprises a plurality of sensors (101) of an automobile, said plurality of sensors (101) are in communication with a processor (102) and said processor (102) executes at least the AI module (1021).

[0016] Method step 201 comprises creating one or more drive cycles for the automobiles by concatenating a plurality of microtrips. Creating drive cycles comprises running the automobile on a dynamometer. A typical drive cycle is constructed by stitching different micro-trips (trip between two consecutive idling periods) together. Several real-world drive cycles have been proposed. Automotive Research Association of India (ARAI) has developed Indian Driving Cycle (IDC) for two-wheelers, three-wheelers and four-wheelers in 1985 (Badusha et al., 1999) and many modifications have been proposed subsequently.

[0017] Method step 202 comprises collecting data comprising values of a plurality of parameters for one or more drive cycles from the plurality of sensors (101) (as elucidated in accordance with figure 1). The plurality parameters comprise at least one or more from the group of engine operating parameters and automobile operating parameters. The plurality of parameters include but are not limited to air-fuel ratio, engine speed, throttle position, engine temperature, mass of air entering the engine, amount of oxygen in the exhaust, ambient pressure, engine RPM, vehicle speed, vehicle power, intake manifold pressure and so on.

[0018] Method step 203 comprises receiving at least one real time emission value. The real time emission value is received cumulatively and as well as individually for total hydrocarbons (THC), carbon monoxide (CO), nitrogen oxides (NOx), carbon-dioxide (CO2), nitrogen dioxide (NO2).

[0019] Method step 204 comprises learning a correlation between the first set of parameters and said at least one real time emission value. The correlation factor learnt is individually different for the various pollutants in emission such as total hydrocarbons (THC), carbon monoxide (CO), nitrogen oxides (NOx), carbon-dioxide (CO2), nitrogen dioxide (NO2) respectively. The correlation factor might further include two different sets of sub-factors for the first set of parameters. The two different sets of sub-factors correspond to the emission value for the pre-catalytic convertor as well as post catalytic convertor emissions respectively. Pre-catalytic convertor emissions are basically untreated and crude emissions. A catalytic convertor an emission control device that converts toxic gases and pollutants coming from an internal combustion engine to less toxic pollutants and gases.

[0020] Method step 205 comprises predicting the emissions using the learnt correlation for the first set of parameters. The prediction emissions include both pre-catalytic convertor as well as post catalytic convertor emissions. The learnt correlation fact is applied on the first set of parameters to derive a value of emission for the various pollutants.

[0021] These training method steps (200) are repeated for various drive cycles with various real time emission values to increase accuracy and robustness of the learnt correlation factors. The AI module (1021) is also fine-tuned after every iteration to increase accuracy and robustness. For example, the network parameters and the hyper parameters of the AI module (1021) are adjusted after every iteration until the best or the most optimum prediction results are achieved.

[0022] Figure 3 illustrates method steps (300) of predicting emissions in the emission prediction system (10). The emission prediction system (10) and its components have been explained in accordance with figure 1 and figure 2. A person skilled in the art would further appreciate that the AI module (1021) trained in accordance with method steps (200) is now deployed to function in a real-time environment in processor (102) of the automobile.

[0023] Method step 301 comprises receiving data comprising values of a plurality of parameters for one or more drive cycles from the plurality of sensors (101). The plurality of sensors (101) include but are not limited to air-fuel ratio meter, engine speed sensor, throttle position sensor, crank position sensor, cam position sensor, knock sensor, Manifold Absolute Pressure or MAP Sensor, Mass Air Flow or MAF Sensor, Oxygen or Lambda sensor, fuel pressure sensor, vehicle speed sensor and the like. Correspondingly, the plurality of parameters comprise at least one or more from the group of engine operating parameters and automobile operating parameters such as air-fuel ratio, engine speed, throttle position, engine temperature, mass of air entering the engine, amount of oxygen in the exhaust, ambient pressure, engine RPM, vehicle speed, vehicle power, intake manifold pressure and so on.

[0024] Method step 302 comprises inputting the collected data (values of the plurality of parameters) to the AI module (1021). Method 303 comprises predicting emission based on output of the AI module (1021). Predicting the emission comprises applying a correlation factor learnt by the AI module (1021) on the values of the first set of parameters. This correlation factor is learnt by the AI module (1021) during training as described in accordance with figure 2 and method steps 200. Since the correlation factor includes two different sets of sub-factors for the plurality of parameters (emission value for the pre-catalytic convertor as well as post catalytic convertor emissions respectively), the predicted emission value also comprises two components i.e. the pre-catalytic emission valve and the post-catalytic emission value.

[0025] A person skilled in the art will appreciate that while these method steps describes only a series of steps to accomplish the objectives, these methodologies may be implemented with custom modification to the training method (200) and the system (10).

[0026] This idea to develop an emission prediction system (10) and a method thereof proposes a smart framework using AI to build predictors for emissions for internal combustion engine (ICE) based vehicles. The input for our framework is fleet data collected from vehicles which is transformed into relevant data for experimentation on the dynamometer. The data (plurality of parameters ) from the dynamometer includes CAN data from the vehicle as well as auxiliary instrumented data. The proposed framework can run on the cloud, server or specialized hardware. Further, the framework is automated and is agnostic to the vehicle type (for 2 wheelers and 4 wheelers both).

[0027] It must be understood that the embodiments explained in the above detailed description are only illustrative and do not limit the scope of this invention. Any modification or adaptation of the Emission prediction system (10) and its methods thereof are envisaged and form a part of this invention. The scope of this invention is limited only by the claims.
, Claims:We Claim:
1. An emission prediction system (10) for an automobile, the emission prediction system (10) comprising at least a processor (102) in communication with a plurality of sensors (101), characterized in that system:
the processor (102) executing an AI module (1021), the processor (102) configured to:
generate one or more drive cycles by concatenating a plurality of microtrips;
receive data comprising values of a plurality of parameters for the one or more drive cycles from the plurality of sensors (101);
input the collected data to the AI module (1021);
predict emission based on output of the AI module (1021).

2. The emission prediction system (10) for an automobile as claimed in claim 1, wherein the plurality of parameters comprise at least one or more from the group of engine operating parameters and automobile operating parameters.

3. The emission prediction system (10) for an automobile as claimed in claim 1, the processor (102) is configured to apply a correlation factor learnt by the AI module (1021) on the values of the first set of parameters to predict emission.

4. A method (200) of training an AI module (1021) for an emission prediction system (10), the system comprising a plurality of sensors (101) of an automobile, said plurality of sensors (101) in communication with a processor (102), said processor (102) executing the AI module (1021), the method steps comprising:
creating (201) one or more drive cycles for the automobiles by concatenating a plurality of microtrips;
collecting (202) data comprising values of a plurality of parameters for one or more drive cycles from the plurality of sensors (101);
receiving (203) at least one real time emission value;
learning (204) a correlation between the first set of parameters and said at least one real time emission value;
predicting (205) the emission using the learnt correlation for the first set of parameters.

5. The method (200) of training an AI module (1021) for an emission prediction system (10) as claimed in claim 4, wherein creating drive cycles comprises running the automobile on a dynamometer.

6. The method (200) of training an AI module (1021) for an emission prediction system (10) as claimed in claim 4, wherein the real time emission value comprises both a pre-catalytic emission value and a post pre-catalytic emission value.

7. The method (200) of training an AI module (1021) for an emission prediction system (10) as claimed in claim 4, wherein the plurality of parameters comprise at least one or more from the group of engine operating parameters and automobile operating parameters.

8. A method (300) of predicting emissions in an automobile, the automobile comprising a processor (102) executing an artificial intelligence (AI) module, said processor (102) in communication with a plurality of sensors (101) of the automobile, the method comprising:
receiving (301) data comprising values of a plurality of parameters for one or more drive cycles from the plurality of sensors (101);
inputting (302) the collected data to the AI module (1021);
predicting (303) emission based on output of the AI module (1021).

9. The method (300) of predicting emissions in an automobile as claimed in claim 8, wherein the plurality of parameters comprise at least one or more from the group of engine operating parameters and automobile operating parameters.

10. The method (300) of predicting emissions in an automobile as claimed in claim 8, wherein predicting the emission comprises applying a correlation factor learnt by the AI module (1021) on the values of the plurality of parameters.