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 of predictive diagnostics. In particular the present invention discloses a

Background of the invention
[0002] Predictive maintenance uses data analysis to identify operational anomalies and potential defects to predict failures in the industrial machines based on its symptoms even before the machine fails. Prediction of remaining useful life (RUL) of various industrial machinery and it’s components has become a critical feature in modern day industrial technology. The topic of remaining useful life prediction is of both practical and scientific interest. Data-driven techniques based on artificial intelligence (AI) such as have attracted more and more attention in the manufacturing sector. These techniques have been applied in large machinery health monitoring. However, small yet critical components such as elastomer sealing used in electronic circuits of such large machinery have been ignored on the Remaining Useful Life (RUL) prediction.

[0003] Elastomer seals are processed rubbers designed to prevent fluid and gas from leaking around mating surfaces. Leakage on mating surfaces can cause serious damage and disrupt electrical or mechanical processes in large machinery. All rubbers are subjected to deterioration at high temperature. Volume change and compression set are both influenced by heat. This physical changes caused owing to heat will reverse when the temperature drops. When such elastomers are exposed to high temperatures for longer durations, chemical changes occur. These generally cause an increase in hardness, volume and change in tensile strength. Being chemical in nature, these changes are not reversible. When an elastomer is used as a gasket at a constant temperature, the elastomer is subjected to either static or dynamic load. When the load is removed the elastomer will return to its original shape if it has a low compression set. However, if the temperatures were not constant, degradation of compression set parameters can be exponential. Hence continuous monitoring of compression set is of paramount importance to determine remaining useful life of the elastomer. Precise RUL prediction of these elastomer sealings can significantly improve the reliability and operational safety of the industrial components or systems, avoid fatal breakdown and reduce the maintenance costs.

[0004] Patent Document DE102017000926 A1 discloses a device with at least one elastically deformable member as a structural part and/or bearing part, on the operational history of changing operating conditions dependent, different deformation forces act on the component to a life-limiting component wear, and with a device for determining the component wear-induced component and a usage time remaining useful life. In accordance with the present invention, a time offset, in each case the same operating state, the predetermined repeating a same deformation force is assigned, through which the elastically deformable component material is deformed. One such predefined operating state is in each case a measurement and evaluation unit is detected and a measurement process automatically started by a start signal, wherein at least one component associated acceleration sensor, the current acceleration of the deformation or characteristic values derived thereof as the characteristic for a current component stiffness is measured and stored in each case in a measurement curve and compared.

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 a system (100) to determine a remaining useful life (RUL) of an elastomer sealing (101);
[0007] Figure 2 illustrates method steps (200) to train a machine learning model (M) (M) to determine remaining useful life (RUL) of an elastomer sealing (101);
[0008] Figure 3 illustrates method steps (300) of determining a remaining useful life (RUL) of an elastomer sealing (101) using the trained machine learning model (M).

Detailed description of the drawings
[0009] Figure 1 depicts a system (100) to determine a remaining useful life (RUL) of an elastomer sealing (101). The system comprises a machine learning model (M), a temperature sensor (104) and at least a processor (103), said processor (103) in communication with a database (102).

[0010] A machine learning model (M) 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 uses different forms of correlation matrices. Using these models and the data from these models, correlations can be established between different types of data to arrive at some logical understanding of the data. A person skilled in the art would be aware of the different types of AI models such as linear regression, naïve bayes classifier, support vector machine, neural networks and the like.

[0011] Some of the typical tasks performed by AI models are classification, clustering, regression etc. Majority of classification tasks depend upon labeled datasets; that is, the data sets are labelled manually in order for a neural network to learn the correlation between labels and data. This is known as supervised learning. Some of the typical applications of classifications are: face recognition, object identification, gesture recognition, voice recognition etc. Clustering or grouping is the detection of similarities in the inputs. The cluster learning techniques do not require labels to detect similarities. Learning without labels is called unsupervised learning. The machine learning model (M) used in the present invention is trained using the unsupervised learning methodology explained in accordance with method steps 200.

[0012] The temperature sensor (104) is adapted to measure real-time value of temperature for the elastomer sealing (101) environment. The database (102) stores a historical data log comprising information regarding previous usage of the elastomer sealing (101) under various temperatures and at least a last observed value of CSV (Compression Set Value). Compression set is irrecoverable deformation on release from a compression. It depends on compression strain (or compression stress), duration, and temperature. The compression set value for an elastomer is defined by the ability of the elastomer to return to it’s original shape or form once the load acting upon the elastomer is removed. For example, a neoprene disc measuring 0.177? inches thick is tested for 70 hours at 212°F, then allowed to cool for 30 minutes. It now measured . 221? inches thick, then compression set value would be 40%, indicating that the amount of the distance compressed that it did not return is 40%.

[0013] The temperature sensor (104) and the database (102) in communication with the processor (103). The processor (103) is in communication with the machine learning model (M). The processor (103) is a logic circuitry and software programs that respond to and processes logical instructions to get a meaningful result. It may be implemented in the system as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuitries, one or more microchips or integrated circuits interconnected using a parent board, hardwired logic, software stored by a memory device and executed by a microprocessor, firmware, an application specific integrated circuit (ASIC), and/or a field programmable gate array (FPGA), and/or any component that operates on signals based on operational instructions. The processor (103) is equipped with dedicated GPUs, scalable architecture, and high-speed networking capabilities, which enable the server computer to accelerate model training and inference while maintaining reliability through redundancy measures and robust security protocols. It is further optimized with software libraries and frameworks for machine learning tasks, alongside monitoring and management tools.

[0014] The processor (103) configured to execute the method step 300 by propagating the real-time value of temperature and the historical data log through the trained machine learning model (M) to determine a compressing set value ( CSV) and predict the remaining useful life (RUL) of the elastomer sealing (101) based on the determined CSV and an experimental data. The historical data log comprises information regarding previous usage of the elastomer sealing (101) under various temperatures and at least the last observed value of CSV. The processor (103) updates the database (102) by recording the real-time value of temperature and the determined CSV. The experimental data information comprises the measured CSV for pre-defined compression force value at predefined temperatures.

[0015] As used in this application, the terms "component," "system," "model," "database," are intended to refer to a computer-related entity or an entity related to, or that is part of, an operational apparatus with one or more specific functionalities, wherein such entities can be either hardware, a combination of hardware and software, software, or software in execution. It must be understood that each of the component of the system may be implemented in different architectural frameworks depending on the applications.

[0016] 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.

[0017] Figure 2 illustrates method steps to train the machine learning model (M) (M) to determine remaining useful life (RUL) of an elastomer sealing (101). The machine learning model (M) resides in the system explained in accordance with figure 1. The components of the system (100) have been spelt out in detail in accordance with figure 1. For clarity it is reiterated that the system comprises a machine learning model (M), a temperature sensor (104) and at least a processor (103), said processor (103) is in communication with a database (102).

[0018] Method step 201 comprises receiving a real-time value of temperature for elastomer sealing (101) environment from the temperature sensor (104). The elastomer sealing (101) could in use in an industrial or automotive electronics.

[0019] Method step 202 comprises retrieving a historical data log for the elastomer sealing (101) from a database (102). The historical data log comprises information regarding previous usage of the elastomer sealing (101) under various temperatures and at least a last observed value of CSV.

[0020] Method step 203 comprises determining an intermediate Compression Set Value (CSV) based on real-time value of temperature and the historical data log by means of the machine learning model (M). The intermediate CSV is a result of initialization of the machine learning model (M) and hence not necessarily accurate.

[0021] Method step 204 comprises defining a loss function for the machine learning model (M) based on a limiting compression set value. The limiting CSV is the limiting value of compression set. This is defined as the range/threshold CSV a particular material should maintain to have operational integrity.

[0022] Method step 205 comprises optimizing the loss function to get a final value of CSV. Optimizing the loss function makes the model learn the CSV in the boundary condition of a limiting CSV. In an exemplary embodiment of the present invention, the machine learning model (M) is a neural network. Neural networks are inspired by the biological neural network or brain cell i.e. neurons. The network parameters include but are not limited to a layers, filter and the like. For simplicity, in computer science, a network of neurons are represented as a set of layers. These layers are categorized into three classes which are input, hidden, and output. Every network has a single input layer and a single output layer. Different layers perform different kinds of transformations/operations on their inputs. Data flows through the network starting at the input layer and moving through the hidden layers until the output layer is reached. Hyperparameter is a parameter whose value is used to control the learning process. While networks parameters are learned during the training stage, hyper parameters are given/chosen. Hyper parameters are typically characterized by the learning rate, learning pattern and the batch size. In method step 205 we basically tune or configure the hyperparameters and network parameters until machine learning model (M) learns to calculate an accurate value of CSV by extrapolation of historical data log in context of the real-time value of temperature subject to the boundary condition of a limiting compression set value.

[0023] Method step 206 comprises predicting RUL based on the CSV to train the machine learning model (M). The real-time value of temperature and the final value CSV are recorded in the database (102). The aforementioned training methodology can be expressed in the form of equations as follows.

[0024] ML model training phase:
? Compression set value (CSV) forecast algorithm as a function of time for a given thermal load o ?????? = ??(????????)?????????????????????? ……… (1)
? Compression set value estimation (CSV) algorithm as a function of Time, thermal load and previous CSV state in time domain o ?????? = ??(????????, ??????????????????????, ???????????????????? ???????????????? ??????)…… (2)
? Define limiting compression set values ???????? at a given thermal load
? Dynamic RUL estimation : Estimate CSV values at the given thermal load from (2). Let the predicted CSV values at ti at temperature T be C.
• The predicted CSV value from (2) at ti-1 at temperature T1 be Ci-1
• C=g(ti,T,Ci-1)
• Estimated RUL : ??-1.(????????) - ??-1(??)|??

o C: This is the current predicted CSV at a specific time ti and temperature T.
o f-1: This represents the inverse function of f, which correlates time and temperature with the CSV. The inverse function f-1 is used to determine the corresponding time for a given CSV value.
o f-1 (YLim): This is the time at which the CSV is expected to reach the limiting value YLim.
o f-1 ( C): This is the time at which the CSV is anticipated to reach the value C.
o The operation f-1 (YLim) - f-1 (C): This calculates the time difference between when the CSV will reach the limiting value and its current value C, effectively providing an estimate of the RUL. The vertical bar notation T specifies that the entire calculation is conducted at a constant temperature T .

[0025] Figure 3 illustrates method steps (300) of determining a remaining useful life (RUL) of an elastomer sealing (101) using a trained machine learning model (M). The machine learning model (M) has been trained in accordance with the aforementioned method steps (300). The machine learning model (M) resides in the system explained in accordance with figure 1.

[0026] Method step 301 comprises measuring a real-time value of value of temperature for the elastomer sealing (101) environment by means of a temperature sensor (104).

[0027] Method step 302 comprises retrieving a historical data log for the elastomer sealing (101) from a database (102). The historical data log comprises information regarding previous usage of the elastomer sealing (101) under various temperatures and at least the last observed value of CSV.

[0028] Method step 303 comprises propagating the real-time value of temperature and the historical data log through the trained machine learning model (M) to determine a compressing set value ( CSV). The machine learning model (M) is trained to calculate an accurate value of CSV by extrapolation of historical data log in context of the real-time value of temperature subject to the boundary condition of a limiting compression set value. The method steps 303 is the inference time function of the machine learning model (M).

[0029] Method step 304 comprises predicting the remaining useful life (RUL) of the elastomer sealing (101) based on the determined CSV and experimental data. The experimental data refers to the measured CSV for pre-defined compression force value at predefined temperatures.

[0030] The proposed methodology and system provides a data driven approach to estimate the remaining useful life of Elastomer sealing (101)s dynamically as opposed to one-time static solutions in the state of the art. The proposed methodology (method steps 200 and 300) accommodate the system dynamics that lead to varied thermal loads impacting the life of the concerned system. The data driven approach incorporating machine learning algorithms estimates the impact on the lifetime of the sealing material considering the dynamic load the system is exposed to at a given duration of time. The proposed invention can be used for elastomer sealing (101) in any electronic components of heavy machinery that require oil cooling. Such sealings are exposed to harsh environments but are critical to safe working of the electronic components and thereby the heavy machinery.

[0031] 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 modification and customizations to the system. 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 ancillary modification to the method of training the machine learning model (M) (M) to determine remaining useful life of an elastomer and the system thereof are envisaged. The scope of this invention is limited only by the claims.
, Claims:We Claim:
1. A method of training a machine learning model (M) to determine remaining useful life (RUL) of an elastomer sealing (101), the method comprising:
receiving a real-time value of temperature for elastomer sealing (101) environment from a temperature sensor (104);
retrieving a historical data log for the elastomer sealing (101) from a database (102);
determining an intermediate Compression Set Value (CSV) based on real-time value of temperature and the historical data log by means of the machine learning model (M);
defining a loss function for the machine learning model (M) based on a limiting compression set value;
optimizing the loss function to get a final value of CSV;
predicting RUL based on the CSV to train the machine learning model (M).

2. The method of training a machine learning model (M) as claimed in claim 1 wherein the historical data log comprises information regarding previous usage of the elastomer sealing (101) under various temperatures and at least a last observed value of CSV.

3. The method of training a machine learning model (M) as claimed in claim 1 wherein the real-time value of temperature and the final value CSV are recorded in the database (102).

4. A method of determining a remaining useful life (RUL) of an elastomer sealing (101) using a trained machine learning model (M), the method comprising:
measuring a real-time value of value of temperature for the elastomer sealing (101) environment by means of a temperature sensor (104);
retrieving a historical data log for the elastomer sealing (101) from a database (102);
propagating the real-time value of temperature and the historical data log through the trained machine learning model (M) to determine a compressing set value ( CSV);
predicting the remaining useful life (RUL) of the elastomer sealing (101) based on the determined CSV and experimental data.

5. The method of determining a remaining useful life (RUL) of an elastomer sealing (101) as claimed in claim 4, wherein the historical data log comprises information regarding previous usage of the elastomer sealing (101) under various temperatures and at least the last observed value of CSV.

6. The method of determining a remaining useful life (RUL) of an elastomer sealing (101) as claimed in claim 4, wherein the machine learning model (M) is trained to calculate an accurate value of CSV by extrapolation of historical data log in context of the real-time value of temperature subject to the boundary condition of a limiting compression set value.

7. The method of determining a remaining useful life (RUL) of an elastomer sealing (101) as claimed in claim 4, wherein the experimental data refers to the measured CSV for pre-defined compression force value at predefined temperatures.

8. A system to determine a remaining useful life (RUL) of an elastomer sealing (101), said system comprising a temperature sensor (104) adapted to measure real-time value of temperature for the elastomer sealing (101) environment, a database (102) for storing historical data log comprising information regarding previous usage of the elastomer sealing (101) under various temperatures and at least a last observed value of CSV, the temperature sensor (104) and the database (102) in communication with a processor (103), characterized in that system:
the processor (103) in communication with a trained machine learning model (M), the processor (103) configured to:
propagate the real-time value of temperature and the historical data log through the trained machine learning model (M) to determine a compressing set value ( CSV);
predict the remaining useful life (RUL) of the elastomer sealing (101) based on the determined CSV and an experimental data.

9. The system to determine a remaining useful life (RUL) as claimed in claim 7, wherein the historical data log comprises information regarding previous usage of the elastomer sealing (101) under various temperatures and at least the last observed value of CSV.

10. The system to determine a remaining useful life (RUL) as claimed in claim 7, wherein the processor (103) updates the database (102) by recording the real-time value of temperature and the determined CSV.