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 an application of artificial intelligence and machine learning for monitoring thermal behavior of control units. In particular, the present invention discloses system to monitor thermal behavior of at least one control unit and a method of training an AI model thereof.

Background of the invention
[0002] Control units used in industrial applications such as automotive sector demand for a safety stringent electronic control unit. In such complex systems, monitoring the thermal behavior of the electronic sub-components constituting the control unit is very essential. The thermal behavior of individual sub-components in electronic control unit like ICs, MOSFETs, microcontrollers can be modelled by its appropriate thermal characterization. Proper thermal characterization of the sub-components is required to assure reduced thermal events. The temperature requirement from the system level is mainly based on exposure range of temperature. Electronic control units are mainly classified based on the range of temperature exposure as Commercial / Domestic: 0°C to 85°C, Industrial: -40°C to 100°C, Automotive: −40°C to 125°C and Military: −55°C to 125°C.

[0003] Currently, there are techniques like inter-dependent temperature monitoring between sub-components for detecting and mitigating thermal events on a sub-component within the electronic control unit as a safety mechanism. But these techniques can fail in real time if certain use-cases are not taken into consideration during the design phase of the electronic control unit. In some instances, the real failure scenario is hidden due to the complex design weakness which is only identified at a later stage when the control unit is deployed in the industrial set-up. It is then expensive & highly time consuming to go for a re design of the control unit. Hence there is a need for precise and accurate method of monitoring thermal behavior of control units that takes into account the complex design weakness.

[0004] Patent Application US2014153607 AA titled “Calibrating thermal behavior of electronics” discloses a method of determining a relationship between indirect thermal data for a processor and a measured temperature associated with the processor, during a calibration process, obtaining the indirect thermal data for the processor during actual operation of the processor, and determining an actual significant temperature associated with the processor during the actual operation using the indirect thermal data for the processor during actual operation of the processor and the relationship.

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 to monitor thermal behavior of at least one control unit (10); and at least
[0007] Figure 2 illustrates method steps for training an AI Model (20) to monitor thermal behavior of a control unit (10).
Detailed description of the drawings

[0008] Figure 1 depicts a system to monitor thermal behavior of at least one control unit (10). The system comprises a trained AI Model (20) in communication with the control unit (10). An AI Model (20) can be defined as reference or an inference set of data, which is use 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. The AI Model (20) is trained in accordance with method step 200.

[0009] It must be understood that this disclosure is not specific to the type of model being executed and can be applied to any AI module irrespective of the AI Model (20) being executed. A person skilled in the art will also appreciate that the AI model 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 can be a hardware embodied within an electronic chip. Such neural network (101) chips are specialized silicon chips, which incorporate AI technology and are used for machine learning.

[0010] The control unit (10) is an electronic logic circuitry that respond to and processes logical instructions to get a meaningful result. The control unit (10) may be deployed in any industrial setup. For the purposes of explanation, the working of the control unit (10) is explained with respect to an automotive set-up, however that in no limits the application of the control unit (10) to any other industrial set-up. The control unit (10) comprises a housing encapsulating a plurality of integrated circuits (IC) and at least a plurality of electronic components interconnected with each other on a Printed Circuit Board (PCB (11)). The control unit (10) has at least one input port and at least one output port and is in connection with at least one battery.

[0011] The most important non-limiting feature of the present invention is the functionality of the control unit (10). The control unit (10) is configured to feed a plurality of parameters as input to the trained AI Model (20). While generating the input, the control unit (10) is configured to aggregate a plurality of parameters before these parameters are fed to the AI Model (20).

[0012] The control unit (10) is configured to calculate temperature values (T1,….Tn) at multiple locations of the PCB (11) at a sampling rate of (S1) with the help of on-board or integrated temperature sensors and relative temperatures of points (T1,…Tn) are calculated. The temperature sensed by negative or positive temperature co-efficient sensors.

[0013] Further control unit (10) is configured to record temperature values (tc1,….tc2) of each of the ICs (14) and electronic components (Q1…Q9)(16) at the sampling rate of (S1). These values are measured by a specialized semiconductor-based temperature sensor is usually incorporated into integrated circuits (ICs) or in other cases it is manually determined and fed into the control unit (10) using conventional thermistors or thermocouples.

[0014] Further control unit (10) records a value of current at each of the input port (s) (12) (ii1,…iin) and output port(s) (13) (io1,…ion). The control unit (10) retrieves a plurality of control unit (10) design parameters. These design parameters comprise but are not limited to resistance (R) of the control unit (10) housing (15), presence or absence of a water protective cover for the control unit (10), ambient temperature of the surrounding wherein the control unit (10) is operating, air-flow in the ambient surroundings of the ECU and the like. The control unit (10) also retrieves a value of voltage supplied by the battery.

[0015] The control unit (10) is further configured to raise an alarm (30) in dependance of an output received from the AI Model (20). The alarm (30) is raised when the output of the AI Model (20) indicates that the control unit (10) is not working in a safe state. The alarm (30) can be an audio/visual indicated relayed through a human machine interface in the industrial set-up. For example, in the automotive set-up a small warning lamp is lit on the dashboard of the vehicle, informing the driver of thermal instability in the particular control unit (10).

[0016] The AI Model (20) is configured to process the input and generate the output comprising a value of state of the control unit (10), hold timing and at least a value of temperature correlation thresholds for the values of (T1,….Tn), (tc1,….tc2). Hold timing is minimum time the system can persist without safety violation. No system reaction will be done in hold time. The AI Model (20) is trained to learn a correlation between the plurality of parameters and the state of control unit (10) to determine an optimum working range between the plurality of parameters of the control unit (10). The training mechanism is elucidated in accordance with method steps (200).

[0017] 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.

[0018] Figure 2 illustrates method steps for training an AI Model (20) to monitor thermal behavior of a control unit (10). The AI Model (20) and the control unit (10) have been explained in accordance with figure 1. For the purposes of clarity, it is reiterated said control unit (10) comprising a housing encapsulating a plurality of integrated circuits (IC) and at least a plurality of electronic components interconnected with each other on a Printed Circuit Board (PCB (11)), said control unit (10) having at least one input port and at least one output port, said control unit (10) in connection with at least one battery.

[0019] Method step 201 comprises creating a training dataset comprising a plurality of parameters. The training dataset is created by aggregating the value of plurality of parameters. This further comprises calculating temperature values (T1,….Tn) at multiple locations of the PCB (11) at a sampling rate of (S1); calculating temperature values (T1,….Tn) at multiple locations of the PCB (11) at a sampling rate of (S1); recording temperature values (tc1,….tc2) of each of the ICs (14) and electronic components (Q1..Q9)(16) at the sampling rate of (S1); recording a value of current at each of the input port (s) (ii1,…iin) and output port(s)(io1,…ion); retrieving a plurality of control unit (10) design parameters; retrieving value of voltage supplied by the battery. The creation of the training dataset has been elucidated in preceding paras.

[0020] Method step 202 comprises feeding the training dataset to the AI Model (20). Method step 203 comprises labelling an output of the AI Model (20) for the fed training dataset. Labelling the output of the AI Model (20) comprises providing a value of state of the control unit (10) and at least a value of temperature correlation thresholds for the values of (T1,….Tn), (tc1,….tc2) as output for the training dataset. For a value of the parameters of the training dataset, a correct value of a state of the control unit (10) (i.e. whether it is operating in a safe state or not), correct value of temperature correlation thresholds for the values of (T1,….Tn), (tc1,….tc2) are provided as output. This is therefore a supervised training mechanism and multiple iterations are carried out for diverse input-output pairs, so that AI Model (20) learns the correlation between parameters for various temperature correlation thresholds for both safe and non-safe state.

[0021] Method step 203 comprises training the AI Model (20) to learn a correlation between the plurality of parameters and the state of control unit (10) to determine an optimum working range between the plurality of parameters of the control unit (10). Based on the labelling of output for different sets of inputs, the AI Model (20) learns the range of values of parameters for the control unit (10) to operate in a safe state. It further learns a value of temperature correlation thresholds for the values of (T1,….Tn), (tc1,….tc2) or hold timings.

[0022] A person skilled in the art will appreciate that while these method steps describe only a series of steps to accomplish the objectives, these methodologies may be implemented with customized modifications to the system.

[0023] This idea to develop a system to monitor thermal behavior of at least one control unit (10) and a training method (200) thereof aims to arrive at accurate detection of safe state of the control unit (10) for different loads and establish new temperature correlation thresholds & new hold timings. Further the proposed system does not need any costly & time-consuming physical testing.

[0024] 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 and adaptation of the system to monitor thermal behavior of at least one control unit (10) and the training method 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. A method (200) of training an AI Model (20) to monitor thermal behavior of a control unit (10), said control unit (10) comprising a housing encapsulating a plurality of integrated circuits (IC (14)) and at least a plurality of electronic components interconnected with each other on a Printed Circuit Board (PCB (11)), said control unit (10) having at least one input port (12) and at least one output port (13), said control unit (10) in connection with at least one battery, the training method comprising:

creating (201) a training dataset comprising a plurality of parameters, creation of the training dataset further comprising;
calculating temperature values (T1,….Tn) at multiple locations of the PCB (11) at a sampling rate of (S1);
recording temperature values (tc1,….tc2) of each of the IC (14)s and electronic components at the sampling rate of (S1);
recording a value of current at each of the input port (s) (ii1,…iin) and output port(s)(io1,…ion);
retrieving a plurality of control unit (10) design parameters;
retrieving value of voltage supplied by the battery;
feeding (202) the training dataset to the AI Model (20);
labelling (203) an output of the AI Model (20) for the fed training dataset;
training (204) the AI Model (20) to learn a correlation between the plurality of parameters and the state of control unit (10) to determine an optimum working range between the plurality of parameters of the control unit (10).

2. The method (200) of training an AI Model (20) to monitor thermal behavior of control unit (10)s as claimed in claim 1, wherein labelling the output of the AI Model (20) comprises providing a value of state of the control unit (10) and at least a value of temperature correlation thresholds for the values of (T1,….Tn), (tc1,….tc2) as output for the training dataset.

3. A system (100) to monitor thermal behavior of at least one control unit (10), said system comprising a trained AI Model (20) in communication with the control unit (10), said control unit (10) comprising a housing encapsulating a plurality of integrated circuits (IC (14)) and at least a plurality of electronic components interconnected with each other on a Printed Circuit Board (PCB (11)), said control unit (10) having at least one input port (12) and at least one output port (13), said control unit (10) in connection with at least one battery, the system characterized by:

the control unit (10) configured to feed a plurality of parameters as input to the trained AI Model (20), while generating the input, the control unit (10) is configured to:
calculate temperature values (T1,….Tn) at multiple locations of the PCB (11) at a sampling rate of (S1);
record temperature values (tc1,….tc2) of each of the IC (14)s and electronic components at the sampling rate of (S1);
record a value of current at each of the input port (s) (ii1,…iin) and output port(s)(io1,…ion);
retrieve a plurality of control unit (10) design parameters;
retrieve a value of voltage supplied by the battery;
raise an alarm in dependance of an output received from the AI Model (20);

the AI Model (20) configured to process the input and generate the output comprising a value of state of the control unit (10) and at least a value of temperature correlation thresholds for the values of (T1,….Tn), (tc1,….tc2).

4. The system (100) to monitor thermal behavior of at least one control unit (10) as claimed in claim 3, wherein the AI Model (20) is trained to learn a correlation between the plurality of parameters and the state of control unit (10) to determine an optimum working range between the plurality of parameters of the control unit (10).