Method and apparatus for temperature prediction

Abstract

A system 1 for predicting thermal behaviour of a component 13, 15 of an electric drive system 11 comprises one or more processors 100. The one or more processors receives at least one input signal 80(

Description

TECHNICAL FIELD The present disclosure relates to a method and apparatus for temperature prediction. Aspects of the invention relate to a control system, a system, a vehicle and a method. BACKGROUND It is known to measure the temperature of a vehicle subsystem (or component) during operation of a vehicle. One or more operating parameter of the vehicle may be modified in dependence on the measured temperature. For example, the operating parameter(s) may be controlled in dependence on detection of a thermal event, such as over- / under-heating of the subsystem. It is also known to model the temperature of a vehicle subsystem (or component) without requiring a temperature sensor. The temperature may be modelled based on a measured time-varying operating parameters) of a related subsystem. For example, the temperature of an exhaust gas treatment system may be modelled in dependence on an operating condition of an internal combustion engine. It would be desirable to be able to predict a temperature, or a change in temperature, of a component or subsystem in the vehicle. The predicted temperature could be used pre-emptively to control one or more vehicle subsystem. The predicted temperature may also be used where a measured temperature of a component is only available intermittently. It is an aim of the present invention to address one or more of the disadvantages associated with the prior art. SUMMARY OF THE INVENTION Aspects and embodiments of the invention provide a control system, a system, a vehicle and a method as claimed in the appended claims. According to an aspect of the present invention there is provided a system for predicting thermal behaviour of a component of an electric drive system, the system comprising one or more processors collectively configured to: receive at least one input signal indicating a time-varying parameter of the electric drive system, the at least one input signal comprising a temperature signal indicating a measured temperature of the component; implement a temperature model to determine a predicted temperature of the component; predict occurrence of a thermal event and an event time of the or each predicted thermal event, wherein predicting the occurrence of the thermal event comprises determining that the predicted temperature crosses one or more predetermined temperature threshold. The system may be configured to output a notification signal indicating the predicted thermal event and the predicted event time. The measured temperature indicated by the temperature signal may be supplied to the temperature model. The temperature model may be configured to update the predicted temperature data in dependence on the measured temperature. The one or more processors may be configured to predict the thermal event and the event time in dependence on the updated temperature data. By predicting the thermal behaviour of the component, the system can predict the occurrence of a thermal event, such as an overheat thermal event or an underheat thermal event. The system may be configured to implement a control strategy to avoid or manage the thermal event. The control strategy may be implemented pre-emptively, i.e. before the thermal event occurs. The one or more processor may be configured to classify the predicted thermal event, for example as an overheat thermal event or as an underheat thermal event. The system may change or modify an operating mode of the electric drive system to prevent occurrence of the predicted thermal event. The system may predict the thermal behaviour for each of a plurality of different control strategies and select one of the control strategies in dependence on the predicted thermal behaviour. The temperature model may re-calibrate the predicted temperature in dependence on the measured temperature of the component. The re-calibration may be performed periodically to improve the accuracy of the predicted temperature(s). The system may comprise one or more controllers collectively comprising at least one electronic processor having an electrical input for receiving an input signal; and at least one memory device electrically coupled to the at least one electronic processor and having instructions stored therein. The at least one electronic processor may be configured to access the at least one memory device and execute the instructions thereon so as to: receive the at least one input signal indicating a time-varying parameter of the electric drive system, the at least one input signal comprising a temperature signal indicating a measured temperature of the component; implement the temperature model to determine a predicted temperature of the component; and predict occurrence of a thermal event in the component and an event time of the or each predicted thermal event, wherein predicting the occurrence of the thermal event comprises determining that the predicted temperature crosses one or more predetermined temperature threshold. The at least one electronic processor may be configured to output the notification signal indicating the predicted thermal event and the predicted event time. The at least one electronic processor may be configured to supply the measured temperature indicated by the temperature signal to the temperature model. The temperature model may be configured to update the predicted temperature data in dependence on the measured temperature; the one or more processors being configured to predict the thermal event and the event time in dependence on the updated temperature data. The at least one electronic processor may be configured to output the or each time-varying parameter indicated by the at least one input signal to the temperature model. The temperature model may be configured to determine the predicted temperature in dependence on the or each time-varying parameter. The at least one electronic processor may be configured to supply the or each time-varying parameterto the temperature model continuously or intermittently. The temperature model may be configured to determine the predicted temperature of the component in dependence on the or each time-varying parameter. The at least one input signal may be received continuously. The temperature signal may be received continuously. The temperature signal may provide a continuous indication of the measured temperature of the component. The temperature model may be operative to predict the temperature of the component in dependence on an interruption in the availability of the measured temperature. Alternatively, the at least one input signal may be received only intermittently. The temperature signal may be received intermittently or periodically. The measured temperature of the component may not be available continuously. The measured temperature of the component may only be available periodically or intermittently. The measured temperature of the component may be interrupted or discontinuous. The measured temperature may be incomplete, for example representing the measured temperature of the component at discrete times, for example discrete times separated by a time interval. The temperature model may be configured to predict the temperature of the component in the time interval when the measured temperature(s) is not available, for example in the time interval between receipt of successive temperature signals. The at least one input signal may indicate an instantaneous value of the time-varying parameter. The at least one electronic processor may be configured to supply the instantaneous value of the or each time-varying parameter to the temperature model. The temperature model may be configured to determine the predicted temperature in dependence on the instantaneous value of the or each time-varying parameter. The temperature model may comprise one or more regression model for predicting the temperature of the component. The one or more regression model may model the temperature of the component in dependence on the or each time-varying parameter. The electric drive system may comprise at least one electric drive unit. The or each electric drive unit may comprise at least one electric machine for generating torque. The time-varying parameter may comprise a torque generated by the one or more electric drive unit. The temperature model may determine the predicted temperature in dependence on the torque generated by the one or more electric drive unit. The at least one input signal may comprise a torque signal indicating the torque generated by the one or more electric drive unit. Alternatively, or in addition, the temperature model may determine the predicted temperature in dependence a power (Watts) supplied to the electric drive unit. The torque generated by the or each electric unit may be measured, for example by a torque sensor. Alternatively, the torque generated by the or each electric drive unit may be modelled, for example in dependence on the electric current supplied to the or each electric machine. The temperature model may be configured to predict the temperature of the component in dependence on the torque generated by the or each electric machine. The temperature model may be configured to predict the temperature of the at least one electric drive unit. The time-varying parameter may comprise the power supplied to the one or more electric drive unit. The at least one input signal may comprise a power signal indicating the power supplied to the one or more electric drive unit. The temperature model may be configured to predict the temperature of the component in dependence on the power supplied to the one or more electric drive unit. The electric drive system may comprise an energy regeneration system for regenerating energy. The timevarying parameter may comprise the energy regenerated by the energy regeneration system. The at least one input signal may comprise an energy regeneration signal providing an indication of the regenerated energy. The regenerated energy may be electrical energy which is regenerated by converting mechanical energy, such as potential energy and / or kinetic energy, into electrical energy. The energy regeneration system may be an electrical regeneration system. The electrical regeneration system may utilise one or more electric machine as a generator which outputs electrical energy. The electrical regeneration system may utilise the one or more electric machine provided in the at least one electric drive unit to generate electrical energy. The energy regeneration signal may provide an indication of the amount of electrical energy regenerated by the one or more electric machine. The regenerated energy may be calculated by integrating the power over time. The regeneration signal may indicate an instantaneous value of the regenerated energy, for example an instantaneous value of the electrical energy regenerated by the electrical regeneration system. The temperature model may be configured to determine the predicted temperature in dependence on the regenerated energy. The temperature model may be configured to predict the temperature of the electrical regeneration system. The temperature model may determine the predicted temperature in dependence on an ambient temperature. The ambient temperature may be measured by a temperature sensor provided on the vehicle. Alternatively, the ambient temperature may be modelled, for example in dependence on meteorological data. The temperature model may comprise a first regression model for predicting the temperature of the component. Alternatively, or in addition, the temperature model may comprise a second regression model for predicting the temperature of the coolant. The temperature model may comprise additional regression models for predicting the temperature of further components in the electric drive system. The temperature model may be configured to determine the predicted temperature of the component at each of a plurality of time steps. The temperature model may predict a temperature change for each time step and / or a magnitude of the temperature for each time step. The plurality of time steps may be successive time steps, i.e. representing a continuous temporal sequence. The predicted temperature at each time step may be used as an input for determining the predicted temperature of the next time step. The temperature model may comprise a recursive model for predicting the temperature of the component over the plurality of time steps. The temperature model may be configured to determine the predicted temperature between receipt of successive temperature signals indicating the measured temperature of the component. The temperature model may predict the temperature of the component in the time interval between successive temperature signals. The temperature model may be configured to predict the temperature for one timestep. The temperature model may use the predicted temperature as an input to predict the temperature for the next timestep. The process is repeated for a plurality of timesteps to model the temperature of the component. The predicted temperature at each time step may be recorded as time-series temperature data. The at least one input signal may comprise a first temperature signal indicating a first measured temperature of the component measured at a first time. The temperature model may generate a first predicted temperature in dependence on the first temperature signal. The first predicted temperature may comprise a temperature prediction in the time period succeeding the first time. The system may determine a first time when the predicted temperature of the component will cross the one or more temperature threshold. The at least one input signal may comprise a second temperature signal indicating a second measured temperature of the component measured at a second time. The temperature model may generate a second predicted temperature in dependence on the second temperature signal. The second predicted temperature may comprise a temperature prediction in the time period succeeding the second time. The system may determine a second time when the predicted temperature of the component will cross the one or more temperature threshold. The temperature model may be trained on a data set comprising substantially continuous temperature measurements of the component. The temperature model may comprise a regression model trained on the substantially continuous temperature measurements of the component. The temperature model may predict the temperature and / or changes in the temperature of the component with respect to time. The temperature model may be configured to generate time-series temperature data representing the predicted temperature of the component. The time-series temperature data may represent the predicted temperature over a period of time. The period of time may span a time interval between successive temperature measurements. The at least one input signal may comprise a power signal indicating a measured power of the at least one electric drive system. The power signal may indicate an instantaneous value of the measured power. The electric drive system may comprise an inverter. The inverter may be configured to supply electrical power to the at least one electric drive unit. The at least one component may comprise the inverter of the electric drive system. The temperature model may be configured to predict the temperature of the inverter. The temperature model may comprise a regression model for predicting the temperature of the inverter. The temperature model may predict the temperature of the inverter in dependence on the predicted temperature of a coolant. The at least one input signal may comprise a measured temperature of the inverter. Alternatively, or in addition, the at least one input signal may comprise the electrical current output by the inverter. The one or more predetermined temperature threshold may comprise a first temperature threshold and a second temperature threshold. The first temperature threshold may be an upper temperature threshold and the second temperature threshold may be a lower temperature threshold. The one or more processors may be configured to predict a first said thermal event and an associated first event time in dependence on a determination that the predicted temperature crosses the first temperature threshold. The first thermal event may comprise an overheat thermal event. Alternatively, or in addition, the one or more processors may be configured to predict a second said thermal event and an associated second event time in dependence on a determination that the predicted temperature crosses the second temperature threshold. The second thermal event may comprise an underheat thermal event. The notification signal may indicate a class (or type) of the predicted thermal event. For example, the notification signal may class the predicted thermal event as one of an overheat thermal event and an underheat thermal event. The electric drive system may comprise one or more electric drive unit. The temperature model may predict the temperature of the component of each of a plurality of electric drive units. The at least one electric drive system may be configured to operate in a plurality of operating modes. The one or more processors may be configured to select one of the plurality of operating modes in dependence on the predicted event time. The plurality of operating modes of the electric drive system may define different rated electrical capabilities of the or each electric drive unit. The plurality of operating modes may define first and second rated electrical capabilities of the electric drive unit. The first rated electrical capability may be higher than the second rated electrical capability. The one or more processors may be configured to select the second operating mode to reduce a rated electrical capability of the or each drive unit. This may de-rate the electric drive system, for example to avoid an overheat thermal event. The one or more processors may be configured to select one of the plurality of operating modes in advance of (i.e., before) the predicted event time. Alternatively, or in addition, the one or more processors may be configured to select one of the plurality of operating modes in dependence on a classification of the predicted thermal event. The temperature model may be configured to predict the temperature of the component in each of the plurality of operating modes. The temperature model may predict the thermal event and the event time for each of the plurality of operating modes. The selection of the one of the plurality of operating modes may be performed in dependence on the predicted thermal event and / orthe predicted event time in each of the operating modes. The one or more processors may be configured to output a control signal to configure the at least one electric drive system to operate in the selected one of the plurality of operating modes. According to an aspect of the present invention there is provided a system for predicting thermal behaviour of a component and / or a coolant of an electric drive system, the system comprising one or more processors collectively configured to: receive at least one input signal indicating a time-varying parameter of the electric drive system, the at least one input signal comprising a temperature signal indicating a measured temperature of at least one of the component and the coolant; implement a temperature model to determine a predicted temperature of the component and / orthe coolant; predict occurrence of a thermal event and an event time of the or each predicted thermal event, wherein predicting the occurrence of the thermal event comprises determining that the predicted temperature of the component and / or the coolant crosses one or more predetermined temperature threshold. The system may be configured to output a notification signal indicating the predicted thermal event and the predicted event time. The measured temperature indicated by the temperature signal may be supplied to the temperature model. The temperature model may be configured to update the predicted temperature data in dependence on the measured temperature. The one or more processor may be configured to predict the thermal event and the event time in dependence on the updated temperature data. According an aspect of the present invention there is provided a method of predicting thermal behaviour of a component of an electric drive system, the method comprising: measuring a temperature of the component; implementing a temperature model to determine a predicted temperature of the component; and predicting occurrence of a thermal event in the component and an event time of the or each predicted thermal event, wherein predicting the occurrence of the thermal event comprises determining that the predicted temperature crosses one or more predetermined temperature threshold. The method is computer-implemented and the temperature model may be implemented by at least one electronic processor. The method may comprise outputting a signal indicating the predicted thermal event and the predicted event time. The temperature model may determine the predicted temperature in dependence on one or more time-varying parameter of the electric drive system. The one or more time-varying parameter may comprise one or more of the following: (i) a torque generated by the electric drive unit; (ii) a regeneration energy of the electric drive unit; (iii) a power (Watts) supplied to the electric drive unit. The one or more time-varying parameter may comprise an ambient temperature. The method may comprise supplying the measured temperature to the temperature model. The temperature model may be configured to update the predicted temperature data in dependence on the measured temperature. The measured temperature may be supplied periodically to the temperature model. According an aspect of the present invention there is provided a control system for controlling operation of an electric drive system, the system comprising one or more processors collectively configured to: implement a temperature model to determine a predicted temperature of a component of the electric drive system; and predict occurrence of a thermal event and an event time of the or each predicted thermal event, wherein predicting the occurrence of the thermal event comprises determining that the predicted temperature crosses one or more predetermined temperature threshold. The temperature model may comprise or consist of a regression model, for example a linear regression model. The temperature model may be a recursive model. The temperature model may determine the predicted temperature for each of a plurality of time steps, the predicted temperature for each time step being used by the temperature model as an input to determine the predicted temperature for the next time step. The control system may be configured to output a control signal to control operation of the electric drive system in dependence on the predicted thermal event and / or the predicted event time. The control signal may be configured the electric drive system to operate in a selected one of a plurality of operating modes. The operating mode may be selected in dependence on the predicted thermal event and / or the predicted event time. The one or more processor may be configured to classify the predicted thermal event, for example as an overheat thermal event or as an underheat thermal event. According to a further aspect of the present invention there is provided computer readable instructions which, when executed by one or more processors, cause the one or more processors to perform the method(s) described herein. Within the scope of this application it is expressly intended that the various aspects, embodiments, examples and alternatives set out in the preceding paragraphs, in the claims and / or in the following description and drawings, and in particular the individual features thereof, may be taken independently or in any combination. That is, all embodiments and / or features of any embodiment can be combined in anyway and / or combination, unless such features are incompatible. The applicant reserves the right to change any originally filed claim or file any new claim accordingly, including the right to amend any originally filed claim to depend from and / or incorporate any feature of any other claim although not originally claimed in that manner. BRIEF DESCRIPTION OF THE DRAWINGS One or more embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings, in which: Figure 1 shows a schematic representation of a system for predicting thermal behaviourof a component and / or a coolant in a vehicle in accordance with an embodiment of the present invention; Figure 2 shows a schematic representation of the subsystems of the vehicle shown in Figure 1; Figure 3 illustrates a vehicle control system for controlling one or more of the vehicle subsystems in dependence on the predicted thermal behaviour of the component and / or the coolant; Figure 4 illustrates a temperature prediction system for predicting the thermal behaviour of the component or the coolant in accordance with an aspect of the present invention; Figure 5 illustrates a computational system for training a temperature model for predicting the thermal behaviour of the component and / or the coolant; Figure 6 is a block diagram representing a generalised computer-implemented method of generating a temperature model; Figure 7 is a block diagram representing a computer-implemented method of generating a temperature model comprising a linear regression model for predicting the thermal behaviour of an inverter in an electric drive unit; Figure 8 is a block diagram representing a computer-implemented method of generating a temperature model comprising a linear regression model for predicting the thermal behaviour of a coolant in an electric drive unit; Figure 9 is a block diagram representing the operation of a temperature model to predict the temperatures of the inverter and coolant using respective temperature models; Figure 10 is a first plot showing a comparison of a measured temperature and a predicted temperature of a coolant without referencing measured temperatures; Figure 11 is a second plot showing a comparison of a measured temperature and a predicted temperature of a coolant with periodic references to measured temperatures; Figure 12 is a third plot illustrating the detection of a thermal event by comparing a predicted temperature to a predetermined temperature threshold in accordance with an aspect of the present invention; and Figure 13 is a block diagram illustrating the generation of a vehicle control signal in dependence on a prediction of a thermal event in accordance with an aspect of the present invention. DETAILED DESCRIPTION A system 1 and a method 400 for predicting thermal behaviour of one or more component of a vehicle 5 in accordance with an embodiment of the present invention is described herein with reference to the accompanying Figures. The vehicle 5 is described herein with reference to a reference frame comprising a longitudinal axis X, a transverse axis Y and a vertical axis Z. The reference signs herein include a suffix in the form of a whole number to differentiate between a plurality of like components on the vehicle 5. The same suffix is applied for components associated with each other, for example components forming part of the same sub-assembly of the vehicle 5. The integer n is used herein to identify a signal or event relating to a corresponding one of a plurality of features of the vehicle 5. As shown in Figure 1, the control system 1 is installed in a vehicle 5 comprising four (4) wheels 6(1)-6(4). The vehicle 5 is a road vehicle, such as an automobile, a sports utility vehicle (SUV) or a utility vehicle. As shown in Figures 1 and 2, the vehicle 5 in the present embodiment is an automobile. The vehicle 5 comprises one or more torque-generating machine 11, such as an internal combustion engine (ICE) and / or an electric drive unit. The vehicle 5 may be a battery electric vehicle (BEV), a hybrid electric vehicle (HEV), a plug-in hybrid electric vehicle (PHEV) or an internal combustion engine (ICE) vehicle. The front wheels 6(1), 6(2) and / or the rear wheels 6(3), 6(4) may be driven by the one or more torque-generating machine. In the present embodiment, the vehicle 5 comprise an electric drive system (10) 10. The electric drive system (10) 10 comprises an electric drive unit (EDU) 11 fordriving the front wheels 6(1), 6(2) of the vehicle 5. Alternatively, or in addition, the rear wheels 6(1), 6(2) of the vehicle 5 may be driven by the electric drive unit 11. The electric drive system (10) 10 may comprise more than one electric drive unit 11. The electric drive unit 11 comprises a torque-generating electric machine 13, an inverter 15 and an EDU control unit 17. The EDU control unit 17 is configured to control operation of the electric machine 13 and the inverter 15. The electric machine 13 (also known as an electric drive motor) is configured to generate a tractive force for propelling the vehicle 5. The tractive force is applied to the front wheels 6(1), 6(2) in the present embodiment. The electric drive unit 11 is electrically connected to a traction battery 19 provided onboard the vehicle 5. The inverter 15 converts direct current from the traction battery 19 to alternating current which is 9 supplied to the electric machine 13. The electric machine 13 comprises a rotor 21 and a stator 23. A coolant 25 is provided for cooling the electric machine 13. The coolant 25 is a liquid, for example oil. The rotor 21 and the stator 23 are disposed in a housing 27 of the electric machine 13. In use, the coolant 25 is circulated through the housing 27. A cooling system 29 is provided for rejecting heat energy from the electric machine 13. The cooling system 29 may, for example, comprise one or more heat exchanger 31. In use, the coolant 25 is circulated through the cooling system 29 to reject thermal energy from the electric machine 13. The cooling system 29 may comprise a pump (not shown) to circulate the coolant 25. The electric drive unit 11 comprises one or more temperature sensor 30(n) for measuring an operating temperature. The one or more temperature sensor 30(n) may measure the operating temperature of one or more component of the electric drive unit 11. The one or more temperature sensor 30(n) may measure an operating temperature of the electric machine 13 orthe inverter 15. The one or more temperature sensor 30(n) may, for example, measure an operating temperature of the rotor 21 and / or the stator 23. Alternatively, or in addition, the one or more temperature sensor 30(n) may measure the operating temperature of the coolant 25. The electric drive unit 11 in the present embodiment comprises a first temperature sensor 30(1) and a second temperature sensor 30(2). The first temperature sensor 30(1) is configured to measure an operating temperature 16 of the inverter 15 (referred to herein as the measured inverter temperature 16). The first temperature sensor 30(1) may, for example, be thermally coupled to the inverter 15. The first temperature sensor 30(1) is configured to output a first temperature signal 80(1) which represents the measured inverter temperature 16. The second temperature sensor 30(2) is configured to measure an operating temperature 18 of the coolant 25 (referred to herein as the measured coolant temperature 18). The second temperature sensor 30(2) may be disposed in the electric drive unit 11 or in the cooling system 29. For example, the second temperature sensor 30(2) may be disposed inside the housing 27 of the electric machine 13 or inside the heat exchanger 31. The second temperature sensor 30(2) is configured to output a second temperature signal 80(2) which represents the measured coolant temperature 18. The electric drive unit 11 may comprise more than two temperature sensors 30(n). The first and second temperature signals 80(1), 80(2) are published to a communication network provided on the vehicle 5. The EDU control unit 17 is configured to output a power signal 81 which provides an indication of an electric power 20 (Watts) of the electric drive unit 11. The power signal 81 represents an instantaneous (discrete) value of a rate of transfer of electrical energy of the electric drive unit 11. The EDU control unit 17 is configured also to output a torque signal 83 indicating a torque 22 (Nm) of the electric machine 13. The torque signal 83 represents an instantaneous (discrete) value of the torque 22 (Nm) generated by the electric machine 13. The EDU control unit 17 publishes the power signal 81 and the torque signal 83 to the communication network provided on the vehicle 5. The EDU control unit 17 may, for example, determine the torque 22 by accessing a look-up table which relates the torque 22 (Nm) to the current (Amps) supplied to the electric machine 13. Other techniques may be used to determine the torque 22 (Nm) generated by the electric machine 13. The torque 22 may be modelled. For example, the EDU control unit 17 may calculate the torque 22 (Nm) of the electric machine 13 using the equation: Torque = Kt.Current Where: Kt is the motor torque coefficient (Nm / Amp); and Current is the electric current (Amps) supplied to the electric motor. Alternatively, or in addition, the EDU control unit 17 may output an electric current signal 84 indicating an instantaneous (discrete) value of the electric current (Amps) supplied to the electric machine 13. The torque 22 generated by the electric machine 13 may be determined in dependence on the electric current indicated by the electric current signal 84. This calculation may be performed by a separate controller, such as the temperature prediction system 200. The vehicle 5 also comprises an air temperature sensor 30(3) for measuring an ambient air temperature 24 (°C). The air temperature sensor 30(3) is configured to output an air temperature signal 80(3) which represents an instantaneous (discrete) value of the ambient air temperature 24. The air temperature signal 80(3) is published to the communication network provided on the vehicle 5. The vehicle 5 also comprises an electrical regeneration system 31 for converting mechanical energy (comprising kinetic energy and potential energy) into electrical energy. Acceleration caused by gravity and / or deceleration of the vehicle 5 may be used to drive an electric motor which outputs regenerated energy in the form of electrical energy. The regenerated energy may be supplied to the traction battery. The electrical regeneration system 31 is configured to output an energy regeneration signal 85 which provides an indication of the regenerated energy 26 (Watts) generated by the electrical regeneration system 31. The regenerated energy 26 may be calculated by integrating the powerovertime. The regenerated energy 26 may be calculated in dependence on the mechanical energy (the sum of the kinetic energy and the potential energy). The calculation may allow for dissipated energy, for example energy dissipated by a braking resistor (braking chopper or shunt). The regenerated energy 26 in the present embodiment is calculated by the electrical regeneration system 31. The calculation may be performed by a different controller. The second temperature sensor 30(2) may optionally be provided in the electrical regeneration system 31 to measure the temperature 18 of the coolant 25. Alternatively, or in addition, a temperature sensor may be provided in the electrical regeneration system 31. The temperature sensor may, for example, measure the temperature of a braking resistor in the electrical regeneration system 31. The system and method(s) described herein may be employed to monitor the temperature of the electrical regeneration system 31 in accordance with a further aspect of the present invention. The system 1 comprises a vehicle control system 100 and a temperature prediction system 200. The vehicle control system 100 is disposed onboard the vehicle 5. In the present embodiment, the temperature prediction system 200 is disposed offboard the vehicle 5, for example in a remote data centre. The vehicle 5 comprises a wireless transceiver 35 configured to communicate between the vehicle control system 100 and the temperature prediction system 200. In a variant, the temperature prediction system 200 may be provided onboard the vehicle 5. For example, the temperature prediction system 200 may be incorporated into the vehicle control system 100. The vehicle control system 100 as illustrated in Figure 3 comprises one controller 110, although it will be appreciated that this is merely illustrative. The controller 110 comprises processing means 120 and memory means 130. The processing means 120 may be one or more electronic processing device 120 which operably executes computer-readable instructions. The memory means 130 may be one or more memory device 130. The memory means 130 is electrically coupled to the processing means 120. The memory means 130 is configured to store instructions, and the processing means 120 is configured to access the memory means 130 and execute the instructions stored thereon. The controller 110 comprises an input means 140 and an output means 150. The input means 140 may comprise an electrical input 140 of the controller 110. The output means 150 may comprise an electrical output 150 of the controller 110. The input 140 is arranged to receive one or more input signal. In the present embodiment, the one or more input signal comprises or consists of a signal from the sensor(s) provided onboard the vehicle 5. The one or more input signal comprises at least one operating parameter of the vehicle 5 and / or the electric drive unit 11. The or each operating parameter is a time-varying parameter (i.e. a parameter which changes with respect to time). In the present embodiment the one or more input signal indicates at least one operating parameter of the electric drive unit 11. The one or more input signal may optionally also indicate an operating parameter of the vehicle 5 and / or the ambient conditions. The one or more input signal may, for example, be read from the communication network on the vehicle 5. In the present embodiment, the input 140 is configured to receive: the first and second temperature signals 80(1), 80(2) from the first and second temperature sensors 30(1), 30(2); the power signal 81; the torque signal 83; and the air temperature signal 30(3). The input 140 is also configured to receive temperature prediction data 255 from the temperature prediction system 200. The output 150 is configured to output a vehicle control signal 125(n) for controlling one or more vehicle subsystems. In the present embodiment, the output 150 is configured to output a vehicle control signal 125(n) to control operation of the electric drive unit 11. The vehicle control signal 125(n) may change an operating mode of the electric drive unit 11, for example to increase or decrease an electrical rating of the capability of the electric machine 13. The output 150 is also configured to transmit data captured by the onboard sensors to the temperature prediction system 200. The output 150 outputs the first and second temperature signals 80(1), 80(2) from the first and second temperature sensors 30(1), 30(2). The vehicle control system 100 is configured to communicate with the temperature prediction system 200 over the wireless communication network. A typical communication cycle comprises two-way communication between the vehicle control system 100 and the temperature prediction system 200. The communication between the vehicle control system 100 and the temperature prediction system 200 may be half-duplex or full-duplex. The vehicle control system 100 is configured to transmit the data received from the onboard sensors and the EDU control unit 17. In the present embodiment, communication between the vehicle control system 100 and the temperature prediction system 200 is performed via the wireless transceiver 35. The vehicle control system 100 is configured to transmit the first and second temperatures 16, 18 measured by the first and second temperature sensors 30(1), 30(2). The vehicle control system 100 is configured also to transmit the electric power 20(Watts) of the electric drive unit 11; and the torque 22 (Nm) generated by the electric machine 13. The vehicle control system 100 may optionally also transmit the air temperature 24 measured by the air temperature sensor 30(3). The data transmitted by the vehicle control system 100 represents at least one instantaneous (discrete) value of the one or more time-varying parameter. The at least one instantaneous value defines the operating state of the vehicle 5 and / or the electric drive unit 11 at a discrete time. In the present embodiment, the communication between the vehicle control system 100 and the temperature prediction system 200 is intermittent. In other words, the communication between the vehicle control system 100 and the temperature prediction system 200 is interrupted or discontinuous. The intermittent communications may be caused by interrupted availability of the wireless communication network, for example due to variations in the network coverage. Alternatively, or in addition, the intermittent communication between the vehicle control system 100 and the temperature prediction system 200 may reduce data transmission over the wireless communication network. The intermittent communication between the vehicle control system 100 and the temperature prediction system 200 results in a time delay between consecutive communication cycles. The time delay may be greater than or equal to 30 seconds, 60 seconds, 90 seconds, 120 seconds, 240 seconds or 300 seconds. The time delay may be longer in certain applications. The duration of the time delay may be predetermined, for example according to a predetermined communication schedule between the vehicle control system 100 and the temperature prediction system 200. Alternatively, the time delay may be of indeterminate length, for example dependent on the availability of the wireless communication network for communications between the vehicle control system 100 and the temperature prediction system 200. The temperature prediction system 200 is configured to implement a temperature model 225 to predict (or simulate) the thermal behaviour of one or more components on the vehicle 5. In the present embodiment the temperature model is configured to predict the temperature of the inverter 15 (referred to herein as the predicted inverter temperature 16’); and the temperature of the coolant 25 (referred to herein as the predicted coolant temperature 18’). The temperature prediction system 200 as illustrated in Figure 3 comprises one controller 210, although it will be appreciated that this is merely illustrative. The controller 210 comprises processing means 220 and memory means 230. The processing means 220 may be one or more electronic processing device 220 which operably executes computer-readable instructions. The memory means 230 may be one or more memory device 230. The memory means 230 is electrically coupled to the processing means 220. The memory means 230 is configured to store instructions, and the processing means 220 is configured to access the memory means 230 and execute the instructions stored thereon. The controller 210 comprises an input means 240 and an output means 250. The input means 240 may comprise an electrical input 240 of the controller 210. The output means 250 may comprise an electrical output 250 of the controller 210. The input 240 of the temperature prediction system 200 is arranged to receive one or more input signal representing an operating parameter of one or more of the following: the electric machine 13, the inverter 15 and the coolant 25. The or each operating parameter is a time-varying parameter. The one or more input signal may comprise or consist of an instantaneous (discrete) value of the oreach operating parameter, i.e. the value of the or each operating parameter at a discrete time. In the present embodiment, the one or more input signal comprise the data signal output from the vehicle control system 100. The one or more input signal comprise the data representing one or more of the following: the measured inverter temperature 16; and the measured coolant temperature 18. The one or more input signal may optionally also comprise data representing one or more of the following: the power 20 of the electric drive unit 11; the torque 22 generated by the electric machine 13; and the measured air temperature 24. The electronic processing device 220 is configured to implement the temperature model 225. The temperature model 225 in the present embodiment is configured to predict the temperature of the inverter 15 and the coolant 25. In particular, the temperature model 225 determines the predicted inverter temperature 16’; and the predicted coolant temperature 18’. The temperature model 225 receives one or more of the following inputs: the measured inverter temperature 16, the measured coolant temperature 18, the power 20, torque 22 and the air temperature T3. In the present embodiment, each of these inputs is used to determine the predicted inverter and coolant temperatures 16’, 18’. The measured inverter temperature 16 and the measured coolant temperature 18 are received from the first and second temperature sensors 30(1), 30(2). The measured inverter and coolant inverter temperature 16, 18 are input to the temperature model 225 and used to define first and second reference (base) temperatures for the inverter 15 and the coolant 25. The temperature model 225 proceeds to determine the predicted inverter and coolant temperatures 16’, 18’ in dependence on the measured inverter and coolant temperatures 16,18. The temperature model 225 is a time-stepping regression model. In the present embodiment, the temperature model 225 comprises at least one linear regression model 227(n). More particularly, the temperature model 225 in the present embodiment is a recursive temperature model. The temperature model 225 is trained to predict changes in the temperature over a series of consecutive time steps. A predicted inverter temperature 16’ and a predicted coolant temperature 18’ are determined for each time step. The predicted inverter temperature 16’ and the predicted coolant temperatures 18’ are then used by the temperature model as a starting point for determining the predicted inverter temperature 16’ and the predicted coolant temperature 18’ after the next (consecutive) time step. For example, the temperature model 225 determines the predicted inverter and coolant temperatures 16’(x), 18’(x) after a time step t(x). The predicted inverter and coolant temperatures 16’(x), 18’(x) are then used by the temperature model 225 to determine the predicted inverter and coolant temperatures 16’(x+1), 18’(x+1) after the next time step t(x+1). The predicted inverter and coolant temperatures 16’(x+1), 18’(x+1) are then used by the temperature model 225 to determine the predicted inverter and coolant temperatures 16’(x+2), 18’(x+2) after the next time step t(x+2). This process is repeated over a plurality of time steps to determine the predicted inverter and coolant temperature 16’, 18’ over an extended time period greaterthan each time step. Each of the plurality of predicted temperatures 16’, 18’ may be determined in dependence on one or more of the preceding predicted temperatures 16’, 18’. The duration of each said time step is less than the time interval between consecutive communication cycles between the vehicle control system 100 and the temperature prediction system 200. The time steps may, for example, have a duration of 0.02 seconds (corresponding to a frequency of 50Hz). The duration of the time steps may be greater than or less than 0.02 seconds. As described herein, the measured inverter and coolant temperatures 16, 18 are periodically supplied to the temperature model 225. The measured inverter and coolant temperatures 16, 18 are used as a reference (base) temperature to recalibrate the temperature model 225. The temperature model 225 in the present embodiment is configured to determine predicted inverter and coolant temperatures 16’, 18’ representing the predicted temperature of the inverter 15 and the coolant 25 respectively. The measured inverter and coolant temperatures 16, 18 are supplied to the temperature model 225. The first and second temperatures 16, 18 are used as first and second reference (base) temperatures 16”, 18” for the temperature model 225. The temperature model 225 calculates predicted inverter and coolant inverter temperature 16’(x), 18’(x) for a first said time unitt(x). The predicted inverter and coolant temperatures 16’(x), 18’(x) forthat time unit t(x) are then used to determine the predicted inverter and coolant temperatures 16’(x+1), 18’(x+1) for the next (consecutive) time unitt(x+1). This process is repeated for each of a plurality of time units t(x, x+1, x+2,...x+n). Each of the plurality of predicted inverter and coolant temperatures 16’(x), 18’(x) is stored as time-series temperature data. The predicted inverter and coolant temperatures 16’, 18’ can be determined over an extended period of time. The predicted inverter and coolant temperatures 16’, 18’ can be determined over a period of time greater than or equal to ten (10) minutes, twenty (20) minutes, thirty (30) minutes or even forty (40) minutes. As described herein, communication between the vehicle control system 100 and the temperature prediction system 200 is intermittent, i.e. discontinuous or interrupted. Accordingly, the measured inverter temperature 16 and the measured coolant temperature 18 are available only periodically. The temperature model 225 is configured to predict the temperature of the inverter 15 and the coolant 25 during the time interval between receipt of the measured inverter and coolant temperatures 16, 18. During these time intervals, the measured inverter and coolant temperatures 16, 18 are not available to the temperature prediction system 200. The measured inverter and coolant temperatures 16, 18 are available only periodically when the temperature prediction system 200 receives the input signal(s) indicating the operating parameter(s). The measured inverter and coolant temperatures 16, 18 are supplied to the temperature model 225 and used as a reference (base) temperature. The temperature model 225 is configured to adjust the current predicted inverter temperature 16’ and the predicted coolant temperature 18’ in dependence on the received measured inverter and coolant temperatures 16,18. The predicted inverter temperature 16’ is updated to align with the measured inverter temperature 16; and the predicted coolant temperature 18’ is updated to align with the measured coolant temperature 18. The temperature model 225 continues to determine the predicted inverter temperature 16’ and the predicted coolant temperature 18’. Upon receipt of the next data set indicating the measured inverter temperature 16 and the measured coolant temperature 18, the temperature model 225 again updates the predicted inverter temperature 16’ and the predicted coolant temperature 18’. This process is repeated each time the measured inverter and coolant temperatures 16, 18 are received. The temperature prediction system 200 is configured to output predicted temperature data 255 comprising the predicted inverter temperature 16’ and the predicted coolant temperature 18’. The predicted temperature data 255 is transmitted wirelessly to the vehicle control system 100. The vehicle control system 100 is configured to compare each of the predicted inverter and coolant temperatures 16’, 18’ to at least one temperature threshold 235(n). The or each temperature threshold 235(n) is associated with a particular thermal event EV, such as an overheat scenario or an underheat scenario. The vehicle control system 100 is configured to predict occurrence of the thermal event 229(n) in dependence on a determination that the predicted temperature TP crosses the associated temperature threshold 235(n). A plurality of temperature thresholds 235(n) may be defined, each of the plurality of temperature thresholds 235(n) being associated with a particular class of thermal event 229(n). The vehicle control system 100 may classify the predicted thermal event 229(n) in dependence on which of the plurality of temperature thresholds 235(n) is crossed by the predicted inverter and coolant temperatures 16’, 18’. The vehicle control system 100 is configured to output a notification signal 135 indicating that the thermal event 229(n) has been predicted (referred to herein as a predicted thermal event 229(n)). The notification signal 135 classifies the predicted thermal event EV, for example to indicate that the predicted thermal event 229(n) is an overheat thermal event or an underheat thermal event. At least in certain embodiments, the notification signal 135 is configured to indicate a predicted event time indicating a time that the predicted thermal event 229(n) will cross the temperature threshold 235(n). Different temperature thresholds 235(n) may be defined for different components of the electric drive unit 11. The vehicle control system 100 in the present embodiment is configured to output a notification signal 135 indicating the predicted thermal event (for example to indicate a type or class of thermal event) and also the predicted event time that the predicted thermal event will occur. At least one inverter temperature threshold 235(n) is defined for the inverter 15. The at least one inverter temperature threshold 235(n) comprises a first (upper) inverter temperature threshold 235(n) and a second (lower) inverter temperature threshold 235(n). The thermal event 229(n) may, for example, represent an overheatscenario when the predicted inverter temperature 16’ exceeds the first inverter temperature threshold; and may represent an underheat scenario when the predicted inverter temperature 16’ is less than the second inverter temperature threshold 235(n). The vehicle control system 100 is configured to determine a predicted thermal event time when the predicted inverter temperature 16’ will cross one of the first and second temperature thresholds 235(n). The vehicle control system 100 is configured to output a notification signal 135 indicating a type or class of thermal event associated with the inverter 15, for example to indicate an overheat scenario or an underheat scenario for the inverter 15. The notification signal 135 also indicates the predicted event time for occurrence of the or each predicted thermal event 229(n). At least one coolant temperature threshold 235(n) is defined for the coolant 25. The at least one coolant temperature threshold 235(n) comprises a first (upper) coolant temperature threshold 235(1) and a second (lower) coolant temperature threshold 235(2). The thermal event 229(n) may, for example, represent an overheat scenario when the predicted inverter temperature 16’ of the coolant 15 exceeds the first coolant temperature threshold 235(1); and may represent an underheat scenario when the predicted inverter temperature 16’ of the coolant 15 is less than the second coolant temperature threshold 235(2). The vehicle control system 100 is configured to determine a predicted thermal event time when the predicted inverter temperature 16’ will cross one of the first and second temperature thresholds 235(n). The vehicle control system 100 is configured to output a notification signal 135 indicating a type or class of the predicted thermal event 229(n) associated with the coolant 25, for example to indicate an overheat scenario or an underheat scenario for the coolant 25. The notification signal 135 also indicates the predicted event time 233 for occurrence of the or each predicted thermal event 229(n). The vehicle control system 100 is configured to generate a vehicle control signal 125(n) in dependence on a predicted thermal event 229(n). The vehicle control signal 125(n) may modify one or more operating parameter of a vehicle subsystem 21A-21G in dependence on the predicted thermal event 229(n). Furthermore, the vehicle control system 100 may determine an appropriate time to modify the one or more operating parameter of the vehicle system in dependence on the predicted event time. By way of example, the vehicle control signal 125(n) may lower a rated electrical capability of the electric drive unit 11 (i.e., de-rate the electric drive unit 11) in dependence on a determination that the predicted inverter temperature 16’ will exceed the first (upper) inverter temperature threshold 235(n); and / or a determination that the predicted coolant temperature 18’ will exceed the first (upper) coolant temperature threshold 235(n). Conversely, the vehicle control signal 125(n) may raise a rated electrical capability of the electric drive unit 11 in dependence on a determination that the predicted inverter temperature 16’ will decrease below the second (lower) inverter temperature threshold; and / or a determination that the predicted coolant temperature 18’ will decrease below the second (lower) coolant temperature threshold. Other vehicle subsystems may be controlled in dependence on the vehicle control signal 125(n). For example, cooling may be increased in dependence on the prediction of a thermal overheat scenario. One or more air inlet may be opened to increase the supply of air to the heat exchanger 31 to promote heat rejection. A computer-implemented method of generating the temperature model 225 will now be described with reference to Figures 5 to 8. The temperature model 225 is trained on a machine learning system 300, as shown schematically in Figure 5. The machine learning system 300 comprises a controller 310 comprising processing means 320 and memory means 330. The processing means 320 may be one or more electronic processing device 320 which operably executes computer-readable instructions. The processing device 320 is configured to implement a machine learning tool set 325 for generating a linear regression model. The memory means 330 may be one or more memory device 330. The memory means 330 is electrically coupled to the processing means 320. The memory means 330 is configured to store instructions, and the processing means 320 is configured to access the memory means 330 and execute the instructions stored thereon. The controller 310 comprises an input means 340 and an output means 350. The input means 340 may comprise an electrical input 340 of the controller 310. The output means 350 may comprise an electrical output 350 of the controller 310. The input 340 is configured to receive at least one training data set 335(n)(n) for training the linear regression model. As described herein, the training data set comprises one or more time-varying parameters. For example, the training data set may comprise the following: the electric power 20 (Watts) of the electric drive unit 11; the torque 22 (Nm) generated by the electric drive unit 11; the measured air temperature 24 (°C); the measured inverter temperature 16 (°C); the measured coolant temperature 18 (°C); the energy regeneration 26 of the electric drive unit 11. An overview of the training of a linear regression model for predicting thermal behaviour will now be described with reference to a block diagram 500 shown in Figure 6. The one or more training data set 335(n) is supplied as input data (BLOCK 505). The or each training data set 335(n) comprises one or more time-varying parameter captured in relation to one or more reference vehicle 5 (equivalent to, or comparable to the vehicle 5). The or each training data set 335(n) define the or each time-varying parameter with respect to time. At least in certain embodiments, the or each training data set 335(n) comprises a continuous (i.e., uninterrupted) representation of the or each time-varying parameter with respect to time. The or each time-varying parameter in the training data set 335(n) may, for example, be defined as time-series data captured during a test period. The one or more time-varying parameter each define an operating parameter of the reference vehicle 5. The one or more time-varying parameter is captured during operation of the reference vehicle 5, typically in a range of different test conditions. The one or more time-varying parameter may, for example, be captured while the reference vehicle 5 is driving along one or more sample driving routes. In the present embodiment, the one or more time-varying parameter comprise one or more of the following: the electric power 20 (Watts) of the electric drive unit 11; a torque 22 (Nm) generated by the electric drive unit 11; the measured air temperature 24 (°C); the measured inverter temperature 16 (°C) of the inverter 15; the measured coolant temperature 18 (°C) of the coolant 25; the energy regeneration 26 of the electric drive unit 11. The time-varying parameters are captured at a frequency of between 100Hzto 200Hz, depending on the capability of the particular sensor. The frequency may be less than 100Hz or greater than 200Hz. The time-varying parameters are stored as a sequence of quantized values in the training data set 335(n). The time-varying parameters are captured at least substantially continuously, i.e. without temporal interruptions, while the reference vehicle 5 is operating. Thus, the time-varying parameters provide a continuous (i.e., uninterrupted) representation of the operating parameter of the reference vehicle 5 (or a subsystem of the vehicle 5) during a capture period. The time-varying parameters are combined into the one or more training data set 335(n). The or each training data set 335(n) may be tuned by adjusting one or more hyper-parameter 345(n) (BLOCK 510). The one or more hyper-parameter 345(n)comprises a time delay 345(1) and an aggregation window 345(2). The time delay 345(1) and / or the aggregation window 345(2) can be adjusted to simulate a heating time of a component, such as the inverter 15, and / or the coolant 25. One or more of the hyper-parameters 345(n)can be tuned respectively for power 20, torque 22 and energy regeneration 26. The tuning of the hyper-parameters 345(n)may be performed in respect of time-varying parameters accumulated under different test conditions or in respect of different test routes. The training data set 335(n) is then supplied to a machine learning tool set 325 for generating a linear regression model (BLOCK 515). An example of a suitable machine learning tool set 325 is the sklearn.linear_model.LinearRegression. The training data set 335(n) is analysed to determine a straight line that will minimize the residual sum of squares between the observed responses in the dataset, and the responses predicted by the linear approximation. The resulting straight line provides a linear regression model 227(n) which relates the temperature to the time-varying parameters. Separate linear regression models 227(n) are generated for the inverter 15 and the coolant 25. The or each linear regression model 227(n) is output as the temperature model 225 (BLOCK 520). The temperature model 225 comprises a first linear regression model 227(1) for modelling the temperature of the inverter 15; and a second linear regression model 227(2) for modelling the temperature of the coolant 25. A linear regression model 227(n) may be generated for other components in the electric drive unit 11. The resulting first and second linear regression models are suitable for predicting the temperature of the inverter 15 and the coolant 25 over a single time step. As described herein, the predicted inverter temperature 16’, 18’ is used as an input for the temperature model 225 to predict the temperature of the next time step. This process is repeated to model the thermal behaviour of the inverter 15 and the coolant 25 with respect to time. The machine learning process for generating the first linear regression model to predict the temperature of the inverter 15 will now be described in more detail with reference to the block diagram 600 shown in Figure 7. The one or more training data set 335(n) is supplied as input data (BLOCK 605). The or each training data set 335(n) comprises a plurality of time-varying parameters captured in relation to the reference vehicle 5. The training data set 335(n) is used to train the linear regression model for predicting the temperature of the inverter 15 comprises the following time-varying parameters: the electric power 20 (Watts) of the electric drive unit 11; a torque 22 (Nm) generated by the electric drive unit 11; the measured air temperature 24 (°C); the measured inverter temperature 16 (°C) of the inverter 15; the measured coolant temperature 18 (°C) of the coolant 25; the energy regeneration 26 of the electric drive unit 11. The measured coolant temperature 18 (°C) represents the (real) measured temperature of the coolant 25. In a variant, the modelled or predicted coolant temperature 18’ may be used in place of the measured coolant temperature 18. The or each time-varying parameter is captured at least substantially continuously, i.e. without temporal interruptions. The time-varying parameters in the present embodiment are captured at a frequency of between 100Hz to 200Hz. The time-varying parameters may be captured at a higher or lower frequency, depending on the operating capability of the associated sensor. The time-varying parameters are combined into one or more training data set 335(n). The or each training data set 335(n) may be tuned by adjusting one or more of a plurality of hyper-parameters 345(n) (BLOCK 610). The hyper-parameters 345(n) are applied to the training data set 335(n) to adjust one or more of the time-varying parameters. The hyper-parameters 345(n) comprise the time delay 345(1) and the aggregation window 345(2). The time delay 345(1) and / or the aggregation window 345(2) can be adjusted to simulate a heating time of the inverter 15 and / or the coolant 25. The aggregation window 345(2) defines a time period during which the timevarying parameters may be aggregated, for example a time period for calculating a moving average of the or each time-varying parameter. The aggregation window 345(2) may define a time period, for example two (2) seconds, three (3) seconds, five (5) seconds or ten (10) seconds, for calculating the moving average of one or more of the time-varying parameters. The aggregation window 345(2) may define a time period for calculating a moving average of the electrical power 20 of the electric drive unit 11 and / or the energy regeneration 26 of the electrical regeneration system 31. In the present embodiment, the aggregation window 345(2) defines a time period of two (2) seconds for calculating a moving average of the electrical power 20 of the electric drive unit 11. The process may comprise separating long-term components and short-term components in one or more of the time-varying parameters. In the present embodiment, the long-term temperature components are separated from the short-term temperature components in the measured inverter temperature 16 and / or the measured coolant temperature 18. The short-term temperature components represent those fluctuations in the temperature which occur over a relatively short time period; and the long-term temperature components comprise or consist of temperature changes which occurs over a longer time frame. It has been determined that the short-term fluctuations in the measured inverter temperature 16 are related to the power 20 supplied to the electric drive unit 11. The hyper-parameters 345(n) in the present embodiment comprise a short-term fluctuation parameter 345(3) for separating the long-term and short-term temperature components of the measured inverter temperature 16 and / or the measured coolant temperature 18. The short-term fluctuation parameter 345(3) comprises a low-pass filter which is applied to the measured inverter temperature 16. The low-pass filter helps to reduce or remove noise in the measured inverter temperature 16. The low-pass filter has a predetermined filter frequency. The filtered inverter temperature data represents a long-term trend (i.e., a smooth trend) in the temperature of the inverter 15. The short-term fluctuations are identified by calculating a difference between the (real) measured inverter temperature 16 and the long-term temperature trend (i.e. the filtered measured inverter temperature). The short-term fluctuations in the predicted inverter temperature 16’ are determined in dependence on the power 20. The hyper-parameters 345(n) comprise a scaling factor 345(4). The scaling factor 345(4) has been identified as correlating the long-term temperature trends of the inverter 15 and the coolant 25. The scaling factor 345(4) scales the predicted coolant temperature 18’ to determine the predicted inverter temperature 16’. The shortterm temperature component determined by the application of the short-term fluctuation parameter 345(3) and the long-term temperature component determined by the application of the scaling factor 345(4) are added together to determine the predicted inverter temperature 16’ (i.e. the sum of the short-term fluctuation and the long-term temperature trend). The one or more training data set 335(n) is processed by the machine learning tool set 325 (BLOCK 615). The machine learning tool set 325 is configured to generate the linear regression model in dependence on the one or more training data set 335(n). The machine learning tool set 325 generates a linear regression model comprising a straight line which relates the temperature of the inverter 15 to the time-varying parameters. The first linear regression model 227(1) is suitable for determining a predicted inverter temperature 16’ over one (1) timestep. The linear regression model 227(1) is integrated into the temperature model 225 (BLOCK 620). The predicted inverter temperature 16’ determined by the temperature model 225 comprises a magnitude of the inverter temperature 16’ (rather than a change in the temperature of the inverter 15). The predicted inverter temperature 16’ comprises the short-term temperature component and the long-term temperature component described herein. The short-term temperature component comprises fluctuations, i.e. changes over a relatively short time period. The long-term temperature component varies over a longer time frame. The long-term component may, for example, be determined in dependence on the predicted coolant temperature 18’ (described below). For example, the long-term component may be calculated as the product of the predicted coolant temperature 18’ and the scaling factor 345(4). The predicted inverter temperature 16’ is modelled by summing the short-term temperature component and the long-term temperature component. The short-term predicted inverter temperature 16’ may be determined independently of the predicted temperate 16’ for the previous timestep. In the final aggregation part, the temperature model 225 does not take account of the data associated with the previous time steps, for example the temperature model 225 does not look at moving average 2 seconds ago. The machine learning process for generating a second linear regression model 227(2) to predict the temperature of the coolant 25 will now be described with reference to the block diagram 700 shown in Figure 8. The one or more training data set 335(n) is supplied as input data (BLOCK 705). The or each training data set 335(n) represents historic data captured during operation of a reference vehicle 5. The or each training data set 335(n) comprises a plurality of time-varying parameters. The training data set 335(n) used to train the linear regression model for predicting the temperature of the coolant 25 comprises the following time-varying parameters: the electric power 20 (Watts) of the electric drive unit 11; the torque 22 (Nm) generated by the electric drive unit 11; the measured inverter temperature 16 (°C); the measured coolant temperature 18 (°C); the measured air temperature 24 (°C); and the energy regeneration 26 of the regeneration system 31. The measured inverter temperature 16 may be omitted fortraining the temperature model 225 to determine the predicted coolant temperature 18’. The time-varying parameters are captured at a frequency of between 100Hz to 200Hz, but may be captured at a higher or lower frequencies. The time-varying parameters are captured at least substantially continuously, i.e. without temporal interruptions. The time-varying parameters are combined into one or more training data set 335(n). The or each training data set 335(n) may be tuned by adjusting one or more of a plurality of hyper-parameters 345(n) (BLOCK 710). The hyper-parameters 345(n)comprise the time delay 345(1) and the aggregation window 345(2). The time delay 345(1) and / or the aggregation window 345(2) can be adjusted to simulate a heating time of the inverter 15. The one or more training data set 335(n) is supplied to the machine learning tool set 325 (BLOCK 715). The training data set 335(n) is then supplied to the machine learning tool set 325 for generating a second linear regression model 227(2). The resulting first linear regression model 227(2) is suitable for determining a predicted coolant temperature 18’ over one (1) timestep. The linear regression model 227(2) is integrated into the temperature model 225 (BLOCK 720). The temperature model 225 is configured to predict a change in the temperature of the coolant 25. The temperature model 225 in the present embodiment is configured to predict a change in the temperature (AT) of the coolant 25. The predicted change in the temperature of the coolant 25 may be added to the predicted temperature for the previous time step to determine a magnitude of the predicted coolant temperature 18’. The temperature model 225 may output a predicted change in the coolant temperature 18 and / or a magnitude of the predicted coolant temperature 18. As outlined above, the first and second linear regression models 227(1), 227(2) are integrated into the temperature model 225 to determine the predicted inverter and coolant temperatures 16’, 18’ of the inverter 15 and the coolant 25 respectively. The operation of the temperature model 225 to determine the predicted inverter and coolant temperatures 16’, 18’will now be described with reference to the block diagram 800 shown in Figure 9. A vehicle parameter data set 215 is supplied to the temperature model 225 (BLOCK 805). The vehicle parameter data set 215 provides a reference set of operating parameters for the vehicle 5. The temperature model 225 is configured to predict the temperature of the inverter 15 and / or the coolant 25 in dependence on the vehicle parameter data set 215. The vehicle parameter data set 215 may, for example, be defined in dependence on historic data relating to the operation of the vehicle 6. The historic data may, for example, be captured during one or more vehicle journey. The vehicle parameter data set 215 in the present embodiment comprises one or more of the following: the power 20, the torque 22 and the energy regeneration 26. The vehicle parameter data set 215 may also comprise data relating to the prevailing operating conditions for the vehicle 5, such as a measured air temperature 24. In a variant, the air temperature may be derived from meteorological data, such as a weather forecast for the current location of the vehicle 5. The vehicle parameter data set 215 is supplied as input data to the pre-trained temperature model 225 (BLOCK 810). If available, a measured inverter temperature 16 and / or a measured coolant temperature 18 may be supplied to the temperature model 225 to define starting (or reference) temperatures for determining the predicted inverter and coolant inverter temperature 16’. 18’. If these are not available, the air temperature 24 may be used as the starting (or reference) temperature. The vehicle parameter data set 215 may optionally provide an indication of an expected or predicted behaviour of the vehicle 5. The vehicle parameter data set 215 may comprise one or more predicted vehicle operating parameter, such as the power 20, the torque 22 or the energy regeneration 26. The one or more vehicle operating parameter may be, for example, be predicted to maintain a defined vehicle speed. The vehicle speed may be defined in dependence on the current operating speed of the vehicle 5, for example a moving (rolling) average of the vehicle speed during the present journey; or the vehicle speed may be defined by a vehicle cruise control system. Alternatively, or in addition, the predicted power, torque and energy regeneration may be modelled in dependence on usage during the present journey. Alternatively, or in addition, the vehicle parameter data set 215 may comprise geolocation data defining a location of the vehicle 5; and / or route planning data defining a route for the vehicle 5. One or more operating parameter of the vehicle 5 may be modelled in dependence on the geolocation data and / or the route planning data. One or more of the power 20, the torque 22 and the energy regeneration 26 may be modelled for the defined route. The temperature model 225 is configured to predict the temperature of the inverter 15 and the coolant 25. The vehicle parameter data set 215 may, for example, comprise the power 20, the torque 22 and the energy regeneration 26 The predicted inverter and coolant temperatures 16’, 18’ are determined for a time step (BLOCK 815). The predicted inverter and coolant temperatures 16’, 18’ are re-introduced into the temperature model 225 and the first and second temperatures predicted for the next time step. The recursive determination of the predicted inverter and coolant temperatures 16’, 18’ continues fora plurality of time steps. The predicted inverter and coolant temperatures 16’, 18’ are output as a predicted temperature data set 255 (BLOCK 820). The predicted inverter and coolant temperatures 16’, 18’ are represented as time-series data in the predicted temperature data set 255. The predicted temperature data set 255 is transmitted from the temperature prediction system 200 to the vehicle control system 100 (BLOCK 825). In the present embodiment, the predicted temperature data set 255 is transmitted over a wireless communication network. The measured inverter and coolant temperatures 16,18 are available periodically. In particular, the measured inverter and coolant temperatures 16, 18 are available when there is communication between the vehicle control system 100 and the temperature prediction system 200. The measured inverter and coolant temperatures 16,18 represent discrete values of the measured temperature of the inverter 15 and the coolant 25. The measured inverter and coolant temperatures 16,18 are supplied to the temperature model 225 as an input (BLOCK 830). The temperature model 225 is configured to update the predicted inverter and coolant temperatures 16’, 18’ to align with the measured inverter and coolant temperatures 16, 18. The temperature model 225 continues to determine the predicted inverter and coolant temperatures 16’, 18’ (BLOCK 815). A first plot 50 representing the temperature (°C) of the coolant 25 with respect to time is shown in Figure 10. The first plot 50 shows the measured coolant temperature 18 and a corresponding predicted coolant temperature 18’ during operation of the vehicle 5. The predicted coolant temperature 18’ in this example is determined by the temperature model 225 without access to the measured coolant temperature 18. The measured coolant temperature 18 is shown in the first plot 50 for comparison purposes but is not supplied to the temperature model 225. As such, no correction is made to the predicted coolant temperature 18’ in dependence on the measured coolant temperature 18. A mean absolute error of 2.94°C is calculated for this scenario. A second plot 60 representing the temperature (°C) of the coolant 25 with respect to time is shown in Figure 11. The second plot 60 shows the measured coolant temperature 18 and a corresponding predicted coolant temperature 18’ during operation of the vehicle 5. The predicted coolant temperature 18’ in this example is determined by the temperature model 225 with intermittent access to the measured coolant temperature 18. In particular, the measured coolant temperature 18 is supplied to the temperature model 225 at time t1 (approximately 350 seconds in the present example), at time t2 (approximately 900 seconds in the present example) and at time t3 (approximately 1800 seconds in the present example). The temperature model 225 is configured to correct the predicted coolant temperature 18’ each time the measured coolant temperature 18 is available. A mean absolute error of 1.99°C is calculated for this scenario. It will be recognised that the intermittent provision of the measured coolant temperature 18 reduces the absolute error of the predicted coolant temperature 18’. A third plot 70 representing the predicted inverter temperature 16’ (°C) of the inverter 15 with respect to time is shown in Figure 12. As described herein, the first predicted temperatures 16’ is compared to the first and second inverter temperature thresholds 235(1), 235(2). The method 400 of monitoring the temperature of the inverter 15 and the coolant 25 in accordance with an embodiment of the present invention will now be described with reference to a block diagram 900 shown in Figure 13. The block diagram 900 represents the operation of the vehicle control system 100 to compare the first predicted temperatures 16’ to the first and second inverter temperature thresholds 235(1), 235(2). The vehicle control system 200 receives the predicted temperature data set 255 from the temperature prediction system 200 (BLOCK 905). The predicted temperature data set 255 comprises time-series data representing the predicted inverter temperature 16’. Athermal event 229(n) is predicted in dependence on a determination that the predicted inverter temperature 16’will cross one of the firstand second inverter temperature thresholds 235(1), 235(2). A determination that the predicted inverter temperature 16’ will cross the first inverter temperature threshold 235(1) is indicative of an overheat thermal event 229(1) in the inverter 15. The vehicle control system 200 compares the predicted inverter temperature 16’ to the first inverter temperature threshold 235(1) (BLOCK 910). The vehicle control system 200 classifies the thermal event 229(n) as an inverter overheat thermal event 229(1) in dependence on a determination that the predicted inverter temperature 16’ will cross the first inverter temperature threshold 235(1) (BLOCK 915). The vehicle control system 200 determines an event time 233 of the predicted thermal event 229(1) (BLOCK 920) The vehicle control system 200 generates a first vehicle control signal 125(1) in dependence on the determined classification and thermal event time 233 of the predicted thermal event 229(1) (BLOCK 925). The first vehicle control signal 125(1) may lower a rated electrical capability of the electric drive unit 11 (i.e., de-rate the electric drive unit 11). The first vehicle control signal 125(1) may be output to the temperature prediction system 200 and the predicted inverter temperature 16’ updated in dependence on the modified operating parameters of the vehicle 5. The vehicle control system 200 continues to monitor the predicted inverter temperature 16’. A determination that the predicted inverter temperature 16’will cross the second inverter temperature threshold 235(2) is indicative of an underheat thermal event 229(1) in the inverter 15. The vehicle control system 200 compares the predicted inverter temperature 16’ to the first inverter temperature threshold 235(1) (BLOCK 930). The vehicle control system 200 classifies the thermal event 229(n) as an inverter underheat thermal event 229(1) in dependence on a determination that the predicted inverter temperature 16’ will cross the second inverter temperature threshold 235(1) (BLOCK 935). The vehicle control system 200 determines an event time 233 of the predicted thermal event 229(1) (BLOCK 940). The vehicle control system 200 generates a second vehicle control signal 125(2) in dependence on the determined classification and thermal event time 233 of the predicted thermal event 229(1) (BLOCK 945). The second vehicle control signal 125(2) may raise a rated electrical capability of the electric drive unit 11. The second vehicle control signal 125(2) may be output to the temperature prediction system 200 and the predicted inverter temperature 16’ updated in dependence on the modified operating parameters of the vehicle 5. The vehicle control system 200 continues to monitor the predicted inverter temperature 16’. The above method(s) of monitoring the predicted inverter temperature 16’ can be used to monitorthe predicted coolant temperature 18’. The predicted coolant temperature 18’ may be compared to the first and second coolant temperature thresholds 235(1), 235(2). It will be understood that the first and second coolant temperature thresholds 235(1), 235(2) may be different from the first and second inverter temperature 235(1), 235(2). At least one inverter temperature threshold 235(n) is defined for the inverter 15. The at least one inverter temperature threshold 235(n) comprises a first (upper) inverter temperature threshold 235(1) and a second (lower) inverter temperature threshold 235(2). A determination that the predicted inverter temperature 16’ crosses the at least one inverter temperature threshold 235(n) indicates a predicted thermal event 229(n). The thermal event 229(n) may, for example, represent an overheat scenario when the predicted inverter temperature 16’ exceeds the first inverter temperature threshold 235(1); and / or the thermal event 229(n) may represent an underheat scenario when the predicted inverter temperature 16’ is less than the second inverter temperature threshold 235(2). The vehicle control system 100 is configured to determine a predicted thermal event time when the predicted inverter temperature 16’ will cross one of the first and second temperature thresholds 235(1), 235(2). The vehicle control system 100 is configured to output a notification signal 135 indicating a type or class of thermal event 229(n) associated with the inverter 15, for example to indicate an overheat scenario or an underheat scenario for the inverter 15. The notification signal 135 also indicates the predicted event time for occurrence of that predicted thermal event 229(n). At least one coolant temperature threshold 235(n) is defined for the coolant 25. The at least one coolant temperature threshold 235(n) comprises a first (upper) coolant temperature threshold 235(1) and a second (lower) coolant temperature threshold 235(2). The thermal event 229(n) may, for example, represent an overheat scenario when the predicted inverter temperature 16’ of the coolant 15 exceeds the first coolant temperature threshold 235(1); and / orthe thermal event 229(n) may represent an underheat scenario when the predicted inverter temperature 16’ of the coolant 15 is less than the second coolant temperature threshold 235(2). The vehicle control system 100 is configured to determine a predicted thermal event time when the predicted inverter temperature 16’ will cross one of the first and second temperature thresholds 235(1), 235(2). The vehicle control system 100 is configured to output a notification signal 135 indicating a type or class of thermal event 229(n) associated with the coolant 25, for example to indicate an overheat scenario or an underheat scenario for the coolant 25. The notification signal 135 also indicates the predicted event time 233 for occurrence of the or each predicted thermal event 229(n). The temperature model 225 described herein comprises first and second linear regression models 227(1), 227(2) for predicting the temperature of the inverter 15 and the coolant 25 respectively. The temperature model 225 could be modified to incorporate only one of the first and second linear regression models 227(1), 227(2). Alternatively, or in addition, the temperature model 225 may comprise linear regressions models for predicting the temperature of one or more other component(s) in the electric drive unit 11. For example, the temperature model 225 may comprise a linear regression model for predicting the temperature of the rotor 21 or the stator 23 of the electric machine 13. Alternatively, or in addition, the temperature model 225 may comprise a linear regression model for predicting the temperature of the traction battery 19 provided on the vehicle 5. The temperature model 225 has been described herein with particular reference to linear regression models. It will be understood that other types of regression model may be implemented to predict temperature(s). The training method(s) described herein for generating the firstand second linear regression models 227(1), 227(2) may be used for generating linear regression models for predicting the temperature of other components in the electric drive unit 11. It will be appreciated that various changes and modifications can be made to the present invention without departing from the scope of the present application.

Claims

1. A system for predicting thermal behaviour of a component of an electric drive system, the systemcomprising one or more processors collectively configured to:receive at least one input signal indicating a time-varying parameter of the component or the electric drive system, the at least one input signal comprising a temperature signal indicating a measured temperature of the component;implement a temperature model to determine a predicted temperature of the component;predict occurrence of a thermal event and an event time of the or each predicted thermal event, wherein predicting the occurrence of the thermal event comprises determining that the predicted temperature crosses one or more predetermined temperature threshold; andoutput a notification signal indicating the predicted thermal event and the predicted event time;wherein the measured temperature indicated by the temperature signal is supplied to the temperature model, the temperature model being configured to update the predicted temperature in dependence on the measured temperature; the one or more processors being configured to predict the thermal event and the event time in dependence on the updated temperature.

2. A system as claimed in claim 1, wherein the temperature model is configured to determine thepredicted temperature for each of a plurality of time steps, the predicted temperature at each time step being used as an input for determining the predicted temperature of the next time step.

3. A system as claimed in claim 1 or claim 2, wherein the at least one input signal is receivedintermittently, the one or more processors being configured to update the predicted thermal event and / or the predicted event time in dependence on an updated value of the time-varying parameter.

4. A system as claimed in claim 3, wherein the temperature model is configured to determine thepredicted temperature between receipt of successive temperature signals indicating the measured temperature.

5. A system as claimed in any one of the preceding claims, wherein the temperature model is trainedon a data set comprising substantially continuous temperature measurements of the component.

6. A system as claimed in claim 5, wherein the temperature model comprises a regression modeltrained on the substantially continuous temperature measurements of the component.

7. A system as claimed in any one of the preceding, wherein the temperature model generates timeseries temperature data representing the predicted temperature of the component.

8. A system as claimed in any one of the preceding claims, wherein the at least one input signalcomprises a power signal indicating a measured power of the at least one electric drive system.

9. A system as claimed in any one of the preceding claims, wherein the at least one input signalcomprises an energy regeneration signal indicating an electrical regeneration of the at least one electric drive system.

10. A system as claimed in any one of the preceding claims, wherein the at least one component comprises an inverter of the electric drive system; wherein the at least one input signal comprises a measured temperature of the inverter; and the temperature model is configured to predict the temperature of the inverter.

11. A system as claimed in claim 10, wherein the temperature model predicts the temperature of the inverter in dependence on a predicted temperature of a coolant of the electric drive system.

12. A system as claimed in any one of the preceding claims, wherein the one or more predetermined temperature threshold comprise a first temperature threshold and a second temperature threshold; wherein the one or more processors is configured to:predict a first said thermal event and an associated first event time in dependence on a determination that the predicted temperature crosses the first temperature threshold, the first thermal event comprising an overheat thermal event; andpredict a second said thermal event and an associated second event time in dependence on a determination that the predicted temperature crosses the second temperature threshold, the second thermal event comprising an underheat thermal event.

13. A system as claimed in any one of the preceding claims, wherein the at least one electric drive system is configured to operate in a plurality of operating modes, the one or more processors being configured to select one of the plurality of operating modes in dependence on the predicted event time.

14. A system as claimed in claim 13, wherein the one or more processors is configured to output a control signal to configure the at least one electric drive system to operate in the selected one of the plurality of operating modes.

15. A vehicle comprising a system as claimed in any one of the preceding claims.s

Images