Motor temperature prediction method and device, vehicle and computer program product

Abstract

The embodiment of the invention provides a motor temperature prediction method and device, a vehicle and a computer program product, and the method comprises the steps: obtaining a temperature influence parameter of a historical time slot according to a motor state, and inputting the temperature influence parameter into a trained temperature change rate prediction model, obtaining the temperature change rate of each node in the motor under the historical time slot; and predicting the motor temperature at the next moment according to the temperature change rate. The training data of the temperature change rate prediction model comes from a three-dimensional finite element model, the temperature change rate in a certain time period is calculated through the temperature influence parameters strongly related to the temperature and the temperature change rate prediction model, and the motor temperature at the next moment is predicted based on the temperature change rate. The computing resource demand can be reduced while the high prediction precision is maintained.

Description

Technical Field

[0001] The present application relates to the field of motor control technology, and in particular, to a method, device, vehicle and computer program product for predicting motor temperature. Background Art

[0002] The performance of the magnetic steel in the permanent magnet synchronous motor is related to the temperature. When the motor temperature reaches a certain value, the magnetic steel will experience irreversible demagnetization. This demagnetization will not only cause a significant decrease in the magnetic properties of the motor, but will also directly affect the output power and efficiency of the motor, and even cause damage to the motor. In addition, motor overheating may also cause other safety issues, such as aging of insulation materials, short circuits, fires, etc. By predicting the motor temperature, potential overheating problems can be discovered and handled in a timely manner, thereby ensuring the safe operation of the motor. However, in the relevant technology, there is currently no method for online real-time prediction of vehicle motor temperature with high calculation accuracy and low computing resource requirements. Summary of the invention

[0003] The purpose of the embodiments of the present application is to provide a method, device, vehicle and computer program product for predicting motor temperature, so as to achieve the technical effect of predicting motor temperature with high accuracy and low computing resource requirements.

[0004] A first aspect of an embodiment of the present application provides a method for predicting motor temperature, the method comprising:

[0005] According to the state of the motor, the temperature influence parameter of the historical time slot is obtained, and the temperature influence parameter is input into the trained temperature change rate prediction model to obtain the temperature change rate of each node in the motor in the historical time slot; the temperature influence parameter includes parameters that affect the temperature of the motor; the nodes include: stator, rotor and winding; the trained temperature change rate prediction model is trained by the input and output data of the trained three-dimensional finite element model;

[0006] The motor temperature at the next moment is predicted according to the temperature change rate.

[0007] In the above implementation process, the training data of the temperature change rate prediction model comes from an accurate three-dimensional finite element model. The temperature change rate within a certain time period is calculated through the temperature influence parameters that are strongly correlated with temperature and the temperature change rate prediction model. Based on the temperature change rate, the motor temperature at the next moment is predicted, which can achieve the goal of reducing computing resource requirements while maintaining a high prediction accuracy.

[0008] Further, predicting the motor temperature at the next moment according to the temperature change rate includes:

[0009] Calculating the minimum time difference between the historical time slot and the next moment;

[0010] The motor temperature is determined based on the temperature change rate and the minimum time difference.

[0011] In the above implementation, on the one hand, by considering the actual time difference between the historical time slot and the next moment, the motor temperature at the next moment can be predicted more accurately. On the other hand, frequent updates of the motor temperature prediction may not be necessary or feasible, especially in an environment with limited computing resources. By calculating the minimum time difference, it is possible to determine when the prediction result needs to be updated, thereby avoiding unnecessary waste of computing resources and achieving optimal resource allocation.

[0012] Furthermore, the motor state is a running state, and the step of obtaining the temperature influence parameter of the historical time slot includes:

[0013] The motor voltage, the motor output speed, the motor output torque, the motor coolant temperature, the motor coolant flow rate and the temperature of each of the nodes are obtained as the temperature influencing parameters.

[0014] In the above implementation process, these parameters are directly related to the operating state and heat generation of the motor during vehicle operation. Among them, the motor voltage, output speed and output torque determine the energy conversion efficiency and loss inside the motor, and these losses are ultimately manifested in the form of heat. Entering these parameters as temperature-influencing parameters into the prediction model can more accurately reflect the internal mechanism of motor temperature changes.

[0015] Furthermore, the motor state is a startup state, and the step of obtaining the temperature influence parameter of the historical time slot includes:

[0016] The motor voltage, the motor output speed, the motor output torque, the motor coolant temperature and the motor coolant flow rate are obtained as the temperature influencing parameters.

[0017] In the above implementation process, by obtaining the key parameters of motor voltage, motor output speed, motor output torque, motor coolant temperature and motor coolant flow as temperature influencing parameters, the temperature change of the motor during the startup phase can be predicted more accurately.

[0018] Furthermore, the temperature change rate prediction model is trained by the following steps:

[0019] Acquire a plurality of preset training operating condition parameters; the training operating condition parameters include motor voltage, motor output speed, motor output torque, motor coolant temperature and motor coolant flow;

[0020] Inputting the training operating condition parameters into the motor heating model to obtain the heat generation of each node of the motor corresponding to each training operating condition parameter;

[0021] Input the training working condition parameters and the corresponding heat generation into a three-dimensional finite element model to obtain the temperature of each node and the temperature change rate of each node;

[0022] The temperature change rate prediction model is supervisedly trained using the training operating condition parameters and their corresponding heating value, temperature, and temperature change rate to obtain the trained temperature change rate prediction model.

[0023] In the above implementation process, by inputting the training operating condition parameters into the motor heating model, the heat generation of each node under different operating conditions can be accurately calculated. By inputting the training operating condition parameters and their corresponding heat generation into the three-dimensional finite element model, factors such as the internal structure, material properties, and heat conduction of the motor can be comprehensively considered to obtain more accurate node temperatures and temperature change rates. This detailed and comprehensive simulation process helps to train a high-precision temperature change rate prediction model. The training process itself involves complex calculations and simulations, but once the trained temperature change rate prediction model is obtained, the temperature change rate can be quickly calculated based on the input parameters without the need for complex physical simulations, thereby achieving high-precision and high-efficiency prediction of motor temperature.

[0024] Furthermore, the three-dimensional finite element model is constructed by the following steps:

[0025] Acquire a motor fluid domain, and set coolant and fluid domain inlets and outlets in the motor fluid domain to obtain a motor fluid model;

[0026] Acquire a motor geometric model, and set a contact surface and a heat value in the motor geometric model to obtain a three-dimensional temperature field model;

[0027] The motor fluid model is coupled with the three-dimensional temperature field model to obtain the three-dimensional finite element model.

[0028] In the above implementation process, by obtaining the motor fluid domain and setting the coolant and its flow path (including inlet and outlet) in detail, the flow state of the coolant inside the motor can be accurately simulated, including flow velocity, flow direction, temperature distribution, etc. The acquisition of the motor geometric model and the setting of the contact surface and heat generation in the model can accurately reflect the heat conduction relationship between the various components inside the motor and the heat distribution generated by current, mechanical friction, etc. This fine setting makes the simulation of the temperature field closer to the actual situation. Coupling the motor fluid model with the three-dimensional temperature field model realizes the interaction simulation of fluid flow and heat transfer, which can more comprehensively reflect the complex heat exchange process inside the motor, thereby improving the accuracy of the prediction results.

[0029] Furthermore, the method is applied to a motor controller mounted on a vehicle, and the method further includes:

[0030] The locally pre-stored trained temperature change rate prediction model is read.

[0031] In the above implementation process, since the temperature change rate prediction model has been trained and pre-stored in the motor controller, the controller can read the model in real time and quickly calculate the temperature change rate of the motor during vehicle operation, which enhances the real-time performance of motor temperature prediction. In addition, compared with online training or real-time calculation, reading the locally pre-stored model can greatly reduce the demand for vehicle computing resources.

[0032] A second aspect of an embodiment of the present application provides a device, the device comprising:

[0033] processor;

[0034] a memory for storing processor-executable instructions;

[0035] Wherein, when the processor calls the executable instruction, any method described in the first aspect is implemented.

[0036] A third aspect of an embodiment of the present application provides a vehicle, and the vehicle applies any method described in the first aspect.

[0037] A fourth aspect of the embodiments of the present application provides a computer program product, wherein the computer program product includes a computer program, and when the computer program is executed by a processor, any method described in the first aspect is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 A schematic flow chart of a method for predicting motor temperature provided in an embodiment of the present application;

[0040] Figure 2 A schematic flow chart of another method for predicting motor temperature provided in an embodiment of the present application;

[0041] Figure 3 A schematic flow chart of another method for predicting motor temperature provided in an embodiment of the present application;

[0042] Figure 4 A schematic diagram of online prediction of motor temperature provided in an embodiment of the present application;

[0043] Figure 5 A structural block diagram of a device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0045] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0046] At present, the research on online prediction methods of motor temperature mainly includes: flux method, data analysis method and thermal network model method.

[0047] The flux method calculates the rotor flux through a simple voltage equation. The error of the calculated rotor temperature is large, and this method is not applicable when the motor is stationary or at low speed.

[0048] The data analysis method uses machine learning to collect motor operating parameters and rotor temperature data to form a data set, and predicts the motor temperature by training the model. However, since the motor temperature is a real-time cumulative quantity, directly using machine learning to predict the temperature is limited by the temperature conditions of the motor at the previous moment, and direct application is difficult.

[0049] Thermal network model method: This method treats points with similar temperatures in the motor as nodes, connects the nodes through thermal resistance, and replaces the heat storage effect of materials in different parts with heat capacity, and finally obtains the thermal network model. However, this method has extremely high requirements on the accuracy of the parameters of the thermal network model, and requires certain experimental calibration work.

[0050] In summary, in the related art, there is currently no method for online real-time prediction of motor temperature with high calculation accuracy and low calculation resource requirements.

[0051] In response to any of the above-mentioned problems, the present application provides a method for predicting motor temperature. Figure 1 , Figure 1 A flow chart of a method for predicting motor temperature provided in an embodiment of the present application.

[0052] In this embodiment, the method includes:

[0053] Step S10: obtaining temperature influence parameters of historical time slots according to the motor state, and inputting the temperature influence parameters into a trained temperature change rate prediction model to obtain the temperature change rate of each node in the motor in the historical time slot; the temperature influence parameters include parameters that affect the motor temperature; the nodes include: a stator, a rotor and a winding; the trained temperature change rate prediction model is trained by input and output data of a trained three-dimensional finite element model;

[0054] It should be noted that the execution subject of this embodiment may be a motor controller of the vehicle, or a server, an embedded chip, or other controllers in the vehicle, such as a thermal management system controller or a domain controller.

[0055] The motor status may include running and just started.

[0056] The historical time slot refers to a certain period of time in the past, for example, the last second, the last minute. The historical time slot can be set according to the sampling interval of the motor controller or according to the predicted demand.

[0057] Temperature-affecting parameters refer to parameters that can directly or indirectly affect the temperature change of the motor. Temperature-affecting parameters usually include a series of variables related to the motor's operating status, environmental conditions, and cooling system performance. These parameters can be obtained in real time through the vehicle's data acquisition system. They can be divided into several categories:

[0058] Motor operating parameters:

[0059] Current: The amount of current in a motor directly affects how much heat the motor generates. The higher the current, the greater the heat generated.

[0060] Voltage: Although voltage itself does not directly generate heat, it works together with current to determine the power and efficiency of the motor, which indirectly affects temperature.

[0061] Speed: Changes in speed will affect the windage loss and mechanical loss of the motor, thereby affecting the temperature.

[0062] Load: The size of the load determines the power that the motor needs to output, which in turn affects the heat generation.

[0063] Environmental condition parameters:

[0064] Ambient temperature: Ambient temperature is the most direct external factor affecting the motor temperature. In a high temperature environment, the heat dissipation efficiency of the motor will decrease, causing the temperature to rise.

[0065] Humidity: Although humidity does not directly affect the temperature of the motor, high humidity environments may affect the efficiency of the cooling system and indirectly affect the temperature.

[0066] Air flow: When the vehicle is running, external air flow can help the motor dissipate heat and reduce the temperature.

[0067] Cooling system parameters:

[0068] Coolant temperature: For motors using liquid cooling, the temperature of the coolant directly affects the heat dissipation effect of the motor.

[0069] Coolant flow rate: The coolant flow rate determines the cooling system's ability to remove heat, thereby affecting the temperature of the motor.

[0070] Cooling fan speed: The speed of the cooling fan determines the speed at which air flows through the radiator, which in turn affects the cooling effect.

[0071] The temperature influencing parameter at least includes a temperature value of each node, and the temperature value may include multiple values.

[0072] It should be understood that the trained temperature change rate prediction model is trained based on the input and output data of the three-dimensional finite element model. The three-dimensional finite element model can accurately simulate the temperature distribution and changes of the motor under different working conditions by comprehensively considering the internal geometric structure, material properties, fluid flow and heat transfer of the motor. During the training stage of the temperature change rate prediction model, the laws and characteristics of the motor temperature changes will be learned.

[0073] Exemplarily, the temperature change rate prediction model may be formed by combining or transforming any one or more of the following:

[0074] 1. Linear regression model

[0075] Linear regression is one of the simplest prediction models, which can be used to predict the linear relationship between one or more independent variables (such as voltage, current, ambient temperature, etc.) and a dependent variable (such as temperature change rate).

[0076] 2. Nonlinear regression model

[0077] In order to more accurately describe the complex relationship between the temperature change rate and the temperature-affecting parameters, a nonlinear regression model can be used. This type of model can capture the nonlinear relationship between variables, such as polynomial regression, exponential regression, logarithmic regression, etc. Nonlinear regression models usually have higher flexibility when fitting complex data sets.

[0078] 3. Neural network model

[0079] Neural networks, especially deep neural networks (DNNs) and convolutional neural networks (CNNs), excel in processing complex nonlinear relationships and pattern recognition. In temperature change rate prediction, neural networks can build prediction models by learning complex features from historical data. These models are able to automatically extract high-level features from raw data and use them to predict future temperature change rates. A major advantage of neural network models is that they can handle large amounts of input data and adapt to different data types and formats.

[0080] 4. Time Series Analysis Model

[0081] Temperature change rate prediction is essentially a time series prediction problem, because temperature changes over time. Therefore, time series analysis models can be used to predict future temperature change rates. Such models include autoregressive models (AR), moving average models (MA), autoregressive moving average models (ARMA), autoregressive integrated moving average models (ARIMA), and seasonal decomposition models (such as STL decomposition). Time series analysis models can capture the time dependence and seasonal changes in time series data, thereby more accurately predicting future temperature change rates.

[0082] 5. Ensemble learning methods

[0083] Ensemble learning methods improve the overall prediction performance by combining the prediction results of multiple base learners (such as decision trees, neural networks, etc.). In the prediction of temperature change rate, ensemble learning methods can be used to combine the advantages of different types of prediction models to obtain more accurate prediction results. Common ensemble learning methods include random forest, gradient boosting machine (GBM), XGBoost, and LightGBM.

[0084] It is understandable that the temperature of the node itself is affected by the temperature-affecting parameters. The nodes can be divided according to actual needs. This embodiment does not limit this. For example, in addition to the three key nodes of stator, rotor and winding, other important components of the motor can also be included, such as bearings and thermal resistance. In addition, the nodes can be further subdivided. Taking the stator as an example, more nodes can be divided based on the geometric shape of the stator, such as a cylindrical stator and a stator with a special shape.

[0085] The temperature change rate refers to the amount of temperature change per unit time, which describes how fast the temperature changes over time. The temperature change rate of each node reflects the temperature change trend at different locations inside the motor.

[0086] Step S20: predicting the motor temperature at the next moment according to the temperature change rate.

[0087] In the specific implementation, the temperature of each node at the next moment is calculated through mathematical operations according to the temperature and temperature change rate of each node. Assuming the time step is Δt (for example, 1 second), the temperature T(t+Δt) at the next moment t+Δt can be calculated by the following formula:

[0088] T(t+Δt)=T(t)+Y*Δt Formula 1;

[0089] In Formula 1, T(t) is the node temperature in the historical time slot, and Y is the temperature change rate of the node.

[0090] The selection of the time step Δt is based on the trade-off between prediction accuracy and computing resources. A smaller Δt can improve the prediction accuracy, but will increase the amount of computation. This embodiment does not limit the time step, and can be set according to actual needs.

[0091] Because the temperature change rate of each node may be different, the above process needs to be repeated for each node in the motor. Finally, the predicted temperature of each node at the next moment is obtained, and the temperature of each node at the next moment can be directly used as the temperature of the motor at the next moment, or the temperature of the motor at the next moment can be derived from the temperature of each node at the next moment after conversion based on the node settings.

[0092] In this embodiment, the prediction model is trained by using the input and output data of the trained three-dimensional finite element model, so that the prediction model can more accurately capture the complex heat conduction and temperature distribution characteristics inside the motor. This includes a detailed simulation of the temperature interactions between different nodes such as the stator, rotor, and winding, thereby improving the prediction accuracy. In addition, compared to directly running the three-dimensional finite element model (which is usually very time-consuming and computationally intensive), using a trained prediction model can greatly speed up the prediction speed and improve the real-time prediction, and the computing resource requirements are low.

[0093] Based on any of the above embodiments, predicting the motor temperature at the next moment according to the temperature change rate includes:

[0094] Calculating the minimum time difference between the historical time slot and the next moment;

[0095] The motor temperature is determined based on the temperature change rate and the minimum time difference.

[0096] Optionally, based on the temperature change rate of each node of the motor, the temperature of each node of the motor is integrated with a fixed step length to determine the temperature prediction value of the entire motor.

[0097] It is understandable that when dealing with motor temperature monitoring, the impact of time changes on temperature needs to be considered. The historical time slot refers to a period of time in the past, which can be the period closest to the current moment, while the next moment refers to a time point in the future that is about to arrive. When predicting future temperatures, more attention is paid to the data closest to the predicted time point, because these data are usually more highly correlated with the predicted results, and the minimum time difference represents the historical data closest to the predicted time point that can be found. Prediction based on the minimum time difference can reduce errors.

[0098] Fixed-step integration is a method of numerical integration that divides the integration interval into a series of subintervals of equal length (i.e., step sizes) and uses a specific approximate method (such as the trapezoidal method, the rectangular method, etc.) in each subinterval to calculate the approximate value of the integral.

[0099] In the specific implementation, time can be divided into a series of steps of equal length, and then in each step, the temperature change is calculated according to the temperature change rate of the node and the step time (i.e., the minimum time difference), and these changes are accumulated to obtain the predicted temperature of each node at the next moment, and the predicted motor temperature is determined based on the predicted temperature of each node. Of course, by using fixed-step integration, the temperature change trajectory during the entire prediction time period can also be obtained, that is, the predicted temperature value of each node at different time points in the future.

[0100] In this embodiment, based on the characteristic that temperature increases over time, the historical data closest to the prediction time point is determined, and then the temperature is accumulated by integration to predict the motor temperature at a future time, thereby improving the accuracy of temperature prediction.

[0101] Based on any of the above embodiments, the motor state is a running state, and obtaining the temperature influence parameter of the historical time slot includes:

[0102] The motor voltage, the motor output speed, the motor output torque, the motor coolant temperature, the motor coolant flow rate and the temperature of each of the nodes are obtained as the temperature influencing parameters.

[0103] It should be noted that motor voltage refers to the voltage applied to both ends of the motor windings, which is an important energy source for driving the motor. The voltage level will directly affect the motor's current, power output, and heat generation. During the operation of the motor, voltage fluctuations may cause changes in the heat generated by the motor's internal resistance, thereby affecting the motor's temperature.

[0104] The motor output speed refers to the number of revolutions that the motor's rotating parts (such as the rotor) make per unit time. The speed is not only related to the motor's design and load, but also directly affects the heat generated by the motor. A high-speed motor will generate more heat due to friction and electromagnetic induction.

[0105] Motor output torque refers to the rotational torque that the motor can generate when it is running, which determines the motor's ability to drive the load. The magnitude of the torque is related to the motor's current, magnetic field strength, and load characteristics. When the motor outputs a large torque, the electromagnetic induction and mechanical friction inside the motor will increase, resulting in increased heat generation.

[0106] Motor coolant temperature refers to the temperature of the coolant used to cool the internal parts of the motor. The coolant reduces the temperature of the motor by absorbing the heat generated inside the motor and dissipating it into the environment. The coolant temperature directly affects the heat dissipation effect of the motor, thereby affecting the temperature state of the motor.

[0107] Motor coolant flow rate refers to the volume of coolant that passes through the motor cooling system per unit time. The coolant flow rate determines the cooling system's ability to remove heat.

[0108] Temperature of each node: Nodes are usually set at different positions in the motor, and the temperature of each node reflects the thermal state of different areas inside the motor. During the operation of the motor, the motor temperature is predicted in real time through the trained temperature change rate prediction model, and the temperature of each node can be determined by the predicted temperature. Of course, the temperature of each node can also be collected by sensors.

[0109] In this embodiment, these parameters are directly related to the heat generation and heat dissipation process during the operation of the motor. By comprehensively considering the changes in multiple parameters, the trend and cause of the change in the motor temperature can be more comprehensively understood. Obtaining these parameters helps to more accurately predict the temperature of the motor.

[0110] Based on any of the above embodiments, the motor state is a startup state, and the step of obtaining the temperature influence parameter of the historical time slot includes:

[0111] The motor voltage, the motor output speed, the motor output torque, the motor coolant temperature and the motor coolant flow rate are obtained as the temperature influencing parameters.

[0112] It should be noted that the startup state refers to the moment when the motor starts, which is a very short process. At this time, the temperature of each node cannot be accurately obtained. Therefore, the coolant temperature is used as the temperature of each node.

[0113] Specifically, when the motor is in the starting state, that is, the motor is in the process of changing from a stationary state to a running state. The starting state is an important stage in the life cycle of the motor. At this time, the mechanical parts inside the motor begin to rotate, the electromagnetic field gradually forms, and the current and voltage will also change accordingly. During the startup process, the motor may experience a large current shock and mechanical stress, which will cause the motor to generate additional heat. "Historical time slot" refers to a specific time period in the past, which can be a few seconds, minutes or longer. Retrieve the motor voltage, motor output speed, motor output torque, motor coolant temperature and motor coolant flow data in the corresponding time period from the data recording system (such as PLC, SCADA system, data acquisition and monitoring system, etc.) or historical archive. These data can help us understand the temperature performance of the motor under different operating conditions, and then analyze the laws and causes of temperature changes.

[0114] In this embodiment, according to the operating state of the motor, corresponding temperature influencing parameters that are strongly correlated with temperature are obtained. During the motor startup phase, the motor voltage, motor output speed, motor output torque, motor coolant temperature and motor coolant flow rate are used as temperature influencing parameters.

[0115] Based on any of the above embodiments, the temperature change rate prediction model is as follows: Figure 2 The training shown in steps S01 to S04 is:

[0116] Step S01: obtaining a plurality of preset training operating condition parameters; the training operating condition parameters include motor voltage, motor output speed, motor output torque, motor coolant temperature and motor coolant flow rate;

[0117] It should be noted that the training of the temperature change rate prediction model can be carried out before actual use. Optionally, the model training process requires strong computing power support. This process can be carried out on a computer. After the temperature change rate prediction model is trained, it is deployed to the execution entity.

[0118] The preset training condition parameters are designed to collect diverse motor operating condition data to cover various operating conditions that the motor may encounter, including the two dimensions of motor cooling conditions and motor operating conditions. Specifically, in addition to motor voltage, motor output speed, motor output torque, motor coolant temperature and motor coolant flow, it can also include motor current, ambient temperature, ambient humidity, motor magnetic field strength, load change rate, etc. These training condition parameters directly affect the operating state and heating of the motor, and are the basis for subsequent analysis and modeling.

[0119] Optionally, five parameters that are strongly correlated with the motor temperature (motor output speed, motor output torque, bus voltage, coolant temperature, coolant flow rate) are identified, and the simulation working conditions are designed using the orthogonal experimental method using the above parameters as five dimensions. During model training, the training working condition parameters in the simulation working conditions are obtained.

[0120] Step S02: inputting the training operating condition parameters into the motor heating model to obtain the heat generation of each node of the motor corresponding to each training operating condition parameter;

[0121] It can be understood that the motor heating model is used to calculate the heat generated by each node of the motor under different working conditions.

[0122] It should be noted that the motor heating model is a physical model used to simulate and predict the heat generated by the motor during operation. This model is based on the physical process and thermal principles inside the motor, and converts the motor's training operating parameters (such as voltage, speed, etc.) into the heat generated by each node inside the motor.

[0123] It should be understood that motor heating mainly comes from the following aspects:

[0124] Copper loss: The heat generated by the resistance when the current passes through the motor winding. Copper loss is proportional to the square of the current and is one of the main sources of motor heating.

[0125] Iron loss: The iron core in the motor will produce eddy current and hysteresis under the action of the alternating magnetic field, thereby losing energy and converting it into heat. Iron loss includes eddy current loss and hysteresis loss, which are related to the frequency, magnetic flux density and material properties of the motor.

[0126] Mechanical loss: During the operation of the motor, mechanical energy is lost due to bearing friction, wind resistance, etc. These losses will also be converted into heat.

[0127] The motor heating model can be constructed by following the steps below:

[0128] Determine the input parameters: including the training operating parameters of the motor, such as voltage, current, speed, load, etc., as well as the structural parameters of the motor, such as winding resistance, core size, heat dissipation conditions, etc.

[0129] Calculate copper loss: Based on the motor current and winding resistance, use Joule's law to calculate the copper loss in the winding.

[0130] Calculating iron loss: The calculation of iron loss is relatively complex because it involves changes in the magnetic field and the material properties of the core. Usually, iron loss can be estimated using experimental data or empirical formulas. In some cases, finite element analysis (FEA) can also be used to more accurately calculate iron loss.

[0131] Consider mechanical losses: It is relatively difficult to estimate mechanical losses because they are related to the specific structure of the motor and the operating environment. In practical applications, mechanical losses can be estimated based on the type and rated power of the motor.

[0132] Heat transfer analysis: The calculated copper loss, iron loss and mechanical loss are used as heat sources, and the temperature distribution inside the motor is analyzed using heat transfer theory (such as Fourier heat conduction equation). This process needs to consider the heat dissipation conditions of the motor (such as fans, radiators, etc.) and parameters such as the thermal conductivity of the motor material.

[0133] Node heat generation calculation: In the motor heat generation model, the motor is usually divided into several small nodes (or units), and the heat generation of each node is calculated. This can be achieved by allocating the total heat to each node in a certain proportion. The basis for the allocation can be factors such as the node volume and heat dissipation conditions.

[0134] In the specific implementation, a motor heating model can be built based on the electromagnetic performance parameters of the motor. The input of the model is the motor voltage, motor output speed, motor output torque, motor coolant temperature and motor coolant flow, and the model output is the heat generated by each node of the motor. Among them, the electromagnetic performance parameters refer to: inductance, resistance, magnetic field strength, magnetic flux, etc. These performance parameters affect the energy conversion efficiency and loss distribution of the motor. During the operation of the motor, changes in these electromagnetic parameters will cause various losses inside the motor, including copper loss, iron loss, mechanical loss, etc. These losses will eventually be dissipated in the form of heat energy, causing the motor temperature to rise.

[0135] Step S03: inputting the training working condition parameters and the corresponding calorific value into a three-dimensional finite element model to obtain the temperature of each node and the temperature change rate of each node;

[0136] It should be noted that a three-dimensional finite element model can be built based on the motor geometry model, cooling method, and motor material properties. The specific building steps are as follows:

[0137] 1. Collect motor geometry data

[0138] Geometry: Get detailed geometry of the motor, including the size and location of components such as the stator, rotor, windings, bearings, cooling system, etc.

[0139] CAD model: If possible, export the 3D model of the motor from the CAD software (such as IGES, STEP and other formats), which will greatly simplify the subsequent modeling work in the finite element software.

[0140] 2. Define motor material properties

[0141] Thermal Conductivity: Specify the thermal conductivity for each material used in the motor (such as copper, iron, insulation, coolant, etc.).

[0142] Specific Heat Capacity: The specific heat capacity of a material determines how much its temperature changes when it absorbs or releases heat.

[0143] Density: Although primarily used in structural analysis, it may also be required in some thermal analyses.

[0144] Other thermal properties: such as thermal expansion coefficient, thermal emissivity, etc., are determined based on the specific needs of the analysis.

[0145] 3. Choose the right finite element software

[0146] Choose a finite element software that supports thermal analysis, such as ANSYS, Abaqus, SolidWorks Simulation, etc.

[0147] Ensure the software has the ability to handle complex geometries, multiple material properties, and advanced thermal boundary conditions.

[0148] 4. Build a 3D finite element model

[0149] Import geometry model: If you already have a CAD model, import it directly into the finite element software. Otherwise, create the geometry model manually in the software.

[0150] Meshing: Select the appropriate mesh type and density based on the complexity of the motor and the analysis requirements. Use denser meshes near heat sources and in areas with large temperature gradients.

[0151] Define Materials: Create a material library in your finite element software and assign the correct material properties to each part in your model.

[0152] Set the boundary conditions:

[0153] Heat Flux: Apply heat flux to heat generating components such as windings to simulate heat generation.

[0154] Convective heat transfer: Set the convection heat transfer coefficient between the cooling system (such as cooling fan, radiator) and the motor surface to simulate the cooling effect.

[0155] Radiation heat transfer (if applicable): Set radiation boundary conditions in areas where radiation heat transfer needs to be considered.

[0156] Initial Conditions: Set the initial temperature distribution for the analysis.

[0157] 5. Solution and post-processing

[0158] Solve: Run a finite element analysis and the software will calculate the temperature field based on the entered geometry, material properties, and boundary conditions.

[0159] Post-processing: View and analyze results, including temperature distribution graphs, temperature rate of change graphs, hot spot identification, etc. Use the tools provided by the software to extract and visualize key data.

[0160] 6. Verification and optimization

[0161] Validation: Compare FEA results with experimental data, other simulation results, or known physical principles to verify the accuracy and reliability of the model.

[0162] Optimization: Make necessary adjustments and optimizations to the model based on the validation results to improve computational accuracy and reliability. This may include modifying mesh density, adjusting boundary conditions, or updating material properties.

[0163] In the specific implementation, the heat value corresponding to the training operating condition parameters is input into the finite element model as a heat source. This is usually achieved by applying a heat flux or heat source to the corresponding nodes or units in the model. The heat value has been calculated in units of nodes, and these values ​​can be directly applied to the corresponding nodes. Boundary conditions are applied to the finite element model, including the temperature and flow rate of the coolant, which will affect the temperature distribution inside the motor. The temperature field is solved using finite element analysis software. The software will calculate the temperature of each node inside the motor through numerical methods (such as finite difference method, finite element method, etc.) based on the input heat value, material properties and boundary conditions. After obtaining the temperature of each node, the temperature change rate of each node can be further calculated. This is usually achieved by comparing the temperature values ​​of adjacent time steps, that is, the temperature change rate = ΔT / Δt, where ΔT is the change in temperature and Δt is the time step.

[0164] Step S04: using the training operating condition parameters and their corresponding heating value, temperature, and temperature change rate, supervised training is performed on the temperature change rate prediction model to obtain the trained temperature change rate prediction model.

[0165] It should be noted that this embodiment uses the temperature change rate prediction model as an example of a neural network model for explanation.

[0166] Specifically, the model simulation working conditions are designed according to the five dimensions of coolant temperature, coolant flow, motor output speed, motor output torque, and motor voltage. According to the designed simulation working conditions, the three-dimensional finite element model is subjected to transient simulation calculation, and the input and output data of the three-dimensional finite element model are retrieved as the database required for neural network model training. Among them, the input data of the three-dimensional finite element model are motor output speed, motor output torque, motor voltage, coolant temperature, and coolant flow, and the output data of the three-dimensional finite element model are the temperature of each node of the motor at different times and the temperature change rate.

[0167] According to the database, the pre-established neural network model is trained. Taking the BP neural network as an example, it is described in detail. The training process of the BP neural network includes two processes: forward propagation of information and back propagation of errors. In the cycle of forward propagation of information and back propagation of errors, the weights and thresholds of each layer are continuously adjusted to complete the learning and training of complex problems. The construction and training of the BP neural network model includes the following steps:

[0168] (1) Dataset division: The database data is divided into training set, validation set, and test set, with a division ratio of 60%, 20%, and 20%;

[0169] (2) Preprocessing: Since the properties, dimensions, magnitudes, availability and other characteristics of each parameter are different, directly using the original value of the parameter for analysis will amplify the effect of the parameter with a high magnitude. Data standardization is performed on all parameters of the training set, test set, and validation set. The formula is as follows:

[0170] X`=(xx min ) / (x max -x min ) Formula 2;

[0171] In Formula 2, X' is the new value of x after standardization, x represents a parameter value in the original data, and x min Indicates that the parameter is the minimum value in the set, x max Indicates that the parameter is the maximum value in the set.

[0172] (3) Model construction:

[0173] Determine the input and output of the BP neural network. In order to realize the online prediction function of the model, the input parameters of the model are set as the motor output speed, motor output torque, motor voltage, coolant temperature, coolant flow, and the current temperature of each node (when the motor is in the starting state, the current temperature of each node is 0). The model output is the temperature change rate of each node;

[0174] Determine the number of hidden layers and nodes. Selecting multiple hidden layers can improve the generalization ability and prediction accuracy of the model. However, considering the computing resources of the motor controller, the number of hidden layers is determined to be 1. The number of hidden layer nodes can be selected according to the following formula 3, where k is the number of hidden layer neuron nodes, n is the number of input layer neuron nodes, m is the number of output layer neuron nodes, and α is a constant between 1 and 10. For example, the number of nodes selected in this embodiment is 7;

[0175]

[0176] Determine the activation function and training function. There are many optional activation functions for BP neural network. This embodiment selects Sigmoid function as the activation function of the hidden layer and linear function as the output layer. The training function selects the gradient descent training function traingdx with momentum back propagation and dynamic adaptive learning rate;

[0177] (4) Model parameter optimization: There are many optimization methods for BP neural network algorithm, mainly including genetic algorithm, particle swarm algorithm, ant colony optimization algorithm, artificial bee colony algorithm, etc. This embodiment takes genetic algorithm as an example to optimize the initial weights and thresholds of BP neural network. The first step is to determine the relevant parameters of the genetic algorithm, including the determination of the chromosome encoding method, the selection of the fitness function, the determination of the selection operation implementation algorithm, the probability of crossover and mutation operations, and the setting of operation parameters. The second step is to encode the initial population. According to the structure of the BP neural network, the original population is randomly generated, and the individuals in the original population are encoded with real numbers. The individual encoding includes the weight and threshold information of the network. The third step is to calculate the individual fitness. This step mainly evaluates the adaptability of each individual in the population. Excellent individuals will be inherited to the next generation population for crossover and mutation. After multiple iterations, the genetic algorithm finally outputs the optimal individual. The fourth step is to decode the optimal individual and assign the optimal initial weight and threshold to the BP neural network;

[0178] (5) Model verification step: The test set data is loaded into the trained BP neural network model, see Formula 4 below, and the relative error (RE), mean relative error (MRE), and root mean square error (RMSE) calculation formulas are introduced to evaluate the model and judge the model performance, and finally the trained temperature change rate prediction model is obtained.

[0179]

[0180]

[0181]

[0182] In formula 4, N is the total amount of data, V i is the temperature change rate data of each node in the test set, V′ i It is the predicted result of the temperature change rate of each node.

[0183] In this embodiment, by obtaining a plurality of preset training operating condition parameters, these parameters cover a variety of possible states when the motor is running (such as different voltages, speeds, torques, and cooling conditions), thereby ensuring the comprehensiveness and diversity of the model training. This wide range of operating condition coverage enables the model to learn the heating characteristics of the motor under more complex conditions, thereby improving the prediction accuracy of the motor temperature change rate under unknown conditions. Since the trained temperature change rate prediction model has learned the direct mapping relationship from the operating condition parameters to the temperature change rate, there is no need to perform complex physical model calculations and finite element simulations in real-time prediction. This greatly simplifies the real-time calculation process and reduces the demand for computing resources.

[0184] Based on any of the above embodiments, the three-dimensional finite element model is Figure 3 Steps S11 to S13 shown construct:

[0185] Step S11: obtaining a motor fluid domain, and setting a coolant and a fluid domain inlet and outlet in the motor fluid domain to obtain a motor fluid model;

[0186] It should be noted that the motor fluid domain is extracted based on the motor geometric model.

[0187] In the specific implementation, it is necessary to first obtain the motor geometry model. In the motor geometry model, all areas involving coolant flow need to be identified. These areas may include cooling water channels, cooling jackets, radiators, etc. Use CAD software or professional fluid dynamics (CFD) pre-processing software to extract these fluid areas from the motor geometry model. This involves removing or hiding non-fluid parts (such as motor cores, windings, etc.) from the model, retaining only the channels and cavities required for fluid flow. Meshing the extracted fluid domain is the basis of CFD simulation. According to the type of coolant used (such as water, oil, ethylene glycol mixture, etc.), set the corresponding physical properties such as density, specific heat capacity, thermal conductivity, viscosity, etc. in the CFD simulation software. Set the initial temperature, pressure and other conditions of the coolant. These conditions should be determined according to the actual operating conditions and cooling requirements of the motor. Determine the inlet and outlet positions of the fluid domain according to the design of the motor's cooling system. The inlet is usually located where the coolant enters the motor, while the outlet is located where the coolant flows out of the motor. In the CFD simulation software, set appropriate boundary conditions for the inlet and outlet of the fluid domain. Inlet boundary conditions may include velocity, flow rate, temperature, etc.; outlet boundary conditions may include pressure, back pressure, etc. These boundary conditions should be determined based on the actual operating conditions of the motor and the design of the cooling system. After completing the construction of the fluid model, a preliminary CFD simulation can be performed to verify the accuracy and rationality of the model. Check whether the mesh quality of the fluid domain, the flow state of the coolant, and the settings of the inlet and outlet boundary conditions meet the requirements.

[0188] Step S12: obtaining a motor geometric model, and setting a contact surface and a heat value in the motor geometric model to obtain a three-dimensional temperature field model;

[0189] It should be noted that in the motor, heat is transferred between different components through contact surfaces. Therefore, in the three-dimensional temperature field model, the location and properties of these contact surfaces need to be clarified. The setting of the contact surface involves the consideration of contact thermal resistance, which is an important factor affecting the efficiency of heat transfer. By reasonably setting the contact surface, the heat transfer process inside the motor can be simulated more accurately.

[0190] The motor will generate heat during operation, which mainly comes from various losses of the motor, such as copper loss, iron loss, mechanical loss, etc. In the three-dimensional temperature field model, this heat needs to be loaded onto the corresponding components as a heat source. The setting of the heat should be calculated or estimated based on the operating conditions and physical characteristics of the motor to ensure the accuracy and reliability of the model.

[0191] Step S13: coupling the motor fluid model with the three-dimensional temperature field model to obtain the three-dimensional finite element model.

[0192] It should be noted that the three-dimensional finite element model includes two parts: a motor fluid model and a motor three-dimensional temperature field model. The fluid model is coupled with the three-dimensional temperature field model to determine the three-dimensional finite element model.

[0193] In this embodiment, by coupling the fluid model with the temperature field model, the flow of the coolant inside the motor and the resulting heat exchange effect can be simulated more accurately, thereby more accurately predicting the temperature distribution of various components of the motor.

[0194] Based on any of the above embodiments, the method is applied to a motor controller mounted on a vehicle, and the method further includes:

[0195] The locally pre-stored trained temperature change rate prediction model is read.

[0196] In the specific implementation, the temperature change rate prediction model is taken as the neural network model for explanation, and the neural network model and the integration module (such as the fixed-step integration function module mentioned above) are encapsulated and embedded in the motor controller. When predicting the motor temperature, the trained temperature change rate prediction model is read, and the temperature change rate of each node in the historical time slot is output through the prediction model. The integration module is then used to perform analysis and calculation based on the temperature change rate to obtain the predicted motor temperature at the next moment.

[0197] like Figure 4 As shown, Figure 4A schematic diagram of online prediction of motor temperature provided by an embodiment of the present application, in which the temperature of the magnetic steel node in the motor is predicted as an example, the neural network model loads the motor output speed, output torque, bus voltage (i.e., motor voltage), coolant flow, coolant temperature, and the temperature of the magnetic steel node at the last moment (when the motor is just started, the temperature is the coolant temperature), calculates the temperature change rate of the magnetic steel node, and then realizes the real-time prediction of the magnetic steel temperature by integration. When predicting the temperature of the motor, the temperature of all nodes can be predicted by repeating the above steps, thereby determining the predicted temperature of the motor.

[0198] In this embodiment, by pre-storing the trained temperature change rate prediction model, the motor controller can quickly call the model for real-time calculation when the vehicle is running. Compared with online training or remote model calling, this method significantly reduces the calculation delay and improves the real-time response of the system. Online training models require a lot of computing resources and time. The temperature change rate prediction model can be trained in advance on a computer using a three-dimensional finite element model, and the motor controller stores the trained prediction model. In fact, the three-dimensional finite element model is reduced to a neural network model (the temperature change rate prediction model takes the neural network model as an example), which not only meets the prediction accuracy requirements but also effectively solves the problem of insufficient on-board computing power, and can realize the real-time prediction function of motor temperature.

[0199] Based on the method described in any of the above embodiments, the present application also provides Figure 5 A schematic diagram of the structure of a device is shown in FIG. Figure 5 At the hardware level, the device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to implement the method described in any of the above embodiments.

[0200] Based on the method described in any of the above embodiments, the present application also provides a vehicle, and the vehicle applies the method described in any of the above embodiments.

[0201] Based on the method described in any of the above embodiments, the present application also provides a computer program product, which includes one or more computer programs or instructions. The computer program or instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. When the computer program is executed by a processor, the method described in any of the above embodiments is implemented.

[0202] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0203] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0204] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0205] The above description is only an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0206] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0207] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

Claims

1. A method for predicting motor temperature, characterized in that: The method comprises: According to the state of the motor, the temperature influence parameter of the historical time slot is obtained, and the temperature influence parameter is input into the trained temperature change rate prediction model to obtain the temperature change rate of each node in the motor in the historical time slot; the temperature influence parameter includes parameters that affect the temperature of the motor; the nodes include: stator, rotor and winding; the trained temperature change rate prediction model is trained by the input and output data of the trained three-dimensional finite element model; The motor temperature at the next moment is predicted according to the temperature change rate.

2. The method according to claim 1, characterized in that The predicting the motor temperature at the next moment according to the temperature change rate includes: Calculating the minimum time difference between the historical time slot and the next moment; The motor temperature is determined based on the temperature change rate and the minimum time difference.

3. The method according to claim 1, characterized in that The motor state is a running state, and the step of obtaining the temperature influence parameter of the historical time slot includes: The motor voltage, the motor output speed, the motor output torque, the motor coolant temperature, the motor coolant flow rate and the temperature of each of the nodes are obtained as the temperature influencing parameters.

4. The method according to claim 1, characterized in that The motor state is a startup state, and the step of obtaining the temperature influence parameter of the historical time slot includes: The motor voltage, the motor output speed, the motor output torque, the motor coolant temperature and the motor coolant flow rate are obtained as the temperature influencing parameters.

5. The method according to claim 1, characterized in that The temperature change rate prediction model is trained by the following steps: Acquire a plurality of preset training operating condition parameters; the training operating condition parameters include motor voltage, motor output speed, motor output torque, motor coolant temperature and motor coolant flow; Inputting the training operating condition parameters into the motor heating model to obtain the heat generation of each node of the motor corresponding to each training operating condition parameter; Input the training working condition parameters and the corresponding calorific value into a three-dimensional finite element model to obtain the temperature of each node and the temperature change rate of each node; The temperature change rate prediction model is supervisedly trained using the training operating condition parameters and their corresponding heating value, temperature, and temperature change rate to obtain the trained temperature change rate prediction model.

6. The method according to claim 1, characterized in that The three-dimensional finite element model is constructed by the following steps: Acquire a motor fluid domain, and set coolant and fluid domain inlets and outlets in the motor fluid domain to obtain a motor fluid model; Acquire a motor geometric model, and set a contact surface and a heat value in the motor geometric model to obtain a three-dimensional temperature field model; The motor fluid model is coupled with the three-dimensional temperature field model to obtain the three-dimensional finite element model.

7. The method according to claim 1, characterized in that The method is applied to a motor controller mounted on a vehicle, and the method further comprises: The locally pre-stored trained temperature change rate prediction model is read.

8. A device, characterized in that: The device comprises: processor; a memory for storing processor-executable instructions; Wherein, when the processor calls the executable instruction, the method described in any one of claims 1-7 is implemented.

9. A vehicle, characterized in that: The vehicle application comprises the method according to any one of claims 1-7.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images