Motor rotor temperature prediction method

Abstract

The invention discloses a motor rotor temperature prediction method, and belongs to the technical field of rotor temperature prediction. The method comprises the following steps: respectively establishing mapping relations between copper loss and iron loss of the motor and parameters such as current, angular velocity and target node temperature, wherein a target node is a node having a heat exchange relation with a rotor node in the motor; constructing an initial rotor temperature model based on the mapping relation and a heat balance model of the target node; based on a preset optimization algorithm and the initial rotor temperature model, determining optimization values of the first coefficient and the second coefficient, and obtaining an optimized rotor temperature prediction model; and determining the current temperature of the rotor node based on the rotor temperature prediction model, the motor current, the motor angular velocity, the temperature of the target node and the temperatures of other nodes. According to the method, the initial rotor temperature model is constructed through the heat balance model of the single node, and the calculation process is simplified and the calculation complexity is reduced by reducing the number of the nodes and simplifying the heat balance model.

Description

Technical Field

[0001] The present application relates to the technical field of rotor temperature prediction, and in particular to a method for predicting the temperature of a motor rotor. Background Art

[0002] Permanent magnet synchronous motors (PMSMs) are widely used in new energy vehicles due to their high efficiency and high torque density. However, these motors also place higher demands on thermal safety. Excessive motor temperature rise can lead to thermal failures, even stator winding insulation failure and irreversible demagnetization of the permanent magnets. Therefore, real-time thermal monitoring of motors is crucial, and rotor temperature monitoring is a key requirement.

[0003] Existing technologies typically rely on a system of heat exchange equations between multiple nodes within the motor to jointly calculate rotor temperature. However, heat exchange within the motor is a complex process involving multiple transfer objects and paths. To ensure energy conservation, all relevant transfer objects and paths must be fully considered, making the overall calculation process extremely tedious and complex. Summary of the Invention

[0004] The embodiments of the present application provide a method, device, equipment and storage medium for predicting the temperature of a motor rotor, which simplifies the motor rotor temperature prediction process and improves prediction efficiency.

[0005] The present invention provides a method for predicting the temperature of a motor rotor, the method comprising:

[0006] Obtaining a first mapping relationship between the motor copper loss and the temperature of a target node, the motor current, and a first coefficient, and a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and a second coefficient, where the target node is a node in the motor that has a heat exchange relationship with the rotor node;

[0007] Constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, and a thermal balance model of the target node;

[0008] Based on a preset optimization algorithm and an initial rotor temperature model, determining optimized values ​​of the first coefficient and the second coefficient, and obtaining an optimized rotor temperature prediction model;

[0009] Based on the rotor temperature prediction model, as well as the motor current, motor angular velocity, the temperature of the target node, and the temperature of other nodes, the current temperature of the rotor node is predicted. The other nodes are nodes in the motor other than the rotor node that have a heat exchange relationship with the target node.

[0010] In some embodiments, the target node is a stator node, and the motor current includes a motor d-axis current and a motor q-axis current;

[0011] Obtaining a first mapping relationship between the motor copper loss and the temperature of the target node, the motor current, and the first coefficient, including:

[0012] Obtaining a first loss mapping relationship between the motor copper loss and the electromagnetic parameters and motor current of the motor, where the electromagnetic parameters include a winding phase resistance of a stator node;

[0013] Based on the first loss mapping relationship, a first mapping relationship is constructed, where the first coefficient includes a first parameter, a second parameter, a third parameter, and a fourth parameter;

[0014] The copper loss is the sum of the first product, the second product, the third product and the fourth product. The first product is the product of the motor d-axis current and the first parameter, the second product is the product of the motor q-axis current and the second parameter, the third product is the product of the motor d-axis current, the stator node temperature and the third parameter, and the fourth product is the product of the motor q-axis current, the stator node temperature and the fourth parameter.

[0015] In some embodiments, obtaining a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and the second coefficient includes:

[0016] Obtaining a second loss mapping relationship between the motor iron loss and the electromagnetic parameters and angular velocity of the motor, where the electromagnetic parameters include a hysteresis loss coefficient, an eddy current loss coefficient, and a magnetic flux at a stator node;

[0017] Based on the second loss mapping relationship, a second mapping relationship is constructed, where the second coefficient includes a fifth parameter and a sixth parameter;

[0018] The iron loss is the sum of the fifth product and the sixth product. The fifth product is the product of the motor d-axis current, the motor angular velocity, and the fifth parameter. The sixth product is the product of the motor q-axis current, the motor angular velocity, and the sixth parameter.

[0019] In some embodiments, further comprising:

[0020] Obtaining a third mapping relationship between the motor mechanical loss, the motor angular velocity, and the third coefficient;

[0021] Constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, the third mapping relationship and the thermal balance model of the target node;

[0022] Based on a preset optimization algorithm and an initial rotor temperature model, optimized values ​​of the first coefficient, the second coefficient, and the third coefficient are determined, and an optimized rotor temperature prediction model is obtained.

[0023] In some embodiments, obtaining a third mapping relationship between the motor mechanical loss, the motor angular velocity, and the third coefficient includes:

[0024] Obtaining a third loss mapping relationship between the motor mechanical loss and the motor friction parameters, motor ventilation parameters, motor angular velocity, and size parameters of the rotor node, where the motor friction parameters include surface roughness coefficient and friction coefficient, the motor ventilation parameters include air density, and the size parameters include the length and radius of the rotor node;

[0025] Based on the third loss mapping relationship, construct a third mapping relationship, where the third coefficient includes a seventh parameter;

[0026] The motor mechanical loss is the product of the motor angular velocity and the seventh parameter.

[0027] In some embodiments, the method further comprises:

[0028] Obtaining a fourth mapping relationship between the heat capacity of the target node, the temperature of the target node, and a fourth coefficient, a fifth mapping relationship between the thermal resistance between the target node and the rotor node, the temperature of the target node, and a fifth coefficient, and a sixth mapping relationship between the thermal resistance between the target node and other nodes, the temperature of the target node, the temperature of the other nodes, and a sixth coefficient;

[0029] Constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, the third mapping relationship, the fourth mapping relationship, the fifth mapping relationship, the sixth mapping relationship, and a thermal balance model of the target node;

[0030] Based on a preset optimization algorithm and an initial rotor temperature model, optimized values ​​of the first coefficient, the second coefficient, the third coefficient, the fourth coefficient, the fifth coefficient, and the sixth coefficient are determined, and an optimized rotor temperature prediction model is obtained.

[0031] In some embodiments, obtaining a fourth mapping relationship between the heat capacity of the target node, the temperature of the target node, and the fourth coefficient includes:

[0032] Obtaining a first preset functional relationship between the heat capacity of the target node and the temperature of the target node;

[0033] Based on the first preset functional relationship, constructing a fourth mapping relationship, where the fourth coefficient includes an eighth parameter;

[0034] The heat capacity of the target node is the product of the temperature of the target node and the eighth parameter.

[0035] In some embodiments, obtaining a fifth mapping relationship between the thermal resistance between the target node and the rotor node, the temperature of the target node, and the fifth coefficient includes:

[0036] Obtaining a second preset functional relationship between the thermal resistance between the target node and the rotor node and the temperature of the target node;

[0037] Based on the second preset functional relationship, constructing a fifth mapping relationship, wherein the fifth coefficient includes a ninth parameter;

[0038] The thermal resistance between the target node and the rotor node is the product of the temperature of the target node and the ninth parameter.

[0039] In some embodiments, obtaining a sixth mapping relationship between the thermal resistance between the target node and the other nodes and the temperature of the target node, the temperature of the other nodes, and the sixth coefficient includes:

[0040] Obtaining a third preset functional relationship between the thermal resistance between the target node and the other nodes, the temperature of the target node, and the temperatures of the other nodes;

[0041] Based on the third preset functional relationship, a sixth mapping relationship is constructed, where the sixth coefficient includes a tenth parameter and an eleventh parameter;

[0042] The thermal resistance between the target node and the other nodes is the sum of the seventh product and the eighth product. The seventh product is the product of the temperature of the other nodes and the tenth parameter. The eighth product is the product of the temperature of the target node and the eleventh parameter.

[0043] In some embodiments, the preset optimization algorithm is the Levenberg-Marquardt algorithm.

[0044] An embodiment of the present application provides a method for predicting the temperature of a motor rotor. The present application obtains a first mapping relationship between the motor copper loss and the temperature of a target node, a motor current, and a first coefficient, and a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and the second coefficient. The target node is a node in the motor that has a heat exchange relationship with the rotor node. Mapping relationships are established between the copper loss and the iron loss and parameters such as the motor current, the angular velocity, and the target node temperature, respectively, thereby reducing the influence of the motor's inherent parameters and making the rotor temperature model more universal. Furthermore, based on the above-mentioned mapping relationship and the thermal balance model of the target node that has a heat exchange relationship with the rotor node, an initial rotor temperature model is constructed. The initial rotor temperature model is constructed using the thermal balance model of a single node. By reducing the number of nodes and simplifying the thermal balance model, the calculation process is simplified, the calculation complexity is reduced, and the temperature prediction is made more efficient. Furthermore, based on a preset optimization algorithm and the initial rotor temperature model, the optimized values ​​of the first coefficient and the second coefficient are determined. By optimizing these coefficients, the model is made more in line with the actual situation, the accuracy of the rotor temperature prediction is improved, and a better match between the prediction result and the actual situation is ensured. Furthermore, based on the rotor temperature prediction model, as well as the motor current, motor angular velocity, temperature of the target node and temperatures of other nodes, the current temperature of the rotor node can be determined, thereby enhancing the thermal monitoring capability of the motor. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] To more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present application. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0046] In order to more completely understand the present application and its beneficial effects, the following description will be given in conjunction with the accompanying drawings, wherein the same drawing numbers represent the same parts in the following description.

[0047] Figure 1 A schematic diagram of a two-node LPTN model provided in an embodiment of the present application;

[0048] Figure 2 A flow chart of a method for predicting motor rotor temperature provided in an embodiment of the present application;

[0049] Figure 3 A schematic diagram of another two-node LPTN model provided in an embodiment of the present application;

[0050] Figure 4 A schematic diagram of an experimental verification provided for an embodiment of the present application. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0052] In the embodiments of this application, "at least one" refers to one or more; "a plurality" refers to two or more. In the description of this application, words such as "first," "second," and "third" are used only for the purpose of distinguishing descriptions and should not be understood as indicating or implying relative importance or order.

[0053] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, the terms "including," "comprising," "having," and their variations in this specification all mean "including but not limited to," unless otherwise specifically stated.

[0054] It should be noted that in the embodiments of the present application, "connection" can be understood as electrical connection, and the connection between two electrical components can be a direct or indirect connection between the two electrical components. For example, the connection between A and B can be either a direct connection between A and B or an indirect connection between A and B through one or more other electrical components.

[0055] A permanent magnet synchronous motor (PMSM) uses permanent magnets to generate a magnetic field. This field interacts with current to synchronize the rotor and stator magnetic fields. It primarily consists of a stator, rotor, and controller. The operating principle of a PMSM is based on the interaction between electromagnetic induction and the magnetic field of permanent magnets. When three-phase alternating current is passed through the stator's three-phase symmetrical windings, a rotating magnetic field is generated. The rotor, composed of permanent magnets, interacts with the stator's magnetic field, generating electromagnetic torque that drives the rotor's synchronous rotation.

[0056] Permanent magnet synchronous motors (PMSMs) are widely used in new energy vehicles due to their high efficiency and high torque density. However, these motors also place higher demands on thermal safety. Excessive motor temperature rise is known to cause thermal failures, even stator winding insulation failure and irreversible demagnetization of permanent magnets. Therefore, real-time thermal monitoring of motors is crucial, with rotor temperature acquisition being a key factor. Accurately estimating rotor temperature can effectively increase the duration of peak torque, improving motor efficiency while reducing the amount of stator magnetic material and manufacturing costs.

[0057] The rotor temperature prediction method can be achieved through LPTN model, electrical parameter tracking method and machine learning methods.

[0058] Among them, electrical parameter tracking methods observe rotor temperature based on electromagnetic parameters, for example, by acquiring the motor's back EMF or high-frequency signal and determining the rotor temperature based on the corresponding relationship between the two. Research has shown that at low speeds, high-frequency signal injection offers better estimation accuracy than back EMF because signal injection eliminates speed-dependent estimation. However, at high speeds, back EMF-based methods offer superior performance in terms of stability and feasibility, which in turn requires extremely high signal acquisition accuracy.

[0059] Machine learning methods use mathematical models, such as multi-layer perceptrons (MLPs), to predict rotational temperature. This method replaces the LPTN with an MLP model, eliminating inherent parameter dependence; however, it requires a large amount of data for offline neural network training. Furthermore, the high complexity of machine learning models results in high prediction accuracy, but low generalization.

[0060] The LPTN (Lumped-Parameter Thermal Network) model divides the object of study into several nodes, assuming uniform temperature distribution. The rotor temperature can be determined by solving the heat balance equation using the measured temperatures at some nodes. The LPTN approach offers advantages in its flexible node nature and low computational overhead. However, it is highly dependent on the geometry and parameters of the motor, making it highly sensitive to parameter uncertainty and lacking generalization capabilities.

[0061] It can be seen that among the three methods of rotor temperature prediction, the LPTN model has a greater advantage in predicting rotor temperature because there is currently no large amount of motor data to train the machine learning model, high-precision signal data cannot be guaranteed, and considering the computational overhead, the LPTN model still has some problems.

[0062] It's important to note that the Lumped-Parameter Thermal Network (LPTN) is a circuit model used to describe the thermal performance of power devices. It represents thermal resistance (R) and thermal capacitance (C) in a circuit format to simulate the device's heat conduction and heat storage characteristics during operation. The LPTN is a fundamental tool for power device thermal design, helping engineers predict device temperature changes during the design phase, optimize heat dissipation solutions, and improve system reliability and efficiency.

[0063] See also Figure 1 As shown, Figure 1 A schematic diagram of a two-node LPTN model provided in an embodiment of the present application.

[0064] The diagram shows two nodes: the stator node and the rotor node. The temperature of the stator node is Ts, and the temperature of the rotor node is Tr. The stator node has a certain heat capacity Cs and a certain heat loss Ps, while the rotor node also has a certain heat capacity Cr and a certain heat loss Pr. Heat transfer occurs between the stator and rotor nodes, resulting in a thermal resistance Rsr between the two nodes, which reflects the difficulty of heat transfer, similar to the resistance in a circuit. In addition, heat transfer also occurs between the stator node and the coolant, with a thermal resistance Rcs between the stator node and the coolant, and the coolant temperature is Tc. Heat transfer also occurs between the rotor and the environment, with a thermal resistance Rra between the rotor and the environment, and the environment temperature is Ta. Consequently, the LPTN model has a series of equations, each of which reflects the energy transfer at each node.

[0065] The equations for the stator and rotor nodes are:

[0066]

[0067] The stator node equation in formula (1) shows that the stator node heat change can be expressed by heat capacity and temperature change (left side of the equation), which also shows that the heat change comes from the stator heat loss P s and heat transfer from other heat sources (right side of the equation). Similarly, the rotor node equation in formula (2) shows that the heat change of the rotor node can be expressed by heat capacity and temperature change (left side of the equation), which also shows that the heat change comes from the rotor heat loss P r and heat transfer from other heat sources (right side of the equation).

[0068] The heat loss of the stator node and the rotor node usually needs to be calculated based on the electromagnetic model, which depends on the electromagnetic parameters of the motor. The loss of the motor includes copper loss P Cu and iron loss P Fe , the calculation formula is as follows:

[0069]

[0070] R s =R s,20 [1+a Cu,20 (T s -20)](5);

[0071]

[0072] Where i d Refers to the motor d-axis current, i q Refers to the motor q-axis current, k h Refers to the hysteresis loss coefficient, k e Refers to the eddy current loss coefficient, w m is the motor angular velocity, ψ s is the designated sub-flux linkage, R s is the phase resistance of the specified node winding, which is related to the stator node temperature; R s,20 It is 20 degrees R s , a Cu,20 is the copper temperature coefficient at 20 degrees; L q Refers to the motor q-axis inductance, L d Refers to the motor d-axis inductance, ψ PM is the flux linkage of a permanent magnet. Hysteresis loss is caused by the hysteresis properties of the core material. When the core is exposed to an alternating magnetic field, friction and rearrangement between magnetic domains occur, resulting in energy loss as heat. Eddy current loss is caused by eddy currents induced in the core by the alternating magnetic field. These eddy currents generate heat as they flow through the core.

[0073] The loss formulas of stator node and rotor node are as follows:

[0074] P s =PCu +kP Fe (7);

[0075] P r =(1-k)P Fe (8);

[0076] Where k is the ratio of stator iron loss to total motor iron loss. The k value can be a preset empirical value that satisfies (0, 1). It should be noted that, generally, the copper loss of a motor comes from the stator nodes.

[0077] Using the above formula, the LPTN model can calculate the temperature relationship between adjacent moments: T(t+1) = f(T(t)), where T(t) represents the temperature of the stator and rotor nodes at time t, and f represents the function solved by the LPTN model. The predicted temperature value T(t) at each moment is then obtained based on the initial temperature value T(0) or the historical temperature T(t-1) and this function f.

[0078] According to the process of the above LPTN model, it can be known that the above LPTN model has the following disadvantages:

[0079] 1. The LPTN model relies on the inherent parameters of the motor. These parameters are often not fixed in the actual motor during use, which results in the LPTN model often not being effective in actual predictions. Specifically, the LPTN model usually requires a series of motor parameters, such as the motor's d-axis and q-axis inductances to calculate the magnetic flux. These parameters are generally considered constants. However, as the motor's usage increases, its materials inevitably age, or the temperature inside the motor rises after actual operation, and the motor parameters will change. For example, the motor's inductance will change with the motor's temperature. When the LPTN model is applied to different motors, if the parameters of these motors vary significantly, the effectiveness of the LPTN model will decrease. For example, the loss calculation formula in the LPTN model is relatively complex, which means that a relatively small change in one of the parameters may lead to very different results.

[0080] 2. The LPTN model assumes that heat is transferred between several objects in the system, and that overall energy conservation is satisfied. However, this assumption is difficult to meet in real-world scenarios. Specifically, based on the two-node LPTN model introduced above, energy transfer only occurs between four parts, including the stator, rotor, coolant, and environment. However, there may be parts other than the stator and rotor that participate in heat transfer, such as the motor casing, which transfers energy with the stator and rotor. If these parts are not added to the LPTN model system, the entire system will no longer satisfy energy conservation. However, it is unrealistic to add all parts involved in energy transfer to the LPTN model for calculation. This would require more sensors, which increases costs, would also rely on more motor parameters, and increase the model's computational complexity.

[0081] 3. The LPTN model can determine the relationship between rotor temperatures at adjacent moments, but it cannot completely eliminate its dependence on historical rotor temperatures. Specifically, the equations used to calculate the LPTN model are essentially differential equations. Solving this system of equations requires initial conditions to determine the solution, namely, the initial or historical values ​​of the stator and rotor temperatures. This means that rotor temperature cannot be directly predicted using the LPTN model alone, and prediction often relies on simulation data or other models.

[0082] 4. The heat capacity and thermal resistance parameters in the LPTN model are often constant, which cannot be generalized to all operating conditions. Specifically, the heat capacity and thermal resistance of the LPTN model are calculated using numerical algorithms based on motor operating data, often resulting in optimal solutions. Traditional methods often produce constant solutions. However, in real-time motor operation, the heat capacity and thermal resistance should be quantities that change over time, and the difficulty of energy transfer should be inconsistent under different conditions. For example, the greater the temperature difference between the stator and rotor, the faster the energy transfer; when there is no temperature difference between the two, there is no energy transfer.

[0083] In view of this, the present application proposes a method for predicting the temperature of a motor rotor, aiming to solve at least one of the above technical problems.

[0084] See also Figure 2 As shown, Figure 2 A flow chart of a method for predicting motor rotor temperature provided in an embodiment of the present application.

[0085] This application proposes a method for predicting the temperature of a motor rotor, which includes:

[0086] S10. Obtain a first mapping relationship between the motor copper loss and the temperature of the target node, the motor current, and the first coefficient, and a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and the second coefficient. The target node is a node in the motor that has a heat exchange relationship with the rotor node.

[0087] It's important to understand that a mapping relationship refers to the association between different variables, which can be expressed using a mathematical formula. Specifically, a mathematical formula can be used to represent a first mapping relationship between copper loss and the target node's temperature, motor current, and a first coefficient. A mathematical formula can also be used to represent a second mapping relationship between iron loss and the motor current, motor angular velocity, and a second coefficient. Based on the first and second mapping relationships, a formula representing copper and iron losses can be derived.

[0088] It is also important to understand that when a motor is running under load, it inevitably generates heat loss. The motor rotor temperature rise is primarily affected by copper loss and iron loss. Therefore, copper loss and iron loss are the primary components of the motor's heat loss. The calculation methods for copper loss and iron loss are shown in equations (3)-(6) above.

[0089] S20 : ​​Constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, and the thermal balance model of the target node.

[0090] Specifically, the initial rotor temperature model is obtained by bringing the characterization formulas of copper loss and iron loss obtained by the first mapping relationship and the second mapping relationship into the thermal balance model of the target node, thereby obtaining the initial rotor temperature model.

[0091] It should be understood that the thermal balance model is the core formula of each node in the LPTN model of the motor, which is used to describe the heat conservation of the node. In this application, a node with a heat exchange relationship with the rotor node is selected as the target node, so that the initial rotor temperature model can be constructed through the thermal balance model of the target node. As a result, the focus is no longer on all nodes in the LPTN model, but on the energy transfer process of a single node. Therefore, it is only necessary to ensure that a certain node satisfies the conservation of energy, which significantly reduces the complexity of the model. In addition, even if other nodes do not satisfy the conservation of energy, it will not affect the analysis results of the target node, thereby improving the fault tolerance and robustness of the model. In addition, the characterization formulas of copper loss and iron loss are obtained based on the first mapping relationship and the second mapping relationship, so as to avoid directly using complex motor parameters, but instead use a simpler loss model to estimate the stator loss value. This method not only reduces the complexity of the model, but also enhances the generalization ability of the model, making it more adaptable under different conditions.

[0092] S30 , based on a preset optimization algorithm and an initial rotor temperature model, determining optimized values ​​of the first coefficient and the second coefficient, and obtaining an optimized rotor temperature prediction model.

[0093] It's important to understand that the initial rotor temperature model only provides a basic framework for predicting rotor temperature. The first and second coefficients represent key parameters in the model. These coefficients directly impact the model's prediction accuracy. By optimizing these coefficients, the model's ability to predict rotor temperature can be improved, making the model more realistic and ensuring that the predicted results better match the actual situation.

[0094] Exemplarily, based on a preset optimization algorithm and an initial rotor temperature model, the optimized values ​​of the first coefficient and the second coefficient are determined, and the optimized rotor temperature prediction model is obtained. The process is as follows:

[0095] Obtain temperature data of rotor nodes at multiple historical moments and input characteristic data corresponding to the rotor temperature model;

[0096] Inputting the input feature data corresponding to each historical moment into the rotor temperature model, and obtaining the predicted temperature data of the rotor node corresponding to the historical moment;

[0097] Based on the error between the predicted temperature data of the rotor node and the temperature data at the corresponding historical moment, the first coefficient and the second coefficient are updated and optimized, thereby updating the optimized rotor temperature model until the error meets the preset training termination condition or the number of iterations meets the preset value, and the optimized rotor temperature prediction model is obtained. At the same time, the optimized values ​​of the first coefficient and the second coefficient in the optimized rotor temperature prediction model are also determined.

[0098] S40. Predicting the current temperature of the rotor node based on the rotor temperature prediction model, the motor current, the motor angular velocity, the temperature of the target node, and the temperature of other nodes. The other nodes are nodes in the motor other than the rotor node that have a heat exchange relationship with the target node.

[0099] It should be understood that the present application can predict the rotor node temperature at the current moment based on the rotor temperature prediction model, combined with the numerical values ​​of the motor current, motor angular velocity, and temperature of the target node required in the rotor temperature prediction model. The present application predicts the rotor node of unknown temperature by using nodes of known temperature, so that the prediction of the current rotor temperature is independent of the predicted value of the historical rotor temperature, thereby avoiding the error accumulation effect. It is understandable that other nodes can be coolant nodes, etc., which are specifically set according to the actual implementation plan and are determined by the temperatures of other nodes involved in the thermal balance model corresponding to the target node.

[0100] Through the above technical solution, an embodiment of the present application provides a method for predicting the temperature of a motor rotor. The present application obtains a first mapping relationship between the motor copper loss and the temperature of the target node, the motor current, and the first coefficient, and a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and the second coefficient. The target node is a node in the motor that has a heat exchange relationship with the rotor node. Mapping relationships are established between the copper loss and the iron loss and parameters such as the current, angular velocity, and the target node temperature, respectively, thereby reducing the influence of the motor's inherent parameters and making the rotor temperature model more universal. Furthermore, based on the thermal balance model of the target node that has a heat exchange relationship with the rotor node in the above mapping relationship, an initial rotor temperature model is constructed. The initial rotor temperature model is constructed through the thermal balance model of a single node. By reducing the number of nodes and simplifying the thermal balance model, the calculation process is simplified, the calculation complexity is reduced, and the temperature prediction is more efficient. Moreover, based on the preset optimization algorithm and the initial rotor temperature model, the optimized values ​​of the first coefficient and the second coefficient are determined. By optimizing these coefficients, the model is made more in line with the actual situation, the accuracy of the rotor temperature prediction is improved, and a better match between the prediction results and the actual situation is ensured. Furthermore, based on the rotor temperature prediction model, as well as the motor current, motor angular velocity, temperature of the target node and temperatures of other nodes, the current temperature of the rotor node can be determined, thereby enhancing the thermal monitoring capability of the motor.

[0101] In some embodiments, the target node is a stator node, and the motor current includes a motor d-axis current and a motor q-axis current;

[0102] Obtaining a first mapping relationship between the motor copper loss and the temperature of the target node, the motor current, and the first coefficient, including:

[0103] Obtaining a first loss mapping relationship between the motor copper loss and the electromagnetic parameters and motor current of the motor, where the electromagnetic parameters include a winding phase resistance of a stator node;

[0104] Based on the first loss mapping relationship, a first mapping relationship is constructed, where the first coefficient includes a first parameter, a second parameter, a third parameter, and a fourth parameter;

[0105] The copper loss is the sum of the first product, the second product, the third product and the fourth product. The first product is the product of the motor d-axis current and the first parameter, the second product is the product of the motor q-axis current and the second parameter, the third product is the product of the motor d-axis current, the stator node temperature and the third parameter, and the fourth product is the product of the motor q-axis current, the stator node temperature and the fourth parameter.

[0106] It should be understood that the first loss mapping relationship between the motor copper loss and the motor's electromagnetic parameters and motor current can obtain the mathematical relationship formula between the motor copper loss and the motor's electromagnetic parameters and motor current, which is another characterization formula for the motor copper loss.

[0107] It is also necessary to understand that based on the first loss mapping relationship, a first mapping relationship is constructed. That is, the electromagnetic parameters and / or fixed parameters in each formula item in the characterization formula of the copper loss corresponding to the first loss mapping relationship are merged and simplified into one parameter, that is, the parameters in each formula item except the temperature of the target node and the motor current are merged and simplified into one parameter. The characterization formula of copper loss includes multiple formula items, and each formula item corresponds to a merged parameter. For example, the first product is the product of the motor d-axis current and the first parameter, and the first parameter is the electromagnetic parameters and / or fixed parameters in the formula item where the first product is located, which are merged and simplified into a parameter.

[0108] For example, based on the above formulas (3) to (6), the characterization formula corresponding to the first loss mapping relationship between the motor copper loss and the electromagnetic parameters and motor current of the motor is:

[0109]

[0110] The characterization formula based on the above copper loss can be simplified into a formula containing four formula items. The electromagnetic parameters and / or fixed parameters in each formula item are combined and simplified into one parameter, and then a first mapping relationship can be constructed based on the first loss mapping relationship. The characterization formula corresponding to the first mapping relationship is:

[0111]

[0112] ∝k1i d 2 +k2i q 2 +k3T s i d 2 +k4T s i q 2 (9);

[0113] Wherein, k1 refers to the first parameter, k2 refers to the second parameter, k3 refers to the third parameter, and k4 refers to the fourth parameter. It can be understood that this application simplifies the copper loss formula and avoids using motor parameters as much as possible to reduce the complexity of the model and increase the generalization ability of the model.

[0114] In some embodiments, obtaining a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and the second coefficient includes:

[0115] Obtaining a second loss mapping relationship between the motor iron loss at the stator node and the electromagnetic parameters and angular velocity of the motor, where the electromagnetic parameters include a hysteresis loss coefficient, an eddy current loss coefficient, and a magnetic flux at the stator node;

[0116] Based on the second loss mapping relationship, a second mapping relationship is constructed, where the second coefficient includes a fifth parameter and a sixth parameter;

[0117] The iron loss is the sum of the fifth product and the sixth product. The fifth product is the product of the motor d-axis current, the motor angular velocity, and the fifth parameter. The sixth product is the product of the motor q-axis current, the motor angular velocity, and the sixth parameter.

[0118] It should be understood that the first loss mapping relationship between the iron loss of the stator node and the electromagnetic parameters and angular velocity of the motor can obtain the mathematical relationship formula between the iron loss of the stator node and the electromagnetic parameters and angular velocity of the motor, which is another characterization formula of the iron loss of the stator node.

[0119] It is also necessary to understand that based on the second loss mapping relationship, a second mapping relationship is constructed. That is, the electromagnetic parameters and / or fixed parameters in each formula item in the characterization formula of the iron loss corresponding to the second loss mapping relationship are merged and simplified into one parameter, that is, the parameters in each formula item except the motor current and the motor angular velocity are merged and simplified into one parameter. The characterization formula of iron loss is a polynomial, then each formula item corresponds to a merged parameter. For example, the fifth product is the product between the motor d-axis current, the motor angular velocity, and the fifth parameter, then the fifth parameter is the electromagnetic parameters and / or fixed parameters in the formula item where the fifth product is located, which are merged and simplified into a parameter.

[0120] For example, based on the above formulas (3) to (6), the characterization formula corresponding to the second loss mapping relationship between the motor iron loss and the electromagnetic parameters and angular velocity of the motor is:

[0121]

[0122] The characterization formula based on the above iron loss can be simplified into a formula containing two formula items. The electromagnetic parameters and / or fixed parameters in each formula item are combined and simplified into one parameter, and then a second mapping relationship can be constructed based on the second loss mapping relationship. The characterization formula corresponding to the second mapping relationship is:

[0123]

[0124] Wherein, k5 refers to the fifth parameter, and k6 refers to the sixth parameter. It can be understood that this application simplifies the iron loss formula and avoids using motor parameters as much as possible to reduce the complexity of the model and increase the generalization ability of the model.

[0125] In some embodiments, the method further comprises:

[0126] Obtaining a third mapping relationship between the motor mechanical loss, the motor angular velocity, and the third coefficient;

[0127] Constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, the third mapping relationship and the thermal balance model of the target node;

[0128] Based on a preset optimization algorithm and an initial rotor temperature model, optimized values ​​of the first coefficient, the second coefficient, and the third coefficient are determined, and an optimized rotor temperature prediction model is obtained.

[0129] It should be understood that during the motor's loaded operation, the motor will inevitably produce corresponding losses, and the motor rotor temperature rise is mainly affected by copper loss, iron loss, and mechanical loss. The calculation method of copper loss and iron loss is shown in the above formulas (3)-(6), and mechanical loss can include friction loss caused by bearing rotation and loss caused by ventilation. Mechanical loss includes friction loss caused by bearing rotation and loss caused by ventilation, and its calculation formula is:

[0130]

[0131] Where k c Refers to the motor surface roughness coefficient, C f Refers to the motor friction coefficient, ρ air refers to the air density, w m is the motor angular velocity, l is the rotor length, and r is the rotor radius. Based on the interaction between the rotor and stator, formula (11) can approximate the mechanical loss of the stator.

[0132] It should also be understood that, based on a preset optimization algorithm and an initial rotor temperature model, the optimized values ​​of the first coefficient, the second coefficient, and the third coefficient are determined, and an optimized rotor temperature prediction model is obtained. Specifically, the initial rotor temperature model is obtained by bringing the characterization formulas for copper loss, iron loss, and mechanical loss obtained from the first mapping relationship, the second mapping relationship, and the third mapping relationship into the thermal balance model of the target node, thereby obtaining the initial rotor temperature model. Furthermore, based on the preset optimization algorithm, the optimized values ​​of the first coefficient, the second coefficient, and the third coefficient can be determined to obtain an optimized rotor temperature prediction model. Based on the rotor temperature prediction model, as well as the motor current, the motor angular velocity, and the temperature of the target node, the current temperature of the rotor node is predicted.

[0133] It is understandable that the characterization formulas for copper loss, iron loss, and mechanical loss are obtained based on the first mapping relationship, the second mapping relationship, and the third mapping relationship, thereby avoiding the direct use of complex motor parameters, reducing the complexity of the model, and increasing the generalization ability of the model. In addition, the motor inevitably produces corresponding mechanical losses. This application obtains a rotor temperature prediction model based on mechanical loss, copper loss, and iron loss, making the prediction of the current temperature of the rotor node more accurate.

[0134] In some embodiments, obtaining a third mapping relationship between the motor mechanical loss, the motor angular velocity, and the third coefficient includes:

[0135] Obtaining a third loss mapping relationship between the motor mechanical loss and the motor friction parameters, motor ventilation parameters, motor angular velocity, and size parameters of the rotor node, where the motor friction parameters include surface roughness coefficient and friction coefficient, the motor ventilation parameters include air density, and the size parameters include the length and radius of the rotor node;

[0136] Based on the third loss mapping relationship, construct a third mapping relationship, where the third coefficient includes a seventh parameter;

[0137] The motor mechanical loss is the product of the motor angular velocity and the seventh parameter.

[0138] It is necessary to understand that the third loss mapping relationship between the motor mechanical loss and the friction parameters, ventilation parameters, motor angular velocity, and the size parameters of the rotor node can obtain the mathematical relationship formula between the mechanical loss and the friction parameters, ventilation parameters, motor angular velocity, and the size parameters of the rotor node, which is the characterization formula of the mechanical loss.

[0139] It is also important to understand that, based on the third loss mapping relationship, a third mapping relationship is constructed. That is, the friction parameter, ventilation parameter, size parameter, and fixed parameter in the mechanical loss characterization formula corresponding to the third loss mapping relationship are combined and simplified into one parameter. That is, all parameters in the formula except the motor angular velocity are combined and simplified into one parameter.

[0140] For example, based on the above formula (11), the characterization formula corresponding to the third loss mapping relationship between mechanical loss and friction parameter, ventilation parameter, motor angular velocity, and rotor node size parameter is:

[0141]

[0142] By combining and simplifying the remaining parameters in the formula except the motor angular velocity into one parameter, a third mapping relationship can be constructed based on the third loss mapping relationship. The corresponding representation formula of the third mapping relationship is:

[0143]

[0144] In the formula, k7 refers to the seventh parameter. It can be understood that this application simplifies the mechanical loss formula and avoids using motor parameters as much as possible to reduce the complexity of the model and increase the generalization ability of the model.

[0145] Through the above technical solution, the present application simplifies the calculation method of the motor loss, thereby getting rid of the dependence on the inherent parameters of the motor, and thus getting rid of the problem of the rotor temperature prediction model being less effective due to the change of motor parameters with different motor operating scenarios.

[0146] See also Figure 3 As shown, Figure 3 A schematic diagram of another two-node LPTN model provided in an embodiment of the present application. Taking the study of motor rotor temperature prediction in an oil-cooled electric drive as an example, the target node is the stator, and the other nodes are the coolant. Heat is transferred between the stator and rotor, and the stator and rotor also rely primarily on the oil pump coolant for heat dissipation.

[0147] It's important to understand that an oil-cooled electric drive system uses oil as a cooling medium, primarily for the drive motors of new energy vehicles. Compared to traditional air- and water-cooling methods, oil-cooling technology leverages the oil's insulating and thermal conductivity properties to directly cool the motor's internal components (such as the stator and rotor). The oil-cooled electric drive system uses an oil pump to pump cooling oil from the oil pan. After heat exchange with the vehicle's coolant through an oil cooler, the cooled oil is then pumped into the motor, where it directly contacts the stator windings and rotor, removing heat before returning to the oil pan.

[0148] like Figure 3 As shown, the expression corresponding to the thermal balance model of the target node is:

[0149]

[0150] The formula converted into the rotor temperature prediction model is:

[0151]

[0152] The stator loss is calculated as follows:

[0153] P s =P Cu +P Fe +P fr (14);

[0154] It can be understood that the original LPTN model transformed based on the above formula is no longer a closed system, but an open system, so this application no longer focuses on whether the entire system satisfies the conservation of energy, but only requires a single node to satisfy the conservation of energy, that is, to satisfy formula (13). Therefore, the initial rotor temperature model can be constructed based on formula (13) combined with the above formulas (9), (10), (12) and (14), and then based on the preset optimization algorithm and the initial rotor temperature model, the optimized values ​​of the first coefficient and the second coefficient can be determined, and the optimized rotor temperature prediction model can be obtained. Finally, the rotor temperature can be predicted based on the numerical values ​​of the variable data in the initial rotor temperature model. It should be noted that in this application, each parameter is determined based on the mapping relationship, that is, the other parameters except the set variable parameters are merged into one parameter as the coefficient of each public item, and then the stator loss can be directly set as the sum of copper loss, iron loss and mechanical loss. The proportion of loss can be compensated by parameter optimization, so as not to affect the prediction of the rotor temperature.

[0155] It's important to note that the thermal capacitance and thermal resistance in the LPTN model are calculated using numerical algorithms based on motor operating data, often yielding optimal solutions. Conventional methods often produce constant solutions. However, in real-time motor operation, the thermal capacitance and thermal resistance should be time-varying quantities, and the ease of energy transfer should vary under different circumstances. For example, a greater temperature difference between the stator and rotor results in faster energy transfer; a zero temperature difference results in no energy transfer. However, the thermal capacitance and thermal resistance parameters in the LPTN model are often constant, which prevents generalization to all operating conditions.

[0156] In some embodiments, the method further comprises:

[0157] Obtaining a fourth mapping relationship between the heat capacity of the target node, the temperature of the target node, and a fourth coefficient; obtaining a fifth mapping relationship between the thermal resistance between the target node and the rotor node, the temperature of the target node, and a fifth coefficient; and obtaining a sixth mapping relationship between the thermal resistance between the target node and other nodes, the temperature of the target node, the temperature of other nodes, and a sixth coefficient, where the other nodes are nodes in the motor that have a heat exchange relationship with the target node, and the other nodes do not include the rotor node.

[0158] Constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, the third mapping relationship, the fourth mapping relationship, the fifth mapping relationship, the sixth mapping relationship, and a thermal balance model of the target node;

[0159] Based on a preset optimization algorithm and an initial rotor temperature model, optimized values ​​of the first coefficient, the second coefficient, the third coefficient, the fourth coefficient, the fifth coefficient, and the sixth coefficient are determined, and an optimized rotor temperature prediction model is obtained.

[0160] It should be understood that the fourth mapping relationship, the fifth mapping relationship, and the sixth mapping relationship establish a mapping relationship between the heat capacity and the thermal resistance and the corresponding node temperature. That is, the fourth mapping relationship between the heat capacity of the target node and the temperature of the target node and the fourth coefficient can be expressed by a mathematical relationship formula. The fifth mapping relationship between the thermal resistance between the target node and the rotor node and the temperature of the target node and the fifth coefficient can also be expressed by a mathematical relationship formula. The sixth mapping relationship between the thermal resistance between the target node and other nodes and the temperature of the target node, the temperature of other nodes, and the sixth coefficient can also be expressed by a mathematical relationship formula. That is, a characterization formula for obtaining the heat capacity of the target node, the thermal resistance between the target node and the rotor node, and the thermal resistance between the target node and other nodes. The present application makes basic assumptions about thermal resistance and heat capacity, so that they are associated with temperature, rather than being a constant, so that they can be generalized to all working conditions.

[0161] In some embodiments, obtaining a fourth mapping relationship between the heat capacity of the target node, the temperature of the target node, and the fourth coefficient includes:

[0162] Obtaining a first preset functional relationship between the heat capacity of the target node and the temperature of the target node;

[0163] Based on the first preset functional relationship, constructing a fourth mapping relationship, where the fourth coefficient includes an eighth parameter;

[0164] The heat capacity of the target node is the product of the temperature of the target node and the eighth parameter.

[0165] It should be understood that in order to make the heat capacity of the target node and the temperature of the target node correlated, an assumption can be made on the relationship between the heat capacity of the target node and the temperature of the target node, thereby obtaining a first preset functional relationship between the heat capacity of the target node and the temperature of the target node. The first preset functional relationship can adopt a linear functional relationship, a nonlinear functional relationship, etc. It is obtained specifically according to the actual implementation plan requirements. For example, a linear relationship is used to represent the first preset functional relationship between the heat capacity of the target node and the temperature of the target node, then the heat capacity of the target node and the temperature of the target node satisfy a linear relationship. Based on the first preset functional relationship between the heat capacity of the target node and the temperature of the target node, a fourth mapping relationship can be constructed, and the characterization formula corresponding to the fourth mapping relationship is:

[0166] C s ≈k8T s (15);

[0167] Wherein, k8 refers to the eighth parameter.

[0168] In some embodiments, obtaining a fifth mapping relationship between the thermal resistance between the target node and the rotor node, the temperature of the target node, and the fifth coefficient includes:

[0169] Obtaining a second preset functional relationship between the thermal resistance between the target node and the rotor node and the temperature of the target node;

[0170] Based on the second preset functional relationship, constructing a fifth mapping relationship, wherein the fifth coefficient includes a ninth parameter;

[0171] The thermal resistance between the target node and the rotor node is the product of the temperature of the target node and the ninth parameter.

[0172] It should be understood that in order to correlate the thermal resistance between the target node and the rotor node with the temperature of the target node, an assumption can be made about the relationship between the thermal resistance between the target node and the rotor node and the temperature of the target node, thereby obtaining a second predetermined functional relationship between the thermal resistance between the target node and the rotor node and the temperature of the target node. The second predetermined functional relationship can be a linear functional relationship, a nonlinear functional relationship, or the like, and is determined based on the actual implementation requirements.

[0173] For example, a linear relationship is used to represent the second preset functional relationship between the thermal resistance between the target node and the rotor node and the temperature of the target node. Then, the thermal resistance between the target node and the rotor node and the temperature of the target node satisfy a linear relationship. Furthermore, a fifth mapping relationship can be constructed based on the second preset functional relationship between the thermal resistance between the target node and the rotor node and the temperature of the target node. The characterization formula corresponding to the fifth mapping relationship is:

[0174] R sr ≈k9T s (16);

[0175] Wherein, k9 refers to the ninth parameter.

[0176] It should be noted that the thermal resistance between the target node and the rotor node should be related to the temperature of the target node and the temperature of the rotor node. However, in general, the stator temperature is higher than the rotor temperature. Therefore, it is believed that the heat transfer between the two is mainly dominated by the stator. Therefore, it is assumed that the thermal resistance between the target node and the rotor node is only related to the temperature of the target node, and the rotor temperature is removed, so that the calculation of formula (13) can be simplified, and there is no rotor temperature on the right side of formula (13).

[0177] In some embodiments, obtaining a sixth mapping relationship between the thermal resistance between the target node and the other nodes and the temperature of the target node, the temperature of the other nodes, and the sixth coefficient includes:

[0178] Obtaining a third preset functional relationship between the thermal resistance between the target node and the other nodes, the temperature of the target node, and the temperatures of the other nodes;

[0179] Based on the third preset functional relationship, a sixth mapping relationship is constructed, where the sixth coefficient includes a tenth parameter and an eleventh parameter;

[0180] The thermal resistance between the target node and the other nodes is the sum of the seventh product and the eighth product. The seventh product is the product of the temperature of the other nodes and the tenth parameter. The eighth product is the product of the temperature of the target node and the eleventh parameter.

[0181] It should be understood that in order to make the thermal resistance between the target node and other nodes correlate with the temperature of the target node and the temperature of other nodes, an assumption can be made about the relationship between the thermal resistance between the target node and other nodes, the temperature of the target node, and the temperature of other nodes, thereby obtaining a third preset functional relationship between the thermal resistance between the target node and other nodes, the temperature of the target node, and the temperature of other nodes. The third preset functional relationship can adopt a linear functional relationship, a nonlinear functional relationship, etc., and is specifically obtained according to the actual implementation requirements.

[0182] For example, a linear relationship is used to represent the third preset functional relationship between the thermal resistance between the target node and the other nodes and the temperature of the target node and the temperature of the other nodes. Then, the thermal resistance between the target node and the other nodes and the temperature of the target node and the temperature of the other nodes satisfy a linear relationship. Then, based on the third preset functional relationship between the thermal resistance between the target node and the other nodes and the temperature of the target node and the temperature of the other nodes, a sixth mapping relationship can be constructed. The characterization formula corresponding to the sixth mapping relationship is:

[0183] R cs ≈k 10 T c +k 11 T s (17);

[0184] Where k 10 Refers to the tenth parameter, k 11 Refers to the eleventh parameter.

[0185] Through the above technical solution, this application makes a basic assumption about thermal resistance and heat capacity, rather than a constant. This application assumes a linear relationship between thermal resistance, heat capacity, and the stator, rotor, and coolant. This assumption is relatively simple and facilitates calculations, and other more reasonable and complex relationships can be used instead.

[0186] In summary, based on formula (13), the rotor temperature can be predicted based on the motor current, motor angular velocity, target node temperature, other node temperatures, etc. Specifically, the rotor temperature can be predicted by the motor current, motor angular velocity, stator temperature, stator temperature change rate, other node temperatures, etc. The stator temperature change rate is determined based on the sampling time interval and the stator temperature value. The smaller the sampling time interval Δt, the closer the stator temperature Ts is to the stator temperature change rate. As a result, the shorter the sampling time interval, the stator temperature change rate approaches the derivative of temperature with respect to time. In addition, there are 11 parameters in formula (13) that can be determined based on the preset optimization algorithm and the initial rotor temperature model.

[0187] In some embodiments, the preset optimization algorithm is the Levenberg-Marquardt algorithm.

[0188] It should be understood that this application uses a nonlinear least squares method to estimate the optimal values ​​of the 11 parameters in equation (13). The goal is to adjust the 11 parameters so that the mean square error between the predicted rotor temperature and the actual rotor temperature is minimized. This application selects the Levenberg-Marquardt algorithm for iterative optimization, which combines the advantages of gradient descent and Newton's method and is suitable for optimization problems of nonlinear functions.

[0189] Specifically, before solving the optimized values ​​of each parameter, the corresponding data needs to be acquired. First, raw data acquisition is performed to obtain data from the vehicle's onboard sensors. This data is synchronously sampled at a certain frequency (in this example, 1 second). Each piece of sampled data can include the sampling time, the temperature of the target node, the motor d-axis current feedback, the motor q-axis current, the motor angular velocity, the temperature of the rotor node, the temperature of other nodes, and other information required by the rotor temperature prediction model. The sampling time interval can be determined based on the sampling time. In order to collect as much data as possible, data from bench tests and data from actual road driving can be collected. Then, driving data acquisition is performed. Considering that the raw data includes data from all driving states of the vehicle, since the motor rotor exchanges more heat with the environment when the motor is powered off, which is different from the heat exchange situation when the motor is powered on, in order to reduce the complexity of rotor temperature prediction, this part of the heat exchange can be ignored. In addition, during real-world driving, the rotor temperature often reaches its maximum value during driving, and the rotor high temperature scenario is often the focus of attention. Therefore, the heat dissipation process of the rotor after power is off is ignored in this application. Furthermore, this application only studies the data when the motor is powered on. The process of filtering all driving data from the raw data is as follows: first, the data obtained when the motor is operating is filtered out using the power-on signal. Then, the motor speed signal and actual torque feedback signal are used to filter out data when the motor is running, that is, data with a non-zero speed or a non-zero absolute torque value. Through these processes, the raw data is transformed into a series of continuous driving segments, which serve as the basis for subsequent model parameter optimization and model effectiveness verification.

[0190] Furthermore, based on the extracted vehicle driving data, the initial rotor temperature model is constructed according to formula (13) combined with formulas (9), (10), (12), (15), (16), and (17), and the sampling time t, stator temperature Ts, coolant temperature Tc, and motor speed w are converted into m 、Motor d-axis current i d 、Motor q-axis current i q A total of 6 features are used as input features, and the rotor temperature T r As the target output. Then, based on the preset optimization algorithm and the initial rotor temperature model, the optimized values ​​of the first coefficient, the second coefficient, the fourth coefficient, the fifth coefficient and the sixth coefficient are determined, and the optimized rotor temperature prediction model is obtained.

[0191] Specifically, the curve_fit function in the scipyoptimize library can be selected for parameter optimization. The input parameters of the curve_fit function mainly include the model function, independent variable data, target data, parameter initial values, optimization method, etc. In this application, the input model function of this function is formula (13); the independent variable data is a series of continuous driving segment data corresponding to the above 6 input features, and the dimension of the independent variable data is the number of segments * segment length * 6; the target data is the rotor temperature data corresponding to the 6 input features; the initial parameter values ​​can be selected such that k1-k11 are all 1, and the optimization method is the Levenberg-Marquardt method. It should be noted that the value of the initial parameter value will not have much impact. The internal optimization process of the curve_fit function refers to the Levenberg-Marquardt algorithm process.

[0192] Finally, the curvefit function outputs the optimized parameters and their covariances. The 11 parameters finally optimized can be used to replace the coefficients in the initial rotor temperature model, that is, to replace the coefficients in the initial rotor temperature model constructed based on formula (13) combined with formulas (9), (10), (12), (15), (16), and (17) to obtain the optimized rotor temperature prediction model. This allows rotor temperature prediction to be performed based on the rotor temperature prediction model and real-time data such as sampling time, motor current, motor angular velocity, target node temperature, and other node temperatures.

[0193] It's important to understand that curve_fit is a function in the SciPy library that performs nonlinear least-squares fitting. It helps us find a set of parameters that makes the rotor temperature prediction model predict the current rotor node temperature as close as possible to the corresponding actual value.

[0194] This application proposes a method for predicting the temperature of a motor rotor. It innovates based on the LPTN model and transforms a closed system into an open system by focusing on the heat transfer mechanism of a single node. The rotor temperature is predicted using the stator temperature and the coolant temperature. At the same time, a dynamic correlation assumption is made for the heat capacity and thermal resistance parameters, and the calculation method of the motor heat loss is simplified, thus getting rid of the dependence on the inherent parameters of the motor. In addition, this application only optimizes and solves the single-node equation, reducing the number of equations and parameters, reducing the complexity of the model, simplifying the loss calculation and not relying on the volatile motor parameters, thereby improving the model's generalization ability, enhancing robustness and facilitating deployment. In terms of error control, the traditional model is prone to error accumulation because the temperature prediction at adjacent moments depends on historical values. However, this application directly calculates the rotor temperature based on the stator temperature, stator temperature change rate and coolant temperature at the current moment, cutting off the dependency between the predicted values, effectively avoiding error transmission and accumulation, and achieving more accurate and stable rotor temperature prediction.

[0195] See also Figure 4 As shown, Figure 4 This is a schematic diagram of an experimental verification of an embodiment of the present application. This application uses two oil-cooled motor vehicles to collect data through multiple days of bench testing and road testing. One day's data from one vehicle is selected as the training set for parameter estimation of the present model, and the remaining days' data from both vehicles are used for performance testing.

[0196] The experimental working conditions of this application include all working conditions with a speed of 15000r / min and a torque below 300N·m. The speed is divided into levels every 2000r / min and the absolute value of the torque is divided into levels every 50. Therefore, all working condition levels include 8*6=48 levels. The road test experiment is the driving data on the real road, including urban roads, viaducts, highways and other road conditions. The bench test is more about testing the motor data under different working conditions, including high speed and high torque, high speed and low torque, low speed and low torque, low speed and high torque, etc.

[0197] In this example, after selecting data from a certain day for model parameter estimation, the results of the other 16 days are shown in Table 1. Table 1 shows the results of the road test and bench test for the other 16 days. The performance indicator is MSE, which is the mean square error between the predicted value and the true value. From Table 1, we can see that the MSE of both the road test set and the bench test set is less than 10, and the MSE of the bench test data is slightly larger because its working conditions are relatively poor. In order to show the model effect in more detail, data under poor working conditions are selected here. Figure 4 As shown, the green curve is the stator temperature, which reaches a maximum of 140 degrees, the red curve is the actual rotor temperature, which reaches a maximum of 80 degrees, the yellow curve is the rotor predicted value, and the purple curve is the oil pump temperature value, which is the minimum value. Figure 4It can be seen that even when the temperature is high, the model can make accurate predictions, and the error does not exceed 5 degrees.

[0198] Table 1

[0199]

[0200]

[0201] It should be noted that the embodiment of the present application predicts the rotor temperature through the energy transfer equation of the stator node on the LPTN model, which only involves the energy transfer between the stator, rotor and coolant. In fact, the number of nodes can be increased to improve the accuracy of the model, such as increasing the casing temperature, which also has energy transfer with the stator and rotor. In addition, the present application assumes that there is a linear relationship between thermal resistance, heat capacity and the stator, rotor and coolant, and this assumption is relatively simple and convenient for calculation. Other more reasonable and complex relationships can also be used instead, such as nonlinear relationships, statistical relationships, etc. It should also be noted that the present application is mainly aimed at the rotor temperature prediction model constructed for oil-cooled electric drive, but it is also applicable to other types such as air-cooled and water-cooled electric drive.

[0202] In a second aspect, an embodiment of the present application further provides a motor rotor temperature prediction device, comprising:

[0203] A first module is configured to obtain a first mapping relationship between the motor copper loss and the temperature of a target node, the motor current, and a first coefficient, and a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and a second coefficient, wherein the target node is a node in the motor that has a heat exchange relationship with the rotor node;

[0204] A second module is configured to construct an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, and a thermal balance model of the target node;

[0205] a third module, configured to determine optimized values ​​of the first coefficient and the second coefficient based on a preset optimization algorithm and the initial rotor temperature model, and to obtain an optimized rotor temperature prediction model;

[0206] The fourth module is used to predict the current temperature of the rotor node based on the rotor temperature prediction model, the motor current, the motor angular velocity, the temperature of the target node, and the temperature of other nodes, where the other nodes are nodes in the motor other than the rotor node that have a heat exchange relationship with the target node.

[0207] It should be noted that the motor rotor temperature prediction device provided in the embodiment of the present application belongs to the same concept as the motor rotor temperature prediction method in the above embodiment. Any method provided in the motor rotor temperature prediction method embodiment can be run on the motor rotor temperature prediction device. The specific implementation process is detailed in the motor rotor temperature prediction method embodiment, which will not be repeated here.

[0208] An embodiment of the present application further provides a storage medium storing a computer program. When the computer program runs on a computer, the computer executes the method in any of the above embodiments.

[0209] In the embodiment of the present application, the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0210] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0211] The above is only a preferred embodiment of the present application and does not constitute any form of limitation to the present application. Although the present application has been disclosed as above with preferred embodiments, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present application. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present application without departing from the content of the technical solution of the present application are still within the scope of the technical solution of the present application.

Claims

1. A method for predicting motor rotor temperature, characterized in that: include: Obtaining a first mapping relationship between the motor copper loss and the temperature of a target node, the motor current, and a first coefficient, and a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and a second coefficient, wherein the target node is a node in the motor that has a heat exchange relationship with the rotor node; constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, and a thermal balance model of the target node; Based on a preset optimization algorithm and the initial rotor temperature model, determining optimized values ​​of the first coefficient and the second coefficient, and obtaining an optimized rotor temperature prediction model; Based on the rotor temperature prediction model, as well as the motor current, the motor angular velocity, the temperature of the target node, and the temperature of other nodes, the current temperature of the rotor node is predicted, where the other nodes are nodes in the motor other than the rotor node that have a heat exchange relationship with the target node.

2. The motor rotor temperature prediction method according to claim 1, characterized in that: The target node is a stator node, and the motor current includes a motor d-axis current and a motor q-axis current; The obtaining of a first mapping relationship between the motor copper loss and the temperature of the target node, the motor current, and the first coefficient includes: Acquire a first loss mapping relationship between the motor copper loss and the electromagnetic parameters of the motor and the motor current, wherein the electromagnetic parameters include the winding phase resistance of the stator node; Based on the first loss mapping relationship, construct the first mapping relationship, where the first coefficient includes a first parameter, a second parameter, a third parameter, and a fourth parameter; The copper loss is the sum of a first product, a second product, a third product and a fourth product, wherein the first product is the product of the motor d-axis current and the first parameter, the second product is the product of the motor q-axis current and the second parameter, the third product is the product of the motor d-axis current, the temperature of the stator node and the third parameter, and the fourth product is the product of the motor q-axis current, the temperature of the stator node and the fourth parameter.

3. The motor rotor temperature prediction method according to claim 2, characterized in that: The obtaining of a second mapping relationship between the motor iron loss and the motor current, the motor angular velocity, and the second coefficient includes: Obtaining a second loss mapping relationship between the motor iron loss and the electromagnetic parameters of the motor and the motor angular velocity, wherein the electromagnetic parameters include a hysteresis loss coefficient, an eddy current loss coefficient, and a magnetic flux of the stator node; Based on the second loss mapping relationship, construct the second mapping relationship, where the second coefficient includes a fifth parameter and a sixth parameter; The iron loss is the sum of the fifth product and the sixth product, the fifth product is the product of the motor d-axis current, the motor angular velocity, and the fifth parameter, and the sixth product is the product of the motor q-axis current, the motor angular velocity, and the sixth parameter.

4. The motor rotor temperature prediction method according to claim 1, characterized in that: Also includes: Acquire a third mapping relationship between the motor mechanical loss, the motor angular velocity, and a third coefficient; constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, the third mapping relationship, and a thermal balance model of the target node; Based on a preset optimization algorithm and the initial rotor temperature model, optimized values ​​of the first coefficient, the second coefficient, and the third coefficient are determined, and an optimized rotor temperature prediction model is obtained.

5. The motor rotor temperature prediction method according to claim 4, characterized in that: The obtaining of a third mapping relationship between the motor mechanical loss, the motor angular velocity, and the third coefficient includes: Obtaining a third loss mapping relationship between the motor mechanical loss and a motor friction parameter, a motor ventilation parameter, the motor angular velocity, and a size parameter of the rotor node, wherein the motor friction parameter includes a surface roughness coefficient and a friction coefficient, the motor ventilation parameter includes an air density, and the size parameter includes a length of the rotor node and a radius of the rotor node; Based on the third loss mapping relationship, construct the third mapping relationship, wherein the third coefficient includes a seventh parameter; The motor mechanical loss is the product of the motor angular velocity and the seventh parameter.

6. The method for predicting motor rotor temperature according to claim 1, wherein: Also includes: Obtaining a fourth mapping relationship between the heat capacity of the target node, the temperature of the target node, and a fourth coefficient, a fifth mapping relationship between the thermal resistance between the target node and the rotor node, the temperature of the target node, and a fifth coefficient, and a sixth mapping relationship between the thermal resistance between the target node and other nodes, the temperature of the target node, the temperature of the other nodes, and a sixth coefficient; constructing an initial rotor temperature model based on the first mapping relationship, the second mapping relationship, the third mapping relationship, the fourth mapping relationship, the fifth mapping relationship, the sixth mapping relationship, and a thermal balance model of the target node; Based on a preset optimization algorithm and the initial rotor temperature model, optimized values ​​of the first coefficient, the second coefficient, the third coefficient, the fourth coefficient, the fifth coefficient and the sixth coefficient are determined, and an optimized rotor temperature prediction model is obtained.

7. The motor rotor temperature prediction method according to claim 6, characterized in that: The acquiring of a fourth mapping relationship between the heat capacity of the target node, the temperature of the target node, and a fourth coefficient includes: Acquire a first preset functional relationship between the heat capacity of the target node and the temperature of the target node; Based on the first preset functional relationship, constructing the fourth mapping relationship, the fourth coefficient includes an eighth parameter; The heat capacity of the target node is a product of the temperature of the target node and the eighth parameter.

8. The motor rotor temperature prediction method according to claim 6, characterized in that: The obtaining of a fifth mapping relationship between the thermal resistance between the target node and the rotor node, the temperature of the target node, and a fifth coefficient includes: Obtaining a second preset functional relationship between the thermal resistance between the target node and the rotor node and the temperature of the target node; Based on the second preset functional relationship, constructing the fifth mapping relationship, the fifth coefficient includes a ninth parameter; The thermal resistance between the target node and the rotor node is the product of the temperature of the target node and the ninth parameter.

9. The method for predicting the motor rotor temperature according to claim 6, wherein: The obtaining of a sixth mapping relationship between the thermal resistance between the target node and the other nodes, the temperature of the target node, the temperature of the other nodes, and the sixth coefficient includes: Obtaining a third preset functional relationship between the thermal resistance between the target node and other nodes, the temperature of the target node, and the temperature of the other nodes; Based on the third preset functional relationship, constructing the sixth mapping relationship, the sixth coefficient includes a tenth parameter and an eleventh parameter; The thermal resistance between the target node and the other nodes is the sum of the seventh product and the eighth product, the seventh product is the product of the temperature of the other nodes and the tenth parameter, and the eighth product is the product of the temperature of the target node and the eleventh parameter.

10. The motor rotor temperature prediction method according to claim 1, characterized in that: The preset optimization algorithm is the Levenberg-Marquardt algorithm.

Images