A neural network-based method for predicting the temperature of a cooling system component body

Abstract

The application provides a kind of based on neural network cooling system component body temperature prediction method, comprising: in step S1, determine associated parameters;In step S2, data grouping processing is carried out, and is divided into training set and verification set;In step S3, using the tool with neural network modeling ability and defining its input layer, output layer and hidden layer, the final neural network model is derived by fitting training;Step S4: the final neural network model is applied to predict the thermal system component body temperature under target working condition;The application can significantly improve the prediction efficiency of cooling system component body temperature, improve prediction accuracy and stability.

Description

Technical Field

[0001] This invention relates to the field of automotive thermal management technology, and more specifically to a method for predicting the body temperature of cooling system components based on neural networks. Background Technology

[0002] As the number of vehicles on the road increases and user scenarios expand, vehicles not only need to meet the requirements of regular road driving, but also need to adapt to harsh driving conditions such as off-road driving, long-term hill climbing under full load, and continuous acceleration and deceleration. Under these conditions, whether the powertrain can continuously output the required power is closely related to whether the cooling system can effectively control the temperature of the powertrain components. One of the main reasons for limited powertrain output is the overheating of powertrain components. Therefore, accurately predicting the temperature of powertrain components is of great significance for evaluating cooling performance under extreme conditions and for developing and optimizing thermal management strategies.

[0003] In existing technologies, OEMs typically lack detailed physical and thermal parameters of the powertrain system, often relying solely on the target inlet coolant temperature required by the supplier. This approach has significant limitations, making it difficult to accurately predict the temperature of powertrain components. Effective prediction is particularly challenging for powertrain temperatures in complex scenarios such as sand driving and mud driving, including engine coolant temperature and the temperature of power devices within the motor controller. Since powertrain temperature is strongly correlated with cooling system strategies, inaccurate predictions can lead to overcooling under certain operating conditions and insufficient cooling under others, ultimately affecting power output, thermal safety, and overall vehicle performance. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method for predicting the body temperature of cooling system components based on neural networks. This method can accurately predict the body temperature of thermal system components under different operating conditions, thereby improving temperature prediction efficiency and operating condition adaptability, and providing effective support for the formulation of thermal management strategies.

[0005] To achieve the above-mentioned technical effects, the present invention adopts the following technical solution: a method for predicting the body temperature of cooling system components based on neural networks, comprising the following steps:

[0006] In step S1, the associated parameters are determined; the associated parameters are factors that affect the heat exchange state, heating state or heat load of the target component, as well as other factors that affect the heat transfer process.

[0007] In step S2, data grouping processing is performed, including:

[0008] S21: The correlation parameters obtained in step S1 and the preset target component body temperature or target temperature data are divided into mutually independent data sets according to the preset sample size requirements.

[0009] S22: Divide the dataset obtained in step S21 into a training set and a validation set;

[0010] In step S3, neural network algorithm fitting is performed, including:

[0011] S31: Use a tool with neural network modeling capabilities and define its input layer, output layer, and hidden layer;

[0012] S32: Fit training is performed on the hidden layers; wherein, the hidden layers include the number of hidden layers, the neural network algorithm, the number of neurons, and the activation function. Based on the set neural network algorithm, the number of neurons, and the activation function, fitting training is performed on different numbers of hidden layers to generate an initial neural network model.

[0013] S33: Use validation set data to verify the accuracy of the initial neural network model; if the model meets the preset accuracy conditions, derive the final neural network model;

[0014] In step S4, the neural network model is applied; the final neural network model is used to predict the body temperature of the thermal system components under the target operating condition.

[0015] Preferably, the associated parameters are three-phase current, cooling water jacket inlet water temperature, and cooling water flow rate.

[0016] Preferably, the number of training sets and validation sets is at least one.

[0017] Preferably, the dataset consists of 5 groups, each containing 3000 samples, and the model training time is 3000 seconds.

[0018] Preferably, the tool with neural network modeling capabilities is Simulink or Amesim.

[0019] Preferably, the number of hidden layers is 4, the neural network algorithm is the Dense algorithm, and the activation function is the ReLU function.

[0020] Preferably, the hidden layer includes a first hidden layer, a second hidden layer, a third hidden layer, and a fourth hidden layer, wherein the first hidden layer is set as the validation set and the third hidden layer is set as the training set.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] 1. By using neural network algorithms to ignore complex heat transfer processes and various boundary factors of the vehicle, there is no need to build a complex and large detailed thermal model, nor is it necessary to rely on complete physical and thermal parameters, which reduces the modeling threshold and development cost.

[0023] 2. It can better predict the body temperature of thermal system components under various complex and extreme working conditions, which is conducive to objectively evaluating the cooling performance and thermal safety margin of the power system;

[0024] 3. It can provide a basis for the formulation and optimization of cooling system strategies, avoid overcooling under some operating conditions and insufficient cooling under other operating conditions, and improve the pertinence and effectiveness of thermal management control. Attached Figure Description

[0025] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0026] Figure 1 This is a flowchart illustrating a method for predicting the body temperature of a cooling system component based on a neural network, as described in the embodiment.

[0027] Figure 2 This is a schematic diagram illustrating the definition of a tool with neural network modeling capabilities in a method for predicting the body temperature of a cooling system component based on a neural network, as described in the embodiment.

[0028] Figure 3 This is a schematic diagram of the hidden layer in a method for predicting the body temperature of a cooling system component based on a neural network, as described in the embodiment.

[0029] Figure 4 This is a schematic diagram of a neuron in a method for predicting the body temperature of a cooling system component based on a neural network, as described in the embodiment.

[0030] Figure 5 This is a comparison chart of the predicted temperature and the measured temperature obtained from the method for predicting the body temperature of a cooling system component based on a neural network as described in the embodiment. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0032] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0033] It should be noted that similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, all directional indications in this application (such as up, down, left, right, front, back, bottom, etc.) are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the figures). If the specific orientation changes, the directional indication will also change accordingly. Furthermore, descriptions involving "first," "second," etc., in this application are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated.

[0034] like Figure 1 - Figure 5 As shown, this embodiment takes motor controller temperature prediction as an example to illustrate the method for predicting the body temperature of cooling system components based on neural networks according to the present invention. The method specifically includes the following steps:

[0035] In step S1, the associated parameters are determined. These associated parameters include direct and indirect parameters. Directly associated parameters are factors that directly affect the heat exchange state, heating state, or heat load of the target component; indirectly associated parameters are factors that affect the operating conditions of the target component, environmental changes, and other factors affecting the heat transfer process. In practical applications, operators can flexibly select the required direct and indirect parameters based on specific operating conditions.

[0036] In this embodiment, the three-phase current, the inlet water temperature of the cooling water jacket, and the cooling water flow rate are defined as directly related parameters.

[0037] In step S2, data grouping processing is performed, mainly involving the following steps:

[0038] Step S21: The correlation parameters obtained in step S1 and the body temperature or target temperature data of the target component preset by the operator are divided into mutually independent data sets according to the preset sample size requirements.

[0039] Step S22: Divide the dataset obtained in step S21 into a training set and a validation set, with at least one training set and one validation set. In practical applications, operators can adjust the preset sample size according to the required prediction accuracy and available computing resources, balancing prediction accuracy and model training efficiency to achieve a reasonable balance between the two.

[0040] In this embodiment, the data related to three-phase current, cooling water jacket inlet water temperature, cooling water flow rate, and IGBT (Insulated Gate Bipolar Transistor) junction temperature are divided into 5 independent datasets, each containing 3000 samples, and the model training time is 3000 seconds.

[0041] In step S3, the neural network algorithm is fitted, which mainly involves the following steps:

[0042] Step S31: Use a tool with neural network modeling capabilities and define its input layer, output layer, and hidden layers;

[0043] Furthermore, tools capable of neural network modeling include simulation modeling software (Simulink) or multiphysics system simulation software (Amesim).

[0044] Step S32: Perform fitting training on the hidden layers; wherein, the hidden layers include the number of hidden layers, the neural network algorithm, the number of neurons, and the activation function. Based on the set neural network algorithm, the number of neurons, and the activation function, perform fitting training on different numbers of hidden layers to generate an initial neural network model.

[0045] Step S33: Use validation set data to verify the accuracy of the initial neural network model; if the model meets the preset accuracy conditions, then derive the final neural network model.

[0046] This embodiment uses Amesim software for neural network algorithm fitting. Four hidden layers are selected: a first hidden layer, a second hidden layer, a third hidden layer, and a fourth hidden layer. The Dense neural network algorithm is used, with the corresponding number of neurons configured and the original data trained and fitted using the ReLU activation function. The first hidden layer is set as the validation set, and the third hidden layer as the training set. Accuracy is verified using the validation set. Once the model meets the preset accuracy, the final neural network model is exported.

[0047] In step S4, the neural network model is applied; the final neural network model is used to predict the body temperature of the thermal system components under the target operating condition.

[0048] In this embodiment, the final neural network model is mainly used to predict the temperature of the motor controller.

[0049] The specific embodiments of the present invention have been described above. Based on the above description, those skilled in the art can make various changes and modifications without departing from the technical concept of the present invention.

Claims

1. A method for predicting the body temperature of cooling system components based on neural networks, characterized in that, Includes the following steps: In step S1, the associated parameters are determined; the associated parameters are factors that affect the heat exchange state, heating state or heat load of the target component, as well as other factors that affect the heat transfer process. In step S2, data grouping processing is performed, including: S21: The correlation parameters obtained in step S1 and the preset target component body temperature or target temperature data are divided into mutually independent data sets according to the preset sample size requirements. S22: Divide the dataset obtained in step S21 into a training set and a validation set; In step S3, neural network algorithm fitting is performed, including: S31: Use a tool with neural network modeling capabilities and define its input layer, output layer, and hidden layer; S32: Fit training is performed on the hidden layers; wherein, the hidden layers include the number of hidden layers, the neural network algorithm, the number of neurons, and the activation function. Based on the set neural network algorithm, the number of neurons, and the activation function, fitting training is performed on different numbers of hidden layers to generate an initial neural network model. S33: Use validation set data to verify the accuracy of the initial neural network model; if the model meets the preset accuracy conditions, derive the final neural network model; In step S4, the neural network model is applied; the final neural network model is used to predict the body temperature of the thermal system components under the target operating condition.

2. The method for predicting the body temperature of cooling system components based on neural networks according to claim 1, characterized in that, The associated parameters are three-phase current, cooling water jacket inlet water temperature, and cooling water flow rate.

3. The method for predicting the body temperature of cooling system components based on neural networks according to claim 1, characterized in that, The number of training sets and validation sets shall be at least one.

4. The method for predicting the body temperature of cooling system components based on neural networks according to claim 1, characterized in that, The dataset consists of 5 groups, each containing 3000 samples, and the model training time is 3000 seconds.

5. The method for predicting the body temperature of cooling system components based on neural networks according to claim 1, characterized in that, The tools with neural network modeling capabilities are Simulink or Amesim.

6. The method for predicting the body temperature of cooling system components based on neural networks according to claim 1, characterized in that, The hidden layers consist of 4 layers, the neural network algorithm is the Dense algorithm, and the activation function is the ReLU function.

7. The method for predicting the body temperature of cooling system components based on neural networks according to claim 6, characterized in that, The hidden layer includes a first hidden layer, a second hidden layer, a third hidden layer, and a fourth hidden layer, wherein the first hidden layer is set as the validation set and the third hidden layer is set as the training set.

Images