Scholarlink

This paper proposes a method that involves establishing a neural network model for the motor thermal management system, using the neural network model as the time-domain prediction model in the model predictive control(MPC) algorithm, and employing the MPC algorithm to optimize the thermal management control strategy, aiming to reduce the system’s energy consumption.This method first utilizes a Long Short Term Memory (LSTM)neural network, which takes the motor heat generation, pump speed, motor outlet temperature, motor temperature, and motor outlet temperature as inputs, and the next motor outlet temperature as output. Then, the optimal pump speed is solved using Particle Swarm Optimization (PSO) algorithm to reduce system energy consumption while maintaining stable control performance. This method does not require the detailed thermal management circuit parameters and physical model required when applying the state space model, avoiding errors that may occur in the later stages of the model due to component aging, changes in electrical component performance, and other factors, thereby affecting the control strategy of the model predictive control algorithm.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on Optimization of Thermal Management Control Strategy for Motor System Based on LSTM Network Prediction Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zeyu Li,
  • Yu Fu,
  • Fengjian Wang

摘要

This paper proposes a method that involves establishing a neural network model for the motor thermal management system, using the neural network model as the time-domain prediction model in the model predictive control(MPC) algorithm, and employing the MPC algorithm to optimize the thermal management control strategy, aiming to reduce the system’s energy consumption.This method first utilizes a Long Short Term Memory (LSTM)neural network, which takes the motor heat generation, pump speed, motor outlet temperature, motor temperature, and motor outlet temperature as inputs, and the next motor outlet temperature as output. Then, the optimal pump speed is solved using Particle Swarm Optimization (PSO) algorithm to reduce system energy consumption while maintaining stable control performance. This method does not require the detailed thermal management circuit parameters and physical model required when applying the state space model, avoiding errors that may occur in the later stages of the model due to component aging, changes in electrical component performance, and other factors, thereby affecting the control strategy of the model predictive control algorithm.