基于循環(huán)神經(jīng)網(wǎng)絡(luò)的工業(yè)機(jī)器人能耗預(yù)測(cè)方法研究

打開文本圖片集

關(guān)鍵詞：工業(yè)機(jī)器人；能耗預(yù)測(cè)；神經(jīng)網(wǎng)絡(luò)中圖分類號(hào)：U414 文獻(xiàn)標(biāo)志碼：ADOI:10.19968/j.cnki.hnkj.1003-5168.2025.14.005文章編號(hào):1003-5168(2025)14-0028-04

Research on Energy Consumption Prediction Method for Industrial RobotsBased on RecurrentNeural Network

YANG Xue (Chongqing Polytechnic University of Electronic Technology, Chongqing 401331, China)

Abstract: [Purposes] To address the issues of high energy consumption and low energy efficiency in industrial robots and achieve energy consumption optimization,an energy consumption prediction model for industrial robots based on a recurrent neural network integrating Long Short-Term Memory (LSTM) and Attention mechanism is proposed.[Methods] First，based on the analysis of energy consumption sources including motor iron loss,copper loss,and joint friction,the total power model and total energy consumption formula of industrial robots were established. Second，the parameters of the LSTMAttention model were configured,with Mean Absolute Error( MAE )，Mean Absolute Percentage Error (MAPE),and Root Mean Square Error (RMSE) employed as evaluation metrics.Finally,the operation data of the robot were collcted,and experiments were conducted using 5-fold cross-validation.[Findings] The experimental results showed that the RMSE was 989.52, the MAE was725.15,andtheMAPE was 6.161% . The relative error of the model's prediction results was low,indicating that it could ffectively predict the energy consumption of industrial robots.[Conclusions] The LSTM-Attention model

收稿日期：2025-04-07

demonstrates significant advantages in energy consumption prediction, which can effectively reduce the prediction errors and improve the prediction accuracy,thus providing substantial support for subsequent energy consumption research.

Keywords: industrial robots; energy consumption optimization; neural network

0 引言

工業(yè)機(jī)器人作為制造系統(tǒng)中的關(guān)鍵裝備，在現(xiàn)代制造業(yè)中發(fā)揮著重要作用。（剩余6778字）