一种战车主减速器温度预测方法研究

目的针对战车主减速器温度预测需求,建立时间序列ARIMA多步预测和BP神经网络预测模型,提出基于BP神经网络修正误差的ARIMA模型温度预测方法。方法结合BP神经网络的非线性能力与ARIMA模型预测能力,分析ARIMA在多步预测时误差产生原因,在神经网络对ARIMA多步误差进行预测基础上计算修正因子,把误差修正因子和BP网络结合,实现对多步预测误差的修正。结果ARIMA模型多步预测时,预测误差随预测步数的逐步增加不断增大,引入了误差修正因子进行修正。通过预测值与实际值进行对比,可有效提高预测准确度。结论 BP神经网络和误差修正因子结合应用可显著提高温度预测效果。

Research on Temperature Prediction Method of Main Reducer for Chariot

Objective Temperature prediction of main reducer is demanded.Methods It establish model of ARIMA multi-step prediction based on time series and BP neural network prediction.It put forward temperature prediction method of ARIMA model based on BP neural network correcting errors. Results The method combined nonlinear ability of BP neural network with prediction ability of ARIMA model.It analyze error caused when ARIMA multi-step prediction was used. Error correction factors is calculated based on neural network predicting ARIMA multi-step error. Multi-step prediction error is corrected by combing BP neural network with error correction factors. Conclusion Prediction accuracy is improved by comparison predictive value with actual value.