模糊系统结合最小二乘的温升预测方法

摘要

针对金具温升过高现象,提出了新的温升预测方法.通过温升试验得到训练数据与测试数据,利用训练数据通过递推最小二乘结合遗传算法的方法对模糊系统进行训练,利用测试数据对训练后的模型进行检验,误差处于合理范围.训练与测试结果说明运用新方法预测金具温升是可行的.通过回归分析对金具温升进行预测,并与新方法进行比较,新方法的预测效果优于传统的回归模型,比较结果体现了新方法在温升预测方面的优势.%For the problem that the rise of temperature of fitting is too high to control , a new method of predicting the rise of temperature has been put forward .The training data and the testing data are obtained from the experiment of rising of temperature .Through training data , the fuzzy system is trained by recursive least square combined genetic algorithm.Then the model is tested by testing data .The error is in reasonable range.The result shows that the new method is feasible to predict the rise of temperature of connection fitting .Regression analysis is used to predict the rise of temperature of connection fitting and the result is compared with that of new model , the result of new model is better than that of traditional regression analysis model , the result of comparison reflects that the new method has advantages for predicting the rise of temperature .