加热炉炉温预测可以确保在正常生产过程中炉温和钢坯温度的稳定性,降低能耗,这对于提高生产效率和优化能源利用具有重要意义.针对加热炉炉温控制耦合参数众多,温度控制受到各种干扰因素的影响,其变化具有复杂的非线性特征,且响应速度较慢、惯性较大的问题,采用加热炉系统的历史生产数据,结合投影小波变换加权孪生支持向量机(projection wavelet weighted twin support vector regression,PWWTSVR)构建了预测加热炉炉温的模型.在建立预测模型的过程中,根据从某钢厂采集到的实际生产数据,将950组数据作为模型的训练数据,将50组数据用来测试模型的准确性.结果表明,在±0.25 ℃的误差容限内,PWWTSVR模型的预测准确率达到98％,优于反向神经网络(back propagation,BP)模型和孪生支持向量机(twin support vector regression,TSVR)模型,因此提出的加热炉炉温预测模型能够更准确地预测加热炉的温度变化,便于决策者决策.
Furnace temperature prediction of heating furnace based on projection wavelet weighted twin support vector regression
The prediction of furnace temperature can ensure the stability of furnace temperature and billet temperature in normal production process and reduce energy consumption,which is of great sig-nificance for improving production efficiency and optimizing energy utilization.Aiming at the prob-lems that the coupling parameters of heating furnace temperature control are numerous,the tempera-ture control is affected by various interference factors,the change has complex nonlinear characteris-tics,the response speed is slow,and the inertia is large,based on the historical production data of the heating furnace system,a projection wavelet weighted twin support vector regression(PWWTSVR)model for predicting heating furnace temperature was constructed.In the process of establishing the prediction model,according to the actual production data collected from a steel mill,950 sets of data are used as the training data of the model,and 50 sets of data are used to test the accuracy of the model.The results show that,within the error tolerance of±0.25 ℃,the prediction accuracy of PW-WTSVR model reaches 98％,which is better than back propagation(BP)model and twin support vector regression(TSVR)model.Therefore,the proposed model can predict the temperature change of heating furnace more accurately,which is convenient for decision makers.
steel rollingheating furnacefurnace temperature predictionprojection wavelet weighted twin support vector regression(PWWTSVR)
万方数据
维普
