Industrial Temperature Prediction by BP Neural Network Based on Parameter Selection

Accurate prediction of future production state of cement rotary kiln is helpful to smooth operation and reasonable control, and plays an important role in ensuring the stability of production link. The data acquisition system of the process industry has obtained a large number of characteristic data of the production process. Due to the huge amount of data, it contains a variety of interrelated links of production, and different characteristic signals exist in the multi-stage process parameter data at different times. Due to the limitations of traditional methods, analysis and prediction are rarely carried out according to the data at a specific moment in combination with actual industrial production process analysis. Most of them only consider the data analysis, but do not consider the internal mechanism of each link of industrial production. Therefore, combining with the production characteristics of cement rotary kiln, we analyzed the data association relationship among the three processes of decomposing furnace, rotary kiln tail and rotary kiln head. When the process data changes smoothly, we used and predicted the important process parameters. The data segment at the time when the head coal, tail coal, the feeding amount is basically the same and the data is basically stable is intercepted for correlation degree analysis to find the correlation relationship of the process parameters at this time, and the parameters with strong correlation with the measured parameters are used as predictive factors. The experimental results show that compared with the traditional BP neural network prediction method, the accuracy of this method is higher.