Neural Network-based Open Mill Rubber Mixing Quality Online Research Techniques To Predict

In recent years, China has become the largest producer of tires, but, there is a biggap between tire manufacturing powerhouse. The quality of the tire depends on thequality the rubber mixing in a large level, for the open mill mixing, due to backwardtechnology, making it difficult on quality mix of online testing, quality of the mixingrubber compound cannot be obtained in a timely, it cannot be adjustment and control ofthe quality parameters of batch mixing process on the quality of mixing rubbercompound in a timely to obtain the desired,’uniform’ mix quality. In response to thissituation, this paper has been open mill mix quality online prediction research, in orderto achieve open mill mixing quality online testing.The main work of this thesis is to study through the rubber mill quality onlineprediction, to improvise the traditional open mill experimental platform; throughexperiments on open mill mixing process parameters on the quality of mixing, to use ofneural network to establish the relationship between process parameters and mixingquality online prediction model; through the MATLAB programming to forecast mix ofquality; experimental study comparing online forecasting techniques and the use ofSPSS software based on neural network regression analysis, predictive mathematicalmodel comparison; the main contents are as follows:(1) Through the advantages and disadvantages between neural network learningand explore a variety of network analysis, to study the use of BP network on the millmixing quality prediction model establish.(2) In order to meet the use of BP neural network forecasting principles mill mixquality off-line mixing of the forecast, there must be a device can detect the device hasvarious process parameters, so we existing traditional open mill mixing plasticexperimental platform has been improved.(3) On the improvement of the experimental platform for open mill to experimentalstudy for steel radial tire tread rubber formulation.(4) To analyze the process parameters which affecting the quality of the Mix, usingBP neural network to determine the number of neurons in the input layer is4, the number of neurons in the output layer is2, using MATLAB software programming,network training and network error test to determine the number of hidden layerneurons is10when the error is the smallest, and the network convergence is the fastest.(5) Comparative experimental study. Research on the open mill mixing qualityonline prediction based on neural networks and statistical analysis software SPSSstepwise regression method (Stepwise Regression) establish rubber mill mixingmathematical model to predict the quality of a comparison, the results of comparingshow that the use of neural network prediction model predictions closer to the measuredvalues, the predicted Mooney viscosity and carbon black dispersion the relative errorwas0.186and0.388respectively, using SPSS statistical analysis software to establishof mathematical models to predict the Mooney viscosity and the dispersion of carbonblack the mean relative error of1.72and1.242, respectively.Research on the success of the technology, not only for online control and as’homogeneity’ mix of research foundation, but also improve China’s tire industry’sinfluence in the international, promote the rapid development of China’s tire industry.Therefore, this technique has important practical significance and practical value.