Application Research Of Data Mining Technology In Cooling Control System After Plate Rolling

With the rapid development of equipment technology in China,the processing capacity of advanced equipment is becoming stronger and stronger.The data produced by steel production line is huge every day.How to extract useful knowledge effectively from these data is a problem that needs to be solved.In the field of plate cooling after rolling,the cooling control system is often disturbed by external factors,the collected data may be abnormal,the control accuracy of the system is not stable enough,the existing cooling control system does not have a special fault diagnosis module.In order to deal with the above problems,the cooling parameters and cooling control system of a medium and heavy plate cooling production line were studied.A data cleaning model for plate cooling after rolling was established.The random forests algorithm and rule-based and case-based fault diagnosis technology were applied to the fault diagnosis of cooling.The cooling schedule calculation software based on grey relational degree was developed.The requirement of control accuracy and stability for cooling system of plate after rolling has been realized.The main works of the thesis are as follows:(1)Based on the inverse distance weight interpolation method,the influence of K coefficients of inverse distance weight interpolation model on the calculation results was studied when K coefficients are 1,2 and 3 respectively.In the process of data legality processing,the density-based outlier detection technology was used to clean the data,which was tested combining the actual cooling data.The influence of different parameters on the final cleaning results was studied.This method provides high quality data for subsequent production.(2)A fault diagnosis model for cooling after rolling was established.Some fault diagnosis rules and cases of cooling after rolling were put forward.According to the characteristics of the random forests,the Python language and Pycharm software were used to implement the model.Processing the data collected in the mills and training the model by setting relevant parameters,the highest accuracy was 91.67%.The results show that the algorithm can predict the faults in cooling after rolling effectively.(3)Combining analytic hierarchy process,grey relational degree and K-means clustering algorithm,the calculation of plate cooling schedule was realized.The attributes of cooling parameters were extracted and the hierarchical structure model was established.The consistency ratio CR<0.1 shows that the overall consistency test was acceptable.Combining with the cooling production line of a steel mill,the K-means clustering algorithm was used to cluster the historical data.The data were divided into several clusters according to the actual situation.Taking the thickness and flow density as examples,when K is greater than 2 and only the data with the greatest correlation degree is selected from one cluster,the amount of calculation is 3.83%-47.5%of the initial calculation amount.Grey relational degree algorithm and K-means clustering algorithm were used to calculate the cooling schedule,the corresponding cooling schedule of plate was calculated.(4)The cooling control system of a domestic plate after rolling was developed.The cooling part of the production line appeared some situations,the control accuracy of a certain layer of steel plate was good or bad and the temperature of a large number of steel plate after rolling were lower than the start cooling temperature.But there were three problems in the application process of the system,including the abnormal scanning pyrometer and the front purge switch,etc.In the later stage,the fault diagnosis model will be further improved to realize the efficient and accurate fault diagnosis of the cooling control system of plate after rolling.The cooling results of sampling in a certain period of time showed that the thickness of steel plate was 12?40 mm,the target temperature is between 650? and 700?,and the average deviation is 6.6?.The results show that the system has good stability and control accuracy.