Optimize Analysis And Application Of Production Parameters Of Steel Rail Rolling Based Big Data Technology

China's high-speed railway construction volume is huge.And has economic plans such as The Silk Road Economic Belt and the 21st-century Maritime Silk Road.At the same time,in recent years,urban rail transit lines are increasing day by day.Rail track is one of the basic equipment and materials of rail transit,is essential elements both in new rail lines and maintenance tracks.so the production quality and yield of the rail are the cornerstone related to the development of logistics economy and the convenience of people's transportation and life.However,in the process of rail rolling production,it still depends on the experience of human experts to modify the pass parameters of universal rolling machine on site.In order to avoid and reduce the accident rate of rail rolling production line,shorten the production cycle and improve the product qualification rate,it is necessary to optimize parameters in the process of rail production by the data-driven way.For optimize rolling production parameters,the experts in the traditional field of mechanical materials,used the differential equations based on the physical changes of stress field,strain field and heat field are often to simulate.On the one hand,it needs the professional knowledge and software skills of material mechanics,On the other hand,the accuracy of simulation is positively related to the number and shape of micro elements,and the precision of equation description.And time consume increases with the number of micro elements.Therefore,this paper uses the multi-objective regression algorithm to fit the data-driven rail rolling production process,establishes the model virtual rolling machine,and then uses the multiple evolution algorithm to search the acceptable Pareto optimal solution set.From this,the problem of sudden increase of bad steel production in the process of rail rolling can be solved quickly.Firstly,the rail production data is preprocessed and analyzed,cleaned,the missing value is filled,and the data is upsampled.This reduces the training difficulty of the model and increases the accuracy of the model.Second,the research of real-time prediction of rail multiple geometric indexes in rail rolling production,aiming at the problem of relying on experts or artificial experience in the production process,this paper uses the machine learning algorithms,after training and fitting,accurately and efficiently predict the geometric index,with the mean error not more than 0.038 mm.Thirdly,according to the requirements of Rail business production,using genetic algorithm framework,self-defined coding mode,fitness function,selection function,etc.,the multi-dimensional parameters are successfully revised and optimized in seconds,and the appropriate Pareto solution is given.With the rise of industry 4.0 and the wave of intelligent manufacturing,manufacturing industry should also make efforts to improve production,improve quality and reduce cost.In this paper,machine learning algorithm is used to fit the knowledge in these data and optimize by search appropriate parameters,which can avoid the theoretical error of rail metal flow equation and the impact of human error.Finally,improve the production capacity and provide more economic benefits.