Research On Comprehensive Evaluation Of Rail Welding Process Quality Based On Multivariate Statistical Analysis

The rapid development of railway technology is inseparable from rail welding technology.Rail welding is a serial working method,which mainly includes eight processes,each of which is inseparably related to the quality of rail welding.In its production process,the most important process is the welding process.For the welding process,it is also the one that consumes the most energy in the entire welding process.By selecting reasonable welding parameters,enterprises can achieve energy saving and emission reduction,and reduce production costs.However,due to the strong coupling,uncertainty and non-linear characteristics of the welding process,it is difficult to model the mechanism.How to determine the relationship between the welding parameters and the amount of misalignment is a hot and difficult problem.The rail welding process includes many indicators,and the quality of each indicator will affect the quality of the overall rail welding.The overall quality of rail welding is related to the energy consumption and cost of the company.The welding process is tedious.The individual evaluation and the overall evaluation are different.How to deal with the evaluation relationship between the individual and the overall is the first problem to be solved.Each process has its own evaluation index.The accuracy of the data in the index is different,which brings great difficulties to the evaluation.How to use the existing data to evaluate the rail welding process is also a problem to be solved.In view of the above problems,the specific research contents of this article are as follows:Firstly,through reading and searching the literature,we mastered the role of each process in the welding process and related technical indicators.By correlating the welding parameters and the amount of misalignment in the welding process and combining the BP neural network and PSO-BP neural network,a prediction model of misalignment is established.The model has a high hit rate and has a strong predictive ability.In order to improve welding parameters and reducing the cost of the enterprise provide a theoretical basis.Secondly,a multi-level evaluation system was constructed using known technical indicators.In order to solve the problem of the importance of each indicator and process in the evaluation process,the weights of each indicator and process were determined according to the analytic hierarchy process and constructed using the triangular membership function Evaluation matrix.Finally,a single-layer fuzzy evaluation is used to construct an evaluation model for each process,and a multi-layer fuzzy evaluation is used to construct an evaluation model for the overall welding quality.