Design And Implementation Of Rule-based Data Analysis And Visualization System

With the development of the digital era,data has become a kind of invisible asset,and the management and research of data have become the hot spot of competition and the core of sustainable development of manufacturing enterprises.At present,Xi 'an xidian high voltage switch operating mechanism of the company's quality management system exist serious problems,such as complex data sources,data fragmentations,data islands,the system can not timely and accurately guide business decisions of firms' management team,can't enhance the efficiency and accuracy of quality analysis,can't satisfy the demand of the enterprise business rules change,can not meet the requirements of the lean enterprise quality management.Rule-based data analysis and visualization system can solve these problems.The specific research work is as follows:This paper has studied the algorithm of the quality control chart pattern recognition.Analyzed the advantages and disadvantages of the traditional BP neural network for pattern recognition,proposed a BP neural network algorithm with adjustable parameters and selfregulated threshold.Using the Monte Carlo method to produce sample data which satisfy the condition of the actual production,through the pretreatment of the sample data,put the data after pretreatment as network input,designed,trained and tested neural networks,compared the network output with other results of pattern recognition algorithms,verified the feasibility of this algorithm.According to the bottleneck of the traditional quality management system can't meet business rules change,this paper introduced the Drools rules engine technology,analyzed business rules of quality control chart anomaly pattern in the key process of the manufacturing process,written business rules into the file with the DRL language,established the rule base which store the rules,realized the fact and rule matching by Rete algorithm of pattern matching algorithm.Drools rules engine technology separates the business rules from the program code,facilitates the update and modification of the rule base,realizes the intelligent reasoning of the business rules,and reduces the difficulty of system development and maintenance.Through the demand analysis of the current quality management system of the company,this paper established the overall framework of the rule-based data analysis and visualization system.On the basis of demand analysis,the system is designed and implemented,And used the system to carry out an example application.With manufacturing process,put the quality of the key working procedure as the research object,using the improved BP neural network algorithm and combining with the business rules matching function of this system,realized recognition of the abnormal quality control chart pattern,and gave the corresponding quality abnormal cause and corrective action,verified the feasibility of the algorithm and the effectiveness and the reliability of this system.This paper contains 40 figures,4 tables and 65 references.