| With the development of the industrial production and science technology, the demand of quality is more increasing. During recent decades, there are many huge equipments emerging, such as satellites, rockets, spacecraft etc, and the phenomenon of demanding of product's reliability, safety, economy and so on is more and more increasing, it is very urgent to solve the problem that how to prove the quality of products. In the process of quality management, whether it is decision analysis, or quality control, we should make a comprehensive study when facing to large number of historical data that can help the managers to make the effective judgment. It is very important meaning that it can identify the class of products efficiently while facing the polydimensional data.During the process of classification of quality products, Enterprises put forward to improve the requirement of quality data with the development of industry. There are polydimensional data which is continuous added so that the requirement of judging the quality class is enhance. However, there are many errors such as data-entry errors, different data source make different expressions that make the data have many problems. But it is very difficult for traditional class methods to solve effectively those problems and assure the accuracy that make us to search a method that can analysis the problems that are pointed above. Now Bayesian network that based on Bayes has many advantage to resolve the class of data.The paper, aim at the set in the data of quality control, study the high efficiency of class that facing large number of data and analysis the results .aim at the set in the data of quality control. The main work of this paper can be summarized as the following four areas:1 introduce the feature and class of quality control and quality data, and summarize the application of Bayesian network in the quality control.2 summary the Bayesian network, and emphasis the methods of the learning of Bayesian network and the type of learning methods, and the end the paper introduce the structure of Bayesian network.3 There is dependency relationship among properties of the quality data, the paper put forward two algorithms basing on the improving the performance of class: one is the based on the mutual information's Bayesian network classification; another is improved classification of Bayesian network. We are analysis the performance of classification and think the latter is based on the former. In the end of the paper, advance a improved algorithm that based on the na(i|¨) ve Bayes classification and analysis the algorithms performance.4 Make the three algorithms apply in the system of quality control and use the data to verify, and the results show that the algorithms is feasible. And then let the algorithms contrast with other algorithms ,the results show the algorithms'superiority and efficiency。... |
