The Research On Quality Prediction And Diagnosis System Based On MES And Data Mining Technology

Product quality is the reflection of customer requirements. In order to position the enterprise for great accomplishment in the future, quality management must be strengthened. With the development of technology, science and maturity of market, consumers have become increasingly more discriminating in regard to quality, the existing quality management methods can no longer meet the demands.Quality data and information are collected by process control system from MES, and stored in database or data warehouse after analyzed by engineers. These data contain process records, monitoring information, etc. With the accumulation of the quality data, the efficiency of data analysis declines as a result of the imitation of analysis ability, the potential value may be ignored, In order to take full advantage of these data and realize TQC, the application of data mining has become a hot topic in quality prediction and diagnosis research. On the basis of previous research on quality prediction and diagnosis, quality data mining platform is established which could realize intellectualized control chart pattern recognition, quality prediction and diagnosis. The conclusions of the study are described as follows:(1) A new decision tree method is utilized into quality prediction system of production. A new impurity measure FCP is proposed which forms the basis of the FCP decision tree algorithm. The new algorithm was tested on various datasets related to quality prediction. The obtained results have been compared to other methods, indicating the superiority in accuracy and computation cost of the proposed method. Based on the FCP decision tree, a control chart pattern recognition method is presented. After training and testing, the performance of this method meets the requirements of pattern recognition task.(2) A quality prediction method based on cluster analysis and FCP decision tree analysis is proposed. Quality data is analyzed by these two algorithms and process quality prediction function is realized by the acquisition of information. The combinations of the quality influence factors are obtained by the cluster analysis. Process quality prediction is realized according to the classification rules which are generated by FCP decision tree. Quality prediction makes the preparation to realize the quality diagnosis by accumulating the quality information.(3) Case-Based Reasoning is introduced into quality diagnosis system to realize the diagnosis knowledge management and self-learning function. The paper demonstrates diagnosis knowledge acquisition and quality diagnosis by analyzing milling process in CNC manufacturing enterprise. Finally, a quality prediction and diagnosis system based on MES is developed. The quality prediction database, diagnosis knowledge database and the functional modules for CNC manufacturing quality control are built. This system realized the control chart pattern recognition in processing, the prediction and diagnosis to quality anomaly control and knowledge self-learning.