Research And Application On Mode Recognition Of Out-of-control Trend And Estimating The Change Time In Manufacturing Process

Statistical Process Control (SPC) is one of the methods most widely applied in manufacturing industry to analyze the quality characteristic value obtained from factory site. SPC is the process of manufacturing data collection, sorting and analysis. It analyzes the quality problems occurred in manufacturing process through objective and quantitative method, improving the quality of products with less resource consumption. Thus it is the most useful tool in quality management. In the real production process, getting out of control is unavoidable, but it also comply with the laws of very complex. SPC can find out the part of the deterministic of the manufacturing process and ensure the qualitative characteristics of the workpiece process are above acceptable levels. Currently under the complex manufacturing environment, the sources of the quality problems are more widely because of the supermatic manufacturing process and the complicacy of products, so it’s more urgent to find out how to identify the reasons for the out of control in the process of manufacturing. It’s a hot spot of current research to study how to use SPC to automatically identify the manufacturing process runaway trend model automatically under the background of computer integrated manufacturing, providing information of process under and out of control of high efficient and high precision and give causes and pointed solutions of runaways.The various modes of the manufacturing process out of control and analyze the change-point trend which causes runaway are explored, embarks from the manufacturing process quality stability, and based on the background of SPC. Furthermore, after considering actual production, pattern recognition and change-point estimation of statistical data of the manufacturing process, a real prediction about process out of control is achieved.Firstly, the classical theory of manufacturing process out-of-control trend recognition and change-point estimation is summarized and the theoretical system of out of control trend is put forward. Recognition of the pattern of runaway trend using fuzzy neural network is proposed. Using fuzzy clustering analysis to perform change-point estimation on this basis can effectively improve quality stability. In the second place, the overall thinking of the paper is showed, which is to design a neural network pattern recognizer based on the features, and complete the feature extraction from the sample function through defining the feature. According to the features of different process model,6types of out of control mode are recognized automatically. Next, based on the fuzzy clustering theory and statistical approach, a new fuzzy statistic clustering method to deal with the real change-point problems and applies this method in different types of control charts is proposed. The results proved that no matter using fixed sampling strategy or variable sampling strategy, this method works well for change-point estimation in the control charts. Both the pattern recognition of out-of-control trend based on features and change-point analysis based on fuzzy clustering to the workpiece process of a cylinder block are applied, it is indicated the mthod can predicted the out-of-control processs plendidly. Finally, based on IDEF and UML, an out-of-control trend analysis system oriented to the manufacturing process is developed. It combined intelligent computing, data collection and many other techniques perfectly, thus providing technical support for the enhancement of the quality stability in manufacturing process.