Research On The Theory And Methodology Of Intelligent Quality Control For Machining Operations

Global market competition among researchers and manufacturers has prompted the rapid developing of advanced manufacturing technology (AMT) such as agile manufacturing (AM) and lean production (LP) and virtual manufacturing (VM) and so on, which provide the products at reduced cost and better quality and rapid response to customers' demands. Under this background, process quality control (PQC) has been playing a more and more important role in the production line. How to implement the automation and intelligence of PQC has also been an important researching issue in the AMT environment.This paper researches on the theory and methodology of intelligent quality control for machining procedure. It mainly includes the following aspects:1. Pattern recognition of abnormal control charts can provide clues to reveal potential quality problems in manufacturing process. It has been a necessary technology to realize the automatic recognition of abnormal patterns with the need of automation and intelligence of PQC. In this paper, a new ANN model named regional supervised feature mapping (RSFM) network is proposed to recognize the control chart patterns, which include six basic patterns and their mixed patterns. Euclid distance discriminance is developed to recognize mixed patterns. Exponentially weighted moving average (EWMA) and fuzzy algorithm for the input samples are also developed to improve its recognition accuracy. Numerical simulation results show this model possesses many advantages, such as good self-adaptive ability, quick training and good recognition performance. With more training samples, this model can improve its recognition accuracy as its flexibility of structure and algorithm. The above-mentioned good performances facilitate the use of the proposed model in an on-line real time mode. The proposed model also provides groundwork on intelligent analysis and diagnosis system for machining quality.2. Estimating the parameters of abnormal control charts can point the abnormal degree of machining process and provide further clues to reveal potential quality problems in manufacturing process. This paper also proposes an intelligent approach based on RSFM network to estimate the parameters of abnormal patterns such as trend slope, shift magnitude, cycle amplitude and cycle length. Simulation results show the proposed network possesses advantages of quick training and good estimation performance.3. It is very difficult to use accurate mathematical model to describe the variation of quality characteristics during manufacturing process because of the complexity, non-linearity and time-variability of manufacturing system, so it is necessary to develop an intelligent quality control approach which can predict the quality characteristics without accurate mathematical model tracking in the change during manufacturing process. This paper tentatively develops an intelligent predictive theory that uses dynamic way to realize the predictive control. An intelligent predictive model based improved supervised linear feature mapping (SLFM) network is proposed to control product quality. Experimental results show the model possesses quite good ability to track in the change of manufacturing processes.4. Since it is difficult to get quality characteristics real time when unequal working face is machined, intelligent predictive control based quality characteristics cannot work under this condition. Error compensation technology (ECT) can meet this situation. The three major types of machining error are geometric, thermal and cutting-force induced error. At the present time, error compensation model usually aims at single type of error and is used for the specific machine tool. This paper develops a universal synthetical model of error compensation for three-axis numerical control (NC) machine tool, which can be updated on time when manufacturing surroundings is changed. It also can be used in multiple-axis one with the same theory.