A Study On Correlations Between AUDIT-GCA Evaluation Systems For Auto-body Defect Diagnosis

Car as a consumer product, its quality is the key factor of success. The process of a customer to interpret the quality of a car starts at the perception of subjective quality. In order to ensure the design intent can be faithfully translated into subjective quality attributes by production process, the major automobile manufacturers have developed their own subjective quality evaluation system. Those subjective quality evaluation systems have different characteristics, resulting in different application areas and evaluation effectiveness.For various reasons, China's own brand manufacturers are sometimes using multiple different subjective quality evaluation systems at the same time. In the actual production process, it's inevitable to develop different results or even contradictions from different systems. In order to solve this problem, this paper completed the following tasks based on the specific situation in SGMW:(1) Conducted in-depth research on several typical existing subjective quality evaluation systems. Through research, it is found that the existing subjective quality evaluation systems have different advantages and disadvantages, thus there is a certain complementarity. So if we can combine different systems as SGMW did, while reducing the contradictions between these systems and improving the correlation and diagnostic capabilities, it will be very beneficial to improve the quality evaluation system capacity.(2) This paper used the AHP-TOPSIS method to improve AUDIT and GCA systems. First, the hierarchical thinking from AHP method is used to normalize the evaluation objects of AUDIT and GCA systems. Based on that process, AHP method is used to normalize the evaluation scale of AUDIT and GCA systems. And finally, an improved AHP-TOPSIS method is use to calculate normalized evaluation results of AUDIT and GCA systems.(3) This paper used the BP neural network to model the transfer relationships between AUDIT and GCA systems. With normalized results of AUDIT and GCA as input and output of the network, the historical data is used to train the BP neural network. After verification, the correlation and reasoning model is shown to have a certain degree of accuracy.(4) Finally, the above method is extended to a generalized situation which involves a subjective quality evaluation system and an objective evaluation system. By analyzing the correlations between those two systems of the fender area of SAIC Roewe 550, the method is proven to be still applicable.