Research On Artificial Intelligence Technologies Applied In Automotive Product Audit System

Automotive product audit system(APAS) is an important means to ensure the stability of product quality and achieve the satisfaction of customers, it plays an increasingly important role in the automotive manufacturing enterprises. With the further development of economic globalization, APAS characterizes complex systems, quick change speed, many uncertainties, a wide range of information, the large volume of data and so on, and the traditional audit methods and audit techniques can not well satisfy the demand for such systems, thus the research on new methods and new technologies of APAS has great practical and economic significance. Artificial intelligence technologies provide a variety of excellent algorithms for APAS under the new situation, and service-oriented architecture(SOA) offers a good system architecture to enterprises'software applications. This dissertation mainly presents the research on the artificial intelligence technologies and SOA using in APAS, the major work is as follows:1. Firstly the dissertation introduces the concept and research status at home and abroad of the product audit using in the automotive manufacturing domain. It also presents the definition, domestic and international research status and trend of APAS, features and problems that nowaday APAS has. A novel approach of constructing intelligent APAS ( IAPAS ) based on artificial intelligence technologies and SOA is put forward.2. Techniques of IAPAS information extraction based on ontology are studied. By analyzing features of information resources and demand of intelligent computing in IAPAS, a information extraction system is constructed based on ontology. Several key techniques of the information extraction system, namely, techniques to construct and apply domain ontology, task ontology, rules ontology and evaluation index are put forward. An example of information extraction is implemented to verify the performance of the information extraction system.3. Techniques of IAPAS intelligent evaluation for automotive defects are studied. By analyzing features and problems of evaluation techniques for automotive defects in product audit system, a intelligent evaluation system for individual defects is constructed based on hierarchical hybrid neural networks, and a intelligent evaluation system for the qualitative defect level of a whole automobile is also built based on entropy weight coefficient method and fuzzy comprehensive evaluation.4. Techniques of IAPAS defect analysis and corrective action decision-making are studied. According to features of defect analysis in IAPAS and through improving traditional FMEA techniques,PAFMEA techniques suitable for defect analysis are presented. Three indices which are the cost of expense, the cost of time and the RPN after correction and their relative information are introduced into PAFMEA, and techniques of corrective action decision-making based on multi-objective optimization algorithm whose optimized objectives are the three indices are presented as well. SAMOACOMV algorithm is created to solve optimization problems of the corrective action decision-making.5. Techniques of IAPAS cost prediction are studied. By analyzing features of cost prediction in IAPAS, a grey-neural network prediction model is put forward. An improved GM (1,1) model and an improved Elman network - MElman network is emphatically studied, and they provides a new excellent approach for cost prediction.6. Finally, a system architecture for enterprises'software applications based on service component architecture(SCA) is studied, and development methodology based on the system architecture is put forward. A prototype of IAPAS based on artificial intelligence technologies and SOA is presented as well, and it verifies the feasibility of approaches to develop enterprises'software applications based on SCA system architecture and the validity of the prototype.