Research On The Extension Design Method For The Automatic Visual Inspection System

Automatic visual inspection technology is widely applied in automation field, which using a large number of automation and intelligent technology. However, With the development of manufacturing industry informatization, the shortening of product life cycle, the diversification of user demands and increasingly stringent quality requirements, the key technology of visual inspection remains to be further improved.The successful completion of the automatic visual inspection task depends on the expression and automatic processing of the visual inspection knowledge.Based on the comprehensive analysis of the automatic visual inspection system, the extenics theory is used to describe the automatic visual inspection information knowledge in this paper.The objective is to find appropriate intelligent knowledge processing way to solve the problem in automatic visual inspection field.The main work of this paper is:(1) Analyzing the theoritic integration between the extenics theory and the automatic visual inspection design.Based on the extenics research findings, the integration of the extension design theory foundation, extension analysis principles, extension transformations principles and the automatic visual inspection design are discussed.(2) On the basis of the modular decomposition and clustering analysis of automatic visual inspection process, the five drawer model and the pyramid data flow model of the automatic visual inspection are put forward.The model is the abstraction expression of the automated visual inspection process, which includes image acquisition, image preprocessing, image segmentation, pattern recognition, and implementation layer. In this model, the change in any layer will lead to the whole system performance change. The designers can complete the reconstruction of new scheme only by changing one of the certain levels in the shortest time, when the detection task or environment changed. Thus the development cycle is shortened and the cost of testing is reduced.(3) The extension matter-element knowledge model is built.The visual inspection process model is described using the extension element representation method.The divergence analysis of the vision detection scheme is carried out using the extended properties of the elements. The the scheme knowledge is generated by the extension reasoning principles.(4) Reconfiguration and evaluation of the visual inspection system.The reconfigurable design method of the automatic visual inspection system is studied by extension transformation and the element combination properties, in order to explore intelligent optimization method for the visual detection problem. The comprehensive evaluation index system of automatic visual inspection system is built by analytic hierarchy process. The various indexes and the corresponding measure conditions are determined, then the dependent functions are established. At last, the index weights are computed by the judgment matrix and the priority degree are given used to rank the automatic visual inspection system.(5) The automatic vision detection reconfiguration platform is developed.The platform can realize the reconfiguration and ranking for the automatic visual detection schemes. Also, it can solve the contradictions between design efficiency and the customization requirement, the design cost and the universal property. The development can establish a bridge of communication between designers and customers, analyze and reasoning the customer requirements effectively.