Research In Study Algorithm For Fuzzy Neural Network And Application In Solder Joint Quality Evaluation

Electronic products, characterized by "lightness, thinness, shortness and pettiness", make great demands upon the micro-miniature and the assemble intensity of the components and parts inside. In view of this, it is extremely important to guarantee the quality of the solder joint, which, as a product of welding operation, is crucial to the quality of the electronic products. However, at present, the experts in our country remain in a visual stage in assessing the integrated circuit plate, but this kind of assessment is unreliable and low in efficiency. Thus, it is urgent to develop a technology of craft automation visible information integration in the field of modern electronic industry.This paper will focus on the application of the technology of the fuzzy neural network to the assessment of the quality of the solder joint. And it will lay emphasis on the method of the combination of fuzzy neural network and the theory of the fuzzy neural consequence as well as the improvement of the clustering arithmetic based on grid. It mainly research in the theory and method of neural network and fuzzy illation, furthermore, performing fuzzy illation by neural network. Meantime, using clustering analyze to pick-up illation rules, solved the problem of the slow rate for clustering result from excessive number of deductive rules caused by the direct experiences of the experts. At the same time, because the introduction of clustering analyze, it remedies such defects as the low precision of clustering, which is owing to the overextension of the clustering. Therefore holistically improve the system's performance.Besides, two methods of fuzzy neural network, based on grid clustering have been designed. Apart from this, both contrast and analysis have been made between the effects of the new methods and those of the fuzzy neural network in the assessment of the solder joint based on the experiences of the experts and clustering arithmetic. It has been proved that the newly-improved fuzzy neural network based on clustering arithmetic is efficient in the assessment of the quality of the solder joint.