The advanced fault diagnostic techniques are the prerequisite for the development of large-scale circuit board. Analog circuit fault diagnosis in the world has been a hot topic of research scholars, and achieved fruitful results, and gradually formed the theory of the system. It has been formed a cross-interdisciplinary about the electronic circuit, artificial intelligence and other fields. Set of test points of the analog circuit optimization techniques aim that selected from the circuit board measuring point which can be measured in some way the fault information contained most easy to distinguish the fault of the set of points. The characteristic of such points is the collection of the test points contains as little as possible, and the number of faults can be isolated as much as possible. The issue is expanded as the optimized set of test points in the analog circuit fault diagnosis technology research under these conditions, combined with the development of the project based on artificial intelligence techniques in the diagnosis of board-level research.Optimal set of test points is constructed as searching one set that can distinguish all known fault minimum set of test points and the automatic generation of technology. The diagnostic path optimization is to find the optimal set of test points. Then it generates test points selection, test signal selection, and a repetitive sequence of the input and processing of test results. Finally, it instructs the user to minimize the cost to diagnose the fault of the actual circuit board. The degree of optimization of this technical has a direct and important impact about the efficiency and the accuracy. The better optimization results can greatly improve the effect of the testing processes. It is a very important key technology for analog circuit testing.Traditional optimization techniques based on integer coding table points is a heuristic process. This process find a standard, and then use the standard to add the point to the optimization point set, until the optimal point set can distinguish between all faults. The key to this process is to find the standard. The standard of this paper is the sensitivity factor and the information gain. The KNN algorithm which instead of the integer coding table based on this standard is optimized. In the KNN algorithm, the optimized method is to segment the KNN training model space to reduce the amount of computation of the algorithm referencing to the idea of divide and conquer. This algorithm can be optimized in some cases half the amount of calculation. Then this heuristic process is implemented to find the optimal testing point. To those found for optimizing testing points according to their corresponding fault set intersection operator, the result removed the points repeated isolation of the fault to obtain the optimized testing points.Finally, through the selection of the active filter circuit, the quad op amp high-pass filter and three circuit logic output control circuit, the acquisition of data using wavelet processing, sensitivity factor calculated sorted and information gain value, enter the KNN algorithm. It calculates a corresponding collection of fault isolation, and finds the optimal set of testing points. Using these for testing points collection, it is effective to calculate the corresponding diagnostic accuracy, verify that the methods described in the clonal selection algorithm for fault diagnosis. |