Research On Fault Detection Of Analog Circuits Based On DBN And GWO-SVM

The development of science and technology makes the scale of circuit system more and more huge.For large-scale integrated electronic equipment,the complex operation mechanism brings great difficulties for circuit fault detection.We presents a method of analog circuit fault diagnosis based on deep learning and support vector machine(SVM).The deep confidence network of deep learning method is used to extract the output signal characteristics of analog circuit,and the intelligent optimization algorithm is used to optimize the parameters of restricted Boltzmann machine,so as to obtain shorter fault diagnosis time and higher fault diagnosis accuracy.The main work of this paper is as follows:Firstly,we summarizes the contribution of Chinese and foreign scholars to analog circuit fault diagnosis technology,introduces the corresponding diagnosis methods for different types,and systematically introduces the process of analog circuit fault diagnosis and the establishment of fault diagnosis model based on the diagnosis methods of analog circuit.Secondly,based on the existing theoretical research of analog circuit fault diagnosis and the difficulty of analog circuit fault diagnosis,a GWO-SVM based analog circuit fault diagnosis model is proposed..Deep belief network of deep learning techniques are used to capture deep feature vectors of analog circuit output,that is,the deep feature of signal.Deep learning network is usually easy to fall into local optimal parameters,and it takes a long time to train.However,deep belief network can effectively overcome such problems,so as to extract the depth features of the target efficiently and accurately.Finally,we presents a GWO-SVM fault diagnosis model to solve some difficult problems.The strong convergence performance of Gray Wolf algorithm(GWO)makes it easy to jump out of the local optimum.GWO method is used to improve the performance of SVM.The algorithm proposed in this paper improves the accuracy of Sallen key band-pass filter circuit to 100% and shortens the fault diagnosis time by about 90%;for four op amp double fourth-order high pass filter,the accuracy is improved to 99.68%,the fault diagnosis time is shortened by about 75% and hundreds of iterations are reduced.The results show that the proposed DBN and gwo-svm methods have faster processing speed,higher fault diagnosis accuracy and shorter fault diagnosis time than the previous methods.