| Analog circuit is an indispensable part of electronic equipment,and its fault diagnosis method has always been a hot topic.At present,the research on fault diagnosis method of analog circuit has been very deep,but there are still deficiencies in solving the component tolerance,parameter continuity,multi fault diagnosis and so on.In order to solve these problems,this paper transforms the problem of fault location and parameter identification into a global optimization problem based on the complex domain circle model,and realizes fault location and parameter identification based on intelligent optimization algorithm.The following work is done in this paper.1)The problem of analog circuit fault location is transformed into a global optimization problem.The continuous fault characteristics of analog circuits are obtained by modeling in complex domain,and fault isolation is realized.The complex domain circle model is verified by PSPICE circuit simulation.Then the tolerance problem is analyzed.The influence of tolerance on fault location is solved by multi frequency simulation.This paper studies how to transform fault location into global optimization.The objective function is established by using transfer function.Fault location can be realized by optimizing the objective function.2)Based on genetic algorithm,the algorithm of multi fault location of analog circuit is designed.The circuit simulation shows that the single measuring point can not locate the double fault,and establishes a three-dimensional model of double faults,and isolates the two faults by double measuring points.The genetic algorithm for the multi fault location of analog circuits is designed,and the accuracy of multi fault location is improved by the operation of range limitation,optimal individual reservation and catastrophic operation.The simulation results show that the algorithm is effective and high accuracy.Finally,the multi fault location of nonlinear analog circuit is realized by co-simulation,which proves that it is suitable for nonlinear analog circuit.3)The population is initialized by value of uniform spacing,and the multi fault parameter identification is realized by genetic algorithm.Different fault parameters can get the same objective function value.The purpose of parameter identification is to get all possible fault parameters.The population of genetic algorithm is initialized by value of uniform spacing in the large neighborhood(search area)of the estimated value of single element parameters.By adjusting the parameters of the fault free components,all the fault element parameters which can minimize the objective function are obtained to determine the range of fault parameters.Simulation experiments are carried out to verify the method,and the shortcomings of the method are studied.4)Based on taking value of uniform spacing,the genetic algorithm is improved to get more accurate parameter recognition results.In order to solve the problem of low precision of parameter identification,a dynamic population method is proposed.By eliminating the individuals with too large objective function value,the region of search is reduced dynamically,and the density of value in unit length is improved,so as to improve the accuracy of parameter identification.Then,the simulation results show that the algorithm can improve the accuracy of parameter recognition.Finally,algorithm before and after improvement is compared to further verify the advantages of the improved algorithm.5)The single fault parameter identification based on the two-objective evolutionary optimization is studied.According to the characteristic curve of the objective function with component parameters,the identification of parameters is transformed into solving the double objective problem according to the solution of boundary points.By group optimization,the single algorithm can solve both the maximum problem and the minimization problem.The simulation results show that the method is effective and the advantages and disadvantages of the method are analyzed. |
