Health Management Of Behavioral Model Circuit Systems

With the development of technology,the use of electronic systems is becoming more and more widespread,and electronic systems are becoming more and more complicated while being intelligent.The failure of key modules or components often leads to system function failure,which has serious consequences.In order to ensure that the electronic system is always in good operating state,it is necessary to carry out health management of the electronic system,estimate the health status while monitoring the status,and monitor the fault when it detects the fault.Otherwise,the electronic system fault is predicted according to its health status.The probability of making an early prediction of the propagation and development of the fault,reducing the impact of the fault.Firstly,with the fault of the component list as the fault unit,combined with the fault model of the component,the circuit model of the normal and potential fault modes is simulated by PSpice A/D and OrCAD Capture software respectively,and the corresponding response waveform file is obtained,and the characteristic attribute data is obtained.The extraction gets the raw data set.Then,the original data set adopts data processing methods such as normalization,dimensionality reduction and clustering to obtain the dataset,which lays a foundation for all aspects of health management.Then,using the hidden state model based on the hidden Markov model,the HMM model of the health circuit is established and trained,and the parameter data set is analyzed to obtain the parameters characterized by the distance and its threshold.The feature attribute data extracted from the circuit to be tested is input into the HMM model,the output likelihood probability of the feature attribute data to the HMM model is calculated,and the degree of similarity between the circuit under test and the health circuit is compared and analyzed to determine the circuit to be tested.Health status.Then,the behavioral level model is combined with system-level circuit fault diagnosis and prediction,so that the behavioral level model becomes the way to establish fault classification model and fault diagnosis method,thus improving the efficiency and accuracy of fault diagnosis.Behavioral level modeling reduces the computational complexity without losing the accuracy of the constructed model.It does not care about the internal structure of the electronic system,making the implementation of health management more convenient.Finally,combine fault diagnosis with machine learning,and use a combination of multiple classifiers in machine learning: AdaBoost and random forest to establish a fault classification model.When the actual circuit fails,the corresponding feature vector is input to the fault classification model.The corresponding output is obtained,and the possible fault source is determined according to the failure mode included in the class data set.The gray model-based fault prediction method is used to predict the future health state of the circuit,so that effective measures can be taken to avoid the loss caused by the fault before the fault occurs.This thesis finally implements the software implementation of the entire process of health management based on the behavioral model circuit system.The software helps users monitor the health of the circuit,diagnose the circuit system or predict the remaining life,enabling the health management of the electronic system.