Research Of Nuclear Detector Circuit Fault Diagnosis Based On Supportv Ectormachine (SVM) Method

Nuclear detector as radiation protection environment monitoring and indispensable part of the nuclear power plant safety monitoring, its reliability and safety is an important safeguard in nuclear device running.But since the detector to work for a long time in high temperature, high humidity and high irradiation environment, radiation damage and temperature and humidity changes can accelerate the aging effect of detector elements, so that the instrument failure is frequent, wrong false positives, reduces the workplace safety.Pulse waveform were introduced in this paper, based on the data analysis of the fault diagnosis method of nuclear detectors, nuclear detector output signal is different from general of sensor signals, it outputs a pulse signal, failure occurs when the detector, the pulse signal of some parameters such as rising along the falling edge, the time and may change accordingly, such as pulse width based on the change of the specific parameters identification, realize the recognition of fault judgment.Combining with the theory of sensor fault diagnosis results and digital signal processing technology, through the simulation fault signal of the detector and normal access to state of two kinds of output data, using independent component analysis(ICA) technology for pretreatment of signal data, to achieve effective separation of the signal and interference signal;To extract the fault feature information and establish the corresponding fault experience model;And based on support vector machine(SVM), which can identify the method of fault feature of fault’s judgment and positioning, studying the fault diagnosis method based on data analysis.Article through software simulation, tried to locate nuclear detector of analog circuit fault diagnosis, through nuclear signal simulation, and forming simulation on specific amplification circuit, after the collection and output signal, signal feature extraction, extracted data, training classification, classifier model is set up, treat the diagnostic data to determine classification, thus complete the circuit fault diagnosis.Simulation results verify the effectiveness of the method of analog circuit fault diagnosis, and explore the nuclear detector circuit automatic fault diagnosis methods.Through existing in this research is trying to reduce nuclear detection instrument fault frequent misjudgment false positives, intelligent fault diagnosis and localization, promote radiation detection instrument digital technology innovation.