NPC Three-level Inverter Fault Diagnosis Method Based on Improved Small Tree Transformation

Abstract

The invention relates to a NPC three-level inverter fault diagnosis method based on improved small tree transformation, comprising: constructing an NPC three-level inverter circuit simulation model, simulating a fault process, and measuring the voltage waveform of a bridge arm as a fault signal; Decompose the fault signal into multiple IMFs s Component; for each IMF s The component is subjected to Hilbert Hilbert transform to obtain its time-frequency distribution and amplitude, and the first 8 IMFs are selected s component; use envelope analysis to fit the envelope signal, and screen out the fault characteristic parameters; use Gaussian kernel function to optimize the small tree transformation to generate independent feature vector samples; divide the sample data into training set and Test set; the training set is used to construct the SVM classifier model, and the test set is used to actually diagnose circuit faults. The invention adopts EEMD to decompose the fault signal, combined with the Hilbert transform, collects more frequency domain features, and is more suitable for dealing with the nonlinear and non-stationary signal generated by the fault of the three-level inverter circuit.

Description

technical field [0001] The invention relates to the technical field of fault diagnosis of circuit devices, in particular to an NPC three-level inverter circuit fault diagnosis method based on improved small tree transformation. Background technique [0002] Compared with the traditional two-level inverter, the three-level inverter has the advantages of low voltage stress on the device, small electromagnetic interference, less harmonics of the output voltage and high inverter efficiency, but the three-level circuit uses More power switching devices will reduce the reliability of the circuit, and any device failure may cause the entire circuit to stop working, and even affect the safety of other circuits, causing immeasurable economic losses. Therefore, how to quickly detect the fault time and type, accurately analyze the fault signal, and realize the accurate positioning of the fault transient state and disturbance time is an urgent problem to be solved at this stage. [000...

AI Technical Summary

Problems solved by technology

In addition, once the basis function is determined, all signals must be processed with this basis function, and it cannot be adaptive

Chen Danjiang and Ye Yinzhong used Fourier transform method for fault feature extraction in "Multi-Neural Network Based Three-Level Inverter Device Open Circuit Fault Diagnosis Method" (Acta Electrotechnical Society, 2013, 28(6):120-126) , Fourier transform is a global transform that can obtain higher resolution in the frequency domain, but it cannot give the change of frequency with time, and cannot meet the requirements of diagnosis in the time domain, making the diagnosis accuracy poor

Images