Abstract:
Aiming at the characteristics of faults signals of the gas valve:strong non-stationary and aperiodic, a fault diagnosis method of reciprocating compressor valve is proposed based on principal component analysis (PCA) and back-propagation (BP) neural network of a genetic algorithm and particle swarm optimization (GA-PSO). First of all, the features of valve faults are extracted by wavelet packet decomposition; then fault feature vectors are dimensionally reduced by using PCA for reducing the scale and computing time of the network. Since the traditional BP algorithm has slow convergence speed and is easy to fall into local minimum, a GA-PSO is employed to optimize the parameters of BP neural network. Finally, using the vibration signal of valve cover of reciprocating compressor as research object, the simulation tests show that PCA and BP neural network of the GA-PSO is feasible and effective for the reciprocating compressor valve fault diagnosis.