Fault Diagnosis for Hydraulic Servo System Using Compressed Random Subspace Based ReliefF

Table 2

Extracted features of the output signals of hydraulic servo system.

Feature	Feature magnitude
	Normal	Amplifier fault	Sensor fault	Hydraulic pump wear fault
Average	−0.0446~−0.0411	0.0004~0.0036	−0.0734~−0.0698	−0.0100~−0.0063
Standard deviation	0.2715~0.2728	0.2745~0.2756	0.2715~0.2727	0.3068~0.3080
Mean square root	0.2746~0.2760	0.2742~0.2754	0.2806~0.2819	0.3067~0.3079
Skewness	0.0080~0.0148	0.0028~0.0099	0.0082~0.0149	0.0085~0.0174
Crest factor	1.4603~1.4608	1.4620~1.4685	1.4599~1.4675	1.8089~1.8189
Maximum amplitude of FFT	95.8092~96.288	96.8611~97.2778	95.8319~96.2598	107.5320~107.94
Total energy of wavelet	99.4925~99.746	99.4695~99.7350	99.5243~99.7664	98.5909~99.0743
Wavelet singular entropy	0.7682~0.8193	0.7701~0.8113	0.7857~0.8401	0.8579~0.9178
Maximum coefficient of wavelet	0.0012~0.0035	0.0005~0.0669	0.0004~0.0034	0.0005~0.0543