基于图正则化堆叠自编码器的风机轴承故障诊断方法

基于图正则化堆叠自编码器的风机轴承故障诊断方法

刘展, 刘健洵, 包琰洋, 李大字

Bearing Faults Diagnosis Method Based on Stacked Auto-Encoder With Graph Regularization for Wind Turbines

Zhan LIU, Jianxun LIU, Yanyang BAO, Dazi LI

表2 特征参数

Tab. 2 Characteristic parameters

时域特征	频域特征
$p_{1} = \frac{\sum_{n = 1}^{N} x (n)}{N}$	$p_{12} = \frac{\sum_{k = 1}^{K} s (k)}{K}$
$P_{2} = \sqrt[]{\frac{\sum_{n = 1}^{N} [x (n) - p_{1}]^{2}}{N - 1}}$	$p_{13} = \frac{\sum_{k = 1}^{K} [s (k) - p_{12}]^{2}}{K - 1}$
$p_{3} = {(\frac{\sum_{n = 1}^{N} \sqrt[]{\|x (n)\|}}{N})}^{2}$	$p_{14} = \frac{\sum_{k = 1}^{K} [s (k) - p_{12}]^{3}}{K {(\sqrt[]{p_{13}})}^{3}}$
$p_{4} = \sqrt[]{\frac{\sum_{n = 1}^{N} [x {(n)]}^{2}}{N}}$	$p_{15} = \frac{\sum_{k = 1}^{K} [s (k) - p_{12}]^{4}}{K p_{13}^{2}}$
$p_{5} = m a x \|x (n)\|$	$p_{16} = \frac{\sum_{k = 1}^{K} f_{k} s (k)}{\sum_{k = 1}^{K} s (k)}$
$p_{6} = \frac{\sum_{n = 1}^{N} [x (n) - p_{1}]^{3}}{(N - 1) p_{2}^{3}}$	$p_{17} = \sqrt[]{\frac{\sum_{k = 1}^{K} (f_{k} - P_{16})^{2} s (k)}{K}}$
$p_{7} = \frac{\sum_{n = 1}^{N} [x (n) - p_{1}]^{4}}{(N - 1) p_{2}^{4}}$	$p_{18} = \sqrt[]{\frac{\sum_{k = 1}^{K} f_{k}^{2} s (k)}{\sum_{k = 1}^{K} s^{s (k)}}}$
$p_{8} = \frac{p_{5}}{p_{4}}$	$p_{19} = \sqrt[]{\frac{\sum_{k = 1}^{K} f_{k}^{4} s (k)}{\sum_{k = 1}^{K} f_{k}^{2} s (k)}}$
$p_{9} = \frac{p_{5}}{p_{3}}$	$p_{20} = \frac{\sum_{k = 1}^{K} f_{k}^{2 s (k)}}{\sqrt[]{\sum_{k = 1}^{K} s (k) \sum_{k = 1}^{K} f_{k}^{4} (k)}}$
$p_{10} = \frac{p_{4}}{\frac{1}{N} \sum_{n = 1}^{N} \|x (n)\|}$	$p_{21} = \frac{p_{17}}{p_{16}}$
$p_{11} = \frac{p_{5}}{\frac{1}{N} \sum_{n = 1}^{N} \|x (n)\|}$	$p_{22} = \frac{\sum_{k = 1}^{K} (f_{k} - p_{16})^{3} s (k)}{K p_{17}^{3}}$
	$p_{23} = \frac{\sum_{k = 1}^{K} (f_{k} - p_{16})^{4} s (k)}{K p_{17}^{4}}$
	$p_{24} = \frac{\sum_{k = 1}^{K} (f_{k} - p_{16})^{1 / 2} s (k)}{K \sqrt[]{p_{17}}}$