Rolling bearing fault diagnosis method based on fast fourier transform coding and lightweight convolutional neural network
Provided is a rolling bearing fault diagnosis method based on FFT coding and L-CNN, including: obtaining original bearing fault vibration data, extracting intrinsic mode components of different frequency bands in the original bearing fault vibration data, calculating a permutation entropy value corresponding to each of the intrinsic mode components, and performing wavelet threshold denoising according to the permutation entropy value to obtain a denoised reconstructed time domain signal; performing fast Fourier transform to the denoised reconstructed time domain signal to obtain a frequency domain signal and a phase angle corresponding to the time domain signal, reconstructing the frequency domain signal according to a preset rule, retaining frequency domain data of features of the phase angle, and drawing FFT-x heat maps of different fault type data according to an amplitude range; and constructing an model L-CNN, and inputting coded data into the L-CNN model, and obtaining fault diagnosis results.
1. A rolling bearing fault diagnosis method based on FFT coding and L-CNN, comprising:
obtaining original bearing fault vibration data, extracting intrinsic mode components of different frequency bands in the original bearing fault vibration data, calculating a permutation entropy value corresponding to each of the intrinsic mode components, and performing wavelet threshold denoising according to the permutation entropy value to obtain a denoised reconstructed time domain signal;
performing fast Fourier transform to the denoised reconstructed time domain signal to obtain a frequency domain signal and a phase angle corresponding to the time domain signal, reconstructing the frequency domain signal according to a preset rule, retaining frequency domain data of features of the phase angle, and drawing FFT-x heat maps of different fault type data according to an amplitude range; wherein
the drawing FFT-x heat maps of different fault type data comprises:
decomposing the denoised reconstructed time domain signal into several single harmonic components by fast Fourier transform, and obtaining a relationship between amplitude, phase, power and frequency domain of each harmonic of the signal; and
when the phase angle is in a first quadrant and a second quadrant, a frequency domain value is positive, and when the phase angle is in a third quadrant and a fourth quadrant, a frequency domain value is negative; and according to this characteristic, reconstructing the frequency domain signal, and according to an amplitude characteristic of a reconstructed frequency domain signal, setting a boundary to (−150, 150) for heat map coding; and
constructing an improved lightweight convolutional neural network model L-CNN, and inputting coded data in the FFT-x heat maps into the L-CNN model for processing and diagnosis, and obtaining fault diagnosis results;
the inputting coded data in the FFT-x heat maps into the L-CNN model for processing and diagnosis, comprising:
capturing different levels of features of the coded data in the FFT-x from different scales by using convolution kernels with different sizes, and weighting captured features for importance of each channel by structures of global average pooling, one-dimensional convolution, channel multiplication and spatial replication through an ECA attention mechanism; and
processing a channel dimension and a spatial dimension of a color feature of an input FFT-x coded map by depth separable convolution to obtain a feature map; exchanging a channel sequence of the feature map between different depths and different groups by using a ChannelSplit module and a Channelshuffle module, and changing a channel arrangement mode; further extracting data features based on the residual and depth separable convolution; and finally, outputting the fault diagnosis results through average pooling and two fully connected layers.
2. The rolling bearing fault diagnosis method based on FFT coding and L-CNN according to claim 1 , wherein the extracting intrinsic mode components of different frequency bands in the original bearing fault vibration data comprises:
S1.1, adding Gaussian white noises with normal distribution to the original bearing fault vibration data to obtain an i-th signal y i (t), and decomposing the i-th signal y i (t) by EMD to obtain a first intrinsic mode function IMF 1 (t) and a residual component r 1 (t) of y i (t);
S1.2, using the residual component r 1 (t) as original data to decompose reconstructed r 1 (t)+β 1 E 1 (ω i (t) to obtain an intrinsic mode function IMF 2 (t) and a residual component r 2 (t), wherein E 1 is EMD decomposition, β 1 is a standard deviation of the Gaussian white noises, and ω i (t) is a Gaussian white noise; and
S1.3, repeating the S1.2 to obtain a (k+1)-th intrinsic mode function IMF k+1 (t) and a k-th residual component r k (t).
3. The rolling bearing fault diagnosis method based on FFT coding and L-CNN according to claim 2 , wherein performing wavelet threshold denoising according to the permutation entropy value comprises:
judging a noisy level of the each of the intrinsic mode components according to the permutation entropy value, and dividing the each of the intrinsic mode components into an effective IMF and a noisy IMF according to the noisy level;
based on data features of the noisy IMF, selecting wavelet bases and decomposition layers to decompose noisy bearing fault vibration data, and calculating wavelet coefficients and selecting a Sqtwolog threshold to calculate a threshold corresponding to each of the wavelet coefficients, and quantizing the wavelet coefficients; and
performing inverse wavelet transform to quantized wavelet coefficients to obtain a denoised IMF, and then reconstructing a bearing vibration signal from the effective IMF and the denoised IMF to obtain the denoised reconstructed signal.
4. The rolling bearing fault diagnosis method based on FFT coding and L-CNN according to claim 3 , wherein a method for calculating the wavelet coefficients is:
I
CWT
(
a
,
b
)
=
1
a
∫
x
(
t
)
ϕ
(
t
-
b
a
)
dt
wherein I CWT is a wavelet coefficient, a is a scale function, b is a position function, and t is a delay time;
a method for calculating a threshold corresponding to each of the wavelet coefficients is:
th
=
mid
0.6745
2
ln
N
wherein th is the threshold corresponding to the each of the wavelet coefficients, mid is a middle value of each decomposed band coefficient, and N is a length of the wavelet coefficient.
5. The rolling bearing fault diagnosis method based on FFT coding and L-CNN according to claim 1 , wherein an expression of the reconstructed frequency domain signal is:
F
(
x
)
=
{
f
(
x
)
θ
>
0
-
f
(
x
)
θ
<
0
wherein F(x) is the reconstructed frequency domain signal, f(x) is an original bearing frequency domain signal, and θ is the phase angle.
6. The rolling bearing fault diagnosis method based on FFT coding and L-CNN according to claim 1 , wherein an objective function of the improved lightweight convolutional neural network model is evaluated as a cross entropy loss function:
Loss
=
-
∑
i
=
1
output
size
y
i
·
log
y
^
i
wherein Loss is a model loss value, y i is a model fault classification prediction value, and ŷ i is a model classification true value.