Motor driving system converter fault diagnosis method based on adaptive sparse filtering
The disclosure discloses a motor driving system converter fault diagnosis method based on adaptive sparse filtering, and belongs to the field of driving system fault diagnosis. The disclosure applies an unsupervised learning algorithm to an application scene of converter fault diagnosis. Effective features are automatically extracted from original data, and the problem of manual feature design based on expert knowledge is solved. Meanwhile, in consideration of current fundamental period change caused by different rotation speed working conditions, rotation speed feedback is introduced, secondary sampling is carried out on current sampled at a constant frequency, it is ensured that the length of a signal input into the deep sparse filtering network is one fundamental wave period, redundant information is better removed from original data, the calculation burden is relieved, and the accuracy and rapidity of the diagnosis algorithm are improved to a certain extent.
1. A motor driving system converter fault diagnosis method based on adaptive sparse filtering, comprising:
S1. sampling current signals and rotation speed signals of a motor driving system when a converter in the motor driving system is in different health states;
S2. performing a secondary sampling on the current signals according to an input dimension of an sparse filtering (SF) network and the rotation speed signals, wherein the step S2 comprises:
a. estimating a fundamental wave period of the current signals according to the sampled rotation speed signals,
T
=
60
np
,
where n is a measured rotation speed, and p is a number of pole pairs of a motor;
b. determining a sampling frequency f corresponding to the one fundamental wave period according to an input dimension of a deep sparse filtering network (DSF),
f
=
N
in
T
,
wherein N in is the input dimension of the DSF network, and T is an estimated fundamental wave period of the current signals; and
c. determining a secondary sampling interval of the current signals,
N
gap
=
f
s
f
-
1
,
wherein f s is a constant sampling frequency of the current signals, and f is a sampling frequency corresponding to the input dimension of the sparse filtering network;
S3. normalizing the current signals after the secondary sampling as a training data set;
S4. adopting the training data set to train a fault diagnosis network; wherein the fault diagnosis network comprises the sparse filtering network and a Softmax classifier; and
S5. sampling the current signals and the rotation speed signals under a current operating state of the converter, using the process of steps S2 and S3 to process the sampled current signals, and inputting the processed signal into the trained fault diagnosis network to realize feature extraction and automatic fault diagnosis.
2. The motor driving system converter fault diagnosis method based on adaptive sparse filtering according to claim 1 , wherein the different health states comprise four states, namely, a normal operating condition, upper-switch open-circuit fault, lower-switch open-circuit fault and single-phase open-circuit fault.
3. The motor driving system converter fault diagnosis method based on adaptive sparse filtering according to claim 2 , wherein the sparse filtering network adopts the deep sparse filtering network.
4. The motor driving system converter fault diagnosis method based on adaptive sparse filtering according to claim 1 , wherein the sparse filtering network adopts the deep sparse filtering network.
5. The motor driving system converter fault diagnosis method based on adaptive sparse filtering according to claim 1 , wherein the step S3 is performing min-max normalization on the current signals after the secondary sampling:
x
~
=
x
-
x
min
x
max
-
x
min
wherein x represents current data after the secondary sampling, x min represents a minimum value of the data, and x max represents a maximum value of the data.
6. The motor driving system converter fault diagnosis method based on adaptive sparse filtering according to claim 5 , wherein the sparse filtering network adopts the deep sparse filtering network.
7. The motor driving system converter fault diagnosis method based on adaptive sparse filtering according to claim 6 , wherein the deep sparse filtering network comprises two layers of sparse filtering sub-networks; the first layer of sparse filtering sub-network maps input data from an input space R N in to a feature space R N mid ; the second layer of sparse filtering sub-network maps features from a space R N mid to a feature space R N out that is more abstract.
8. A motor driving system converter fault diagnosis system based on adaptive sparse filtering, comprising: a computer-readable storage medium and a processor;
wherein the computer-readable storage medium is configured to store executable instructions;
the processor is configured to read the executable instructions stored in the computer-readable storage medium, and execute the motor driving system converter fault diagnosis method based on adaptive sparse filtering according to claim 1 .