A motor abnormal vibration early warning method and system
By constructing the reference feature vector matrix of different load intervals of the motor and calculating the Mahayana distance index, the single variable and manual intervention problems of traditional motor state monitoring methods are solved, and accurate detection and timely warning of abnormal motor vibrations are achieved, and monitoring is improved. Accuracy and reliability.
Patent Information
- Application Number
- CN202110886276.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-08-03
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2041-08-03
AI Technical Summary
Traditional motor status monitoring methods rely on a single variable and manually set thresholds, making it difficult to accurately detect equipment vibration changes, and do not consider the load status, resulting in deviations in monitoring results and it is difficult to timely detect irreversible damage to key components.
By collecting the motor's vibration data and electrical data, a reference characteristic vector matrix of different load intervals is constructed, the Marshallow distance index and monitoring threshold are calculated, and whether the motor is abnormal is adaptively judged, the load state is considered, and manual operation is simplified.
It improves the accuracy of motor abnormal vibration detection and model accuracy, reduces misjudgment, and promptly issues early warnings to avoid economic losses caused by long-term shutdown of equipment.
Smart Images

Figure CN113609964B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of motor monitoring, and in particular to a method and system for early warning abnormal vibration of a motor. Background Art
[0002] The main mechanical parts of the motor are the rotor, stator and bearings. Due to the life of the unit and load changes, the motor will have different vibration characteristics. Vibration is an important manifestation of the motor's operating characteristics.
[0003] During the operation of the motor, damage or abnormality of various components will cause the vibration of the motor to change. In the early stage of damage, the amplitude of this vibration change is often not very obvious and difficult to detect. Eventually, it will lead to irreversible damage to key components, causing the unit to shut down for a long time and causing serious economic losses.
[0004] Traditional condition monitoring usually only monitors a single characteristic quantity and judges the operating status of the equipment by manually setting a threshold. This method requires a lot of manual intervention and relies on experience. And a single variable often cannot fully reflect the operating status of the equipment. At the same time, when considering the vibration of the equipment, its load condition is not considered. However, due to different processes in industrial sites, the load status of the equipment generally changes in real time. In addition, different load conditions will cause different vibration conditions of the equipment. Therefore, if the load status of the equipment is not considered and only its vibration index is monitored, the monitoring results may be biased. Summary of the invention
[0005] The object of the present invention is to provide a method and system for early warning abnormal vibration of a motor with accurate monitoring results.
[0006] In an embodiment of the present invention, a method for early warning of abnormal vibration of a motor is provided, which comprises:
[0007] The vibration data and electrical data of the motor are collected, the load range of the motor is obtained according to the electrical data of the motor, and the reference eigenvector matrix x of the vibration data of the motor in different load ranges is constructed. m×n , where m is the number of statistical samples and n is the characteristic dimension of each sample data;
[0008] According to the baseline eigenvector matrix X for each load interval m×n Calculate the monitoring threshold in each load interval respectively;
[0009] Calculate the current load interval of the sample to be tested and the matrix X of the sample to be tested and the benchmark feature vector in the current load interval m×n The Mahalanobis distance indicator;
[0010] According to the sample to be tested and the benchmark feature vector matrix xm×n The Mahalanobis distance index and the monitoring threshold of the current load range are used to determine whether the motor is abnormal.
[0011] In the embodiment of the present invention, according to the reference eigenvector matrix X of each load interval m×n Calculate the monitoring thresholds in each load interval separately, including:
[0012] First, the base eigenvector matrix X m×n Normalization obtains the normalized eigenvector matrix Z;
[0013] Then, the sample is calculated by Mahalanobis distance Where C represents the covariance matrix of matrix Z;
[0014] Finally, the monitoring threshold MD is calculated by the 3Sigma criterion threshold =MD train ×3.
[0015] In the embodiment of the present invention, the current load interval of the sample to be tested and the matrix x of the sample to be tested and the reference feature vector in the current load interval are calculated. m×n The Mahalanobis distance indicators include:
[0016] Calculate the load rate of the sample, and obtain the load range of the sample according to the load rate of the sample;
[0017] Extract the feature vector y = [y1, y2, ..., yN] from the sample to be tested and perform normalization processing, and calculate its reference feature vector matrix X between the current load interval and the feature vector matrix X m×n The corresponding Mahalanobis distance index of the normalized eigenvector matrix Z is: MD = yC - 1 y T .
[0018] In an embodiment of the present invention, obtaining the load range of the motor according to the electrical data of the motor includes:
[0019] Calculate the motor load factor based on the motor's electrical data;
[0020] The load range of the sample is obtained according to its load rate, where the load range is divided according to the following rules:
[0021] The load rate ranges from 30% to 100%, and the load interval is selected according to the principle of 5 percentage points. The load interval is closed on the left and open on the right.
[0022] In the embodiment of the present invention, according to the sample to be tested and the reference feature vector matrix X m×n The Mahalanobis distance index is used to determine whether the motor is abnormal, including:
[0023] Compare the collected samples to be tested with the benchmark feature vector matrix X m×n The Mahalanobis distance index is compared with the monitoring threshold of the corresponding load interval. When the Mahalanobis distance index of three consecutively collected samples to be tested in the same load interval and the benchmark eigenvector matrix exceeds the monitoring threshold, it is determined that the motor vibration state is abnormal and an alarm is issued.
[0024] In the embodiment of the present invention, the vibration data of the motor includes vibration data of the driving end and the non-driving end of the motor.
[0025] In the embodiment of the present invention, the electrical data of the motor includes three-phase voltage and current data.
[0026] In an embodiment of the present invention, there is also provided an early warning system for abnormal vibration of a motor, which comprises:
[0027] An acquisition module is used to collect vibration data and electrical data of the motor;
[0028] The reference eigenvector matrix construction module is used to obtain the load range of the motor according to the electrical data of the motor, and to construct the reference eigenvector matrix X of the vibration data of the motor in different load ranges. m×n , where m is the number of statistical samples and n is the characteristic dimension of each sample data;
[0029] The monitoring threshold calculation module is used to calculate the baseline feature vector matrix X of each load interval. m×n Calculate the monitoring threshold in each load interval respectively;
[0030] The Mahalanobis distance index calculation module is used to calculate the current load interval of the sample to be tested and the matrix X between the sample to be tested and the benchmark feature vector in the current load interval. m×n The Mahalanobis distance indicator;
[0031] The motor state judgment module is used to determine the state of the motor according to the sample to be tested and the reference feature vector matrix X m×n The Mahalanobis distance index and the monitoring threshold of the current load range are used to determine whether the motor is abnormal.
[0032] In the embodiment of the present invention, the acquisition module includes vibration acceleration sensors respectively arranged at the driving end and the non-driving end of the motor, and respectively used to monitor the vibration signals of the driving end and the non-driving end of the motor.
[0033] In the embodiment of the present invention, the acquisition module further includes a voltage transformer and a current transformer, which are respectively used to obtain the three-phase voltage and three-phase current data of the motor.
[0034] Compared with the prior art, in the technical solution of the present invention, according to the reference eigenvector matrix X of each load interval m×nCalculate the monitoring threshold in each load interval respectively, according to the sample to be tested and the benchmark feature vector matrix X m×n The Mahalanobis distance index and the monitoring threshold of the current load range are used to determine whether the motor is abnormal. In statistical process control, the operating status of the equipment is monitored through multi-dimensional features, and the monitoring indicators are adaptively constructed. The threshold is automatically calculated, and the load status of the equipment during operation is taken into account, which simplifies manual operations. The multi-dimensional characteristics of the application signal can avoid the specific defects of the traditional status monitoring threshold and improve the accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 It is a flow chart of the early warning method for abnormal vibration of a motor according to an embodiment of the present invention.
[0036] Figure 2 It is a structural schematic diagram of an early warning system for abnormal vibration of a motor according to an embodiment of the present invention. DETAILED DESCRIPTION
[0037] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0038] The implementation of the present invention is described in detail below in conjunction with specific embodiments.
[0039] like Figure 1 As shown, in an embodiment of the present invention, a method for early warning of abnormal vibration of a motor is provided, which includes steps S1-S5, which are described below respectively.
[0040] Step S1: Data acquisition.
[0041] It should be noted that before judging whether the motor vibration is abnormal, it is necessary to first collect the vibration data of the motor during normal operation, and extract the vibration data characteristics of the motor during normal operation from the vibration data of the motor during normal operation. Therefore, in this step, the electrical data and vibration data of the motor under various load conditions are collected.
[0042] Specifically, a vibration signal monitoring point is arranged at the motor drive end and the non-drive end respectively, which are used to monitor the operation of the motor drive end and the non-drive end respectively. The vibration sensor adopts a vibration acceleration sensor, and the sampling frequency of the sensors at the two measuring points is set above 20kHz. Voltage transformers and current transformers are used to collect the three-phase voltage and current signals input by the motor, and the sampling frequency is set at 5k~15kHz.
[0043] Step S2: Construct feature matrix.
[0044] In this step, first, the load factor of the motor is calculated based on the electrical data of the motor;
[0045] The load range of the sample is obtained according to its load rate, where the load range is divided according to the following rules:
[0046] The load rate ranges from 30% to 100%, and the load interval is selected according to the principle of 5 percentage points. The load interval is closed on the left and open on the right.
[0047] After obtaining the load range of the motor, the reference eigenvector matrix X of the vibration data of the motor in different load ranges is constructed respectively. m×n , where m is the number of statistical samples and n is the characteristic dimension of each sample data.
[0048] Step S3: Calculate the monitoring threshold of the motor in different load ranges.
[0049] It should be noted that the monitoring threshold of motor vibration is different in different load ranges. The specific calculation method is as follows:
[0050] First, the base eigenvector matrix X m×n Normalization obtains the normalized eigenvector matrix Z;
[0051] Then, the sample is calculated by Mahalanobis distance Where C represents the covariance matrix of matrix Z;
[0052] Finally, the monitoring threshold MD is calculated by the 3Sigma criterion threshol d=MD train ×3.
[0053] Step S4: Calculate the current load interval of the sample to be tested and the matrix x of the sample to be tested and the reference feature vector in the current load interval m×n The Mahalanobis distance indicator.
[0054] The specific process includes:
[0055] First, calculate the load rate of the sample, and obtain the load range of the sample according to the load rate of the sample;
[0056] Then, the feature vector y = [y1, y2, ..., yN] is extracted from the sample to be tested and normalized, and the baseline feature vector matrix X between it and the current load interval is calculated. m×n The corresponding Mahalanobis distance index of the normalized eigenvector matrix Z is: MID = yC -1 y T .
[0057] Step S5: Determine whether the motor is abnormal.
[0058] Specifically, the sample to be tested calculated in step S4 is compared with the reference feature vector matrix X m×n The Mahalanobis distance index of the motor is compared with the monitoring threshold of the corresponding load interval. When the Mahalanobis distance index of the three consecutive samples to be tested in the same load interval and the reference feature vector matrix exceeds the monitoring threshold, it is determined that the motor vibration state is abnormal to avoid misjudgment. After determining that the motor vibration state is abnormal, an alarm is issued.
[0059] like Figure 2 As shown, in an embodiment of the present invention, a warning system for abnormal vibration of a motor is also provided, which includes an acquisition module 1, a reference feature vector matrix construction module 2, a monitoring threshold calculation module 3, a Mahalanobis distance index calculation module 4 and a motor state judgment module 5.
[0060] The acquisition module 1 is used to collect vibration data and electrical data of the motor. In an embodiment of the present invention, the acquisition module 1 includes vibration acceleration sensors respectively arranged at the motor drive end and the non-drive end, respectively used to monitor the vibration signals at the motor drive end and the non-drive end. The acquisition module also includes a voltage transformer and a current transformer, respectively used to obtain the three-phase voltage and three-phase current data of the motor.
[0061] The reference eigenvector matrix construction module 2 is used to obtain the load range of the motor according to the electrical data of the motor, and to construct the reference eigenvector matrix X of the vibration data of the motor in different load ranges. m×n , where m is the number of statistical samples and n is the characteristic dimension of each sample data.
[0062] The monitoring threshold calculation module 3 is used to calculate the baseline characteristic vector matrix X of each load interval. m×n Calculate the monitoring threshold in each load interval respectively.
[0063] The Mahalanobis distance index calculation module 4 is used to calculate the current load interval of the sample to be tested and the matrix X between the sample to be tested and the reference feature vector in the current load interval. m×n The Mahalanobis distance indicator.
[0064] The motor state judgment module 5 is used to determine the state of the motor according to the sample to be tested and the reference feature vector matrix X m×n The motor is judged to be abnormal by the Mahalanobis distance index and the monitoring threshold of the current load interval. When the Mahalanobis distance index of the three consecutive samples to be tested in the same load interval and the reference feature vector matrix exceeds the monitoring threshold, it is determined that the motor vibration state is abnormal and an alarm is issued.
[0065] In summary, in the technical solution of the present invention, according to the reference eigenvector matrix X of each load interval m×nCalculate the monitoring threshold in each load interval respectively, according to the sample to be tested and the benchmark feature vector matrix X m×n The Mahalanobis distance index and the monitoring threshold of the current load range are used to determine whether the motor is abnormal. In statistical process control, the operating status of the equipment is monitored through multi-dimensional features, and the monitoring indicators are adaptively constructed. The threshold is automatically calculated, and the load status of the equipment during operation is taken into account, which simplifies manual operations. The multi-dimensional characteristics of the application signal can avoid the specific defects of the traditional status monitoring threshold and improve the accuracy of the model.
[0066] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for early warning of abnormal vibration of a motor, characterized in that: include: The vibration data and electrical data of the motor are collected, the load range of the motor is obtained according to the electrical data of the motor, and the reference eigenvector matrix X of the vibration data of the motor in different load ranges is constructed. m×n , where m is the number of statistical samples and n is the characteristic dimension of each sample data; According to the baseline eigenvector matrix X for each load interval m×n Calculate the monitoring threshold in each load interval respectively; Calculate the current load interval of the sample to be tested and the matrix X of the sample to be tested and the benchmark feature vector in the current load interval m×n The Mahalanobis distance indicator; According to the sample to be tested and the benchmark feature vector matrix X m×n The Mahalanobis distance index and the monitoring threshold of the current load range are used to determine whether the motor is abnormal. According to the baseline eigenvector matrix X for each load interval m×n Calculate the monitoring thresholds in each load interval separately, including: First, the base eigenvector matrix X m×n Normalization obtains the normalized eigenvector matrix Z; Then, the sample is calculated by Mahalanobis distance Where C represents the covariance matrix of matrix Z; Finally, the monitoring threshold MD is calculated by the 3Sigma criterion threshold =MD train ×3, Calculate the current load interval of the sample to be tested and the matrix X of the sample to be tested and the benchmark feature vector in the current load interval m×n The Mahalanobis distance indicators include: Calculate the load rate of the sample, and obtain the load range of the sample according to the load rate of the sample; Extract the feature vector y = [y1, y2, ..., yN] from the sample to be tested and perform normalization processing, and calculate its reference feature vector matrix X between the current load interval and the feature vector matrix X m×n The corresponding Mahalanobis distance index of the normalized eigenvector matrix Z is: MD = yC -1 y T .
2. The method for early warning of abnormal vibration of a motor according to claim 1, characterized in that: The load range of the motor is obtained based on the electrical data of the motor, including: Calculate the motor load factor based on the motor's electrical data; The load range of the sample is obtained according to its load rate, where the load range is divided according to the following rules: The load rate ranges from 30% to 100%, and the load interval is selected according to the principle of 5 percentage points. The load interval is closed on the left and open on the right.
3. The method for early warning of abnormal vibration of a motor according to claim 1, characterized in that: According to the sample to be tested and the benchmark feature vector matrix X m×n The Mahalanobis distance index is used to determine whether the motor is abnormal. include: Compare the collected samples to be tested with the benchmark feature vector matrix X m×n The Mahalanobis distance index is compared with the monitoring threshold of the corresponding load interval. When the Mahalanobis distance index of three consecutively collected samples to be tested in the same load interval and the benchmark eigenvector matrix exceeds the monitoring threshold, it is determined that the motor vibration state is abnormal and an alarm is issued.
4. The method for early warning of abnormal vibration of a motor according to claim 1, characterized in that: It is characterized in that The vibration data of the motor includes the vibration data of the motor drive end and the non-drive end.
5. The method for early warning of abnormal vibration of a motor according to claim 1, characterized in that: The electrical data of the motor includes three-phase voltage and current data.
6. An early warning system for abnormal vibration of a motor, characterized in that: include: An acquisition module is used to collect vibration data and electrical data of the motor; The reference eigenvector matrix construction module is used to obtain the load range of the motor according to the electrical data of the motor, and to construct the reference eigenvector matrix X of the vibration data of the motor in different load ranges. m×n , where m is the number of statistical samples and n is the characteristic dimension of each sample data; The monitoring threshold calculation module is used to calculate the baseline feature vector matrix X of each load interval. m×n Calculate the monitoring threshold in each load interval respectively; The Mahalanobis distance index calculation module is used to calculate the current load interval of the sample to be tested and the matrix X between the sample to be tested and the benchmark feature vector in the current load interval. m×n The Mahalanobis distance indicator; The motor state judgment module is used to determine the state of the motor according to the sample to be tested and the reference feature vector matrix X m×n The Mahalanobis distance index and the monitoring threshold of the current load range are used to determine whether the motor is abnormal. The monitoring threshold calculation module is based on the reference feature vector matrix X of each load interval. m×n Calculate the monitoring thresholds in each load interval separately, including: First, the base eigenvector matrix X m×n Normalization obtains the normalized eigenvector matrix Z; Then, the sample is calculated by Mahalanobis distance Where C represents the covariance matrix of matrix Z; Finally, the monitoring threshold MD is calculated by the 3Sigma criterion threshold =MD train ×3, The Mahalanobis distance index calculation module calculates the current load interval of the sample to be tested and the matrix X between the sample to be tested and the reference feature vector matrix X in the current load interval. m×n The Mahalanobis distance indicators include: Calculate the load rate of the sample, and obtain the load range of the sample according to the load rate of the sample; Extract the feature vector y = [y1, y2, ..., yN] from the sample to be tested and perform normalization processing, and calculate its reference feature vector matrix X between the current load interval and the feature vector matrix X m×n The corresponding Mahalanobis distance index of the normalized eigenvector matrix Z is: MD = yC -1 y T .
7. The early warning system for abnormal vibration of a motor as claimed in claim 6, characterized in that: The acquisition module includes vibration acceleration sensors respectively arranged at the motor driving end and the non-driving end, and are used to monitor the vibration signals of the motor driving end and the non-driving end respectively.
8. The early warning system for abnormal vibration of a motor according to claim 7, characterized in that: The acquisition module also includes a voltage transformer and a current transformer, which are used to obtain the three-phase voltage and three-phase current data of the motor respectively.
