Fault diagnosis method based on fault simulation experiment platform
By using a fault simulation experimental platform and machine learning classification model in mechanical equipment fault diagnosis, the problems of poor adaptability and parameter sensitivity of fault diagnosis models in the prior art are solved, and higher prediction accuracy and robustness are achieved.
Patent Information
- Application Number
- CN202510128906.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-27
- Publication Date
- 2025-05-30
AI Technical Summary
The prior art is difficult to improve the prediction accuracy and robustness of mechanical equipment fault diagnosis, especially when the system structure changes or new types of faults occur, the model is poorly adaptable and the parameters are sensitive.
By establishing a fault simulation experimental platform, vibration data in normal and fault states are collected, order analysis and feature index calculations are performed, feature indexes are standardized, and training sets and test sets are divided, which are used to train supervised machine learning classification models.
Ensure that machine learning models perform well in the face of various failures, improve the generalization ability and accuracy of the model, enhance the effectiveness of features, and make the model focus more on fault diagnosis tasks.
Smart Images
Figure CN120067798A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of mechanical equipment fault diagnosis, and particularly to a fault diagnosis method based on a fault simulation experimental platform. Background Art
[0002] With the rapid development of modern industrial technology, mechanical equipment fault diagnosis has become increasingly important because it is directly related to production efficiency, safety, and maintenance costs. Effective fault diagnosis can detect potential problems in advance, avoid unexpected shutdowns, reduce repair time and costs, and extend the service life of equipment.
[0003] Data-driven methods are usually widely used in the field of fault diagnosis. By collecting data in normal working conditions and fault conditions, a model is trained to identify different fault states. However, in practical applications, it is relatively difficult to obtain sufficient and representative fault data. Therefore, how to effectively generate data samples for training and improve the generalization ability of the model has become a key issue. In related technologies, mechanism-based and data-driven solutions are adopted. Among them, for mechanism-based fault models, their adaptability is poor. Once the system structure changes or new types of unexpected faults occur, the original model may no longer be applicable, and many mechanism models are very sensitive to parameter settings. Incorrect parameter selection may lead to misdiagnosis, and thus the prediction accuracy and robustness cannot be improved. Summary of the Invention
[0004] This application provides a fault diagnosis method based on a fault simulation experimental platform to solve the problem of inability to improve prediction accuracy and robustness.
[0005] This application provides a fault diagnosis method based on a fault simulation experimental platform, and the method includes:
[0006] Establish a fault simulation experimental platform according to the mechanical equipment to be diagnosed;
[0007] Collect vibration data of the fault simulation experimental platform; the vibration data is a sequence of running vibration data in normal and fault states;
[0008] Label the vibration data according to the state category to obtain a labeled sequence;
[0009] Perform order analysis on the vibration data and calculate characteristic indicators based on the fault mechanism;
[0010] Calculate the root mean square value of the vibration data and standardize the characteristic indicators to obtain standardized characteristic indicators;
[0011] Divide the standardized characteristic indicators and the labeled sequence into a training set and a test set to train a supervised machine learning classification model;
[0012] Predict the faults of the mechanical equipment to be diagnosed based on the real data collected from the mechanical equipment to be diagnosed and the trained machine learning classification model.
[0013] This application simulates various types of faults through a fault simulation experimental platform, which can ensure that the machine learning model has good performance in the face of various faults; collecting a large amount of fault data helps to improve the generalization ability and accuracy of the machine learning model; calculating feature indicators based on fault mechanisms for training the machine learning model not only has good scalability and versatility, but also can improve the effectiveness of features, making the machine learning model more focused on the fault diagnosis task and solving the problem of inability to improve prediction accuracy and robustness.
[0014] Optionally, the vibration data is:
[0015]
[0016] where k is the number of state categories; n is the number of sampling groups of each state data.
[0017] Optionally, the feature indicators are:
[0018] feature m =[feature 1 ,feature 2 ,…,feature m ;
[0019] where m is the number of data features in each group.
[0020] Optionally, the root mean square value of the vibration data is:
[0021]
[0022] where x i is a randomly sampled vibration data; n is the number of sampling groups of each state data.
[0023] Optionally, the formula for the standardized feature indicator is:
[0024]
[0025] where feature m is the feature indicator; rms is the root mean square value of the vibration data.
[0026] Optionally, the steps for predicting the faults of the mechanical equipment to be diagnosed based on the real data collected from the mechanical equipment to be diagnosed and the trained machine learning classification model include:
[0027] Collect the real data of the mechanical equipment to be diagnosed and perform noise reduction processing to obtain the processed real data;
[0028] Perform order analysis on the processed real data and calculate real characteristic indicators based on the fault mechanism;
[0029] Calculate the root mean square value of the processed real data and standardize the real characteristic indicators to obtain standardized real characteristic indicators;
[0030] Put the standardized real characteristic indicators and the annotation sequence into the trained machine learning classification model to predict the faults of the mechanical equipment to be diagnosed; wherein, the standardized real characteristic indicators are used as inputs and the annotation sequence is used as the output.
[0031] By performing noise reduction processing on the collected real data, removing background noise and interference signals, and retaining useful information related to faults, it helps to improve the accuracy of subsequent analysis; then perform order analysis on the noise-reduced data, identify signal components in different frequency bands, calculate real characteristic indicators that can reflect the operating state of the equipment based on the fault mechanism, and then eliminate the dimensional differences between different characteristic indicators through standardization processing.
[0032] Optionally, the structure of the fault simulation experimental platform is the same as that of the mechanical equipment to be diagnosed, and the powers of the fault simulation experimental platform are scaled down proportionally according to the powers of the mechanical equipment to be diagnosed.
[0033] The structure of the fault simulation experimental platform being the same as that of the mechanical equipment to be diagnosed can ensure that the operating state and working environment of the actual mechanical equipment can be truly simulated during the experiment, which helps to improve the accuracy and reliability of the experimental results. The powers of the fault simulation experimental platform being scaled down proportionally according to the powers of the mechanical equipment to be diagnosed can reduce the experimental cost and risk on the premise of ensuring the experimental effect.
[0034] Optionally, the steps of performing order analysis on the vibration data and calculating characteristic indicators based on the fault mechanism include:
[0035] Synchronously collect the rotation speed sequence of the fault simulation experimental platform and interpolate the length of the rotation speed sequence to be the same as that of the vibration data sequence;
[0036] Integrate the rotation speed sequence to obtain an angle sequence;
[0037] Taking the angle sequence as the horizontal axis, calculate the angle threshold sampling frequency of the vibration data sequence;
[0038] Perform fast Fourier transform on the vibration data sampled at equal angles to obtain an order spectrum;
[0039] Obtain the characteristic index according to the order spectrum.
[0040] Optionally, the angle sequence is:
[0041]
[0042] where speed 1,1 is the rotational speed sequence; i is the length of the vibration data sequence.
[0043] Optionally, the angular threshold sampling frequency of the vibration data sequence is:
[0044]
[0045] where fs is the sampling frequency of the vibration data sequence; speed 1,1 is the rotational speed sequence.
[0046] As can be seen from the above technical solutions, the present application provides a fault diagnosis method based on a fault simulation experiment platform. A fault simulation experiment platform is established according to the mechanical equipment to be diagnosed; vibration data of the fault simulation experiment platform is collected; the vibration data is an operating vibration data sequence in a normal state and a fault state; the vibration data is labeled according to the state category to obtain a labeled sequence; order analysis is performed on the vibration data, and characteristic indexes are calculated based on the fault mechanism; the root mean square value of the vibration data is calculated and the characteristic indexes are standardized to obtain standardized characteristic indexes; the standardized characteristic indexes and the labeled sequence are divided into a training set and a test set to train a supervised machine learning classification model; the fault of the mechanical equipment to be diagnosed is predicted according to the real data collected from the mechanical equipment to be diagnosed and the trained machine learning classification model, so as to solve the problem that the prediction accuracy and robustness cannot be improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] In order to more clearly illustrate the technical solutions of the present application, the drawings required for the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.
[0048] Figure 1 is a flowchart of the fault diagnosis method based on a fault simulation experiment platform provided by an embodiment of the present application;
[0049] Figure 2 is a flowchart of predicting a fault with real data in the fault diagnosis method based on a fault simulation experiment platform provided by an embodiment of the present application;
[0050] Figure 3Schematic diagram of calculating characteristic indexes in the fault diagnosis method based on a fault simulation experiment platform provided by an embodiment of the present application;
[0051] Figure 4 Spectrum amplitude diagram of the first 4 orders before the misalignment fault of the fault simulation experiment platform provided by an embodiment of the present application;
[0052] Figure 5 Schematic diagram of the trend change of the characteristic indexes of the first to fourth orders before the misalignment fault of the real data provided by an embodiment of the present application. Detailed implementation manners
[0053] The embodiments will be described in detail below, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following embodiments do not represent all implementation manners consistent with the present application. They are only examples of systems and methods consistent with some aspects of the present application.
[0054] To solve the problem of being unable to improve the prediction accuracy and robustness, refer to Figure 1 , taking the centrifugal fan as an example of the mechanical equipment to be diagnosed, an embodiment of the present application provides a fault diagnosis method based on a fault simulation experiment platform, and the method includes:
[0055] S100: Establish a fault simulation experiment platform according to the mechanical equipment to be diagnosed.
[0056] It should be understood that in some embodiments, the structure of the fault simulation experiment platform is the same as the structure of the mechanical equipment to be diagnosed, and the powers of each item of the fault simulation experiment platform are reduced proportionally according to the powers of each item of the mechanical equipment to be diagnosed.
[0057] The same structure of the fault simulation experiment platform as the structure of the mechanical equipment to be diagnosed can ensure that the running state and working environment of the actual mechanical equipment can be truly simulated during the experiment, which helps to improve the accuracy and reliability of the experimental results. The powers of each item of the fault simulation experiment platform are reduced proportionally according to the powers of each item of the mechanical equipment to be diagnosed, which can reduce the experimental cost and risk on the premise of ensuring the experimental effect.
[0058] In some other embodiments, if there are some complex structures in the mechanical equipment to be diagnosed that are irrelevant to fault diagnosis (such as the decorative patterns on the surface of the mechanical equipment), the structure of the fault simulation experiment platform can only be similar to the structure of the mechanical equipment to be diagnosed, and there is no need to be exactly the same. For example, the centrifugal fan fault simulation test platform shown in the embodiment of the present application uses a 1.5kw motor, a 4-jaw elastic coupling, and a centrifugal fan with 12 blades.
[0059] S200: Collect the vibration data of the fault simulation experiment platform.
[0060] The vibration data is a sequence of operating vibration data in normal and faulty states. It should be understood that the vibration data sampling settings of the faulty simulation experiment platform should be consistent with those of the mechanical equipment to be diagnosed. For example, in this case, the sampling frequency is 2560 Hz and the sampling length is 8192 points. Due to the diversity of faulty states, multiple sequences of operating vibration data need to be collected for different faults.
[0061] In some embodiments, the vibration data is:
[0062]
[0063] where k is the number of state categories; n is the number of sampling groups for each type of state data.
[0064] For example, the number of state categories k of the centrifugal fan faulty simulation test platform can be preset to 5. Among them, the vibration data in the normal state of the faulty simulation experiment platform is collected as basic data, and the basic data sequence data 1,n = [data 1,1 , data 1,2 , …, data 1,n ; counterweights are added to the shafting to simulate an imbalance fault, and the imbalance fault sequence data 2,n = [data 2,1 , data 2,2 , …, data 2,n is collected; the axial positions of the motor and the fan are changed to form a parallel misalignment fault state to simulate a misalignment fault, and the misalignment fault sequence data 3,n = [data 3,1 , data 3,2 , …, data 3,n is collected; the fastening bolts of the centrifugal fan to the foundation are loosened to simulate a foundation bolt loosening fault, and the loosening fault sequence data 4,n = [data 4,1 , data 4,2 , …, data 4,n is collected; the fan impeller is replaced with a faulty impeller with one damaged impeller to simulate an impeller fault, and the impeller fault sequence data 5,n = [data 5,1 , data 5,2 , …, data 5,n is collected.
[0065] S300: Label the vibration data according to the state categories to obtain a labeled sequence.
[0066] It should be understood that the annotation sequence is label k = [label 1 , label 2 , …, label k ; where the number of fault status categories is k - 1, and the first category is the vibration data under normal conditions. For example, set the type label of data 1,n to 0, that is, label 1 = 0; set the type label of data 2,n to 1, that is, label 2 = 1; set the type label of data 3,n to 2, that is, label 3 = 2; set the type label of data 4,n to 3, that is, label 4 = 3; set the type label of data 5,n to 4, that is, label 5 = 4.
[0067] S400: Perform order analysis on the vibration data and calculate characteristic indicators based on the fault mechanism.
[0068] It should be understood that even though the fault simulation experimental platform can collect data under continuously variable rotational speed conditions, the rotational speed of the equipment in the actual field may not be consistent with the rotational speed of the data collected by the fault simulation experimental platform. By performing order analysis on the vibration data, the influence of different rotational speeds can be eliminated, and the generalization ability of the model can be improved. The order calculation scheme of equal-angle resampling can be adopted in the embodiments of the present application.
[0069] The characteristic indicators in the embodiments of the present application are summarized from the fault mechanism. It is pointed out in many literatures that the data 2,n , data 3,n , data 4,n , data 5,n in step S200 have certain characteristics. Refer to Figure 4 . In the misalignment fault data of the fault simulation experimental platform, the vibration data collected by the fault simulation experimental platform shows that among the amplitudes of the first four orders, the amplitude of the second order increases significantly. At the same time, refer to Figure 5 . The change trend of the amplitudes of the first four orders of the misalignment fault actually occurring on site. The present application does not elaborate on its principle, but the simulation waveform made according to the fault generation mechanism is close to the actual fault waveform, so the characteristic indicators extracted according to the fault mechanism are reasonable. In other fault types, different characteristic indicators can also be extracted according to different characteristics.
[0070] In some embodiments, the characteristic indicators are:
[0071] feature m = [feature 1 , feature 2 , …, feature m ;
[0072] Among them, m is the number of data features in each group.
[0073] S500: Calculate the root mean square value of the vibration data and standardize the feature index to obtain a standardized feature index;
[0074] It should be understood that standardizing the feature index aims to reduce the vibration amplitude difference caused by the differences in power and load between the fault simulation test platform and the actual equipment. After standardization, the feature index removes the dimension, and its actual meaning becomes the ratio of the energy of this order component to the total amplitude.
[0075] In some embodiments, the root mean square value of the vibration data is:
[0076]
[0077] where x i is a vibration data sampled randomly; n is the number of sampling groups of each type of state data.
[0078] In some embodiments, the formula for the standardized feature index is:
[0079]
[0080] where feature m is the feature index; rms is the root mean square value of the vibration data.
[0081] S600: Divide the standardized feature index and the annotation sequence into a training set and a test set to train a supervised machine learning classification model.
[0082] It should be understood that the standardized feature index and the annotation sequence can be divided into a training set and a test set according to a ratio of 8:2. The machine learning classification model is not specifically limited. An XGBoost model with good test results after multiple tests can be selected. Use the standardized feature index sequence feature_normalize k,n as the input, and the label sequence label k as the output to train the machine learning classification model.
[0083] S700: Predict the faults of the mechanical equipment to be diagnosed based on the real data collected from the mechanical equipment to be diagnosed and the trained machine learning classification model.
[0084] This application simulates various types of faults through a fault simulation experimental platform, which can ensure that the machine learning model has good performance in the face of various faults; collecting a large amount of fault data helps to improve the generalization ability and accuracy of the machine learning model; calculating characteristic indicators based on the fault mechanism for training the machine learning model not only has good scalability and generality, but also can improve the effectiveness of the features, enabling the machine learning model to focus more on the fault diagnosis task and solving the problem of inability to improve the prediction accuracy and robustness.
[0085] In some embodiments, referring to Figure 2 , the steps of predicting the faults of the mechanical equipment to be diagnosed based on the real data collected from the mechanical equipment to be diagnosed and the trained machine learning classification model include:
[0086] S710: Collect the real data of the mechanical equipment to be diagnosed and perform noise reduction processing to obtain the processed real data;
[0087] S720: Perform order analysis on the processed real data and calculate the real characteristic indicators based on the fault mechanism;
[0088] S730: Calculate the root mean square value of the processed real data and standardize the real characteristic indicators to obtain the standardized real characteristic indicators;
[0089] S740: Put the standardized real characteristic indicators and the annotation sequence into the trained machine learning classification model to predict the faults of the mechanical equipment to be diagnosed.
[0090] Among them, the standardized real characteristic indicators are used as the input, and the annotation sequence is used as the output.
[0091] It should be understood that when collecting the real data of the mechanical equipment to be diagnosed on site, noise reduction processing needs to be carried out first, because the vibration data of the fault simulation experimental platform has relatively less noise, while the real data may have more noise. Using noise reduction processing can eliminate the difference between the vibration data of the fault simulation experimental platform and the real data as much as possible.
[0092] By performing noise reduction processing on the collected real data, removing background noise and interference signals, and retaining the useful information related to faults, it helps to improve the accuracy of subsequent analysis; then performing order analysis on the noise-reduced data to identify the signal components in different frequency bands, calculating the real characteristic indicators that can reflect the operating state of the equipment based on the fault mechanism, and then through standardization processing, eliminating the dimensional difference between different characteristic indicators.
[0093] In some embodiments, referring to Figure 3 , the steps of performing order analysis on the vibration data and calculating characteristic indicators based on the failure mechanism include:
[0094] S410: Synchronously collect the rotation speed sequence of the fault simulation experiment platform, and interpolate the length of the rotation speed sequence to be the same as the length of the vibration data sequence.
[0095] It should be understood that if the lengths of the two are the same, no transformation is required; if the lengths are different, transformation is needed. For example, the vibration data sequence data 1,1 = [x 1 , x 2 , …, x 8192 has a length of 8192, and the rotation speed sequence speed 1,1 = [rpm 1 , rpm 2 , …, rpm 160 has a length of 160. Since the lengths are different, the cubic spline interpolation method can be used to interpolate the rotation speed sequence to expand its length to 8192:
[0096] speed 1,1 = p cubic spline (speed 1,1 ) = [rpm 1 , rpm 2 , …, rpm 8192 .
[0097] S420: Integrate the rotation speed sequence to obtain an angle sequence.
[0098] In some embodiments, the angle sequence is:
[0099]
[0100] where speed 1,1 is the rotation speed sequence; i is the length of the vibration data sequence.
[0101] S430: Using the angle sequence radio 1,1 as the horizontal axis, calculate the angle threshold sampling frequency of the vibration data sequence data 1,1 .
[0102] In some embodiments, the angle threshold sampling frequency of the vibration data sequence is:
[0103]
[0104] where fs is the sampling frequency of the vibration data sequence; speed1,1 is the rotational speed sequence.
[0105] It should be understood that in the embodiments of the present application, fs is 2560. According to rad fs The angular domain is resampled at an equal-angle resampling interval of rad, and in this embodiment, rad fs = 256, that is, 256 points are sampled per revolution. The points in the vibration sequence that cannot be resampled are fitted by cubic spline interpolation. The resampled vibration sequence data is:
[0106] data_order 1,1 = [order 1 , orderx 2 , …, order i ;
[0107] S440: Perform a fast Fourier transform on the vibration data sampled at equal angles to obtain an order spectrum.
[0108] It should be understood that the fast Fourier transform (FFT) is an efficient algorithm for calculating the discrete Fourier transform (DFT) and its inverse transform. By reducing the amount of calculation and optimizing the calculation process, the efficiency of spectrum analysis is significantly improved.
[0109] S450: Obtain the characteristic index according to the order spectrum.
[0110] It should be understood that the order selection calculation can calculate the amplitudes of orders 1 to 8 and the amplitudes of 1 and 2 times the blade passing order.
[0111] The calculations of 1 and 2 times the blade passing order are respectively:
[0112] order blp_1 = 1 × num blade ;
[0113] order blp_2 = 1 × num blade × 2;
[0114] In the embodiments of the present application, num blade = 12, that is, the amplitudes of 1 and 2 times the blade passing order are the amplitudes of orders 12 and 24 respectively. The amplitudes of orders 1 to 8, 12, and 24 in the final order spectrum form the final characteristic sequence:
[0115] feature 1,1 = [feature 1 , feature 2 , …, feature m ;
[0116] In the embodiments of the present application, m = 10.
[0117] As can be seen from the above technical solutions, the embodiments of the present application provide a fault diagnosis method based on a fault simulation experiment platform. A fault simulation experiment platform is established according to the mechanical equipment to be diagnosed; vibration data of the fault simulation experiment platform is collected; the vibration data is an operation vibration data sequence in a normal state and a fault state; the vibration data is labeled according to the state category to obtain a labeled sequence; order analysis is performed on the vibration data, and characteristic indexes are calculated based on the fault mechanism; the root mean square value of the vibration data is calculated and the characteristic indexes are standardized to obtain standardized characteristic indexes; the standardized characteristic indexes and the labeled sequence are divided into a training set and a test set to train a supervised machine learning classification model; the fault of the mechanical equipment to be diagnosed is predicted according to the real data collected from the mechanical equipment to be diagnosed and the trained machine learning classification model, so as to solve the problem that the prediction accuracy and robustness cannot be improved.
[0118] For the similarities between the embodiments provided in the present application, reference can be made to each other. The specific embodiments provided above are only several examples under the general concept of the present application and do not constitute a limitation on the protection scope of the present application. For those skilled in the art, any other implementation manner extended based on the solution of the present application without creative efforts belongs to the protection scope of the present application.
Claims
1. A fault diagnosis method based on a fault simulation experiment platform, characterized in that: The method comprises: Establish a fault simulation experimental platform based on the mechanical equipment being diagnosed; Collecting vibration data of the fault simulation experiment platform; the vibration data is a sequence of operating vibration data under normal state and fault state; annotating the vibration data according to the state category to obtain an annotated sequence; performing order analysis on the vibration data and calculating characteristic indicators based on the fault mechanism; Calculating a root mean square value of the vibration data and normalizing the characteristic index to obtain a normalized characteristic index; Dividing the standardized feature index and the labeled sequence into a training set and a test set to train a supervised machine learning classification model; The fault of the diagnosed mechanical equipment is predicted based on the real data collected from the diagnosed mechanical equipment and the trained machine learning classification model.
2. The fault diagnosis method based on the fault simulation experiment platform according to claim 1 is characterized in that: The vibration data are: Among them, k is the number of state categories; n is the number of sampling groups of each type of state data.
3. The fault diagnosis method based on the fault simulation experiment platform according to claim 1 is characterized in that: The characteristic indicators are: feature m =[feature1,feature2,…,feature m ]; Among them, m is the number of features in each group of data.
4. The fault diagnosis method based on the fault simulation experiment platform according to claim 3 is characterized in that: The root mean square value of the vibration data is: Among them, x i is a randomly sampled vibration data; n is the number of sampling groups for each type of state data.
5. The fault diagnosis method based on the fault simulation experiment platform according to claim 4 is characterized in that: The formula of the standardized characteristic index is: Among them, feature m is the characteristic index; rms is the root mean square value of the vibration data.
6. The fault diagnosis method based on the fault simulation experiment platform according to claim 1 is characterized in that: The steps of predicting the fault of the diagnosed mechanical equipment based on the real data collected from the diagnosed mechanical equipment and the trained machine learning classification model include: Collect the real data of the diagnosed mechanical equipment and perform noise reduction processing to obtain processed real data; Perform order analysis on the processed real data and calculate the real characteristic indicators based on the fault mechanism; Calculating the root mean square value of the processed real data and standardizing the real feature index to obtain a standardized real feature index; The standardized real feature index and the labeled sequence are placed into a trained machine learning classification model to predict the diagnosed mechanical equipment fault; wherein the standardized real feature index is used as input and the labeled sequence is used as output.
7. The fault diagnosis method based on the fault simulation experiment platform according to claim 1 is characterized in that: The structure of the fault simulation experiment platform is the same as that of the mechanical equipment to be diagnosed, and the power of each item of the fault simulation experiment platform is reduced in proportion to the power of each item of the mechanical equipment to be diagnosed.
8. The fault diagnosis method based on the fault simulation experiment platform according to claim 1 is characterized in that: The steps of performing order analysis on the vibration data and calculating characteristic indicators based on the fault mechanism include: Synchronously collecting a speed sequence of the fault simulation experiment platform, and interpolating the length of the speed sequence to the same length as the vibration data sequence; Integrating the rotation speed sequence to obtain an angle sequence; Taking the angle sequence as the horizontal axis, calculating the angle threshold sampling frequency of the vibration data sequence; Perform fast Fourier transform on the vibration data sampled at equal angles to obtain the order spectrum; The characteristic index is obtained according to the order spectrum.
9. The fault diagnosis method based on the fault simulation experiment platform according to claim 8 is characterized in that: The angle sequence is: Among them, speed 1,1 is the speed sequence; i is the length of the vibration data sequence.
10. The fault diagnosis method based on the fault simulation experiment platform according to claim 9 is characterized in that: The vibration data sequence angle threshold sampling frequency is: Where, fs is the sampling frequency of the vibration data sequence; speed 1,1 The speed sequence.