An intelligent recognition system for the center orbit of a sliding bearing

Abstract

The present invention discloses an intelligent recognition system for the center orbit of a sliding bearing. The key points of the technical solution are as follows: including the following steps: Step 1, after starting, batch-simulate the displacement signals of the center orbits of two sliding bearings; Step 2, convert the data generated in batch in Step 1 into a two-dimensional matrix; Step 3, add labels to the two-dimensional matrix in Step 2; Step 4, train the intelligent recognition model for the center orbit; Step 5, obtain the intelligent recognition model after testing; Step 6, collect the displacement signals of two mutually perpendicular directions of the sliding bearing; Step 7, respectively fit the signals containing only the 1st and 2nd harmonics from the two signals; Step 8, transform the fitted signals into a two-dimensional matrix; the two-dimensional matrix is input into the intelligent recognition model for the recognition of the center orbit shape; this system can complete the training of the model only with simulated signals without the need for a large number of real samples, and the recognition accuracy of the signals adapted to the recognition model obtained by fitting is high and significantly improved compared with the traditional method.

Description

Technical Field

[0001] The present invention relates to the field of identification systems, and particularly to an intelligent identification system for the center orbit of a sliding bearing. Background Art

[0002] The main shaft supports of rotating equipment generally fall into two categories: contact bearings and non-contact bearings. A sliding bearing is a non-contact bearing. Under liquid lubrication conditions, it has many advantages such as a small friction coefficient, allowing a very high rotational speed to be achieved, and oil film vibration absorption. And the monitoring of the center orbit is widely applied in the fault diagnosis of sliding bearings. The center orbit refers to the movement orbit of a point on the axis relative to the bearing housing, and this orbit is in a plane perpendicular to the axis. Therefore, the center orbit is generally synthesized from the vibration displacement signals monitored by two mutually perpendicular displacement sensors on the same cross-section (the two displacement sensors are perpendicular to the axis pairwise).

[0003] Center orbits of different shapes of sliding bearings often correspond to different health states. For the rotor imbalance fault, the time-domain waveform of its displacement signal is a sinusoidal waveform with equal amplitude, and the corresponding center orbit is a stable ellipse with a small difference between the major and minor axes. For the rotor misalignment fault, the time-domain waveform is a sinusoidal waveform with a double peak, and the center orbit is banana-shaped or crescent-shaped, and will become an external "8" shape when the fault is serious. For the oil film whirl fault, its center orbit is an internal "8" shape.

[0004] Due to the complex industrial environment, the monitored center orbit signals are often accompanied by relatively large noise. At present, the research difficulty of the center orbit of a sliding bearing is how to obtain a clear center orbit and classify and identify it. In order to obtain a clear center orbit image, scholars generally use principles such as wavelet transform, harmonic wavelet, and EEMD to purify the center orbit. However, the purification effect is seriously affected by the signal-to-noise ratio of the original signal. In terms of the automatic identification and classification of the center orbit, there are two relatively classic methods: one is to use feature extraction based on the Hu invariant moment theory and the Mahalanobis distance as a classifier to map different types of center orbits, or to change the Mahalanobis distance to use classifiers such as support vector machines and neural networks as variants of this method; the second is to collect different displacement signals, synthesize different center orbit images as training samples, train an intelligent identification model for the center orbit based on a convolutional neural network, and directly classify and identify the center orbit images; the third is to collect different displacement signals, synthesize different center orbit images as training samples, train an intelligent identification model for the center orbit based on a convolutional neural network, and directly classify and identify the center orbit images.

[0005] Generally speaking, the current technical means have the following defects:

[0006] First, invariant moments are extracted based on the Hu invariant moment theory as features. These features have extremely strong instability. The different signal acquisition durations and noise levels will seriously affect the recognition accuracy. Therefore, the recognition accuracy of this method is relatively low.

[0007] Second, displacement signals under different health conditions are collected, and different shaft center orbit signals are synthesized for training the neural network model. This method is very feasible in theory. However, it is difficult to collect a large amount of real fault shaft center orbit data in the actual industrial environment, resulting in insufficient training samples and relatively single training data. Therefore, the trained model does not have good generalization performance.

[0008] Third, for some equipment, the vibration noise at the workshop site is extremely large. Sometimes, even with the human eye, it is difficult to distinguish which shape the purified shaft center orbit belongs to. At this time, it is obviously meaningless to input the signal into the model; (because the trained neural network is difficult to reach the level of expert diagnosis, and even experts are difficult to classify it into a certain shape). Summary of the Invention

[0009] In view of the problems mentioned in the background art, the purpose of the present invention is to provide an intelligent recognition system for the shaft center orbit of a sliding bearing to solve the problems mentioned in the background art.

[0010] The above technical object of the present invention is achieved through the following technical solutions:

[0011] An intelligent recognition system for the shaft center orbit of a sliding bearing includes the following steps:

[0012] Step 1: After starting, batch-simulate the displacement signals of the shaft center orbit of two sliding bearings;

[0013] Step 2: Convert the data produced in batch in Step 1 into a two-dimensional matrix;

[0014] Step 3: Add labels to the two-dimensional matrix in Step 2;

[0015] Step 4: Train the intelligent recognition model for the shaft center orbit;

[0016] Step 5: Obtain the intelligent recognition model after testing;

[0017] Step 6: Collect the displacement signals of two perpendicular directions of the sliding bearing;

[0018] Step 7: Fit the signals containing only the 1st and 2nd harmonics respectively from the two signals;

[0019] Step 8: Transform the fitted signals to obtain a two-dimensional matrix;

[0020] Step 9: Input the two-dimensional matrix in Step 8 into the intelligent recognition model in Step 5 for recognition;

[0021] Step 10: Identify the shape of the shaft center locus to obtain the recognition result, and then end.

[0022] Preferably, in Step 1, the batch simulation only includes displacement signals composed of 1 times and 2 times frequencies, and the formula for simulating the displacement signal is as follows:

[0023] (1);

[0024] where f is a set value, and by changing the value, two-way displacement signals corresponding to the shaft center locus shape are batch-generated.

[0025] Preferably, in Step 2, the two generated two-way displacement signals are transformed to obtain a two-dimensional matrix that can characterize the shaft center locus shape, and the two-dimensional matrix is composed of 0 and 1; the transformation method for changing the two signals into a two-dimensional matrix is as follows:

[0026] S21: Denote the peak-to-peak value and the minimum value of as , respectively, and denote the peak-to-peak value and the minimum value of as , respectively. Let , , where is a regulation coefficient, and its value range is [1.2, 2] to prevent the shaft center locus image from falling on the edge of the two-dimensional matrix and avoid information loss during the convolution operation of the generated matrix;

[0027] S22: Initialize the two-dimensional matrix representing the shaft center locus shape, denoted as a square matrix of width , which contains elements all equal to 0;

[0028] S23: Traverse the values x and y in and in pairs, and let:

[0029] (2);

[0030] where, has a value range of [0.1, 0.2], and the purpose is to prevent information loss during the convolution operation of the generated matrix; the formula (2) is to assign the value of 1 to the element corresponding to the matrix at the th row and the th column;

[0031] After the traversal ends, the obtained matrix is the two-dimensional matrix that can represent the shape of the shaft center locus.

[0032] Preferably, the two-dimensional matrix obtained by batch conversion in step 3 is added with a label, and the label is encoded in the One-Hot manner.

[0033] Preferably, in step 4, the two-dimensional matrix generated by the analog signal is used to train the intelligent recognition model of the shaft center locus, and the intelligent recognition model is tested to ensure the accuracy rate.

[0034] Preferably, in step 6, the two-channel shaft center locus displacement signals monitored are processed by the least square method to fit out the displacement signals containing only the fundamental frequency and the second harmonic; the two displacement signals are respectively fitted, and the specific fitting method is as follows:

[0035] Denote one of the two displacement signals collected as , the number of sampling points is , the signal to be fitted is denoted as , and The th data points in are respectively denoted as and , the to be fitted is written as the expression:

[0036] (3);

[0037] Where is the rotational frequency of the shaft. Assuming it has been obtained, when the fitting effect is the best, it is and The distance between is the smallest, that is, when the following formula (4) obtains the minimum value:

[0038] (4);

[0039] Expand the above formula (3) to get:

[0040] (5);

[0041] Let:

[0042] (6);

[0043] Substitute (6) into (5), and the expression of the to be fitted is abbreviated as:

[0044] (7);

[0045] Substituting (7) into (4) gives (8), that is, the goal is to minimize equation (8):

[0046] (8);

[0047] To minimize equation (8), according to the minimum value theorem of continuous functions, we have:

[0048] (9);

[0049] Expanding equation (9) gives:

[0050] (10);

[0051] Rearranging equation (10) gives a system of equations:

[0052] (11);

[0053] Let:

[0054] , ,

[0055] Then equation (11) can be written as:

[0056] (12);

[0057] Furthermore, the solutions of each parameter are obtained:

[0058] (13);

[0059] Since the value has been obtained, from (6), we have

[0060] , , , ;

[0061] Above, the fitting of one displacement signal is completed, and the other collected displacement signal is fitted in the same way.

[0062] Preferably, the transformation method used in step 8 to transform the signal obtained from fitting into a two-dimensional matrix is the same as the calculation method in step 2 for converting the produced data into a two-dimensional matrix.

[0063] In summary, the present invention mainly has the following beneficial effects:

[0064] The intelligent recognition system for the center orbit of the sliding bearing does not require a large number of difficult-to-obtain real samples and can complete the training of the model only with simulated signals. It can fit the signals suitable for the recognition model from high-noise signals, improving the generalization of the model. The recognition accuracy is significantly improved compared with the traditional method. The whole solution is obtained through the combined use of the model training method and the fitting method of the collected center orbit signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 is the system block diagram of the present invention;

[0066] Figure 2 is the diagram showing two-way displacement signals collected in the present invention;

[0067] Figure 3 is the original signal center orbit diagram in the present invention;

[0068] Figure 4 is the diagram showing two-way displacement signals fitted in the present invention;

[0069] Figure 5 is the fitted signal center orbit diagram in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0071] Embodiment 1

[0072] Refer to Figures 1 to 5 , an intelligent recognition system for the center orbit of a sliding bearing, comprising the following steps:

[0073] Step 1: After starting, batch-simulate two-way displacement signals of the center orbit of the sliding bearing;

[0074] Step 2: Convert the data produced in batch in Step 1 into a two-dimensional matrix;

[0075] Step 3: Add labels to the two-dimensional matrix in Step 2;

[0076] Step 4: Train the intelligent recognition model for the center orbit;

[0077] Step 5: Obtain the intelligent recognition model after testing;

[0078] Step 6: Collect two-way displacement signals perpendicular to each other of the sliding bearing;

[0079] Step 7: Fit the signals containing only the 1st and 2nd harmonics from the two signals respectively;

[0080] Step 8: Transform the fitted signals to obtain a two-dimensional matrix;

[0081] Step 9: Input the two-dimensional matrix in Step 8 into the intelligent recognition model in Step 5 for recognition;

[0082] Step 10: Identify the shape of the shaft center orbit to obtain the recognition result, and then end.

[0083] Among them, in Step 1, a batch of displacement signals containing only the 1st and 2nd harmonic components are simulated. The formula for simulating the displacement signals is as follows:

[0084] (1);

[0085] where f is a set value, and by varying the value, two displacement signals corresponding to the shaft center orbit shape are generated in batches.

[0086] Among them, in Step 2, the two generated displacement signals are transformed to obtain a two-dimensional matrix that can represent the shaft center orbit shape. The two-dimensional matrix consists of 0s and 1s; the transformation method for changing the two signals into the two-dimensional matrix is as follows:

[0087] S21: Denote the peak-to-peak value and the minimum value of as , respectively, and denote the peak-to-peak value and the minimum value of as , respectively. Let , , where is a regulation coefficient, and its value range is [1.2, 2], to avoid the shaft center orbit image falling on the edge of the two-dimensional matrix and prevent information loss during the convolution operation of the generated matrix;

[0088] S22: Initialize the two-dimensional matrix representing the shaft center orbit shape, denoted as a square matrix of width , containing elements all equal to 0;

[0089] S23: Traverse the values x and y in and in pairs, and let:

[0090] (2);

[0091] Among them, The value range of is [0.1, 0.2], aiming to prevent information loss during the convolution operation of the generated matrix; Equation (2) is to make the value of the element corresponding to the th row and

[0092] the matrix obtained after traversal is the two-dimensional matrix that can represent the shape of the shaft center orbit.

[0093] Among them, in Step 3, the two-dimensional matrix obtained by batch conversion is added with a label, and the label is encoded in the One-Hot manner.

[0094] Among them, in Step 4, the two-dimensional matrix generated by the analog signal is used to train the intelligent shaft center orbit recognition model, and the intelligent recognition model is tested to ensure the accuracy rate.

[0095] Among them, in Step 6, the least squares method is used to process the two monitored shaft center orbit displacement signals, and the displacement signals containing only the fundamental frequency and the second harmonic are fitted; the two displacement signals are fitted separately, and the specific fitting method is as follows:

[0096] Denote one of the two collected displacement signals as , the number of sampling points is , the signal to be fitted is denoted as , and the th data points in and are respectively denoted as and

[0097] (3);

[0098] Among them is the rotational frequency of the shaft, assuming it has been obtained. When the fitting effect is the best, it is when and the distance between them is the smallest, that is, when the following formula (4) obtains the minimum value:

[0099] (4);

[0100] Expand the above formula (3) to get:

[0101] (5);

[0102] Let:

[0103] (6);

[0104] Substitute (6) into (5), and the expression to be fitted is abbreviated as: :

[0105] (7);

[0106] Substitute (7) into (4) to get (8), that is, the goal is to minimize equation (8):

[0107] (8);

[0108] To minimize equation (8), according to the minimum value theorem of continuous functions, we have:

[0109] (9);

[0110] Expand equation (9) to get:

[0111] (10);

[0112] Rearrange equation (10) to obtain a system of equations:

[0113] (11);

[0114] Let:

[0115] , ,

[0116] Then equation (11) can be written as:

[0117] (12);

[0118] Furthermore, solve for the parameters:

[0119] (13);

[0120] Since has been obtained, from (6), we have

[0121] , , , ;

[0122] Above, the fitting of one displacement signal is completed. Fit the other collected displacement signal in the same way.

[0123] Among them, the transformation method used in step 8 to transform the fitted signal into a two-dimensional matrix is the same as the calculation method for converting the data generated in step 2 into a two-dimensional matrix.

[0124] Among them, the intelligent recognition system for the center orbit of the sliding bearing does not require a large number of difficult-to-obtain real samples, and can complete the training of the model only with simulation signals; it fits the signals suitable for the recognition model from high-noise signals to improve the generalization of the model; the recognition accuracy is significantly improved compared with the traditional method; the whole solution is obtained by the combined use of the model training method and the fitting method of the collected center orbit signals.

[0125] Embodiment 2

[0126] Reference Figures 1 to 5 , this embodiment is mainly used to identify four different operating health states of the sliding bearing, namely imbalance, misalignment, oil film whirl, and normal; the ideal center orbits corresponding to these 4 health states are mainly composed of the fundamental frequency and the second harmonic, so the present invention first simulates the center orbit images in various health states under ideal conditions to train the intelligent recognition model for the center orbit; secondly, extracts the time-domain signals containing only the fundamental frequency and the second harmonic from the collected original vibration displacement signals, then converts them into center orbit images, and inputs them into the trained model for recognition to obtain their health states; the specific steps of the solution are as follows:

[0127] Step 1: Batch-simulate the displacement signals composed only of the fundamental frequency and the second harmonic. The formula for simulating the displacement signals is as follows:

[0128] (1);

[0129] where f is a set value, and by changing the value, two-way displacement signals corresponding to the shape of the center orbit are generated in batches.

[0130] Step 2: Transform the two-way displacement signals generated in Step 1 to obtain a two-dimensional matrix that can represent the shape of the center orbit. The two-dimensional matrix is composed of 0 and 1; the following is a specific transformation method for changing the two-way signals into a two-dimensional matrix:

[0131] Step 21: Denote the peak-to-peak value and the minimum value of as , , respectively, and denote the peak-to-peak value and the minimum value of as , , respectively. Let , , where is a regulation coefficient, and its value range is [1.2, 2] to avoid the center orbit image falling on the edge of the two-dimensional matrix. The meaning is to prevent information loss during the convolution operation of the generated matrix.

[0132] Step 22: Initialize a two-dimensional matrix representing the shape of the shaft center orbit, denoted as a square matrix of order width , which contains elements all equal to 0.

[0133] Step 23: Traverse the values x and y in and in pairs, and let:

[0134] (2);

[0135] Among them, has a value range of [0.1, 0.2], and the purpose is also to prevent information loss during the convolution operation of the generated matrix; the meaning of formula (2) is to assign the value of 1 to the element corresponding to the matrix at the th row and the th column.

[0136] After the traversal, the obtained matrix is the two-dimensional matrix that can represent the shape of the shaft center orbit.

[0137] Step 3: Add labels to the two-dimensional matrices batch-converted in Step 2, and the labels are encoded in the One-Hot manner.

[0138] Step 4: Use the two-dimensional matrices generated by the analog signal to train the intelligent shaft center orbit recognition model, and test the intelligent shaft center orbit recognition model to ensure the accuracy rate.

[0139] Step 5: After the intelligent shaft center orbit recognition model passes the test and ensures the passing rate, obtain the required intelligent recognition model.

[0140] Step 6: Install two mutually perpendicular eddy current displacement sensors in a certain cross-section perpendicular to the motor spindle to collect two mutually perpendicular displacement signals of the sliding bearing.

[0141] Step 7: Process the monitored two-way shaft center orbit displacement signals by the least squares method to fit out the displacement signals containing only the fundamental frequency and the second harmonic; the two displacement signals are fitted separately, and the specific fitting method is as follows:

[0142] Denote one of the two collected displacement signals as , the number of sampling points is , the signal to be fitted is denoted as , and The th data points in and are denoted as and

[0143] (3);

[0144] Here is the rotational frequency of the shaft, which is assumed to have been obtained; when the fitting effect is the best, it must be and the distance between them is the smallest, that is, the following formula (4) obtains the minimum value:

[0145] (4);

[0146] Expand the above formula (3) to get:

[0147] (5);

[0148] Let:

[0149] (6);

[0150] Substitute (6) into (5), and the expression of to be fitted is abbreviated as:

[0151] (7);

[0152] Substitute (7) into (4) to get (8), that is, the goal is to make formula (8) obtain the minimum value:

[0153] (8);

[0154] And to make formula (8) obtain the minimum value, according to the minimum value theorem of continuous functions, we have:

[0155] (9);

[0156] Expand formula (9) to get:

[0157] (10);

[0158] Rearrange formula (10) to get the system of equations:

[0159] (11);

[0160] Let:

[0161] , , ;

[0162] Then formula (11) is written as:

[0163] (12);

[0164] Furthermore, the solutions of the respective parameters are obtained:

[0165] (13);

[0166] Since the value of has been obtained, from (6), we have

[0167] , , , ;

[0168] Above, the fitting of one-way displacement signal is completed, and the collected other displacement signal is fitted in the same way.

[0169] Step 8: Transform the two fitted displacement signals in Step 7 to obtain a two-dimensional matrix that can represent the shape of the shaft center locus. The transformation method is the same as that in Step 2.

[0170] Step 9: Input the two-dimensional matrix obtained in Step 8 into the intelligent shaft center locus recognition model trained in Step 5.

[0171] Step 10: Identify the shape of the shaft center locus to obtain the recognition result, and then end.

[0172] There is a rolling mill equipment. The bearing used in its drive motor is a sliding bearing. Two mutually perpendicular eddy current displacement sensors are installed in a cross-section perpendicular to the motor spindle; the two-way displacement signals collected are as Figure 2 .

[0173] According to the collected original signals, the shaft center locus diagram is drawn as Figure 3 , and it can be seen that the shaft center locus has been completely submerged by noise.

[0174] The rotational speed monitored on-site is 336 RPM. Therefore, the rotational frequency f = 5.6. Using the signal fitting method proposed by the present invention, the displacement signals containing only the fundamental frequency and the second harmonic are fitted as shown in Figure 4 .

[0175] The two fitted signals are synthesized into the shaft center locus as shown in Figure 5 , and it can be seen that the shaft center locus is in the shape of an outer "8".

[0176] The two fitted signals are converted into a two-dimensional matrix and input into the trained model, and the result recognized by the model is also in the shape of an outer "8".

[0177] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent recognition system for the center orbit of a sliding bearing, characterized in that: It includes the following steps: Step 1: After starting, batch-simulate the axial center locus displacement signals of two sliding bearings; Step 2: Convert the data produced in batch in Step 1 into a two-dimensional matrix; Step 3: Add labels to the two-dimensional matrix in Step 2; Step 4: Train the intelligent recognition model for axial center locus; Step 5: Obtain the intelligent recognition model after testing; Step 6: Collect the two displacement signals perpendicular to each other of the sliding bearing; Step 7: Respectively fit the signals containing only the 1st and 2nd harmonics from the two signals; Step 8: Transform the fitted signals to obtain a two-dimensional matrix; Step 9: Input the two-dimensional matrix in Step 8 into the intelligent recognition model in Step 5 for recognition; Step 10: Identify the shape of the axial center locus to obtain the recognition result, and then end; In the above Step 1, batch-simulate the displacement signals composed of only the 1st and 2nd frequencies. The formula for simulating the displacement signals is as follows: (1)； where f is a set value, and by varying the value, two displacement signals corresponding to the shape of the shaft center orbit are generated in batches; In step 2, the two generated displacement signals are transformed to obtain a two-dimensional matrix that can represent the shape of the shaft center locus. The two-dimensional matrix consists of 0 and 1; the transformation of the two signals The transformation method of the two-dimensional matrix is as follows: S21. Denote the peak-to-peak value and the minimum value of as , respectively. Denote the peak-to-peak value and the minimum value of as and respectively. Let , , where is a regulation coefficient with a value range of [1.2, 2], to avoid the center orbit image falling on the edge of the two-dimensional matrix and prevent information loss during the convolution operation of the generated matrix; S22. Initialize a two-dimensional matrix representing the shape of the shaft center orbit, denoted as a square matrix of order width , which contains elements all equal to 0; S23. Traverse in pairs and the values x and y in, let: (2)； Among them, The value range of is [0.1, 0.2], aiming to prevent information loss during the convolution operation of the generated matrix; the formula (2) is to make The th row of the matrix, and the value of the element corresponding to the th column is assigned to 1; After the traversal ends, the matrix is the two-dimensional matrix that can represent the shape of the shaft center orbit.

2. The intelligent recognition system for the center orbit of a sliding bearing according to claim 1, wherein: In the above Step 3, add labels to the two-dimensional matrix converted in batch. The labels are encoded in the One-Hot manner.

3. The intelligent recognition system for the center orbit of a sliding bearing according to claim 1, characterized in that: In the above Step 4, use the two-dimensional matrix generated by the simulated signals to train the intelligent recognition model for axial center locus, and test the intelligent recognition model to ensure the accuracy rate.

4. The intelligent recognition system for the center orbit of a sliding bearing according to claim 1, characterized in that: In the above Step 6, use the least squares method to process the monitored two axial center locus displacement signals to fit the displacement signals containing only the 1st and 2nd harmonics; the two displacement signals are respectively fitted. The specific fitting method is as follows: One of the two displacement signals collected is denoted as , the number of sampling points is , the signal to be fitted is denoted as , and The th data points in and are respectively denoted as . The expression of the signal to be fitted is written as: （3）； where is the rotational frequency of the shaft, assuming it has been obtained. When the fitting effect is the best, it is and when the distance between them is the smallest, that is, when the following formula (4) obtains the minimum value: （4）； Expand the above formula (3) to: （5）； Let: （6）； Substitute (6) into (5), and the expression to be fitted is abbreviated as: （7）； Substitute (7) into (4) to get (8), that is, the goal is to minimize formula (8): （8）； To minimize formula (8), according to the minimum value theorem of continuous functions, there is: （9）； Expand formula (9) to: （10）； Arrange formula (10) to obtain a system of equations: （11）； Let: ， ， ； Then formula (11) is written as: (12)； Furthermore, obtain the solutions of each parameter: (13)； Since the value has been obtained, from (6) we have ， ， ， ； Above, the fitting of one displacement signal is completed. Fit the other collected displacement signal in the same way.

5. The intelligent recognition system for the center orbit of a sliding bearing according to claim 3, characterized in that: The transformation method used in the above Step 8 to transform the fitted signals to obtain a two-dimensional matrix is the same as the transformation method for converting the data produced in Step 2 into a two-dimensional matrix.

Images