Rotating speed identification method of rotating mechanical equipment

Abstract

The invention discloses a rotating speed identification method for rotating mechanical equipment, and relates to the technical field of signal identification, and the method comprises the following steps: S1, carrying out the vibration data collection of the rotating mechanical equipment; s2, historical start-stop acceleration vibration signal data of the equipment are collected, and start-stop of the equipment is judged by adopting a start-stop recognition algorithm; s3, screening candidate frequencies and performing initial processing to obtain S candidate frequency lists [f1, f2,... fs]; s4, multiple harmonic amplitudes of the S candidate frequencies are calculated, and S groups of frequencies and corresponding harmonic amplitude lists.According to the rotating speed recognition method, non-contact recognition can be carried out based on the equipment vibration signals to obtain the equipment rotating speed information, dependence on a traditional contact sensor is eliminated, and the reliability of the equipment rotating speed recognition method is improved. The method is suitable for high-temperature, high-voltage, high-speed rotation, strong electromagnetic interference or limited space and other scenes where physical sensors cannot be installed, the method is low in cost and easy to deploy, the equipment monitoring cost is remarkably reduced, and meanwhile the reliability and maintenance efficiency of the system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of signal recognition, and in particular to a method for identifying the rotational speed of rotating mechanical equipment. Background Art

[0002] In the field of industrial production and equipment monitoring, the operating status of rotating equipment (such as motors, turbines, generators, engines, etc.) directly affects the safety, efficiency, and reliability of the system. As one of the core parameters of rotating equipment, speed can reflect important information such as the equipment's operating conditions, load changes, and mechanical failures (such as imbalance, bearing wear, and shaft misalignment). Therefore, high-precision speed identification technology is a key foundation for equipment health management (PHM), fault diagnosis, and predictive maintenance.

[0003] Traditional speed measurement relies on physical sensors (such as photoelectric encoders, Hall sensors, and magnetoelectric tachometers). These methods require direct contact with the device or the installation of specialized hardware, which poses challenges in complex industrial environments, such as high cost, limited installation, and susceptibility to interference. With the development of the Industrial Internet of Things (IIoT) and intelligent sensing technologies, non-contact speed identification algorithms based on vibration, sound, or current signals have become a research hotspot. These methods analyze the physical signal characteristics of the device during operation (such as periodic vibration harmonics and current fluctuations) and combine them with signal processing to achieve low-cost, highly robust speed estimation. They are particularly suitable for scenarios where sensors cannot be installed (such as enclosed equipment and high-speed rotating objects). Summary of the Invention

[0004] The purpose of the present invention is to solve the problems in the prior art and to propose a method for identifying the rotation speed of a rotating mechanical device.

[0005] A method for identifying the rotation speed of a rotating mechanical device comprises the following steps:

[0006] S1. Collect vibration data of rotating machinery and equipment;

[0007] S2. Collect historical start-up and shutdown acceleration and vibration signal data of the equipment, and use a start-up and shutdown recognition algorithm to determine whether the equipment is started or shut down;

[0008] S3, filter the candidate frequencies and preliminarily process them to obtain a list of S candidate frequencies [f 1 ,f 2 ,...f s ];

[0009] S4. Calculate the multiple harmonic amplitudes of the S candidate frequencies to obtain S groups of frequencies and their corresponding harmonic amplitude lists, f s The corresponding harmonic amplitude list is [a s1,....a s p ];

[0010] S5. Weight the S groups of frequency amplitudes according to the weighted effective value calculation algorithm to obtain the weighted effective values ​​of S frequencies, f s The corresponding weighted effective value is f rms s ;

[0011] S6, according to f s The corresponding harmonic amplitude list [a s 1,....a s p ] and the weighted effective value is f rms s , and adopts a comprehensive judgment algorithm to determine the final switching frequency fn;

[0012] S7. Calculate the rotational speed W = 60*fn.

[0013] Preferably, in step S2, using a start-stop identification algorithm to determine whether the equipment is started or stopped includes the following steps:

[0014] (1) Calculation: Collect N groups of shutdown data and running data in a 1:1 ratio, calculate the RMS value of all acceleration vibration signal data, and record them;

[0015] (2) Data sorting: Arrange the RMS data of shutdown and operation in ascending order respectively;

[0016] (3) Merge candidate points: merge two sets of data and sort them to generate candidate thresholds;

[0017] (4) Calculate the number of misclassifications: For each candidate threshold t, count:

[0018] a. The number of ≥t in the downtime data;

[0019] b. The number of < t in the running data;

[0020] c. Total number of misclassifications = the sum of the above two;

[0021] Traverse all possible t and find the threshold t that minimizes the total number of misclassifications. The mathematical expression is:

[0022]

[0023] (5) Select the optimal threshold: The t value when the total number of misclassifications is the smallest is the optimal boundary value. If there are multiple values, take the middle value.

[0024] Preferably, the step S3 includes the following steps:

[0025] (1) Perform Fourier transform on the vibration time series signal to obtain the frequency amplitude spectrum;

[0026] (2) Find the maximum amplitude corresponding to a frequency every k Hz in the frequency range f1 to f2 and record the frequency N of the maximum amplitude. max ;

[0027] (3) Finally, the amplitudes of N segments and their corresponding frequencies f are screened out. max 1 ~f max N , and filter these N amplitudes, discard the frequencies corresponding to amplitudes less than d% of the maximum value, and finally obtain a list of S candidate frequencies [f 1 ,f 2 ,...f s ].

[0028] Preferably, in step S4, the harmonic frequency in the calculation of the multiple harmonic amplitudes of the S candidate frequencies is less than f2, a s p Represents the amplitude of the pth harmonic of the sth candidate frequency.

[0029] Preferably, in step S5, the weighted effective value calculation algorithm includes the following steps:

[0030] (1) Weight list generation: Generate the weights of the 1st to Nth harmonics according to the function y = 1 / x, where x = 1 represents the weight of the fundamental frequency;

[0031] (2) Weight normalization: normalize the weights of all harmonics;

[0032] (3) Calculate the weighted effective value: multiply the normalized weight by the square of the harmonic amplitude element by element, then sum and take the square root.

[0033] Preferably, in step S6, the comprehensive determination algorithm includes the following steps:

[0034] (1) A frequency list with the largest weighted effective value is fuzzy calculated in the S group of frequencies, and there are m frequencies in the frequency list.

[0035] (2) A frequency list with the largest amplitude is fuzzy calculated in the S group of frequencies, and there are n frequencies in the frequency list.

[0036] (3) Merge the final frequencies of (1) and (2): If there are fuzzy identical frequencies, then this frequency is the transfer frequency; if there is a fuzzy frequency doubling relationship between the frequencies, then the smaller frequency is the transfer frequency; if the two frequency lists do not have the above two situations, then the frequency with the smallest frequency in the frequency list with the maximum fuzzy weighted effective value is taken as the transfer frequency fn.

[0037] Compared with the existing technology, the advantages of the present invention are:

[0038] The speed identification method proposed in the present invention can obtain equipment speed information through non-contact identification based on the equipment vibration signal, getting rid of the dependence on traditional contact sensors. It is suitable for scenarios where physical sensors cannot be installed, such as high temperature, high pressure, high-speed rotation, strong electromagnetic interference or limited space. In addition, this method is low-cost and easy to deploy, significantly reducing equipment monitoring costs while improving system reliability and maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the process of the present invention.

[0040] Figure 2 This is a simulation diagram of the experimental platform of the present invention. DETAILED DESCRIPTION

[0041] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0042] Reference Figure 1 As shown, a method for identifying the rotation speed of a rotating mechanical device includes the following steps:

[0043] S1. Collect vibration data of rotating machinery and equipment;

[0044] S2. Collect historical start-up and shutdown acceleration and vibration signal data of the equipment, and use a start-up and shutdown recognition algorithm to determine whether the equipment is started or shut down;

[0045] In step S2, using the start-stop identification algorithm to determine whether the equipment is started or stopped includes the following steps:

[0046] (1) Calculation: Collect N groups of shutdown data and running data in a 1:1 ratio, calculate the RMS value of all acceleration vibration signal data, and record them;

[0047] (2) Data sorting: Arrange the RMS data of shutdown and operation in ascending order respectively;

[0048] (3) Merge candidate points: merge two sets of data and sort them to generate candidate thresholds;

[0049] (4) Calculate the number of misclassifications: For each candidate threshold t, count:

[0050] a. The number of ≥t in the shutdown data (misjudged as running);

[0051] b. The number of times < t in the operating data (misjudged as shutdown);

[0052] c. Total number of misclassifications = the sum of the above two;

[0053] Traverse all possible t and find the threshold t that minimizes the total number of misclassifications. The mathematical expression is:

[0054]

[0055] (5) Select the optimal threshold: The t value when the total number of misclassifications is the smallest is the optimal boundary value. If there are multiple values, take the middle value.

[0056] S3, filter the candidate frequencies and preliminarily process them to obtain a list of S candidate frequencies [f 1 ,f 2 ,...f s ];

[0057] The step S3 includes the following steps:

[0058] (1) Perform Fourier transform on the vibration time series signal to obtain the frequency amplitude spectrum;

[0059] (2) Find the maximum amplitude corresponding to a frequency every k Hz in the frequency range f1 to f2 and record the frequency N of the maximum amplitude. max , where the value of k is 5*sampling frequency ÷ number of sampling points;

[0060] (3) Finally, the amplitudes of N segments and their corresponding frequencies f are screened out. max 1 ~f max N , and filter these N amplitudes, discard the frequencies corresponding to amplitudes less than d% of the maximum value, and finally obtain a list of S candidate frequencies [f 1 ,f 2 ,...f s ].

[0061] S4. Calculate the multiple harmonic amplitudes of the S candidate frequencies to obtain S groups of frequencies and their corresponding harmonic amplitude lists, f s The corresponding harmonic amplitude list is [a s 1,....a s p ];

[0062] In step S4, the harmonic frequency in the multiple harmonic amplitude calculation of the S candidate frequencies is less than f2, a s p Represents the amplitude of the pth harmonic of the sth candidate frequency.

[0063] S5. Weight the S groups of frequency amplitudes according to the weighted effective value calculation algorithm to obtain the weighted effective values ​​of S frequencies, f s The corresponding weighted effective value is f rmss ;

[0064] In step S5, the weighted effective value calculation algorithm includes the following steps:

[0065] (1) Weight list generation: Generate the weights of the 1st to Nth harmonics according to the function y = 1 / x, where x = 1 represents the weight of the fundamental frequency;

[0066] (2) Weight normalization: normalize the weights of all harmonics;

[0067] (3) Calculate the weighted effective value: multiply the normalized weight by the square of the harmonic amplitude element by element, then sum and take the square root.

[0068] S6, according to f s The corresponding harmonic amplitude list [a s 1,....a s p ] and the weighted effective value is f rms s , and adopts a comprehensive judgment algorithm to determine the final switching frequency fn;

[0069] In step S6, the comprehensive determination algorithm includes the following steps:

[0070] (1) A frequency list with the largest weighted effective value is fuzzy calculated in the S group of frequencies, and there are m frequencies in the frequency list.

[0071] (2) A frequency list with the largest amplitude is fuzzy calculated in the S group of frequencies, and there are n frequencies in the frequency list.

[0072] (3) Merge the final frequencies of (1) and (2): If there are fuzzy identical frequencies, then this frequency is the transfer frequency; if there is a fuzzy frequency doubling relationship between the frequencies, then the smaller frequency is the transfer frequency; if the two frequency lists do not have the above two situations, then the frequency with the smallest frequency in the frequency list with the maximum fuzzy weighted effective value is taken as the transfer frequency fn.

[0073] S7. Calculate the rotational speed W = 60*fn.

[0074] Example

[0075] Experimental platform such as Figure 2 As shown in the figure, eight acceleration sensors were deployed on the test bench to collect acceleration vibration data, and one motor speed sensor was used to measure the motor speed. The experimental data was mainly collected from 2024 / 12 / 08 to 2025 / 05 / 20. The speed information and speed vibration signal were collected once every hour every day. The sampling frequency of the speed vibration signal was 2560Hz. A total of 3900 sets of speed vibration signals and speed information were collected. The speed information is shown in the following table:

[0076] Table 1: Speed ​​information statistics

[0077] Speed Number of data items 1500rpm 2016 0rpm 1884

[0078] Due to the frequency resolution of the speed vibration sensor, the speed identification result is considered correct if it is within ±0.12 Hz of the actual speed, and the statistics are shown in the following table:

[0079] Table 2: Statistics of frequency conversion recognition results

[0080]

[0081]

[0082] As can be seen from the table above, when the device is in the shutdown state, there are a total of 1884 data items, 1821 of which are correctly identified, with an accuracy rate of 96.6%; when the device is in the power-on state, the total number of data items is 2016, 1733 of which are correctly identified, with an accuracy rate of 85.9%; after integrating all the data, the total number of correctly identified items is 3557, the total number of data items is 3900, and the accuracy rate is 91.2%.

[0083] In summary, the rotational speed identification method of the present invention has high accuracy and anti-interference performance, and can achieve low-cost and high-robustness rotational speed estimation.

[0084] It is understood from common technical knowledge that the present invention may be implemented by other embodiments that do not depart from its spirit or essential features. Therefore, the embodiments disclosed above are, in all respects, merely illustrative and not exclusive. All modifications within the scope of the present invention or equivalent to the scope of the present invention are intended to be encompassed by the present invention.

Claims

1. A method for identifying the rotational speed of a rotating mechanical device, characterized in that: The following steps are involved: S1. Collect vibration data of rotating machinery and equipment; S2. Collect historical start-up and shutdown acceleration and vibration signal data of the equipment, and use a start-up and shutdown recognition algorithm to determine whether the equipment is started or shut down; S3, filter the candidate frequencies and preliminarily process them to obtain a list of S candidate frequencies [f 1 ,f 2 ,...f s ]; S4. Calculate the multiple harmonic amplitudes of the S candidate frequencies to obtain S groups of frequencies and their corresponding harmonic amplitude lists, f s The corresponding harmonic amplitude list is [a s 1,....a s p ]; S5. Weight the S groups of frequency amplitudes according to the weighted effective value calculation algorithm to obtain the weighted effective values ​​of S frequencies, f s The corresponding weighted effective value is f rms s ; S6, according to f s The corresponding harmonic amplitude list [a s 1,....a s p ] and the weighted effective value is f rms s , and adopts a comprehensive judgment algorithm to determine the final switching frequency fn; S7. Calculate the rotational speed W = 60*fn.

2. The method for identifying the rotation speed of a rotating mechanical device according to claim 1, wherein: In step S2, using the start-stop identification algorithm to determine whether the equipment is started or stopped includes the following steps: (1) Calculation: Collect N groups of shutdown data and running data in a 1:1 ratio, calculate the RMS value of all acceleration vibration signal data, and record them; (2) Data sorting: Arrange the RMS data of shutdown and operation in ascending order respectively; (3) Merge candidate points: merge two sets of data and sort them to generate candidate thresholds; (4) Calculate the number of misclassifications: For each candidate threshold t, count: a. The number of ≥t in the downtime data; b. The number of < t in the running data; c. Total number of misclassifications = the sum of the above two; Traverse all possible t and find the threshold t that minimizes the total number of misclassifications. The mathematical expression is: (5) Select the optimal threshold: The t value when the total number of misclassifications is the smallest is the optimal boundary value. If there are multiple values, take the middle value.

3. The method for identifying the rotation speed of a rotating mechanical device according to claim 1, wherein: The step S3 includes the following steps: (1) Perform Fourier transform on the vibration time series signal to obtain the frequency amplitude spectrum; (2) Find the maximum amplitude corresponding to a frequency every k Hz in the frequency range f1 to f2 and record the frequency N of the maximum amplitude. max ; (3) Finally, the amplitudes of N segments and their corresponding frequencies f are screened out. max 1 ~f max N , and filter these N amplitudes, discard the frequencies corresponding to amplitudes less than d% of the maximum value, and finally obtain a list of S candidate frequencies [f 1 ,f 2 ,...f s ].

4. The method for identifying the rotation speed of a rotating mechanical device according to claim 3, wherein: In step S4, the harmonic frequency in the multiple harmonic amplitude calculation of the S candidate frequencies is less than f2, a s p Represents the amplitude of the pth harmonic of the sth candidate frequency.

5. The method for identifying the rotation speed of a rotating mechanical device according to claim 1, wherein: In step S5, the weighted effective value calculation algorithm includes the following steps: (1) Weight list generation: Generate the weights of the 1st to Nth harmonics according to the function y = 1 / x, where x = 1 represents the weight of the fundamental frequency; (2) Weight normalization: normalize the weights of all harmonics; (3) Calculate the weighted effective value: multiply the normalized weight by the square of the harmonic amplitude element by element, then sum and take the square root.

6. The method for identifying the rotation speed of a rotating mechanical device according to claim 1, wherein: In step S6, the comprehensive determination algorithm includes the following steps: (1) A frequency list with the largest weighted effective value is fuzzy calculated in the S group of frequencies, and there are m frequencies in the frequency list. (2) A frequency list with the largest amplitude is fuzzy calculated in the S group of frequencies, and there are n frequencies in the frequency list. (3) Merge the final frequencies of (1) and (2): If there are fuzzy identical frequencies, then this frequency is the transfer frequency; if there is a fuzzy frequency doubling relationship between the frequencies, then the smaller frequency is the transfer frequency; if the two frequency lists do not have the above two situations, then the frequency with the smallest frequency in the frequency list with the maximum fuzzy weighted effective value is taken as the transfer frequency fn.

Images