Strong non-stationary equipment health monitoring method and system based on high-order time-frequency analysis

Through a method based on higher-order time-frequency analysis, combined with data preprocessing and machine learning algorithms, the characteristic parameters of the device signal are extracted and the health monitoring model is established, which solves the complexity and noise sensitivity problems in the existing technology and improves the fault diagnosis. Accuracy and overall reliability of the equipment.

CN118936858APending Publication Date: 2024-11-12CHONGQING SPECIAL EQUIP INSPECTION & RES INST
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411009597.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing high-order time-frequency analysis methods have complexity in equipment health monitoring, high computing resource requirements, high noise sensitivity and poor real-time performance, resulting in inaccurate fault diagnosis and filtering low-frequency characteristic information, affecting the health monitoring effect.

Method used

Using advanced time-frequency analysis methods, through data preprocessing, time-frequency analysis and machine learning algorithms, the characteristic parameters of the equipment signal are extracted, and the equipment health monitoring model is established to realize real-time monitoring and fault warning. Specific steps include data acquisition, preprocessing, time-frequency analysis, model establishment and equipment monitoring, using short-time Fourier transform and trend filtering algorithms, and combining Apriori algorithms learned by association rules.

Benefits of technology

It improves the sensitivity of fault feature extraction and the accuracy of fault diagnosis, realizes real-time data acquisition and analysis, reduces the probability of equipment failure, reduces operation and maintenance costs, and optimizes production efficiency and management efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118936858A_ABST
    Figure CN118936858A_ABST
Patent Text Reader

Abstract

The invention discloses a method and a system for monitoring the health of strong non-stationary equipment based on high-order time-frequency analysis, which are applied to the technical field of health monitoring of mechanical equipment. The method comprises the following steps: acquiring historical fault data of equipment and historical vibration signal data corresponding to the historical fault data through preset monitoring equipment; preprocessing the collected historical fault data and historical vibration signal data; performing time-frequency analysis on the preprocessed historical fault data and historical vibration signal data, and extracting characteristic parameters according to a time-frequency analysis result; establishing a relation between the extracted feature parameters and the equipment health state by adopting a machine learning algorithm, and establishing an equipment health monitoring model according to the relation; and the equipment health is monitored in real time through the equipment health monitoring model. According to the invention, transient characteristics and frequency domain changes in equipment signals can be captured, and the sensitivity of fault characteristic extraction is improved, so that the accuracy of fault diagnosis is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical equipment health monitoring, and more specifically to a method and system for monitoring the health of strongly non-stationary equipment based on high-order time-frequency analysis. Background Art

[0002] High-order time-frequency analysis methods (such as high-order instantaneous frequency analysis, time-frequency spectrum estimation, short-time Fourier transform, etc.) usually involve complex mathematical models and algorithms, and their implementation and debugging are relatively difficult. For non-professionals, it may be difficult to understand and operate these technologies. Since high-order time-frequency analysis needs to process a large amount of time-frequency data, the amount of calculation is huge. Especially in real-time monitoring systems, the demand for computing resources and storage space is large, which may lead to low system operation efficiency. High-order analysis methods are highly sensitive to noise and interference. Under complex working conditions, the signals of the equipment may be affected by various noises, resulting in unstable extracted features, which in turn affects the accuracy of fault diagnosis. When performing feature extraction and fault identification, such methods often need to set parameters and models based on historical data or expert experience. If there is a lack of sufficient prior knowledge, the results may be unsatisfactory.

[0003] Although time-frequency analysis can provide high-resolution frequency information, in practical applications, especially in rapid monitoring scenarios, the real-time nature and response speed of information are still a bottleneck. In high-order time-frequency analysis, some important low-frequency feature information may be filtered out during feature extraction, which may lead to missed detection or misdiagnosis in health monitoring. Some high-order time-frequency analysis methods may only be effective for specific types of signals or fault modes, but may fail in other types of equipment or faults, and need to be adapted for specific applications.

[0004] Therefore, it is an urgent problem for technical personnel in this field to propose a health monitoring method and system for strongly non-stationary equipment based on high-order time-frequency analysis to solve the difficulties existing in the prior art. Summary of the invention

[0005] In view of this, the present invention provides a method and system for monitoring the health of strongly non-stationary equipment based on high-order time-frequency analysis, which can capture the transient characteristics and frequency domain changes in equipment signals, enhance the sensitivity of fault feature extraction, and thus improve the accuracy of fault diagnosis.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis includes the following steps:

[0008] S1, data acquisition step: collecting historical fault data of the equipment and historical vibration signal data corresponding to the historical fault data through a preset monitoring device;

[0009] S2, data processing step: pre-processing the collected historical fault data and historical vibration signal data, including filtering and denoising, data alignment and calibration;

[0010] S3, time-frequency analysis step: performing time-frequency analysis on the pre-processed historical fault data and historical vibration signal data, and extracting characteristic parameters according to the results of the time-frequency analysis;

[0011] S4, model building step: using machine learning algorithm to establish a connection between the characteristic parameters extracted in S3 and the health status of the equipment, and establishing an equipment health monitoring model based on the connection;

[0012] S5. Equipment monitoring step: Monitor equipment health in real time through the equipment health monitoring model.

[0013] In the above method, optionally, the historical fault data in S1 includes fault type data, fault time data, fault frequency data, fault cause data and maintenance plan data.

[0014] In the above method, optionally, the time-frequency analysis of the historical vibration signal data in S3 specifically includes:

[0015] The signal is divided into multiple short time segments by short-time Fourier transform, and then the signal in each segment is Fourier transformed to obtain fault time-frequency characteristic data, including fault signal frequency, frequency spectrum characteristic data, power spectrum characteristic data and energy spectrum characteristic data.

[0016] In the above method, optionally, S4 further includes:

[0017] The trend filtering algorithm is used to eliminate random interference on the trend when establishing a connection and obtain an accurate trend. The formula is as follows:

[0018]

[0019] Among them, y f is a standard time series, f=1,…,n,s f As a basic trend, s f-1 and f+1 Respectively represent s f The front and back points of ; λ is a non-negative parameter used to control the smoothness of s and balance the size of the residual.

[0020] In the above method, optionally, in S4, the extracted characteristic parameters are linked to the health status of the equipment through the Apriori algorithm in the association rule learning, and an equipment health monitoring model is established based on the link.

[0021] A health monitoring system for strongly non-stationary equipment based on high-order time-frequency analysis, using any of the above-mentioned methods for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis, comprising a data acquisition module, a data processing module, a time-frequency analysis module, a model building module, and an equipment monitoring module connected in sequence; wherein,

[0022] A data acquisition module, used to collect historical fault data of the equipment and historical vibration signal data corresponding to the historical fault data through a preset monitoring device;

[0023] A data processing module is used to pre-process the collected historical fault data and historical vibration signal data, including filtering and denoising, data alignment and calibration;

[0024] The time-frequency analysis module is used to perform time-frequency analysis on the pre-processed historical fault data and historical vibration signal data, and extract characteristic parameters based on the results of the time-frequency analysis;

[0025] A model building module is used to establish a connection between the extracted characteristic parameters and the health status of the equipment using a machine learning algorithm, and to establish an equipment health monitoring model based on the connection;

[0026] The equipment monitoring module is used to monitor the equipment health in real time through the equipment health monitoring model.

[0027] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method and system for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis, which has the following beneficial effects:

[0028] (1) The high-order time-frequency analysis adopted by the present invention can capture the transient characteristics and frequency domain changes in the equipment signal, improve the sensitivity of fault feature extraction, and thus improve the accuracy of fault diagnosis; the system can realize real-time data collection and analysis, timely monitor the equipment operation status, and issue early warnings when abnormal situations occur, helping users to take quick measures; by regularly monitoring and analyzing the health status of the equipment, potential faults can be discovered in time, reducing the probability of equipment failure, thereby improving the overall reliability of the equipment. Accurate health monitoring and fault prediction systems reduce unnecessary maintenance and overhaul, reduce operation and maintenance costs, and optimize the allocation of maintenance resources.

[0029] (2) Provide real-time and accurate data analysis to help management make more scientific decisions on equipment maintenance and upgrades, thereby improving overall management efficiency. High-order time-frequency analysis is suitable for processing strong non-stationary signals, and can effectively cope with complex and dynamic working environments, ensuring the comprehensiveness and accuracy of health monitoring. Through in-depth understanding of equipment status, it can reduce downtime, optimize production arrangements, and improve overall production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0031] Figure 1 A flow chart of a method for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis provided by the present invention;

[0032] Figure 2 A structural diagram of a health monitoring system for strongly non-stationary equipment based on high-order time-frequency analysis provided by the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] Reference Figure 1 As shown, the present invention discloses a method for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis, comprising the following steps:

[0035] S1, data acquisition step: collecting historical fault data of the equipment and historical vibration signal data corresponding to the historical fault data through a preset monitoring device;

[0036] S2, data processing step: pre-processing the collected historical fault data and historical vibration signal data, including filtering and denoising, data alignment and calibration;

[0037] S3, time-frequency analysis step: performing time-frequency analysis on the pre-processed historical fault data and historical vibration signal data, and extracting characteristic parameters according to the results of the time-frequency analysis;

[0038] S4, model building step: using machine learning algorithm to establish a connection between the characteristic parameters extracted in S3 and the health status of the equipment, and establishing an equipment health monitoring model based on the connection;

[0039] S5. Equipment monitoring step: Monitor equipment health in real time through the equipment health monitoring model.

[0040] Furthermore, the historical fault data in S1 includes fault type data, fault time data, fault frequency data, fault cause data and maintenance plan data.

[0041] Furthermore, the time-frequency analysis of historical vibration signal data in S3 specifically includes:

[0042] The signal is divided into multiple short time segments by short-time Fourier transform, and then the signal in each segment is Fourier transformed to obtain fault time-frequency characteristic data, including fault signal frequency, frequency spectrum characteristic data, power spectrum characteristic data and energy spectrum characteristic data.

[0043] Specifically, Short-Time Fourier Transform (STFT) is a signal processing technique used to analyze the spectrum of a time-varying signal. Unlike the standard Fourier transform, STFT captures the time-varying characteristics of the signal by dividing the signal into short time windows and analyzing the frequency components within each window. The specific formula is as follows:

[0044]

[0045] Where X(t,f) is the short-time Fourier transform result at time t and frequency f, x(τ) is the original signal, w(t-τ) is the window function, and e -j2πfτ are the basis functions of the Fourier transform.

[0046] Furthermore, S4 also includes:

[0047] The trend filtering algorithm is used to eliminate random interference on the trend when establishing a connection and obtain an accurate trend. The formula is as follows:

[0048]

[0049] Among them, y f is a standard time series, f=1,…,n,s f As a basic trend, s f-1 and f+1 Respectively represent s f The front and back points of ; λ is a non-negative parameter used to control the smoothness of s and balance the size of the residual.

[0050] Furthermore, in S4, the extracted feature parameters are linked to the equipment health status through the Apriori algorithm in association rule learning, and an equipment health monitoring model is established based on the link.

[0051] Specifically, the Apriori algorithm is a classic association rule learning algorithm used to mine frequent item sets and generate association rules from large databases. The specific steps include:

[0052] 1. Set the support threshold: Determine a minimum support (min_support), which is the key parameter for identifying frequent itemsets.

[0053] 2. Generate initial candidate item set (C1): Extract all single items (1-item set) from the database to form candidate item set C1.

[0054] 3. Calculate support: Scan the database, calculate the support of all candidate item sets in the database, and extract the item sets whose support satisfies min_support to form frequent item sets (L1).

[0055] 4. Iteratively generate candidate sets: Use the frequent itemsets obtained in the previous step to generate new candidate itemsets:

[0056] Join operation: join two frequent k-item sets to generate a candidate (k+1)-item set.

[0057] Pruning operation: For the generated candidate item set, check whether all its subsets are in the frequent item set. If any subset is not in the frequent item set, delete the candidate item set.

[0058] 5. Support calculation: Scan the database again, calculate the support of the new candidate item set, and extract the frequent item set (Lk) that meets the minimum support.

[0059] 6. Repeat steps 4 and 5: Repeat the above steps until no new frequent itemsets are generated.

[0060] 7. Generate association rules: Generate association rules from frequent item sets, calculate the confidence of each rule, and filter out the rules that meet the minimum confidence.

[0061] like Figure 2 As shown, a health monitoring system for strongly non-stationary equipment based on high-order time-frequency analysis, a health monitoring method for strongly non-stationary equipment based on high-order time-frequency analysis using any of the above, including a data acquisition module, a data processing module, a time-frequency analysis module, a model building module, and an equipment monitoring module connected in sequence; wherein,

[0062] A data acquisition module, used to collect historical fault data of the equipment and historical vibration signal data corresponding to the historical fault data through a preset monitoring device;

[0063] A data processing module is used to pre-process the collected historical fault data and historical vibration signal data, including filtering and denoising, data alignment and calibration;

[0064] The time-frequency analysis module is used to perform time-frequency analysis on the pre-processed historical fault data and historical vibration signal data, and extract characteristic parameters based on the results of the time-frequency analysis;

[0065] A model building module is used to establish a connection between the extracted characteristic parameters and the health status of the equipment using a machine learning algorithm, and to establish an equipment health monitoring model based on the connection;

[0066] The equipment monitoring module is used to monitor the equipment health in real time through the equipment health monitoring model.

[0067] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0068] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis, characterized in that: The following steps are involved: S1, data acquisition step: collecting historical fault data of the equipment and historical vibration signal data corresponding to the historical fault data through a preset monitoring device; S2, data processing step: pre-processing the collected historical fault data and historical vibration signal data, including filtering and denoising, data alignment and calibration; S3, time-frequency analysis step: performing time-frequency analysis on the pre-processed historical fault data and historical vibration signal data, and extracting characteristic parameters according to the results of the time-frequency analysis; S4, model building step: using machine learning algorithm to establish a connection between the characteristic parameters extracted in S3 and the health status of the equipment, and establishing an equipment health monitoring model based on the connection; S5. Equipment monitoring step: Monitor equipment health in real time through the equipment health monitoring model.

2. According to claim 1, a method for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis is characterized in that: The historical fault data in S1 includes fault type data, fault time data, fault frequency data, fault cause data and maintenance plan data.

3. According to claim 1, a method for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis is characterized in that: The time-frequency analysis of historical vibration signal data in S3 specifically includes: The signal is divided into multiple short time segments by short-time Fourier transform, and then the signal in each segment is Fourier transformed to obtain fault time-frequency characteristic data, including fault signal frequency, frequency spectrum characteristic data, power spectrum characteristic data and energy spectrum characteristic data.

4. According to claim 1, a method for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis is characterized in that: S4 also includes: The trend filtering algorithm is used to eliminate random interference on the trend when establishing a connection and obtain an accurate trend. The formula is as follows: Among them, y f is a standard time series, f=1,…,n,s f As a basic trend, s f-1 and f+1 Respectively represent s f The front and back points of ; λ is a non-negative parameter used to control the smoothness of s and balance the size of the residual.

5. According to claim 1, a method for health monitoring of strongly non-stationary equipment based on high-order time-frequency analysis is characterized in that: In S4, the extracted feature parameters are linked to the equipment health status through the Apriori algorithm in association rule learning, and an equipment health monitoring model is established based on the link.

6. A health monitoring system for strongly non-stationary equipment based on high-order time-frequency analysis, characterized in that: A method for monitoring the health of strongly non-stationary equipment based on high-order time-frequency analysis as described in any one of claims 1 to 5 is applied, comprising a data acquisition module, a data processing module, a time-frequency analysis module, a model building module, and an equipment monitoring module connected in sequence; wherein, A data acquisition module, used to collect historical fault data of the equipment and historical vibration signal data corresponding to the historical fault data through a preset monitoring device; A data processing module is used to pre-process the collected historical fault data and historical vibration signal data, including filtering and denoising, data alignment and calibration; The time-frequency analysis module is used to perform time-frequency analysis on the pre-processed historical fault data and historical vibration signal data, and extract characteristic parameters based on the results of the time-frequency analysis; A model building module is used to establish a connection between the extracted characteristic parameters and the health status of the equipment using a machine learning algorithm, and to establish an equipment health monitoring model based on the connection; The equipment monitoring module is used to monitor the equipment health in real time through the equipment health monitoring model.

Citation Information

Patent Citations

  • Pumped storage unit vibration trend prediction method

    CN108875841A

  • Method for monitoring and diagnosing large-scale high-speed rotor shaft crack

    CN110441063A

  • Non-stationary working condition fault diagnosis method and system and medium

    CN117633634A

  • Power equipment state monitoring method and system based on multi-source data fusion

    CN117874467A

  • Method for predicting faults in electrical equipment based on multi-dimension time series

    WO2015176565A1