A mechanism and data-based dynamic equipment numerical model dual-drive fault diagnosis method

Abstract

The application provides a mechanism and data-based dynamic equipment numerical and modal double-driven fault diagnosis method, comprising the following steps: S1: obtaining dynamic equipment structure characteristics, referring to typical fault types, and respectively performing fault mechanism analysis and high-dimension feature extraction; S2: performing mechanism-based diagnosis according to the fault mechanism analysis; S3: performing data-based diagnosis according to the high-dimension feature extraction; S4: establishing a mechanism and data-based dynamic equipment numerical and modal double-driven fault diagnosis model; and S5: using the mechanism and data-based dynamic equipment numerical and modal double-driven fault diagnosis model to perform fault diagnosis on the dynamic equipment. The model is used to diagnose the fault of the dynamic equipment, helps to combine the facts and rules stored in the working memory, establishes a reasoning model, reasons out new information, and enables the data-driven equipment automatic diagnosis system to be continuously updated.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to a fault diagnosis method for dynamic equipment based on mechanism and data-driven digital-analog dual-drive. Background Technology

[0002] Rotating equipment refers to equipment that is driven by a motor (i.e., equipment that consumes energy), such as pumps and fans. The energy source can be electric power, pneumatic power, steam power, etc. Failures of rotating equipment are usually caused by fatigue, wear, etc., and these relatively clear failure mechanisms are closely related to the physical parameters of the equipment, especially vibration data.

[0003] Currently, diagnostic methods for equipment faults mainly include machine learning-based methods and fault mechanism-based methods. However, these methods have the following problems:

[0004] 1) Machine learning-based diagnostic methods face the problem of extremely imbalanced data samples when applied to real-world engineering projects. Equipment failure is a low-probability event, and some failures may never occur during the entire lifecycle of a piece of equipment, making it impossible to guarantee the completeness of the sample space. The number of healthy data samples far exceeds the number of failed data samples, and there are even no failed data samples.

[0005] 2) Fault mechanism-based diagnostic methods suffer from insufficient generalization ability when applied to practical engineering. Individual differences such as equipment manufacturing and installation errors, usage environment, and changes in operating conditions make the application environment complex and variable. Even slight changes in the equipment's state can lead to incorrect diagnostic conclusions, resulting in poor robustness of the diagnostic model. Summary of the Invention

[0006] The purpose of this invention is to provide a mechanism- and data-based method for fault diagnosis of moving equipment using both digital and analog drives, which solves the problems of unbalanced sample data and weak generalization ability in the practical application of current intelligent fault diagnosis methods for equipment.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A mechanism- and data-based method for fault diagnosis of dynamic equipment using both digital and analog drives includes the following steps:

[0009] S1: Obtain the structural characteristics of the moving equipment, and perform fault mechanism analysis and high-dimensional feature extraction respectively, referring to typical fault types;

[0010] S2: Based on the fault mechanism analysis, perform mechanism-based diagnosis;

[0011] S3: Perform data-driven diagnosis based on high-dimensional feature extraction;

[0012] S4: Establish a fault diagnosis model for dynamic equipment based on both mechanism and data;

[0013] S5: Use the mechanism and data-based dual-drive fault diagnosis model for moving equipment to diagnose faults in the moving equipment.

[0014] Typical failure types include shaft misalignment, impeller failure, cavitation, bushing failure, rolling bearing failure, and impeller eccentricity.

[0015] The mechanism-based diagnosis specifically includes:

[0016] S21: Study the failure mechanisms of various equipment failure types;

[0017] S22: Analyze the vibration signal based on the mechanism to achieve fault diagnosis.

[0018] S22 specifically includes: selecting typical fault diagnosis tasks for key components of equipment from the task library, activating the association rules of the task, calculating the changes in feature values ​​based on feature extraction, determining the diagnostic rule model that needs to be matched, and ending the reasoning process when no matching rule model is found, thus drawing a fault conclusion.

[0019] The mechanism- and data-based dual-drive fault diagnosis model for dynamic equipment includes the imbalance model of large generator sets, the misalignment model of large generator sets, the surge model of large generator sets, and the oil film vortex model of large generator sets.

[0020] The mechanism- and data-based dual-drive fault diagnosis model for moving equipment includes a cylinder collision model for reciprocating compressors, a piston rod fracture model for reciprocating compressors, a cylinder scoring model for reciprocating compressors, a crankcase fault model for reciprocating compressors, and a valve fault model for reciprocating compressors.

[0021] The mechanism- and data-based dual-drive fault diagnosis model for dynamic equipment includes early / mid-term / late-term bearing fault models for key pumps and machinery, bearing imbalance models for key pumps and machinery, bearing misalignment models for key pumps and machinery, and bearing cavitation / vaporization models for key pumps and machinery.

[0022] The data-based diagnosis specifically includes:

[0023] S31: After collecting various fault case data for the same type of equipment, train a neural network, and input the new data monitored in real time into the trained neural network for multi-class discrimination;

[0024] S32: After fault identification and on-site inspection and confirmation, once the fault conclusion is confirmed, the fault data can be added to the training data;

[0025] S33: Repeated iterations enable the data-driven automatic diagnostic system for equipment to be continuously updated.

[0026] Compared with existing technologies, the mechanism- and data-based digital-analog dual-drive fault diagnosis method for moving equipment provided by this invention has the following advantages:

[0027] The mechanism- and data-based dual-drive fault diagnosis method for moving equipment provided by this invention establishes a mechanism- and data-based dual-drive fault diagnosis model for moving equipment. Using this model to diagnose faults in moving equipment helps to combine facts and rules stored in the working memory to establish a reasoning model, deduce new information, and enable the data-driven automatic equipment diagnosis system to be continuously updated, effectively solving the problem of poor robustness of the diagnosis model. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0029] Figure 1 This is a flowchart of the mechanism and data-based digital-analog dual-drive fault diagnosis method for moving equipment provided in an embodiment of the present invention;

[0030] Figure 2 This is a flowchart of a mechanism-based diagnostic process provided in an embodiment of the present invention;

[0031] Figure 3 This is a technical roadmap for diagnosis and confidence assessment provided in the embodiments of the present invention. Detailed Implementation

[0032] The following detailed description provides further details on specific implementation methods.

[0033] like Figures 1 to 3 As shown, this invention provides a method for fault diagnosis of dynamic equipment based on both mechanism and data-driven digital-analog dual-drive systems, comprising the following steps:

[0034] S1: Obtain the structural characteristics of the moving equipment, and refer to typical failure types (shaft misalignment, impeller failure, cavitation, bushing failure, rolling bearing failure, and impeller eccentricity) to conduct failure mechanism research and high-dimensional feature extraction research. The moving equipment refers to rotating equipment driven by a motor, such as a pump. The main components of a pump include: casing, pump cover, bottom cover, pump core (which includes a rotor), drive-end bearing, non-drive-end bearing, mechanical seal assembly, discharge pipe assembly, drain pipe assembly, and balance water pipe assembly.

[0035] Vibration characteristic values ​​include high-frequency peak value, high-frequency peak factor, peak value, peak factor, skewness, deviation, kurtosis, kurtosis index, high-frequency RMS value, acceleration RMS value, velocity RMS value, intensity value, acceleration envelope RMS value, velocity envelope RMS value, velocity envelope noise value, acceleration envelope noise value, acceleration noise value, velocity noise value, root square amplitude, waveform index, impulse index, and margin index. Based on vibration standards and relevant industry standards, the calculation methods for 12 vibration characteristic values ​​are mainly selected and introduced, as shown in Table 1.

[0036] Table 1 Formulas for Calculating Vibration Characteristic Values

[0037]

[0038]

[0039] S2: Based on the fault mechanism analysis, conduct mechanism-based diagnosis; the fault mechanism of a pump refers to the physical and chemical, electrical and mechanical processes that induce faults in components and equipment systems, that is, the internal evolution process and causal principles of a certain fault in the equipment before it reaches the surface. Studying the fault mechanism is of great significance for fault diagnosis.

[0040] S3: Data-driven diagnostics are performed based on high-dimensional feature extraction analysis. The high-dimensional feature extraction method is based on kernel independent component analysis (KEMI). KEMI is a nonlinear mapping method that uses kernel functions to map low-dimensional data to a high-dimensional space. The significance is that data that is difficult to decompose or classify in low dimensions becomes easier to decompose or classify after being mapped to a higher dimension.

[0041] S4: Establish a fault diagnosis model for dynamic equipment based on both mechanism and data;

[0042] S5: Use the mechanism and data-based dual-drive fault diagnosis model for moving equipment to perform fault diagnosis on the moving equipment.

[0043] In this embodiment, intelligent diagnosis of moving equipment is achieved through a dual-drive approach of data and mechanism models. The intelligent diagnosis models include models of imbalance, misalignment, surge, and oil film eddy in large units; models of cylinder collision, piston rod breakage, cylinder scoring, crankcase failure, and valve failure in reciprocating compressors; and models of early / mid / late-stage bearing failures, imbalance, misalignment, cavitation, and other issues in key pumps and motors.

[0044] The mechanism-based diagnosis in S2 specifically includes:

[0045] S21: Analyze the failure mechanisms of various equipment failure types;

[0046] S22: Analyze the vibration signal based on the mechanism to achieve fault diagnosis.

[0047] like Figure 2 As shown, the specific steps for analyzing vibration signals based on the mechanism are as follows: select typical fault diagnosis tasks of key components of equipment from the task library, activate the association rules of the task, calculate the changes in feature values ​​based on feature extraction, determine the diagnostic rule model that needs to be matched, and continue the reasoning until no rule model is matched, and then draw a fault conclusion.

[0048] Specifically, the reasoning module is the reasoning engine of the expert system, employing forward reasoning techniques. Rule-based reasoning combines facts and rules stored in the working memory to build a reasoning model and deduce new information.

[0049] like Figure 2 As shown, equipment types include pumps and similar devices. Different types of moving equipment are prone to different types of failures. Selecting the appropriate moving equipment type facilitates subsequent fault diagnosis and judgment. The process includes: Executing consequence set operations: After task initialization, consequence set operations are executed for the selected moving equipment type. Traversing rule models: The rule models are traversed to confirm their suitability for this moving equipment type. Activating rule models: Typical fault diagnosis tasks for key equipment components are selected from the task library, and the associated rules for that task are activated. Pattern matching: Based on feature extraction, changes in feature values ​​are calculated to determine the diagnostic rule model to be matched. Fault diagnosis: The collected data is analyzed and processed, and combined with model matching and intelligent analysis and diagnosis, to arrive at a conclusion on the equipment's health status. If a fault is identified, the cause of the fault is pointed out, and maintenance suggestions are provided.

[0050] The data-based diagnostics in S3 are as follows:

[0051] S31: After collecting various fault case data for the same type of equipment, a neural network is trained. New data monitored in real time is input into the trained neural network for multi-class classification. After classifying and classifying the data of various faults, the data is then further divided according to the severity of the faults to achieve accurate fault identification.

[0052] S32: After fault identification and on-site inspection and confirmation, once the fault conclusion is confirmed, the fault data can be added to the training data; if the fault conclusion is inconsistent with the on-site inspection opinion, the case database data will be corrected or improved.

[0053] S33: This iterative process allows the data-driven automatic diagnostic system for equipment to be continuously updated, further improving the accuracy of fault diagnosis.

[0054] like Figure 3As shown, the system's pump fault diagnosis model analyzes input vibration and non-vibration data to output the cause of the fault and the confidence level of the diagnosis result. Users interact with the system based on the actual fault conditions of the equipment, providing feedback on the accuracy of the diagnosis result. If the diagnosis is correct, the pump fault diagnosis model will be optimized based on the correct data, increasing the confidence level of the corresponding fault diagnosis result for subsequent similar data inputs. If the diagnosis is incorrect, the diagnostic model is retrained based on user input to output the diagnosis result. This iterative process continuously updates the data-driven automatic equipment diagnosis system, further improving the accuracy of fault diagnosis.

[0055] This invention acquires the structural characteristics of moving equipment, refers to typical fault types, and conducts fault mechanism research and high-dimensional feature extraction respectively; based on the fault mechanism, it performs mechanism-based diagnosis; based on the high-dimensional feature extraction, it performs data-based diagnosis; it establishes a mechanism- and data-driven digital-analog dual-drive fault diagnosis model for moving equipment; and it uses this mechanism- and data-driven digital-analog dual-drive fault diagnosis model to diagnose faults in moving equipment. By establishing a mechanism- and data-driven digital-analog dual-drive fault diagnosis model for moving equipment, this invention helps to combine facts and rules stored in the working memory to establish a reasoning model and deduce new information; thus enabling the data-driven automatic equipment diagnosis system to be continuously updated.

[0056] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A fault diagnosis method for dynamic equipment based on both mechanism and data-driven digital-analog dual-drive systems, characterized in that, Includes the following steps: S1: Obtain the structural characteristics of the moving equipment, and perform fault mechanism analysis and high-dimensional feature extraction respectively, referring to typical fault types; S2: Based on the fault mechanism analysis, perform mechanism-based diagnosis; including: S21: Analyze the failure mechanisms of various equipment failure types; S22: Analyze the vibration signal based on the mechanism to achieve fault diagnosis; select typical fault diagnosis tasks of key components of equipment from the task library, activate the association rules of the task, calculate the changes of vibration feature values ​​by extracting features from the vibration signal, determine the diagnostic rule model that needs to be matched, and the reasoning ends when no rule model is matched, and draw a fault conclusion. S3: Data-driven diagnostics based on high-dimensional feature extraction; including: S31: After collecting various fault case data for the same type of equipment, train a neural network, and input the new data monitored in real time into the trained neural network for multi-class discrimination; S32: After fault identification and on-site inspection and confirmation, once the fault conclusion is confirmed, the fault case data can be added to the training data; S33: Repeated iterations enable the data-driven automatic diagnostic system for equipment to be continuously updated; S4: Establish a fault diagnosis model for dynamic equipment based on both mechanism and data; S5: Use the mechanism and data-based dual-drive fault diagnosis model for moving equipment to diagnose faults in the moving equipment.

2. The method for fault diagnosis of moving equipment based on mechanism and data-driven digital-analog dual-drive according to claim 1, characterized in that, Typical failure types include shaft misalignment, impeller failure, cavitation, bushing failure, rolling bearing failure, and impeller eccentricity.

3. The method for fault diagnosis of moving equipment based on mechanism and data-driven digital-analog dual-drive according to claim 1, characterized in that, The mechanism- and data-based dual-drive fault diagnosis model for dynamic equipment includes the imbalance model of large generator sets, the misalignment model of large generator sets, the surge model of large generator sets, and the oil film vortex model of large generator sets.

4. The method for fault diagnosis of moving equipment based on mechanism and data-driven digital-analog dual-drive according to claim 1, characterized in that, The mechanism- and data-based dual-drive fault diagnosis model for moving equipment includes a cylinder collision model for reciprocating compressors, a piston rod fracture model for reciprocating compressors, a cylinder scoring model for reciprocating compressors, a crankcase fault model for reciprocating compressors, and a valve fault model for reciprocating compressors.

5. The method for fault diagnosis of moving equipment based on mechanism and data-driven digital-analog dual-drive according to claim 1, characterized in that, The mechanism- and data-based dual-drive fault diagnosis model for dynamic equipment includes early / mid-term / late-term bearing fault models for key pumps and machinery, bearing imbalance models for key pumps and machinery, bearing misalignment models for key pumps and machinery, and bearing cavitation / vaporization models for key pumps and machinery.

Images