Intelligent fault diagnosis method and system for pump valve device

Abstract

The application discloses an intelligent fault diagnosis method and system of a pump valve device, and relates to the technical field of fault diagnosis.The method comprises the following steps: collecting operation data; cleaning and verifying the data; extracting key features; diagnosing the running state of the device; receiving the fault diagnosis result, combining the historical data and the running conditions of the device to establish a dynamic early warning model; storing and managing the professional knowledge, historical fault cases, diagnosis rules and expert experience configuration knowledge base related to the device; providing a user interface to perform system configuration, operation monitoring, fault alarm and diagnosis result display; and storing and managing the data generated by the system.By adopting the dynamic threshold early warning model, the application can adaptively adjust the early warning threshold, effectively reduce false positives and false negatives, and significantly improve the accuracy of fault early warning, which enables the device maintenance personnel to timely discover potential problems, avoids the device running with diseases, and improves the accuracy of fault early warning.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to an intelligent fault diagnosis method and system for pump and valve equipment. Background Technology

[0002] Pumps and valves are critical components of large industrial facilities, and their operational status directly impacts the safety and efficiency of the entire system. Fault diagnosis technology for pumps and valves aims to monitor and analyze operating parameters to promptly identify potential faults, prevent equipment failure, and thus ensure stable system operation. An effective fault diagnosis system can significantly improve equipment reliability, reduce unplanned downtime, and lower maintenance costs.

[0003] Currently, fault diagnosis of pump and valve equipment mainly relies on fixed threshold alarms and manual experience judgment. This method has many limitations: First, fixed thresholds are difficult to adapt to the dynamic changes of equipment under different operating conditions, which can easily lead to false alarms or missed alarms; second, traditional diagnostic methods often only focus on a single parameter or a simple combination of parameters, which is difficult to fully reflect the complex state of the equipment; third, most existing diagnostic systems lack adaptive learning capabilities and are difficult to continuously optimize the diagnostic model as the equipment state changes. In addition, existing technologies also lack targeted diagnostic strategies for the different characteristics of different types of equipment such as valves and pumps. Therefore, we provide an intelligent fault diagnosis method and system for pump and valve equipment. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent fault diagnosis method and system for pump and valve equipment to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent fault diagnosis method for pump and valve equipment, comprising the following steps: Step 1: Collect operational data from pump and valve equipment; Step 2: Clean and validate the collected data; Step 3: Extract key features from the processed data; Step 4: Based on the extracted features, perform fault diagnosis on the operating status of the equipment; Step 5: Receive the fault diagnosis results and, in conjunction with historical equipment data and operating conditions, establish a dynamic early warning model; Step 6: Configure a knowledge base with expertise in storage and management devices, historical failure cases, diagnostic rules, and expert experience to provide knowledge support for fault diagnosis and early warning; Step 7: Provide a user interface to display system configuration, operation monitoring, fault alarms, and diagnostic results; Step 8: Store and manage the data generated by the system.

[0006] Preferably, the data collected in step 1 includes, but is not limited to, actuator operating current, motor temperature, thrust, torque, open position signal, closed position signal, input signal, output signal, stroke, operating flow rate, vibration, valve seat internal leakage, packing external leakage, and hysteresis error. The actuator operating current is used to reflect the load during valve opening and closing; the motor temperature is used to monitor the motor's operating status and determine if it is overheating; the thrust is used to reflect the mechanical resistance during valve opening and closing; the torque is used to determine if the valve is stuck; the open position signal is used to indicate the valve's open state; the closed position signal is used to indicate the valve's closed state; the input signal is used to adjust the valve opening degree; the output signal is used to indicate the actual valve opening degree; the stroke is used to reflect the actual valve opening degree; the operating flow rate is used to reflect the valve's actual operating state; the vibration is used to determine if the valve is vibrating abnormally; the valve seat internal leakage is used to detect the valve's sealing performance; the packing external leakage is used to detect the valve's packing sealing performance; the housing external leakage is used to detect the overall sealing performance of the valve; and the hysteresis error is used to reflect the valve's positioning accuracy.

[0007] Preferably, the abnormal data cleaned in step 2 includes values ​​that have not fluctuated for a long time, data with abnormal fluctuation amplitude, data that exceeds the preset range, and data with abnormal rate of change. The data verification includes data integrity verification, data consistency verification, data anomaly detection, statistical feature analysis, data normalization, spectrum analysis, expert rule verification, and data trend verification.

[0008] Preferably, the key features in step 3 include valve features and pump features. The valve features include current change rate, temperature change trend, torque and torque fluctuation range, response time of switch position signal, linearity of input and output signals, and spectral features of vibration signal. The pump features include time domain features, frequency domain features, time-frequency domain features, and current features.

[0009] Preferably, the fault diagnosis in step 4 includes valve fault diagnosis and pump fault diagnosis, and the method steps for valve fault diagnosis include: Establish a logical expression fault library for rapid diagnosis of common faults; Establish normal working patterns using historical data, and diagnose faults by comparing the deviation between real-time data and normal patterns by calculating the Pearson correlation coefficient. For specific types of valves, specialized mathematical models are developed for fault diagnosis. The method for diagnosing pump and motor faults includes the following steps: Construct a diagnostic decision tree that includes multiple fault types; By training a vector machine model using extracted features, multi-class fault diagnosis can be achieved. Using deep learning methods, fault characteristics are automatically learned.

[0010] Preferably, step 5 includes the following specific steps: Step 5.1: Construct a Dirichlet process mixture model: ; ; ; in, For the Dirichlet process, For concentration parameters, Basic distribution; Step 5.2, Maximize the variational lower bound: ; in, It is an approximate posterior distribution; The iterative formula for variational Bayesian inference is: ; in, Indicates to Update; Step 5.3: Calculate the likelihood function value as an anomaly indicator; Step 5.4: Dynamically adjust the warning threshold based on the equipment's operating status and historical data.

[0011] Preferably, the knowledge base in step 6 includes an equipment parameter base, a fault case base, a diagnostic rule base, and a maintenance strategy base.

[0012] Preferably, the interactive interface in step 7 is a visual human-computer interaction interface based on a B / S architecture.

[0013] Preferably, the data in step 8 is stored using distributed storage.

[0014] The present invention also provides an intelligent fault diagnosis system for pump and valve equipment, implementing the intelligent fault diagnosis method for pump and valve equipment described in any one of the above claims, comprising: A data acquisition module, which is used to acquire operating data from pump and valve equipment; A data preprocessing module, which is used to clean and verify the collected data; The feature extraction module is used to extract key features from the processed data; The fault diagnosis module diagnoses faults in the operating status of the equipment based on the extracted features. The early warning model module is used to receive fault diagnosis results and establish a dynamic early warning model by combining historical equipment data and operating conditions. The knowledge base module is used to store and manage equipment-related professional knowledge, historical fault cases, diagnostic rules, and expert experience to configure the knowledge base, providing knowledge support for fault diagnosis and early warning. The human-computer interaction module provides a user interface for displaying system configuration, operation monitoring, fault alarms, and diagnostic results. A data storage module is used to store and manage data generated by the system.

[0015] The technical effects and advantages of this invention are as follows: (1) By adopting a dynamic threshold early warning model, the present invention can adaptively adjust the early warning threshold, effectively reduce false alarms and missed alarms, and significantly improve the accuracy of fault early warning. This enables equipment maintenance personnel to discover potential problems in a timely manner, avoid equipment from operating with defects, and improve the accuracy of fault early warning. (2) The present invention utilizes the application of multi-source data fusion technology to enable the system to fully perceive the equipment status, not only considering a single parameter, but also capturing the interrelationship between parameters, thereby providing a more comprehensive and accurate assessment of equipment health status, realizing comprehensive equipment status perception. Combined with the intelligent diagnostic algorithm of expert rules and data-driven methods, the system has expert-level diagnostic capabilities and continuous learning characteristics, which not only improves the accuracy of diagnosis, but also enables the system to continuously adapt to new fault modes. (3) The present invention utilizes a configurable fault model library, which not only improves the flexibility of the system, but also provides an effective way for knowledge accumulation and experience transfer, enabling the system to be continuously optimized and improved to adapt to the ever-changing equipment conditions and diagnostic needs. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a method for intelligent fault diagnosis of pump and valve equipment according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] This invention provides, for example Figure 1 The intelligent fault diagnosis method for pump and valve equipment shown includes the following steps: Step 1: Collect operational data from pump and valve equipment; Step 2: Clean and validate the collected data; Step 3: Extract key features from the processed data; Step 4: Based on the extracted features, perform fault diagnosis on the operating status of the equipment; Step 5: Receive the fault diagnosis results and, in conjunction with historical equipment data and operating conditions, establish a dynamic early warning model; Step 6: Configure a knowledge base with expertise in storage and management devices, historical failure cases, diagnostic rules, and expert experience to provide knowledge support for fault diagnosis and early warning; Step 7: Provide a user interface to display system configuration, operation monitoring, fault alarms, and diagnostic results; Step 8: Store and manage the data generated by the system.

[0019] Preferably, the data collected in step 1 includes, but is not limited to, actuator operating current, motor temperature, thrust, torque, open position signal, closed position signal, input signal, output signal, stroke, operating flow rate, vibration, valve seat internal leakage, packing external leakage, and hysteresis error. The actuator operating current reflects the load during valve opening and closing; the motor temperature monitors the motor's operating status and determines if it is overheating; the thrust reflects the mechanical resistance during valve opening and closing; the torque determines if the valve is jammed; the open position signal indicates the valve's open state; the closed position signal indicates the valve's closed state; the input signal adjusts the valve opening degree; the output signal indicates the actual valve opening degree; the stroke reflects the actual valve opening degree; and the operating flow rate... The parameters are used to reflect the actual working state of the valve; the vibration is used to determine whether there is abnormal vibration in the valve; the valve seat internal leakage is used to detect the valve sealing performance; the packing external leakage is used to detect the valve packing sealing performance; the shell external leakage is used to detect the overall sealing performance of the valve; and the hysteresis error is used to reflect the valve positioning accuracy. When collecting vibration data, the acquisition frequency is usually set to more than 10 times the maximum speed of the pump to ensure that high-frequency fault characteristics can be captured. For example, for a pump with a speed of 3000 rpm, the acquisition frequency should be at least 500 Hz; for a system with a power supply frequency of 50 Hz, the acquisition frequency of current data should be at least 100 Hz to capture the second harmonic. Process parameters such as temperature, pressure, and flow rate change relatively slowly, so the acquisition frequency can be lower, usually 1 Hz or lower.

[0020] Preferably, the abnormal data cleaned in step 2 includes values ​​that have not fluctuated for a long time, data with abnormal fluctuation amplitude, data that exceeds the preset range, and data with abnormal rate of change. The data verification includes data integrity verification, data consistency verification, data anomaly detection, statistical feature analysis, data normalization, spectrum analysis, expert rule verification, and data trend verification.

[0021] The system ensures data integrity through the following steps: a. Timestamp continuity check: Check whether the time series of data collection is continuous; b. Numerical reasonableness check: Check whether the data is within the expected range; c. Missing value identification: Identify and label missing values ​​in the dataset.

[0022] For multi-source data, the system performs cross-validation to ensure data consistency: a. Comparison of relevant parameters: Compare related parameters, such as current and power; b. Redundant sensor data comparison: For key measurement points, compare the data from redundant sensors.

[0023] Based on the characteristics of pump and valve equipment, the system focuses on the following anomalies: a. Mutation detection: Identifying sudden changes in data; b. Trend Anomalies: Detects abnormal changes in data trends; c. Periodic anomalies: For parameters with periodicity, detect periodic anomalies.

[0024] Calculate and analyze the statistical characteristics of the data, including: a. Mean and standard deviation; b. Skewness and kurtosis; c. Quantiles; These statistical characteristics help identify potential data anomalies.

[0025] To facilitate the comparison and analysis of different types of data, the system performs data normalization processing: ; in, For the normalized data, This is the original data. and These are the minimum and maximum values ​​of the data, respectively.

[0026] For vibration data, the system performs spectral analysis to verify the validity of the data. a. Perform Fast Fourier Transform (FFT); b. Check the presence and amplitude of characteristic frequencies.

[0027] Based on the knowledge of domain experts, the system has established a series of expert rules for data validation: a. Parameter relationship rules: such as the relationship between pressure and flow rate; b. Operating condition restriction rules: such as parameter range restrictions under specific operating conditions.

[0028] The system performs trend analysis on long-term data to verify the data's reasonableness. a. Sliding window analysis: Using the sliding window technique to analyze short-term trends in data; b. Long-term trend comparison: Compare the current data with historical long-term trends; This module can effectively ensure the quality of pump and valve equipment operation data, providing a reliable data foundation for subsequent feature extraction and fault diagnosis.

[0029] Preferably, the key features in step 3 include valve features and pump features. The valve features include current change rate, temperature change trend, torque and torque fluctuation range, response time of switch position signal, linearity of input and output signals, and spectral features of vibration signal. Feature extraction adopts a combination of time domain analysis and frequency domain analysis. Time domain features include mean, standard deviation, peak factor, etc.; frequency domain features are obtained through fast Fourier transform and include amplitude and phase information of characteristic frequency. The pump features include time domain features, frequency domain features, time-frequency domain features, and current features.

[0030] Preferably, the fault diagnosis in step 4 includes valve fault diagnosis and pump fault diagnosis, and the method steps for valve fault diagnosis include: Establish a logical expression fault library for rapid diagnosis of common faults; Establish normal working patterns using historical data, and diagnose faults by comparing the deviation between real-time data and normal patterns by calculating the Pearson correlation coefficient. For specific types of valves, specialized mathematical models can be established for fault diagnosis. For example, for control valves, a valve characteristic curve model can be established, and faults can be diagnosed by comparing the deviation between the actual curve and the theoretical curve. The method for diagnosing pump and motor faults includes the following steps: Construct a diagnostic decision tree that includes multiple fault types; By training a vector machine model using extracted features, multi-class fault diagnosis can be achieved. By using deep learning methods, fault characteristics can be automatically learned. Neural network models can capture nonlinear relationships in data and are suitable for handling complex fault modes.

[0031] Preferably, step 5 includes the following specific steps: Step 5.1: Construct a Dirichlet process mixture model: ; ; ; in, For the Dirichlet process, For concentration parameters, Basic distribution; Step 5.2, Maximize the variational lower bound: ; in, It is an approximate posterior distribution; The iterative formula for variational Bayesian inference is: ; in, Indicates to Update; Step 5.3: Calculate the likelihood function value as an anomaly indicator; Step 5.4: Dynamically adjust the warning threshold based on the equipment's operating status and historical data.

[0032] This method, based on the Dirichlet process mixture model and variational Bayesian inference, can adaptively adjust the warning threshold, effectively reduce false alarms and missed alarms, and improve the accuracy of fault warning. Through this method, the system can better adapt to the dynamic changes of equipment under different operating conditions, providing strong protection for the safe operation of pump and valve equipment.

[0033] Preferably, the knowledge base in step 6 includes an equipment parameter base, a fault case base, a diagnostic rule base, and a maintenance strategy base.

[0034] Preferably, the interactive interface in step 7 is a visual human-computer interaction interface based on a B / S architecture.

[0035] Preferably, the data in step 8 is stored using distributed storage.

[0036] When encountering actual problems, such as a sudden increase in the valve's current measurement, the system's handling process is as follows: (1) Data acquisition: The system collects valve current data in real time through an intelligent positioner. When the detected current value exceeds the normal range, subsequent processing is triggered.

[0037] (2) Data preprocessing: Assuming the normal current average is 10A and the standard deviation is 0.5A, the current measurement is 12.5A, which satisfies the condition that |12.5-10|>3*0.5. Therefore, it is confirmed as an outlier.

[0038] (3) Feature extraction: Calculate the rate of change of current: Rate of change = (Current value - Previous value) / Time interval (4) Early warning: The Dirichlet process mixture model calculates the likelihood function value of the current state. Assuming the obtained anomaly index is 0.85, which exceeds the preset threshold of 0.7, fault diagnosis is triggered.

[0039] (5) Fault diagnosis: The system first queries the logical expression fault database and matches the following rule: IF (Sudden Current Increase) AND (Increased Thrust) THEN Fault Type = "Valve Jamming" Meanwhile, the system calls the SVM model for multi-class diagnosis, and the result also points to "valve jamming" fault, with a probability of 92%.

[0040] (6) Results Display: A warning window pops up in the human-computer interaction interface, displaying: "Warning: Inlet valve No. 1 is suspected of being stuck; probability of failure is 92%; symptoms are a sudden increase in current and increased thrust; recommended actions: 1. Check the valve stem for foreign objects; 2. Check if the packing seal is too tight; 3. Disassemble and inspect the valve if necessary." (7) Knowledge update: The system adds the diagnostic results to the fault case library and updates the characteristic patterns of this type of fault for subsequent model optimization.

[0041] The present invention also provides an intelligent fault diagnosis system for pump and valve equipment, implementing the intelligent fault diagnosis method for pump and valve equipment described in any one of the above claims, comprising: A data acquisition module, which is used to acquire operating data from pump and valve equipment; A data preprocessing module, which is used to clean and verify the collected data; The feature extraction module is used to extract key features from the processed data; The fault diagnosis module diagnoses faults in the operating status of the equipment based on the extracted features. The early warning model module is used to receive fault diagnosis results and establish a dynamic early warning model by combining historical equipment data and operating conditions. The knowledge base module is used to store and manage equipment-related professional knowledge, historical fault cases, diagnostic rules, and expert experience to configure the knowledge base, providing knowledge support for fault diagnosis and early warning. The human-computer interaction module provides a user interface for displaying system configuration, operation monitoring, fault alarms, and diagnostic results. A data storage module is used to store and manage data generated by the system.

[0042] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent fault diagnosis of pump and valve equipment, characterized in that, Includes the following steps: Step 1: Collect operational data from pump and valve equipment; Step 2: Clean and validate the collected data; Step 3: Extract key features from the processed data; Step 4: Based on the extracted features, perform fault diagnosis on the operating status of the equipment; The fault diagnosis in step 4 includes valve fault diagnosis and pump fault diagnosis. The method steps for valve fault diagnosis include: Establish a logical expression fault library for rapid diagnosis of common faults; Establish normal working patterns using historical data, and diagnose faults by comparing the deviation between real-time data and normal patterns by calculating the Pearson correlation coefficient. For specific types of valves, specialized mathematical models are developed for fault diagnosis. The method for diagnosing pump and motor faults includes the following steps: Construct a diagnostic decision tree that includes multiple fault types; By training a vector machine model using extracted features, multi-class fault diagnosis can be achieved. Using deep learning methods, fault characteristics are automatically learned; Step 5: Receive the fault diagnosis results and, in conjunction with historical equipment data and operating conditions, establish a dynamic early warning model; The specific steps of step 5 include: Step 5.1: Construct a Dirichlet process mixture model: ； ； ； in, For the Dirichlet process, For concentration parameters, Basic distribution; Step 5.2, Maximize the variational lower bound: ； in, It is an approximate posterior distribution; The iterative formula for variational Bayesian inference is: ； in, Indicates to Update; Step 5.3: Calculate the likelihood function value as an anomaly indicator; Step 5.4: Dynamically adjust the early warning threshold based on the equipment's operating status and historical data; Step 6: Configure a knowledge base containing expertise in storage and management devices, historical failure cases, diagnostic rules, and expert experience to provide knowledge support for fault diagnosis and early warning. Step 7: Provide a user interface to display system configuration, operation monitoring, fault alarms, and diagnostic results; Step 8: Store and manage the data generated by the system.

2. The intelligent fault diagnosis method for pump and valve equipment according to claim 1, characterized in that, The data collected in step 1 includes, but is not limited to, actuator operating current, motor temperature, thrust, torque, open position signal, closed position signal, input signal, output signal, stroke, operating flow rate, vibration, valve seat internal leakage, packing external leakage, and hysteresis error. The actuator operating current is used to reflect the load during valve opening and closing; the motor temperature is used to monitor the motor's operating status and determine if it is overheating; the thrust is used to reflect the mechanical resistance during valve opening and closing; the torque is used to determine if the valve is stuck; the open position signal is used to indicate the valve's open state; the closed position signal is used to indicate the valve's closed state; the input signal is used to adjust the valve opening degree; the output signal is used to indicate the actual valve opening degree; the stroke is used to reflect the actual valve opening degree; the operating flow rate is used to reflect the actual working state of the valve; the vibration is used to determine if the valve is vibrating abnormally; the valve seat internal leakage is used to detect the valve's sealing performance; the packing external leakage is used to detect the valve's packing sealing performance; the housing external leakage is used to detect the overall sealing performance of the valve; and the hysteresis error is used to reflect the valve's positioning accuracy.

3. The intelligent fault diagnosis method for pump and valve equipment according to claim 1, characterized in that, The abnormal data cleaned in step 2 includes values ​​that have not fluctuated for a long time, data with abnormal fluctuation amplitude, data that exceeds the preset range, and data with abnormal rate of change. The data verification includes data integrity verification, data consistency verification, data anomaly detection, statistical feature analysis, data normalization, spectrum analysis, expert rule verification, and data trend verification.

4. The intelligent fault diagnosis method for pump and valve equipment according to claim 1, characterized in that, The key features in step 3 include valve features and pump features. The valve features include current change rate, temperature change trend, torque and torque fluctuation range, response time of switch position signal, linearity of input and output signals, and spectral features of vibration signal. The pump features include time domain features, frequency domain features, time-frequency domain features, and current features.

5. The intelligent fault diagnosis method for pump and valve equipment according to claim 1, characterized in that, The knowledge base in step 6 includes an equipment parameter library, a fault case library, a diagnostic rule library, and a maintenance strategy library.

6. The intelligent fault diagnosis method for pump and valve equipment according to claim 1, characterized in that, In step 7, the interactive interface is a visual human-computer interaction interface based on a B / S architecture.

7. The intelligent fault diagnosis method for pump and valve equipment according to claim 1, characterized in that, In step 8, the data is stored using distributed storage.

8. An intelligent fault diagnosis system for pump and valve equipment, implementing the intelligent fault diagnosis method for pump and valve equipment according to any one of claims 1-7, characterized in that, include: A data acquisition module, which is used to acquire operating data from pump and valve equipment; A data preprocessing module, which is used to clean and verify the collected data; The feature extraction module is used to extract key features from the processed data; The fault diagnosis module diagnoses faults in the operating status of the equipment based on the extracted features. The early warning model module is used to receive fault diagnosis results and establish a dynamic early warning model by combining historical equipment data and operating conditions. The knowledge base module is used to store and manage equipment-related professional knowledge, historical fault cases, diagnostic rules, and expert experience to configure the knowledge base, providing knowledge support for fault diagnosis and early warning. The human-computer interaction module provides a user interface for displaying system configuration, operation monitoring, fault alarms, and diagnostic results. A data storage module is used to store and manage data generated by the system.

Images