An artificial intelligence-assisted valve fault diagnosis system

Abstract

The present invention relates to the field of artificial intelligence technology, and specifically to an artificial intelligence-assisted valve fault diagnosis system. The system includes a non-linear fault feature analysis module, an operating pressure anomaly identification module, a structural stress impact assessment module, a fault impact range determination module, and a valve fault trend prediction module. In the present invention, by comprehensively collecting and analyzing vibration signals, opening and closing time series, fluid pressure, and drive load responses, accurate capture of fault features and efficient screening of key feature points are achieved. By accurately calculating the fluid pressure offset amplitude and meticulously evaluating the stress state of the sealing structure, the accuracy of diagnosis and the judgment of sealing performance are improved. Through vibration trend analysis and pressure change monitoring, the reliability of fault prediction is enhanced, and a scientific basis is provided for preventive maintenance. By optimizing the data analysis process, the accuracy of fault diagnosis and the adaptability of the system are significantly improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-assisted valve fault diagnosis system. Background Art

[0002] The field of artificial intelligence technology includes multiple directions such as machine learning, deep learning, neural networks, computer vision, and natural language processing. Its core content is to simulate human intelligence in a data-driven manner to achieve automated decision-making, prediction, and analysis. In the artificial intelligence technology system, pattern recognition and data analysis based on machine learning are one of the key technologies. It relies on a large amount of data to train models to identify complex patterns and provide optimization suggestions. Computer vision technology enables artificial intelligence to process image and video information, and through methods such as feature extraction and object detection, it realizes intelligent perception of the real world. Natural language processing is used to analyze and understand text data, enabling the computer to understand human language and interact. The applications of artificial intelligence cover multiple industries, such as medical diagnosis, autonomous driving, intelligent manufacturing, financial risk control, etc., relying on deep learning and neural networks to improve data analysis and decision-making efficiency.

[0003] Among them, an artificial intelligence-assisted valve fault diagnosis system refers to the real-time monitoring and analysis of the valve operating state based on artificial intelligence technology to determine whether there are abnormalities and identify specific fault types. This system mainly uses neural network algorithms to train and classify the sensor data collected to achieve automatic recognition of valve fault characteristics. At the same time, combined with pattern recognition technology to analyze historical data, a fault library is established to improve the accuracy of diagnosis. The system usually adopts multi-sensor fusion technology to obtain the operating parameters of the valve, such as pressure, temperature, vibration signals, etc., and uses signal processing methods to extract feature information, and then classifies and identifies through the trained model. In addition, this system can also integrate a cloud computing architecture, upload data to a remote server for calculation, improve the diagnosis efficiency and support remote monitoring and maintenance, so as to achieve valve fault diagnosis based on artificial intelligence.

[0004] In the prior art, the processing method of sensor data during feature extraction is relatively single, and the dynamic changes of different signals cannot be effectively combined, resulting in some key features not being fully utilized, which affects the accuracy of diagnosis. The identification of abnormal pressure usually relies on fixed thresholds or simple statistical analysis, and the dynamic changes of the operating environment are not fully considered, resulting in errors in the determination of abnormal pressure points. The force evaluation mainly relies on single-point measurement data, lacking an analysis of the overall stress state of the sealing structure, making it difficult to accurately judge the degree of stress deviation and possibly causing local damaged areas to be overlooked. In terms of determining the damaged range, traditional methods lack fine-grained quantification of the damaged degree, affecting the judgment of the stability of the actuator and key components. The prediction of fault trends fails to comprehensively consider vibration trends, pressure rate offsets, and long-term force changes, and only relies on historical data for speculation, making it difficult to effectively identify long-term fault evolution trends and reducing the early warning ability. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art and propose an artificial intelligence-assisted valve fault diagnosis system.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: An artificial intelligence-assisted valve fault diagnosis system includes:

[0007] The non-linear fault feature analysis module collects vibration signals, opening and closing time series, fluid pressure, and driving load, calculates the time derivative of vibration, analyzes the deviation of the opening and closing trend, extracts pressure fluctuations, identifies the amplitude of load changes, constructs the valve dynamic feature path, screens key feature points, analyzes the gradient change, eliminates irrelevant features, and obtains the fault feature distribution value;

[0008] The operating pressure abnormality identification module, based on the fault feature distribution value, extracts pressure parameters, fluid pressure offset, opening and closing response time difference, and load change rate, calculates the pressure offset amplitude, analyzes the pressure time series distribution, identifies abnormal pressure points, calculates the distribution interval, and obtains the operating pressure abnormality interval;

[0009] The structural stress impact assessment module, based on the operating pressure abnormality interval, extracts the sealing force, the offset of the valve body force, and the material strain rate, calculates the force balance, analyzes the force change, judges the material strain amplitude, screens the force offset area, and obtains the structural force offset value;

[0010] The fault impact range determination module, based on the force offset value, extracts the component damage level, the stability of the actuator, and the valve stem force gradient, calculates the damage offset, analyzes the force balance, screens the damaged components, and obtains the fault impact range coefficient.

[0011] As a further solution of the present invention, the fault feature distribution value includes vibration derivative features, opening and closing trend features, pressure fluctuation features, load change features, and gradient change features; the abnormal operation pressure range includes cavity pressure offset, fluid pressure abnormal points, opening and closing response deviation, load fluctuation range, and abnormal pressure range; the structural force offset value includes seal force deviation, valve body force offset amount, material strain deviation, force balance index, and maximum force area; the fault influence range coefficient includes component damage degree, actuator stability level, valve stem force gradient range, damage offset degree, and force distribution balance.

[0012] As a further solution of the present invention, the non-linear fault feature analysis module includes:

[0013] The vibration signal analysis sub-module collects the vibration signals of the actuator, analyzes the time derivative, obtains the change rate at different time points, screens the data points exceeding the vibration change threshold, calculates the abnormal fluctuation range index, and obtains the mean square deviation of abnormal vibration fluctuation.

[0014] The fluid pressure fluctuation extraction sub-module calls the mean square deviation of abnormal vibration fluctuation, divides the fluid pressure data, extracts the fluctuation interval, calculates the pressure extreme value within the interval, and screens the intervals exceeding the fluid pressure change threshold to obtain the fluid pressure fluctuation amplitude.

[0015] The driving load change identification sub-module calculates the driving load response amplitude according to the fluid pressure fluctuation amplitude, screens the characteristic points with load change exceeding the threshold, calculates the change rate of the characteristic points, and obtains the fault feature distribution value.

[0016] As a further solution of the present invention, the abnormal operation pressure identification module includes:

[0017] The pressure offset calculation sub-module extracts the cavity pressure parameter and fluid pressure offset value based on the fault feature distribution value, calculates the offset amplitude index at different time points, screens the time periods with offset amplitude exceeding the pressure offset threshold, and analyzes the offset mean value and range within the screened time periods to obtain the fluid pressure offset amplitude.

[0018] The load fluctuation identification sub-module calls the fluid pressure offset amplitude, calculates the opening and closing response time difference and load dynamic change rate, compares the change trends of the two, screens the time periods with load fluctuation exceeding the load change threshold, and calculates the change rate to obtain the load fluctuation change result.

[0019] The abnormal pressure screening sub-module identifies the operation stages with prominent load fluctuations according to the load fluctuation change result, screens the abnormal pressure points, calculates the distribution index of the abnormal pressure points, and obtains the abnormal operation pressure range.

[0020] As a further solution of the present invention, the calculation formula of the pressure offset amplitude index is specifically:

[0021] ;

[0022] wherein, represents the pressure offset amplitude index at time point , represents the measured pressure value at time point at position , represents the reference pressure value at time point at position , represents the total number of measurement points, represents the mean value of the measured pressure values at time point , represents the total pressure offset of all measurement points at time point , represents the total pressure variance of all measurement points at time point , represents the standard deviation of the measured pressure values at time point .

[0023] As a further solution of the present invention, the structural stress influence assessment module includes:

[0024] The force balance calculation sub-module extracts the stress state of the sealing structure and the strain rate of the sealing material based on the abnormal operation pressure range, calculates the stress distribution index at the differential stress points of the sealing structure, screens the areas with uneven force, calculates the stress difference within the area and normalizes it to obtain the force balance index of the sealing structure;

[0025] The valve body force analysis sub-module calls the force balance index of the sealing structure, analyzes the force change of the valve body within the abnormal operation pressure range, calculates the force change index, compares the change amplitude of the differential stress points, screens the areas where the force offset exceeds the force offset threshold, and obtains the force offset value of the valve body;

[0026] The abnormal force screening sub-module judges the strain amplitude of the sealing material according to the force offset value of the valve body, compares the component force data, screens the area with the largest force offset, calculates the offset amplitude index, and obtains the structural force offset value.

[0027] As a further solution of the present invention, the calculation formula of the normalized regional stress difference is specifically:

[0028] ;

[0029] wherein, represents the normalized regional stress difference, represents the stress value of the th stress point within the region, represents the average stress of all stress points within the region, represents the number of stress points within this region, represents the total absolute deviation of all stress points from the average regional stress, represents the sum of the squares of the stress deviations of all stress points, represents the standard deviation of the stress in this region.

[0030] As a further aspect of the present invention, the fault influence range determination module includes:

[0031] The damaged offset calculation sub-module extracts the component damage level and the valve stem stress gradient based on the structural stress offset value, analyzes the damaged offset degree of different components, screens the components with an offset degree exceeding the damaged offset threshold, calculates the average damage value of the screened components, and extracts the fluctuation range to obtain the component damaged offset degree index;

[0032] The execution stability evaluation sub-module calls the component damaged offset degree index, calculates the force balance index of the actuator in different damaged states, compares the stress gradients of different parts of the actuator, screens the regions with a force balance offset exceeding the valve stem stress gradient threshold to obtain the actuator force balance threshold;

[0033] The influence area screening sub-module screens the components with significant damage effects according to the actuator force balance threshold, calculates the damaged offset amplitude of the components, screens the region with the widest influence range, calculates the damage coefficient of the influence area, and obtains the fault influence range coefficient.

[0034] As a further aspect of the present invention, the system includes a valve fault trend prediction module;

[0035] The valve fault trend prediction module extracts the vibration trend, pressure offset, and force stability based on the influence range coefficient, calculates the vibration attenuation rate, analyzes the pressure change amplitude, screens the trend patterns, and generates a fault diagnosis plan;

[0036] The fault diagnosis plan includes the vibration trend pattern, pressure offset rate, structural force change trend, vibration attenuation rate, and fault mode category.

[0037] As a further aspect of the present invention, the valve fault trend prediction module includes:

[0038] The vibration trend calculation sub-module extracts the vibration change trend based on the fault influence range coefficient, calculates the change rate of the vibration signal at different time points, screens the time interval where the rate change exceeds the vibration change threshold, and obtains the vibration trend time decay rate.

[0039] The pressure rate analysis sub-module calls the vibration trend time decay rate, calculates the change amplitude of the pressure rate offset in different time periods, screens the interval where the amplitude exceeds the pressure rate offset threshold, and obtains the pressure rate change amplitude.

[0040] The fault mode screening sub-module compares the long-term change of the structural stress stability according to the pressure rate change amplitude, calculates the trend change under different fault modes, screens the mode where the trend change exceeds the stability offset threshold, and obtains the fault diagnosis scheme.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0042] In the present invention, by comprehensively collecting and analyzing vibration signals, opening and closing time series, fluid pressure, and driving load responses, accurate capture of fault characteristics and efficient screening of key feature points are achieved. By accurately calculating the fluid pressure offset amplitude and carefully evaluating the stress state of the sealing structure, the accuracy of diagnosis and the judgment of sealing performance are improved. Through vibration trend analysis and pressure change monitoring, the reliability of fault prediction is enhanced, and a scientific basis for preventive maintenance is provided. By optimizing the data analysis process, the accuracy of fault diagnosis and the adaptability of the system are significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 is the system flow chart of the present invention;

[0045] Figure 2 is the sub-module flow chart of the present invention;

[0046] Figure 3 is the flow chart of the non-linear fault feature analysis module of the present invention;

[0047] Figure 4 is the flow chart of the operating pressure anomaly identification module of the present invention;

[0048] Figure 5 is the flow chart of the structural stress influence assessment module of the present invention;

[0049] Figure 6 It is the flowchart of the fault influence scope determination module of the present invention;

[0050] Figure 7 It is the flowchart of the valve fault trend speculation module of the present invention. Specific embodiments

[0051] The technical solutions in the present invention will be described below with reference to the accompanying drawings.

[0052] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as an "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two can be selected.

[0053] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when their differences are not emphasized, the meanings they express are the same. "(of)", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when their differences are not emphasized, the meanings they express are the same.

[0054] In the embodiments of the present invention, sometimes subscripts such as W1 may be written in a non-subscript form such as W1. When their differences are not emphasized, the meanings they express are the same.

[0055] To make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0056] Please refer to Figure 1 and Figure 2 , an artificial intelligence-assisted valve fault diagnosis system includes:

[0057] The non-linear fault feature analysis module collects the vibration signal of the actuator, the opening and closing time series, the fluid pressure change, and the driving load response, calculates the time derivative of the vibration signal, analyzes the trend deviation of the opening and closing time series, extracts the fluctuation range of the fluid pressure, identifies the change amplitude of the driving load in the differential opening and closing stages, constructs the valve dynamic feature path, screens the key feature points, analyzes the gradient change between the feature points, eliminates the irrelevant features, and obtains the fault feature distribution value;

[0058] Based on the fault feature distribution values, the operating pressure anomaly recognition module extracts the cavity pressure parameters, fluid pressure offset values, opening and closing response time differences, and load dynamic change rates, calculates the offset amplitude index of the fluid pressure, analyzes the time series distribution of the cavity pressure, compares the opening and closing response times with the load change trend, identifies the operating stages with significant load fluctuations, screens out the abnormal pressure points, calculates the distribution interval of the abnormal pressure, and obtains the abnormal operating pressure interval;

[0059] Based on the abnormal operating pressure interval, the structural stress impact assessment module extracts the stress state of the sealing structure, the force offset of the valve body, and the strain rate of the sealing material, calculates the force balance index of the sealing structure, analyzes the force change of the valve body within the abnormal operating pressure interval, determines the strain amplitude of the sealing material, compares the force data of the components, screens out the area with the largest force offset, and obtains the structural force offset value;

[0060] Based on the structural force offset value, the fault impact range determination module extracts the damage level of the components, the stability of the actuator, and the force gradient of the valve stem, calculates the damage offset degree index of the components, analyzes the force balance of the actuator, screens out the components with the most significant damage impact, and obtains the fault impact range coefficient;

[0061] Based on the fault impact range coefficient, the valve fault trend prediction module extracts the vibration change trend, pressure rate offset, and structural force stability, calculates the time decay rate of the vibration trend, analyzes the change amplitude of the pressure rate, compares the long-term change of the structural force stability, screens out the fault modes with trend changes, and generates a fault diagnosis plan.

[0062] The fault feature distribution values include vibration derivative features, opening and closing trend features, pressure fluctuation features, load change features, and gradient change features; the abnormal operating pressure interval includes cavity pressure offset, fluid pressure abnormal points, opening and closing response deviation, load fluctuation interval, and abnormal pressure range; the structural force offset value includes sealing force deviation, valve body force offset amount, material strain deviation, force balance index, and maximum force area; the fault impact range coefficient includes component damage degree, actuator stability level, valve stem force gradient range, damage offset degree, and force distribution balance; the fault diagnosis plan includes vibration trend mode, pressure offset rate, structural force change trend, vibration decay rate, and fault mode category.

[0063] Please refer to Figure 3 and Figure 2 , the non-linear fault feature analysis module includes:

[0064] The vibration signal analysis sub-module collects the vibration signals of the actuator, analyzes the time derivative, obtains the change rate at different time points, screens out the data points exceeding the vibration change threshold, calculates the abnormal fluctuation range index, and obtains the mean square deviation of the abnormal vibration;

[0065] First, collect the vibration signals of the actuator, set the time interval for data sampling, and obtain the vibration amplitudes of the actuator at different time points through a high-precision sensor. Record the vibration signal sequence. These signals may come from mechanical components inside the actuator, such as motors, connecting rods, bearings, etc. After collecting the vibration signals, the system needs to calculate the time derivative of them, that is, calculate the change rate of the vibration amplitude at adjacent time points. This calculation can be obtained through the difference between consecutive data points. At the same time, the influence of signal noise needs to be considered. During the calculation process, by setting an appropriate time step, ensure the smoothness and accuracy of the calculation. Then, perform further difference calculation on the change rate of the vibration signal to obtain the second-order change rate to reflect the acceleration characteristics of the vibration signal. During the analysis process, it is necessary to set the threshold for vibration change. This threshold needs to be set based on the historical operation data of the actuator. The mean and standard deviation can be calculated through the vibration data under long-term operating conditions, and an appropriate coefficient is selected for amplification to ensure that abnormal vibration points can be effectively identified. Subsequently, the system screens the vibration data exceeding this threshold, identifies abnormal vibration points, and further calculates the abnormal fluctuation range index, that is, statistically analyzes the distribution of abnormal points in time, combines their quantity, position, and the degree of change in vibration amplitude, and calculates the mean square deviation of the entire abnormal interval as an index to measure the degree of abnormal vibration fluctuation. For example, during the operation of a certain actuator, if there are multiple consecutive high-amplitude fluctuations in the collected vibration data, the system should mark these data points and calculate their mean square deviation to ensure the accuracy of abnormal vibration identification. Finally, obtain the mean square deviation of abnormal vibration fluctuation.

[0066] The fluid pressure fluctuation extraction sub-module calls the mean square deviation of abnormal vibration fluctuation, divides the fluid pressure data, extracts the fluctuation interval, calculates the pressure extreme difference within the interval, and screens the intervals exceeding the fluid pressure change threshold to obtain the fluid pressure fluctuation amplitude.

[0067] First, set the time interval for fluid pressure sampling and obtain the fluid pressure data sequence. During the operation of the system, the pressure data may be affected by factors such as the external environment, pipeline status, and actuator working status. Therefore, during data processing, it is necessary to use a fixed time window to block the fluid pressure data. Each time window contains a certain number of pressure data points. Subsequently, the system analyzes the pressure fluctuation situation within each time window, calculates the maximum and minimum pressures within this interval, and obtains the pressure extreme difference as a measure of the fluctuation degree. To screen out the pressure intervals with significant fluctuation characteristics, it is necessary to set a threshold for fluid pressure change. This threshold can calculate the normal pressure fluctuation range through historical operation data and set a reasonable multiple range according to its standard deviation. For example, under normal operating conditions, the extreme difference of pressure fluctuation usually remains within a certain range. If the pressure extreme difference of a certain time window exceeds this range, then this interval is determined as an abnormal fluctuation interval. During the specific calculation process, the system needs to traverse all time windows, screen out the intervals where the pressure extreme difference exceeds the threshold, and record its start time, end time, and fluctuation amplitude. For example, if the pressure of a certain device rises from 1.02 MPa to 1.15 MPa and then drops to 1.05 MPa within a certain period of time, then its pressure fluctuation amplitude is 0.13 MPa. If this value exceeds the set threshold range, the system marks this interval as an abnormal pressure fluctuation interval and records the relevant data, finally obtaining the fluid pressure fluctuation amplitude.

[0068] The drive load change identification sub-module calculates the drive load response amplitude according to the fluid pressure fluctuation amplitude, screens out the characteristic points where the load change exceeds the threshold, calculates the change rate of the characteristic points, and obtains the fault characteristic distribution value;

[0069] First, it is necessary to establish a mapping relationship between fluid pressure and driving load. The corresponding relationships between pressure and load for different devices may vary and need to be fitted based on actual operating data. By sampling the operating data of the device, the load change value corresponding to the pressure change is obtained, and the conversion coefficient is calculated. After obtaining the amplitude of the driving load change, the system needs to screen it to identify the characteristic points exceeding the set threshold. The setting method of this threshold is usually based on the normal load change range of the device, calculating the standard deviation of the load fluctuation in combination with historical data, and setting an appropriate multiple as the screening basis. For example, if the normal load change amplitude of a certain device is within 2 N·m, the load change threshold can be set to 6 N·m. During the analysis process, the system traverses the load data at all time points, screens out the characteristic points where the load change exceeds the threshold, and further calculates the change rate of the characteristic points. The change rate of the characteristic points can be obtained by calculating the ratio of the load change value of the characteristic point to the time interval of the adjacent time point. For example, if the load of a certain device changes from 5 N·m to 12 N·m within 0.5 s, its change rate is 14 N·m / s. After the system completes the calculation, it statistically analyzes the distribution of all characteristic points to obtain the distribution value of the fault characteristics, ensuring that the moment and amplitude of abnormal load changes can be accurately identified, and finally obtaining the fault characteristic distribution value.

[0070] Please refer to Figure 4 and Figure 2 , the abnormal operating pressure identification module includes:

[0071] Based on the fault characteristic distribution value, the pressure offset calculation sub-module extracts the cavity pressure parameter and the fluid pressure offset value, calculates the offset amplitude index at different time points, screens the time periods where the offset amplitude exceeds the pressure offset threshold, analyzes the offset mean and range within the screened time periods, and obtains the fluid pressure offset amplitude.

[0072] The specific calculation formula for the pressure offset amplitude index is:

[0073] ;

[0074] where represents the pressure offset amplitude index at time point , represents the measured pressure value at time point at position , represents the reference pressure value at time point at position , represents the total number of measurement points, represents the mean value of the measured pressure values at time point , represents the total pressure offset of all measurement points at time point , Represents the total pressure variance of all measurement points at time point , represents the standard deviation of the measured pressure values at time point :

[0075] The formula includes two main parts: mean deviation and standard deviation. First, define each parameter:

[0076] is the measured pressure value at time point and position .

[0077] is the reference pressure value at time point and position , usually based on pressure measurements under normal operating conditions.

[0078] is the number of positions measured at each time point.

[0079] is the average value of all pressure measurements at time point .

[0080] Apply the above parameters to a specific example for demonstration:

[0081] Suppose the pressure values measured at three positions at time point are 101 kPa, 103 kPa, and 98 kPa respectively, while the corresponding reference pressure values are 100 kPa, 100 kPa, and 100 kPa respectively. Then the calculation process is as follows:

[0082] Calculate the mean deviation:

[0083] ;

[0084] Calculate the average value of the pressure values :

[0085] ;

[0086] Calculate the standard deviation:

[0087] ;

[0088] ;

[0089] ;

[0090] Calculation of the total pressure deviation index :

[0091] ;

[0092] This result indicates that at the measurement time point , considering the average offset and variability comprehensively, the pressure offset index is 4.05 kPa. This value represents the degree of deviation of the pressure state at this time point from the normal pressure state. The larger the value, the greater the deviation, which may indicate potential system problems or the need for further adjustment and analysis.

[0093] The load fluctuation identification sub-module calls the fluid pressure offset amplitude, calculates the opening and closing response time difference and the load dynamic change rate, compares the change trends of the two, screens the time periods when the load fluctuation exceeds the load change threshold, calculates the change rate, and obtains the load fluctuation change result;

[0094] First, calculate the opening and closing response time difference, that is, the time interval for the actuator to respond after the control signal is issued. This time interval can be calculated by collecting the time difference between the control signal timestamp and the start time of the load change. On this basis, the system obtains the load dynamic change rate, which is calculated by dividing the change amplitude of the load at adjacent time points by the time interval. To judge the load fluctuation situation, the system needs to compare the trends of the opening and closing response time difference and the load dynamic change rate. First, the data of the two need to be normalized to ensure reasonable comparison in the case of inconsistent units. Then, by calculating their change rates, analyze whether the fluctuation trends of the two are synchronous or lagging. On this basis, the system screens out the time periods when the load fluctuation exceeds the load change threshold. This threshold can be set by the standard deviation of the historical load change of the device. If the standard deviation of the load fluctuation of a device during normal operation is 3 N·m, the threshold can be set to 9 N·m. After screening out the time periods that exceed this threshold, further calculate the change rate. The change rate can be calculated by the change amplitude of the load at adjacent time points and the time interval. For example, within a certain time period, the load changes from 8 N·m to 20 N·m, and the time interval is 2 s, then the change rate is 6 N·m / s. Finally, the load fluctuation change result is obtained.

[0095] The abnormal pressure screening sub-module identifies the operation stages with prominent load fluctuations according to the load fluctuation change result, screens out the abnormal pressure points, calculates the distribution index of the abnormal pressure points, and obtains the abnormal operating pressure range;

[0096] First, identify the operating stage with prominent load fluctuations. The prominent load fluctuation stage can be determined by calculating the mean and standard deviation of the load change rate, setting a threshold, and screening the time periods that exceed this threshold. For example, if the mean load change rate of a certain device is 4 N·m / s and the standard deviation is 2 N·m / s, the threshold can be set to 10 N·m / s. After screening out the time periods with prominent load fluctuations, the system further screens for abnormal pressure points within this time period. The screening criteria for abnormal pressure points are set based on the normal range of historical pressure data. For example, if the standard deviation of the operating pressure of a certain device is 0.08 MPa, the threshold for abnormal pressure points can be set to 0.24 MPa. After screening out the pressure points that exceed this threshold, the system calculates the distribution index of abnormal pressure points. This index includes the density of abnormal pressure points on the time axis, the maximum deviation amplitude of abnormal points, and their correlation with load fluctuations. For example, if multiple pressure abnormal points are detected within a certain abnormal time period and the interval time is less than 1 s, it can be judged that there is abnormal pressure fluctuation within this time period, and finally the abnormal interval of operating pressure is obtained.

[0097] Please refer to Figure 5 and Figure 2 , the structural stress impact assessment module includes:

[0098] Based on the abnormal interval of operating pressure, the force balance calculation sub-module extracts the stress state of the sealing structure and the strain rate of the sealing material, calculates the stress distribution index of the sealing structure at different stress points, screens out the areas with uneven stress, calculates the stress difference within the area and normalizes it to obtain the force balance index of the sealing structure;

[0099] The specific formula for the normalized regional stress difference is:

[0100] ;

[0101] Where, represents the normalized regional stress difference, represents the stress value of the th stress point within the area, represents the average stress value of all stress points within the area, represents the number of stress points within this area, represents the total absolute deviation of all stress points from the regional stress mean, represents the total sum of the squares of the stress deviations of all stress points, represents the stress standard deviation of this area:

[0102] This formula is used to evaluate the stress distribution balance of the stress points of the sealing structure. It combines two statistical indicators, the mean deviation and the standard deviation, to measure the balance degree of the stress distribution. The following is the detailed explanation of the parameters, the specific calculation steps, and an example with specific values substituted:

[0103] Define each parameter:

[0104] It refers to the stress value of the th stress point within a given area.

[0105] It refers to the average stress value of all stress points within the same area.

[0106] It refers to the total number of stress points within this area.

[0107] Assume data acquisition:

[0108] Suppose in a test, the stress values (unit: MPa) of three stress points within an area are respectively: 10, 12, 8.

[0109] The total number of measured stress points .

[0110] Calculate the average stress :

[0111] ;

[0112] Calculate the average deviation and standard deviation:

[0113] Average deviation:

[0114] Sum of squared deviations:

[0115] Standard deviation:

[0116] Calculate the normalized stress difference ( ):

[0117] ;

[0118] This result indicates that the stress distribution of the sealing structure within the test area has a certain degree of non-uniformity. The normalized stress difference is 2.96 MPa, indicating that the fluctuation of the stress distribution within the area has a relatively large difference from the average value. This may mean that the sealing material or structure within this area requires further inspection or optimization to improve its stress balance. Through this method, areas with uneven stress can be identified and corresponding measures can be taken to prevent seal failure or other structural problems.

[0119] The valve body force analysis sub-module calls the force balance index of the sealing structure, analyzes the force change of the valve body in the abnormal operating pressure range, calculates the force change index, compares the change amplitudes of different force points, screens out the areas where the force offset exceeds the force offset threshold, and obtains the valve body force offset value;

[0120] First, analyze the force change of the valve body in the abnormal operating pressure range. This force change is calculated by comparing the force values of the valve body at different time points. The force of the valve body is mainly affected by fluid pressure, structural support force, and external environmental factors. During the data acquisition process, the system needs to obtain force data for different measurement points and calculate the force change index. This index measures the stability of the valve body force by comparing the change amplitudes of the force values at adjacent time points. Then, the system compares the force change amplitudes of different force points, obtains the positions with large force fluctuations, and further screens out the areas where the force offset exceeds the force offset threshold. The setting of this threshold can be based on the force change range under normal operating conditions. For example, if the normal operating force change range of a certain valve body is within 2 MPa, the force offset threshold can be set to 6 MPa. After screening out the areas that exceed this threshold, the system calculates the force offset value in this area. For example, if the force changes from 4 MPa to 12 MPa in a certain area, the force offset value is 8 MPa. Finally, the valve body force offset value is obtained.

[0121] The abnormal force screening sub-module judges the strain amplitude of the sealing material according to the valve body force offset value, compares the force data of components, screens out the area with the largest force offset, calculates the offset amplitude index, and obtains the structural force offset value;

[0122] First, judge the strain amplitude of the sealing material. This strain amplitude is obtained by calculating the deformation amount of the sealing material and its original size. When calculating, the system needs to read the thickness, force area, and elastic coefficient of the sealing material, and calculate the strain amplitude of the sealing material in combination with the force offset value. On this basis, compare the force data of different components and select the area with the largest force offset. The selection of this area is based on the force change rate and the degree of force value offset. For example, if the force of a certain component changes from 6 MPa to 14 MPa within 1 s, its force offset rate is 8 MPa / s. If this value exceeds the set threshold, it is determined that the force offset in this area is large. Subsequently, the system calculates the offset amplitude index of this area. This index is used to measure the relative degree of force offset. When calculating, it needs to combine the overall force change range and the local force change degree. For example, if the maximum force value of a certain structure is 20 MPa and the minimum force value is 5 MPa, the offset amplitude index can be calculated as 0.6. Finally, the structural force offset value is obtained.

[0123] Please refer to Figure 6 and Figure 2 , the fault influence range determination module includes:

[0124] The damaged offset calculation sub-module extracts the component damage level and the stem force gradient based on the structural stress offset value, analyzes the damaged offset degree of different components, filters out the components whose offset degree exceeds the damaged offset threshold, calculates the average damage value of the filtered components, and extracts the fluctuation range to obtain the component damaged offset degree index;

[0125] First, extract the component damage level, which is determined by the force condition, material fatigue limit, and deformation degree of the component. During data acquisition, the system needs to obtain the force data of each component at different time points and combine historical usage data to calculate the damage degree of the component after long-term operation. Subsequently, extract the stem force gradient, which represents the force change rate of the stem along the length direction. During the measurement process, the system needs to analyze the force value distribution at different positions of the stem, calculate the force value difference between adjacent measurement points, and use this to judge whether the stem force is balanced. After obtaining the force data, the system compares the damaged offset degrees of different components, which is obtained by comparing the current damaged state of the component with its historical damaged trend. If the current damage value of a certain component deviates from the historical average by more than the set range, its damaged offset degree is relatively large. The system further filters out the components whose offset degree exceeds the damaged offset threshold, which can be set according to the damage change range under the normal operation state of the equipment. For example, if the normal damage offset range of a certain component is within 0.02 mm, the damaged offset threshold can be set to 0.06 mm. After filtering out the components that exceed this threshold, the system calculates the average damage value of the filtered components, which is obtained by statistically calculating the damage data of multiple measurement points of the filtered components. For example, if the damage data of a certain component is {0.08 mm, 0.12 mm, 0.09 mm, 0.11 mm}, the average damage value is 0.10 mm. Then, extract the fluctuation range, which represents the maximum change amplitude of the damage value at different time points. Finally, obtain the component damaged offset degree index.

[0126] The execution stability evaluation sub-module calls the component damaged offset degree index, calculates the force balance index of the actuator under different damaged states, compares the stress gradients of different parts of the actuator, filters out the areas where the force balance offset exceeds the stem force gradient threshold, and obtains the actuator force balance threshold;

[0127] First, calculate the force balance index of the actuator in the differential damage state. This index is used to evaluate whether the forces on different parts of the actuator are balanced. The system needs to collect force data at different measurement points and calculate the stress mean and standard deviation of each measurement point to measure the uniformity of the force distribution. Subsequently, the system compares the stress gradients of the differential parts of the actuator. This gradient is obtained by calculating the stress change rate between adjacent measurement points. For example, if the stress values of an actuator at adjacent measurement points are 6 MPa and 10 MPa respectively, and the distance between them is 2 cm, then the stress gradient is 2 MPa / cm. The system needs to calculate the stress gradient for all measurement points and screen out the areas where the force balance deviation exceeds the threshold of the valve stem force gradient. The setting of this threshold is based on the stress gradient change range during normal operation of the equipment. For example, if the standard deviation of the stress gradient during normal operation is 0.5 MPa / cm, then the force balance deviation threshold can be set to 1.5 MPa / cm. After screening out the areas that exceed this threshold, the system calculates the force balance difference in this area and statistics its change range, and finally obtains the force balance threshold of the actuator.

[0128] The influence area screening sub-module screens out the components with significant damage effects according to the force balance threshold of the actuator, calculates the damage offset amplitude of the components, screens out the area with the widest influence range, calculates the damage coefficient of the influence area, and obtains the fault influence range coefficient;

[0129] First, screen out the components with significant damage effects. This screening process is judged based on the force balance deviation value of each component. The system calculates the force change rate of each component and screens out the components whose force change exceeds the set range. For example, if the force change rate of a component is 5 MPa / s, exceeding the set threshold of 3 MPa / s, then this component is marked as a component with significant damage effects. Subsequently, the system calculates the damage offset amplitude of the component. This amplitude represents the change of the damage value of the component at different time points. For example, if the damage value of a component changes from 0.05 mm to 0.15 mm within 1 hour, then the damage offset amplitude is 0.10 mm. The system screens out the area with the widest influence range among all damaged components. This area is determined by statistically analyzing the spatial distribution of the damaged components. For example, if the number of damaged components in a certain area accounts for 70% of the whole, then this area is marked as the area with the widest influence range. Finally, the system calculates the damage coefficient of the influence area. This coefficient is obtained by calculating the ratio of the damage offset amplitude to the force balance deviation value. For example, if the damage offset amplitude of a certain area is 0.12 mm and the force balance deviation value is 2 MPa / cm, then the damage coefficient is calculated as 0.06. Finally, obtain the fault influence range coefficient.

[0130] Please refer to Figure 7 and Figure 2 , the valve fault trend prediction module includes:

[0131] The vibration trend calculation sub-module extracts the vibration change trend based on the fault influence range coefficient, calculates the change rate of the vibration signal at different time points, screens the time intervals where the rate change exceeds the vibration change threshold, and obtains the vibration trend time decay rate;

[0132] First, extract the vibration change trend, which is obtained by measuring the vibration signals of the actuator or equipment at different time points. The system needs to collect parameters such as vibration amplitude, frequency, and phase, and perform trend analysis on the vibration data at different time points. Subsequently, calculate the change rate of the vibration signal at different time points. This rate represents the increase or decrease of the vibration signal over time. When calculating, take the difference in vibration amplitude between adjacent time points and divide it by the time interval. For example, within a certain time period, the vibration amplitude increases from 0.02 mm to 0.08 mm, and the time interval is 0.5 s, then the change rate is calculated as 0.12 mm / s. Then, screen the time intervals where the rate change exceeds the vibration change threshold. The setting of this threshold is based on the normal vibration change range of the equipment. If the normal vibration rate change range of a certain equipment is within 0.05 mm / s, then the vibration change threshold can be set to 0.15 mm / s. The system traverses all time periods and screens out the time intervals that exceed this threshold. Subsequently, obtain the vibration trend time decay rate. This decay rate represents the degree of decrease in the change rate of the vibration signal in different time periods. When calculating, compare the change rates at different time points and calculate the rate decrease amplitude. For example, if the initial change rate in a certain time period is 0.20 mm / s and the change rate drops to 0.05 mm / s at the end, then the decay rate is calculated as 75%. Finally, obtain the vibration trend time decay rate.

[0133] The pressure rate analysis sub-module calls the vibration trend time decay rate, calculates the change amplitude of the pressure rate offset in different time periods, screens the intervals where the amplitude exceeds the pressure rate offset threshold, and obtains the pressure rate change amplitude;

[0134] First, calculate the change amplitude of the pressure rate offset in different time periods. This change amplitude represents the change trend of the pressure rate in different time periods. When calculating, collect the pressure rate data for each time period and compare the rate changes between adjacent time periods. For example, in a certain time period, the pressure rate increases from 3 MPa / s to 6 MPa / s, then the change amplitude is calculated as 3 MPa / s. Then, screen the intervals where the amplitude exceeds the pressure rate offset threshold. The setting of this threshold is based on the normal pressure change rate range of the system. For example, if the normal pressure rate change amplitude does not exceed 2 MPa / s, then the pressure rate offset threshold can be set to 5 MPa / s. The system traverses all time intervals, screens out the intervals that exceed this threshold, and records the pressure rate change situation in this time period. Finally, obtain the pressure rate change amplitude.

[0135] The fault mode screening sub-module compares the long-term changes in the structural force stability based on the magnitude of the pressure rate change, calculates the trend changes under different differential fault modes, screens the modes with trend changes exceeding the stability offset threshold, and obtains the fault diagnosis scheme;

[0136] First, compare the long-term changes in the structural force stability. This comparison process is completed by calculating the force change data of the structure at different time periods. The system needs to collect long-term operation data and calculate the standard deviation of the force at each time point to measure the force stability. Subsequently, calculate the trend changes under different differential fault modes. This calculation involves the force fluctuation situation of the fault modes. The system needs to analyze the increase or decrease of the force change rate under different fault modes, and screen the modes with trend changes exceeding the stability offset threshold. The setting of this threshold is based on the long-term operation force change range. For example, if the standard deviation of the normal structural force fluctuation rate is 1 MPa / s, the stability offset threshold can be set to 3 MPa / s. The system traverses all fault modes and screens out the modes exceeding this threshold, and finally obtains the fault diagnosis scheme.

[0137] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An artificial intelligence-assisted valve fault diagnosis system, characterized in that: The system includes: The non - linear fault feature analysis module collects vibration signals, opening and closing time series, fluid pressure, driving load, calculates the time derivative of vibration, analyzes the deviation of the opening and closing trend, extracts pressure fluctuations, identifies the amplitude of load changes, constructs the dynamic feature path of the valve, screens key feature points, analyzes gradient changes, eliminates irrelevant features, and obtains the fault feature distribution value; The operating pressure anomaly identification module, based on the fault feature distribution value, extracts pressure parameters, fluid pressure offset, opening and closing response time difference, load change rate, calculates the pressure offset amplitude, analyzes the pressure time series distribution, identifies abnormal pressure points, calculates the distribution interval, and obtains the operating pressure anomaly interval; The structural stress impact assessment module, based on the operating pressure anomaly interval, extracts the sealing force, the offset of the valve body force, the material strain rate, calculates the force balance, analyzes the force change, judges the material strain amplitude, screens the force offset area, and obtains the structural force offset value; The fault impact range determination module, based on the force offset value, extracts the component damage level, the stability of the actuator, the force gradient of the valve stem, calculates the damage offset, analyzes the force balance, screens the damaged components, and obtains the fault impact range coefficient.

2. The artificial intelligence-assisted valve fault diagnosis system according to claim 1, wherein: The fault feature distribution value includes vibration derivative features, opening and closing trend features, pressure fluctuation features, load change features, gradient change features; the operating pressure anomaly interval includes cavity pressure offset, fluid pressure anomaly points, opening and closing response deviation, load fluctuation interval, abnormal pressure range; the structural force offset value includes sealing force deviation, valve body force offset amount, material strain deviation, force balance index, maximum force area; the fault impact range coefficient includes component damage degree, actuator stability level, valve stem force gradient range, damage offset degree, force distribution balance.

3. The artificial intelligence-assisted valve fault diagnosis system according to claim 1, wherein: The non - linear fault feature analysis module includes: The vibration signal analysis sub - module collects the vibration signals of the actuator, analyzes the time derivative, obtains the change rate of different time points, screens the data points exceeding the vibration change threshold, calculates the abnormal fluctuation range index, and obtains the mean square deviation of vibration abnormal fluctuation; The fluid pressure fluctuation extraction sub - module calls the mean square deviation of vibration abnormal fluctuation, divides the fluid pressure data, extracts the fluctuation interval, calculates the pressure extreme value within the interval, and screens the intervals exceeding the fluid pressure change threshold, and obtains the fluid pressure fluctuation amplitude; The driving load change identification sub - module calculates the driving load response amplitude according to the fluid pressure fluctuation amplitude, screens the feature points with load changes exceeding the threshold, calculates the change rate of the feature points, and obtains the fault feature distribution value.

4. The artificial intelligence-assisted valve fault diagnosis system according to claim 1, wherein: The operating pressure anomaly identification module includes: The pressure offset calculation sub - module, based on the fault feature distribution value, extracts the cavity pressure parameter and the fluid pressure offset value, calculates the offset amplitude index of different time points, screens the time periods with the offset amplitude exceeding the pressure offset threshold, analyzes the offset mean value and range within the screened time periods, and obtains the fluid pressure offset amplitude; The load fluctuation identification sub-module calls the fluid pressure offset amplitude, calculates the opening and closing response time difference and the load dynamic change rate, compares the change trends of the two, screens the time periods with load fluctuations exceeding the overload change threshold, calculates the change rate, and obtains the load fluctuation change result; The abnormal pressure screening sub-module identifies the operation stages with prominent load fluctuations according to the load fluctuation change result, screens the abnormal pressure points, calculates the distribution index of the abnormal pressure points, and obtains the operation pressure abnormal range.

5. The artificial intelligence-assisted valve fault diagnosis system according to claim 4, wherein: The specific calculation formula of the pressure offset amplitude index is: ； wherein, represents the pressure offset amplitude index at the time point ; represents the measured pressure value at the time point at the position ; represents the reference pressure value at the time point at the position ; represents the total number of measurement points ; represents the mean value of the measured pressure values at the time point ; represents the total pressure offset of all measurement points at the time point ; represents the total pressure variance of all measurement points at the time point ; represents the standard deviation of the measured pressure values at the time point 6. The artificial intelligence-assisted valve fault diagnosis system according to claim 1, wherein: The structure stress influence assessment module includes: The force balance calculation sub-module extracts the stress state of the sealing structure and the strain rate of the sealing material based on the operation pressure abnormal range, calculates the stress distribution index of the sealing structure at different stress points, screens the areas with uneven force, calculates the stress difference within the area and normalizes it to obtain the force balance index of the sealing structure; The valve body force analysis sub-module calls the force balance index of the sealing structure, analyzes the force change of the valve body within the operation pressure abnormal range, calculates the force change index, compares the change amplitudes of different stress points, screens the areas where the force offset exceeds the force offset threshold, and obtains the force offset value of the valve body; The abnormal force screening sub-module judges the strain amplitude of the sealing material according to the force offset value of the valve body, compares the force data of the components, screens the areas with the largest force offset, calculates the offset amplitude index, and obtains the structure force offset value.

7. The artificial intelligence-assisted valve fault diagnosis system according to claim 6, wherein: The specific calculation formula of the normalized regional stress difference is: ； Among them, represents the normalized regional stress difference value, represents the stress value of the th stress point within the region, represents the average stress value of all stress points within the region, represents the number of stress points within this region, represents the total absolute deviation of all stress points from the average regional stress, represents the sum of the squares of the stress deviations of all stress points, represents the standard deviation of the stress in this region.

8. The artificial intelligence-assisted valve fault diagnosis system according to claim 1, wherein: The fault influence range determination module includes: The damaged offset calculation sub-module extracts the damaged level of the components and the force gradient of the valve stem based on the structure force offset value, analyzes the damaged offset degree of different components, screens the components with the offset degree exceeding the damaged offset threshold, calculates the average damaged value of the screened components, and extracts the fluctuation range to obtain the component damaged offset degree index; The execution stability evaluation sub-module calls the component damaged offset degree index, calculates the force balance index of the actuator under different damaged states, compares the stress gradients of different parts of the actuator, screens the areas where the force balance offset exceeds the valve stem force gradient threshold, and obtains the force balance threshold of the actuator; The influence area screening sub-module screens the components with significant damage influence according to the force balance threshold of the actuator, calculates the damaged offset amplitude of the components, screens the area with the widest influence range, calculates the damaged coefficient of the influence area, and obtains the fault influence range coefficient.

9. The artificial intelligence-assisted valve fault diagnosis system according to claim 1, wherein: The system includes a valve fault trend prediction module; The valve fault trend prediction module extracts the vibration trend, pressure offset, and force stability based on the influence range coefficient, calculates the vibration attenuation rate, analyzes the pressure change amplitude, screens the trend pattern, and generates a fault diagnosis plan; The fault diagnosis plan includes the vibration trend pattern, pressure offset rate, structure force change trend, vibration attenuation rate, and fault mode category.

10. The artificial intelligence-assisted valve fault diagnosis system according to claim 9, wherein: The valve fault trend prediction module includes: The vibration trend calculation sub-module extracts the vibration change trend based on the fault influence range coefficient, calculates the change rate of the vibration signal at different time points, filters the time intervals where the rate change exceeds the vibration change threshold, and obtains the vibration trend time decay rate; The pressure rate analysis sub-module calls the vibration trend time decay rate, calculates the change amplitude of the pressure rate offset in different time periods, filters the intervals where the amplitude exceeds the pressure rate offset threshold, and obtains the pressure rate change amplitude; The fault mode screening sub-module compares the long-term change of the structural force stability according to the pressure rate change amplitude, calculates the trend change under different fault modes, filters the modes where the trend change exceeds the stability offset threshold, and obtains the fault diagnosis solution.

Images