A fluid valve actuator monitoring and diagnostic system based on AI intelligence

Abstract

The present invention relates to the field of fluid valve actuator equipment diagnosis technology, and relates to a fluid valve actuator monitoring and diagnosis system based on AI intelligence. The present invention generates a target opening control instruction by matching the fluid medium characteristic data, thereby realizing precise control of the fluid valve; the deployed sensor group captures the opening change data set and the torque time series data set in real time, performs multi-dimensional dynamic fluctuation analysis, and realizes intelligent identification and screening of fluid valve anomalies; the medium temperature, pipeline pressure, and medium flow rate in the dynamic working condition parameters are substituted into the multivariate regression linear model to generate reference torque and opening, and the dynamic diagnosis of the stuck anomaly is realized by combining the fusion calculation of the torque fluctuation coefficient and the opening deviation coefficient; the accurate positioning of the abnormal source is realized through the correlation between the opening and torque time series data and the traceability analysis of the current dynamic working condition parameters, which provides a systematic solution for the state diagnosis and reliability evaluation of the fluid valve actuator.

Description

Technical Field

[0001] The present invention relates to the technical field of fluid valve actuator equipment diagnosis, and in particular to an AI-based intelligent fluid valve actuator monitoring and diagnosis system. Background Art

[0002] Fluid valves are widely used in numerous fields, including industrial production, energy distribution, and municipal engineering, to control the flow, pressure, and direction of fluids. The stable operation of fluid valve actuators is crucial to the proper functioning of the entire system. Failures in fluid valves, such as sticking, can not only affect the accuracy and efficiency of fluid control but can also lead to production interruptions, equipment damage, and even safety incidents. Therefore, accurate and timely monitoring and diagnosis of fluid valve actuators are of great practical significance.

[0003] Several patents exist for diagnosing intermittent sticking in fluid valves. For example, Chinese Patent Publication No. CN118043584A describes a data processing system that receives input data, including at least time-series data on the valve opening of a main valve and time-series data on the pressure at the output side of the solenoid valve, which supplies and discharges the driving fluid from the solenoid valve to a driver. The system then infers time-series data on the applied torque corresponding to the input data from a learned model, which has learned through machine learning the correlation between the input data and output data, including time-series data on the applied torque acting on the valve stem. The system then performs prescribed processing on the inferred applied torque time-series data and, based on the processing results, determines whether at least one of the main valve and the driver is abnormal.

[0004] However, the existing technology has the following problems: 1. The operating state of the fluid valve is affected by many factors. In the existing technology, it is difficult to fully reflect the operating state of the actuator through a single torque data. At the same time, the dynamic impact of dynamic working conditions on the operation of the fluid valve is not fully considered. These changes will affect the normal operation and diagnostic results of the fluid valve, resulting in the diagnostic results being seriously disturbed by working condition fluctuations, resulting in an increased probability of misjudgment of stuck faults.

[0005] 2. After determining that a fluid valve is abnormal, the existing technology does not conduct in-depth tracing analysis of the source of the abnormality. It is difficult to accurately determine whether the fault is caused by abnormal reasons such as lubricant failure, foreign object obstruction, or mechanical transmission abnormality based solely on torque data. This is not conducive to quickly taking targeted maintenance measures, which prolongs equipment downtime and increases maintenance costs. Summary of the Invention

[0006] The present invention aims to provide an AI-based intelligent fluid valve actuator monitoring and diagnosis system to solve the problems of low diagnostic accuracy, poor adaptability to dynamic working conditions, and difficulty in tracing the source of abnormalities in the existing technology of fluid valve actuator monitoring and diagnosis, so as to achieve more accurate and comprehensive monitoring and diagnosis of fluid valve actuators and improve the reliability and stability of equipment operation.

[0007] The present invention solves its technical problems by adopting a technical solution: an AI-based fluid valve actuator monitoring and diagnosis system, comprising an execution instruction generation module, a multidimensional data monitoring module, an anomaly judgment module, a dynamic fault diagnosis module, and an anomaly source tracing module. The execution instruction generation module is connected to the multidimensional data monitoring module, which is connected to the anomaly judgment module, the dynamic fault diagnosis module is connected to the anomaly judgment module and the anomaly source tracing module, respectively, and the database is connected to the execution instruction generation module and the dynamic fault diagnosis module, respectively.

[0008] The database is used to store the required openings of fluid valves for different types of fluid media at different densities and viscosities, and to store historical execution records of fluid valve actuators.

[0009] The execution instruction generation module is used to generate the fluid valve target opening control instruction according to the fluid medium characteristic data and send the instruction to the actuator.

[0010] The multi-dimensional data monitoring module is used to collect the opening change data set and torque time series data set of different fluid valves in real time through the sensor group deployed in the fluid valve as the time series response data.

[0011] The abnormality judgment module is used to extract the opening start change time and the steady-state time error from the opening change data set, and judge whether the fluid valve has an abnormality based on the opening start change time and the steady-state time error.

[0012] The dynamic fault diagnosis module is used to perform dynamic deviation analysis on the timing response data of abnormal fluid valves and the reference data affected by preset dynamic working conditions, and screen out stuck abnormal fluid valves based on the deviation results.

[0013] The abnormal source tracing module is used to combine the timing response data of the stuck abnormal fluid valve and the current dynamic operating parameters to perform traceability analysis and generate abnormal source diagnosis results.

[0014] Compared with the prior art, the present invention has the following beneficial effects:

[0015] (1) The present invention adopts multi-dimensional data acquisition and analysis methods, and integrates the opening change data set and torque time series data set of different fluid valves to make abnormal judgments, thereby solving the problem of diagnostic accuracy, reflecting the operating status of the fluid valve actuator more comprehensively and accurately, improving the reliability of fault judgment, and ensuring stable operation of the equipment.

[0016] (2) The present invention is based on the reference data of dynamic working condition influence compensation fitted by multivariate regression linear equation. Through normalization processing and associated compensation term generation, the interference of working condition parameters such as temperature, pressure, flow rate on the diagnosis results is effectively eliminated, and the probability of misjudgment of stuck fault is reduced.

[0017] (3) The present invention performs dynamic deviation analysis on the timing response data of abnormal fluid valves and the reference data affected by preset dynamic working conditions, thereby achieving refined diagnosis of fluid valves under different working conditions, accurately screening out stuck abnormal fluid valves, reducing the risk of diagnostic errors caused by changes in working conditions, and significantly improving the accuracy and reliability of diagnosis.

[0018] (4) The present invention combines the timing response data of the stuck abnormal fluid valve with the current dynamic operating parameters to conduct a traceability analysis and generate abnormal source diagnosis results. Through timestamp alignment and analysis of the corresponding relationship between opening and torque, the fault source can be quickly located, providing clear guidance for optimizing the operation and maintenance measures of the fluid valve, thereby reducing maintenance costs and downtime. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 This is a schematic diagram of system module connections of the present invention.

[0021] Figure 2 Schematic diagram of the flow chart for determining whether a fluid valve is abnormal in the present invention.

[0022] Figure 3 Schematic diagram of the abnormal source diagnosis result generation and determination steps in the present invention. DETAILED DESCRIPTION

[0023] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement of components and steps, numerical expressions, and numerical values described in these embodiments do not limit the scope of the present invention. It should also be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn to scale.

[0024] The following description of at least one exemplary embodiment is merely illustrative in nature and is not intended to limit the invention, its application, or uses. Technologies, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such technologies, methods, and apparatus should be considered part of the specification.

[0025] In all examples shown and discussed herein, any specific values should be interpreted as merely exemplary and not limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0026] The present invention proposes an AI-based intelligent fluid valve actuator monitoring and diagnosis system, which generates target opening control instructions by matching fluid medium characteristic data to achieve precise control of the fluid valve; uses a deployed sensor group to capture the opening change data set and torque time series data set in real time, performs multi-dimensional dynamic fluctuation analysis, and achieves intelligent identification and screening of fluid valve anomalies; substitutes the medium temperature, pipeline pressure, and medium flow rate in the dynamic operating parameters into a multivariate regression linear model to generate reference torque and opening, and combines the fusion calculation of the torque fluctuation coefficient and the opening deviation coefficient to achieve dynamic diagnosis of sticking anomalies; and achieves precise positioning of the anomaly source through the correlation between the opening and torque time series data and the traceability analysis of the current dynamic operating parameters. This solution breaks through the bottleneck of abnormal diagnosis of traditional methods under dynamic operating condition interference, and provides a systematic solution for the status diagnosis and reliability assessment of fluid valve actuators.

[0027] See also Figure 1 As shown, the present invention provides an AI-based fluid valve actuator monitoring and diagnostic system, comprising an execution instruction generation module, a multidimensional data monitoring module, an anomaly determination module, a dynamic fault diagnosis module, an anomaly source tracing module, and a database. The modules are connected as follows: the execution instruction generation module is connected to the multidimensional data monitoring module, the multidimensional data monitoring module is connected to the anomaly determination module, the dynamic fault diagnosis module is connected to the anomaly determination module and the anomaly source tracing module, respectively; and the database is connected to the execution instruction generation module and the dynamic fault diagnosis module, respectively.

[0028] The database is used to store the required openings of fluid valves for different types of fluid media at different densities and viscosities, and to store historical execution records of fluid valve actuators.

[0029] The required fluid valve opening for different types of fluid media at different densities and viscosities can be obtained through experimental testing. In a laboratory environment, by changing the density and viscosity of different types of fluid media (water, crude oil, alcohol, molten metal, etc.), the fluid valve actuator is used to control the fluid valve opening at each density-viscosity combination, and the optimal required opening when the target flow or pressure is reached is measured. After repeated experiments, the data repeatability is ensured and outliers are eliminated.

[0030] The execution instruction generation module is used to generate a fluid valve target opening control instruction according to the fluid medium characteristic data, and send the instruction to the actuator.

[0031] The fluid valve target opening control instruction is generated by retrieving the required valve openings of the same type of fluid at different densities and viscosities from a database based on the fluid type in the fluid medium characteristic data. This opening is then matched with the fluid density and viscosity in the fluid medium characteristic data to obtain the matched required openings. Based on the matched required openings, the fluid valve target opening control instruction is generated in a specific instruction format and sent to the actuator via a communication interface, achieving precise control of the fluid valve opening.

[0032] The multi-dimensional data monitoring module is used to collect, in real time, data sets of opening variation and torque time series of different fluid valves as time series response data through a sensor group deployed in the fluid valves. The sensor group includes a magnetic induction sensor and a strain gauge torque sensor.

[0033] The abnormality judgment module is used to extract the opening start change time and the steady-state time error from the opening change data set, and judge whether the fluid valve has an abnormality based on the opening start change time and the steady-state time error.

[0034] See also Figure 2 As shown, it should be noted that the determination of whether there is an abnormality in the fluid valve specifically includes: extracting the start time of the opening change and the stable time of the opening change from the opening change data set, performing data dynamic fluctuation analysis on the start time of the opening change of different fluid valves, and obtaining the time deviation fluctuation. The analysis method of the time deviation fluctuation is: performing mean calculation on the start time of the opening change of different fluid valves to obtain the average start time of the opening change, substituting the start time of the opening change of different fluid valves and the average start time of the opening change into the standard deviation calculation formula to obtain the time deviation fluctuation. By analyzing the fluctuation of the start time of the opening change of different fluid valves, the stability of the fluid valve during operation can be determined. If the time deviation fluctuation is large, it means that the start time of the opening change of the fluid valve is unstable and there may be a delay abnormality.

[0035] Opening change start time: refers to the time point corresponding to the start of the opening change in the opening change data set.

[0036] Opening change stabilization time: refers to the time point in the opening change data set when the opening reaches a stable state (no longer changes).

[0037] The difference between the opening stabilization time and the opening start time of different fluid valves is analyzed with the standard time required for the corresponding fluid valve to reach the target opening to obtain the steady-state time error. The steady-state time error reflects the control accuracy of the fluid valve when reaching the target opening. A smaller steady-state time error indicates that the valve can more accurately reach and maintain the target opening. An increase in the steady-state time error may indicate a potential fault in the fluid valve, such as wear of mechanical components or abnormal control signals.

[0038] When the time deviation fluctuation exceeds a preset time deviation fluctuation threshold or the steady-state time error exceeds a preset time error threshold, it is determined that the fluid valve is abnormal.

[0039] The present invention adopts multi-dimensional data collection and analysis methods, and integrates the opening change data sets and torque timing data sets of different fluid valves to make abnormality judgments, thereby solving the problem of diagnostic accuracy, reflecting the operating status of the fluid valve actuator more comprehensively and accurately, improving the reliability of fault judgment, and ensuring stable operation of the equipment.

[0040] The dynamic fault diagnosis module is used to perform dynamic deviation analysis on the timing response data of abnormal fluid valves and reference data affected by preset dynamic working conditions, and screen out stuck abnormal fluid valves based on the deviation results.

[0041] It should be noted that the reference data for the preset dynamic operating condition influence is set by retrieving the operating condition parameters, torque time series dataset, and aperture change dataset from multiple execution records of the fluid valve actuator stored in the database. Multiple regression fitting is performed on the average torque value in the torque time series dataset and the stable aperture value in the aperture change dataset with the operating condition parameters of medium temperature, pipeline pressure, and medium flow rate, respectively. Multiple regression linear equations for the operating condition parameter-torque and the operating condition parameter-application are obtained. The stable aperture value corresponds to the value in the aperture change dataset at which the aperture reaches a stable state.

[0042] The fitting process of the multivariate regression linear equation is as follows: the torque average value, opening stability value, medium temperature, pipeline pressure and medium flow rate recorded for multiple executions are normalized, and the normalized torque average value and opening stability value are respectively fitted with the medium temperature, pipeline pressure and medium flow rate to form a multivariate linear regression equation; if the multivariate linear regression equation fitting is unsuccessful, the medium temperature, pipeline pressure and medium flow rate are subjected to correlation analysis to generate correlation compensation terms, and the multivariate regression linear equations of operating parameter-torque and operating parameter-opening are obtained after refitting.

[0043] The normalization processing method can be a minimum-maximum normalization formula Or Z-score normalization formula ,in is the normalized data, For data that needs to be normalized, are the minimum and maximum values of the data that need to be normalized, and The mean and standard deviation of the data that need to be normalized. Each parameter is processed separately, for example, the opening stability value is converted to the minimum-maximum normalization. , the average value of torque, medium temperature, pipeline pressure and medium flow rate are normalized using Z-score.

[0044] In a specific embodiment, the torque average value, medium temperature, pipeline pressure and medium flow rate after normalization of multiple execution records are fitted with a multivariate linear regression equation, and the fitting equation is: , are the normalized average torque value, medium temperature, pipeline pressure and medium flow rate, is the intercept, is the regression coefficient, is the regression coefficient corresponding to the medium temperature. When the temperature increases by 1°C, the torque change may be negative due to the decrease in viscosity. is the regression coefficient corresponding to the pipeline pressure. When the pressure increases by 1 MPa, the torque change affected by the pressure difference driving effect is positive. is the regression coefficient corresponding to the medium flow rate. When the flow rate increases by 1 m / s, the torque change may be positive due to the dominance of the inertia force.

[0045] If the solution If all of them are unique solutions, the multivariate linear regression equation of working parameter-torque is fitted successfully.

[0046] If the solution If any one of them is not the only solution, the multivariate linear regression equation fitting of the working parameter-torque is unsuccessful. The Pearson correlation coefficient analysis formula is used to perform a pairwise correlation analysis on the medium temperature, pipeline pressure and medium flow rate, output the correlation coefficient, determine the linear relationship between temperature-pressure, temperature-flow rate and pressure-flow rate, generate the correlation compensation term for the parameter ratio in the combination with the linear relationship, and combine the correlation compensation term to perform the multivariate linear regression equation. Refit, is the regression coefficient corresponding to the associated compensation term, is an associated compensation item. There are one or more associated compensation items. is the error term, which covers all the torque changes that are not explained by temperature, pressure, flow rate and related compensation terms, so that the model can more realistically reflect the complexity of real data and solve , the solution is fitted to generate a multiple linear regression equation of working condition parameter-torque. At the same time, the multiple linear regression equation of working condition parameter-opening is fitted in the same way as above.

[0047] Furthermore, taking the correlation analysis of medium temperature and pipeline pressure as an example, the medium temperature and pipeline pressure of n execution records are obtained, and the temperature-pressure correlation coefficient is analyzed by the Pearson correlation coefficient analysis formula. When the correlation coefficient is , then the linear relationship between temperature and pressure is positive linear correlation; when the correlation coefficient is 0, the linear relationship between temperature and pressure is non-linear; when the correlation coefficient is , then the linear relationship between temperature and pressure is a negative linear correlation.

[0048] Substituting the current dynamic operating parameters of the fluid valve into the multivariate regression linear equation, the reference torque and reference opening of the fluid valve affected by the dynamic operating conditions are output. The current dynamic operating parameters, including medium temperature, pipeline pressure, and medium flow rate, are measured using an infrared thermometer, a piezoelectric pressure sensor, and an electromagnetic flowmeter, respectively.

[0049] The present invention is based on the reference data of dynamic working condition influence compensation fitted by multivariate regression linear equation, and through normalization processing and associated compensation item generation, effectively eliminates the interference of working condition parameters such as temperature, pressure, flow rate on the diagnosis result, and reduces the probability of misjudgment of stuck fault.

[0050] It should be noted that the method for screening stuck abnormal fluid valves is: screening the maximum torque, minimum torque and torque mean from the torque time series data set in the time series response data of the abnormal fluid valve, and calculating the torque fluctuation coefficient by combining the fusion of the reference torque.

[0051] The fusion calculation method of the torque fluctuation coefficient is as follows: the torque time series data set is divided into windows of fixed time length, the torque corresponding to all windows is extracted, the maximum torque, minimum torque and torque mean are screened, the difference between the maximum torque and the minimum torque is compared with the reference torque to calculate the output instantaneous fluctuation intensity, the absolute difference between the torque mean and the reference torque is compared with the reference torque to calculate the output steady-state fluctuation intensity, and adjacent feature matching is performed on the instantaneous fluctuation intensity and the steady-state fluctuation intensity to obtain the torque fluctuation coefficient.

[0052] Among them, weighted average is a mathematical means to achieve adjacent feature matching. The instantaneous fluctuation intensity measures the ratio of the absolute fluctuation range of the torque within the time window to the reference torque. Under normal working conditions, the greater the instantaneous fluctuation intensity, the greater the torque fluctuation, and the presence of severe vibration or seizure. The steady-state fluctuation intensity quantifies the degree to which the average torque deviates from the reference value, reflecting long-term friction and wear or system drift. When the steady-state fluctuation intensity is greater, the torque fluctuation is greater, and there is lubricant failure or mechanical structure deformation. In order to adapt to different working conditions, adaptive weight adjustment can be used. The weight factor of the instantaneous fluctuation intensity reflects the importance of the instantaneous peak-to-peak value. The weight factor of the instantaneous fluctuation intensity is the instantaneous fluctuation intensity divided by the sum of the instantaneous fluctuation intensity and the steady-state fluctuation intensity. The weight factor of the steady-state fluctuation intensity reflects the importance of the mean shift. The weight factor of the steady-state fluctuation intensity is the difference between 1 and the weight factor of the instantaneous fluctuation intensity.

[0053] Based on the deviation between the opening at the end of the opening change data set and the reference opening, an opening deviation coefficient is generated. The opening deviation coefficient is expressed as the ratio of the deviation to the reference opening.

[0054] The torque fluctuation coefficient and the opening deviation coefficient are comprehensively statistically analyzed to output the comprehensive deviation of the abnormal fluid valve. The abnormal fluid valves with comprehensive deviation greater than the set deviation threshold are screened and regarded as stuck abnormal fluid valves. At the same time, the abnormal fluid valves with comprehensive deviation less than or equal to the set deviation threshold are marked as reminders to remind personnel to check the abnormal fluid valve, so as to capture potential hidden dangers and prevent fault escalation.

[0055] The statistical formula for the comprehensive deviation of the abnormal fluid valve is: , where are the torque fluctuation coefficient and the opening deviation coefficient respectively, are weight coefficients of torque fluctuation coefficient and opening deviation coefficient respectively, which are assigned by expert experience. For example, if the sticking is more sensitive to torque fluctuation, , .

[0056] The present invention performs dynamic deviation analysis on the timing response data of abnormal fluid valves and reference data affected by preset dynamic working conditions, thereby achieving refined diagnosis of fluid valves under different working conditions, accurately screening out stuck abnormal fluid valves, reducing the risk of diagnostic errors caused by changes in working conditions, and significantly improving the accuracy and reliability of diagnosis.

[0057] The abnormality source tracing module is used to perform a tracing analysis based on the timing response data of the abnormally stuck fluid valve and the current dynamic operating condition parameters to generate an abnormality source diagnosis result.

[0058] See also Figure 3As shown, it should be noted that the generation of abnormal source diagnosis results includes: aligning the opening change data set and torque time series data set of the stuck abnormal fluid valve through timestamps to obtain the opening change trend and torque change trend under the same time series, and obtaining the correlation between the opening and torque according to the opening change trend and torque change trend of the stuck abnormal fluid valve under all time series, and performing traceability analysis in combination with the current dynamic operating parameters of the stuck abnormal fluid valve to obtain the abnormal source diagnosis results.

[0059] Furthermore, the correlation between the opening and the torque is obtained as follows: if the opening change trend and the torque change trend of the stuck abnormal fluid valve at all time sequences are both upward trends, then the correlation between the opening and the torque is a linear relationship; if the torque change trend of the stuck abnormal fluid valve at any time sequence is an upward trend and the opening change trend is a stable trend, then the correlation between the opening and the torque is a nonlinear relationship.

[0060] The relationship between opening and torque is linear. The judgment principle is: the increase in opening requires overcoming the continuously increasing friction resistance. For example, insufficient or failed lubrication leads to increased uniform friction, and the torque demand increases linearly with the opening.

[0061] The correlation between opening and torque is a nonlinear relationship. The judgment principle is: when the valve movement suddenly encounters intermittent resistance, such as local obstruction by foreign matter or abnormal deformation of mechanical transmission, the torque will suddenly increase and the opening will stagnate, showing a nonlinear mutation.

[0062] The traceability analysis is performed in combination with the current dynamic operating parameters of the stuck abnormal fluid valve. The specific process is as follows: when the correlation between the opening degree and torque of the stuck abnormal fluid valve is a linear relationship, the medium temperature in the current dynamic operating parameters is extracted, and the medium temperature is compared with the adaptive temperature range of the fluid valve lubricant. Based on the comparison results, it is determined whether the abnormal source of the stuck abnormal fluid valve is lubricant failure.

[0063] The basis for judging lubricant failure is that the medium temperature is not within the applicable temperature range of the fluid valve lubricant, because the lubricant carbonizes due to high temperature or solidifies due to low temperature. The applicable temperature range of the fluid valve lubricant is determined by material property tables or laboratory tests.

[0064] When the correlation between the opening and torque of the stuck abnormal fluid valve is nonlinear, the medium flow rate is extracted from the current dynamic operating parameters of the stuck abnormal fluid valve, and the deviation degree of the medium flow rate and the fluid medium reference flow rate corresponding to the fluid valve opening at the current time sequence are analyzed. When the deviation degree is greater than the set deviation degree threshold, the abnormal source of the fluid valve is diagnosed as foreign body obstruction; otherwise, the abnormal source of the fluid valve is diagnosed as mechanical transmission abnormality.

[0065] The deviation degree analysis method is to compare the medium flow rate with a reference flow rate corresponding to the fluid valve opening at the current time sequence, and use the ratio of the resulting deviation to the reference flow rate corresponding to the fluid valve opening at the current time sequence as the deviation degree. The reference flow rate corresponding to the fluid valve opening at the current time sequence is generated using a fluid medium flow-opening characteristic curve or numerical simulation.

[0066] The present invention combines the timing response data of the stuck abnormal fluid valve with the current dynamic operating parameters to perform traceability analysis and generate abnormal source diagnosis results. Through timestamp alignment and analysis of the corresponding relationship between opening and torque, the fault source can be quickly located, providing clear guidance for optimizing the operation and maintenance measures of the fluid valve, thereby reducing maintenance costs and downtime.

[0067] The above formulas are all dimensionless and numerically calculated, and the preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0068] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0069] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0070] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0071] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0072] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An AI-based fluid valve actuator monitoring and diagnostic system, characterized in that: include: The execution instruction generation module generates the fluid valve target opening control instruction according to the fluid medium characteristic data and sends the instruction to the actuator; The multi-dimensional data monitoring module collects the opening change data set and torque time series data set of different fluid valves in real time through the sensor group deployed in the fluid valve as the time series response data; The abnormality judgment module extracts the opening start change time and the steady-state time error from the opening change data set, and judges whether the fluid valve has an abnormality based on the opening start change time and the steady-state time error; The dynamic fault diagnosis module performs dynamic deviation analysis on the timing response data of abnormal fluid valves and the reference data affected by preset dynamic working conditions, and screens out abnormally stuck fluid valves based on the deviation results; The abnormality source tracing module combines the timing response data of the stuck abnormal fluid valve and the current dynamic operating parameters to perform traceability analysis and generate abnormality source diagnosis results; Generating the abnormality source diagnosis result includes: By aligning the opening change dataset and torque time series dataset of the stuck abnormal fluid valve with timestamps, the opening change trend and torque change trend at the same time series are obtained. Based on the opening change trend and torque change trend of the stuck abnormal fluid valve at all time series, the correlation between the opening and torque is obtained. Combined with the current dynamic operating condition parameters of the stuck abnormal fluid valve, the source of the abnormality is traced and analyzed to obtain the abnormality source diagnosis result. The specific process of tracing the source of the abnormal stuck fluid valve by combining the current dynamic working condition parameters is as follows: When the correlation between the opening degree and torque of the abnormally stuck fluid valve is a linear relationship, the medium temperature in the current dynamic working condition parameters is extracted, and the medium temperature is compared with the temperature range suitable for the fluid valve lubricant. Based on the comparison result, it is determined whether the abnormal source of the abnormally stuck fluid valve is lubricant failure; When the correlation between the opening and torque of the stuck abnormal fluid valve is nonlinear, the medium flow rate is extracted from the current dynamic operating parameters of the stuck abnormal fluid valve, and the deviation degree of the medium flow rate and the fluid medium reference flow rate corresponding to the fluid valve opening at the current time sequence are analyzed. When the deviation degree is greater than the set deviation degree threshold, the abnormal source of the fluid valve is diagnosed as foreign body obstruction; otherwise, the abnormal source of the fluid valve is diagnosed as mechanical transmission abnormality.

2. The AI-based fluid valve actuator monitoring and diagnostic system according to claim 1, characterized in that: The fluid valve target opening control instruction is generated in the following manner: According to the fluid type in the fluid medium characteristic data, the required fluid valve opening of the same type of fluid medium at different densities and viscosities is retrieved from the database, matched with the fluid density and fluid viscosity in the fluid medium characteristic data, and the matched required opening is obtained. The fluid valve target opening control instruction is generated based on the required opening.

3. The AI-based fluid valve actuator monitoring and diagnostic system according to claim 1, characterized in that: The determining whether the fluid valve is abnormal specifically includes: Extract the opening change start time and opening change stabilization time from the opening change data set, perform data dynamic fluctuation analysis on the opening change start time of different fluid valves, and obtain the time deviation fluctuation; The difference between the opening change stabilization time and the opening change start time of different fluid valves is analyzed with the standard time for the corresponding fluid valve to reach the target opening to obtain the steady-state time error; When the time deviation fluctuation exceeds a preset time deviation fluctuation threshold or the steady-state time error exceeds a preset time error threshold, it is determined that the fluid valve is abnormal.

4. The AI-based fluid valve actuator monitoring and diagnostic system according to claim 3, characterized in that: The analysis method of the time deviation fluctuation is: The average opening change starting time of different fluid valves is calculated by averaging, and the opening change starting time of different fluid valves and the average opening change starting time are substituted into the standard deviation calculation formula to obtain the time deviation fluctuation.

5. The AI-based fluid valve actuator monitoring and diagnostic system according to claim 1, characterized in that: The reference data affected by the preset dynamic working conditions is set as follows: The operating parameters, torque time series data set, and opening change data set of multiple execution records of the fluid valve actuator are retrieved from the historical execution records of the fluid valve actuator. The torque average value of the torque time series data set and the opening stability value of the opening change data set are respectively subjected to multivariate regression fitting with the medium temperature, pipeline pressure, and medium flow rate in the operating parameters to obtain the multivariate regression linear equations of the operating parameter-torque and the operating parameter-opening; The current dynamic working condition parameters of the fluid valve are substituted into the multivariate regression linear equation, and the reference torque and reference opening of the fluid valve affected by the dynamic working condition are output.

6. The AI-based fluid valve actuator monitoring and diagnostic system according to claim 5, characterized in that: The fitting process of the multiple regression linear equation is: The average torque value, stable opening value, medium temperature, pipeline pressure and medium flow rate recorded in multiple executions are normalized, and the normalized average torque value and stable opening value are fitted with the medium temperature, pipeline pressure and medium flow rate respectively to form a multiple linear regression equation. If the multiple linear regression equation fitting is unsuccessful, the medium temperature, pipeline pressure and medium flow rate are correlated and analyzed to generate correlation compensation terms. After re-fitting, the multiple regression linear equations of operating parameter-torque and operating parameter-opening are obtained.

7. The AI-based fluid valve actuator monitoring and diagnostic system according to claim 5, characterized in that: The screening method for abnormal stuck fluid valves is as follows: The maximum torque, minimum torque and torque mean are selected from the torque time series data set in the time series response data of the abnormal fluid valve, and the torque fluctuation coefficient is calculated by combining with the reference torque; Generate an opening deviation coefficient based on the deviation between the opening at the end of the opening change data set and the reference opening; The torque fluctuation coefficient and the opening deviation coefficient are comprehensively counted to output the comprehensive deviation of the abnormal fluid valve. The abnormal fluid valves with comprehensive deviation greater than the set deviation threshold are screened and regarded as stuck abnormal fluid valves.

8. The AI-based fluid valve actuator monitoring and diagnostic system according to claim 7, characterized in that: The fusion calculation method of the torque fluctuation coefficient is: The torque time series dataset is divided into windows of fixed length, and the torques corresponding to all windows are extracted. The maximum torque, minimum torque, and mean torque are screened out. The difference between the maximum and minimum torques is ratioed with the reference torque to calculate the output instantaneous fluctuation intensity. The absolute difference between the mean torque and the reference torque is ratioed with the reference torque to calculate the output steady-state fluctuation intensity. The torque fluctuation coefficient is obtained by weighted averaging the instantaneous and steady-state fluctuation intensities.

Images