A regulating valve fault state early warning method and device

Abstract

This invention discloses a method and apparatus for early warning of control valve malfunctions. The method includes: collecting historical operating data of the control valve during normal operation, including time series of inlet pressure, inlet flow rate, and valve opening; cleaning the historical operating data to form a benchmark dataset; constructing a standardized regression model of pressure-opening-flow rate based on the benchmark dataset to eliminate the influence of inlet pressure changes and data fluctuations on the flow characteristic curve; collecting real-time operating data of the control valve, calculating the theoretical inlet flow rate based on the standardized regression model and the current inlet pressure; comparing the deviation between the theoretical inlet flow rate and the measured inlet flow rate with a dynamic threshold, and triggering an early warning when the deviation exceeds the dynamic threshold. The control valve malfunction early warning apparatus disclosed in this invention includes modules for data acquisition, data cleaning, model construction, and online diagnosis, and can execute the aforementioned malfunction early warning method to provide an early warning.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, specifically to a method and device for early warning of control valve malfunction status. Background Technology

[0002] As the core actuator of fluid control systems, the operating status of control valves directly affects the safety and economy of industrial production. In high-risk fields such as petrochemicals, power, and nuclear energy, unplanned shutdowns caused by control valve malfunctions such as jamming, leakage, or wear can lead to major safety accidents and huge economic losses.

[0003] Current technologies primarily rely on fitting the characteristic curves of valve opening and flow rate, and setting fixed thresholds for deviation warnings. However, they face significant drawbacks in actual industrial applications: First, changes in inlet pressure and data fluctuations lead to model inaccuracies. The theoretical flow rates corresponding to the same valve opening vary significantly under different pressure conditions. Current technologies do not standardize pressure parameters, and when the system pressure deviates from the modeling baseline, the model prediction error increases sharply, easily triggering false alarms. Second, data noise suppression and adaptive capabilities are insufficient. Field instruments frequently generate zero values, negative values, or abnormally fluctuating data, and current technologies lack effective cleaning mechanisms. Furthermore, fixed alarm thresholds do not consider individual equipment differences and performance degradation; thresholds that are too strict are prone to false alarms, while those that are too lenient are prone to missed alarms, resulting in poor warning reliability and high maintenance costs.

[0004] In summary, existing technologies suffer from poor adaptability to operating conditions and weak anti-interference capabilities, resulting in problems such as low early warning accuracy and frequent false alarms, making it difficult to meet the urgent needs of predictive maintenance for high reliability and low operation and maintenance costs. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for early warning of control valve malfunctions, comprising: Collect historical operating data of the control valve during normal operation, including time series of inlet pressure, inlet flow, and valve opening. The historical operating data is cleaned to remove zero values, negative values, and abnormal jump data caused by sampling anomalies, forming a benchmark dataset; A standardized regression model of pressure-opening-flow is constructed based on the benchmark dataset. The standardized regression model is used to eliminate the influence of inlet pressure changes and data fluctuations on the flow characteristic curve. Real-time acquisition of the operating data of the control valve; calculation of the theoretical inlet flow rate based on the standardized regression model and the current inlet pressure. The deviation between the theoretical inlet flow and the measured inlet flow is compared with a dynamic threshold, and an early warning is triggered when the deviation exceeds the dynamic threshold.

[0006] Optionally, constructing the standardized regression model includes: Calculate the average inlet pressure P0 in the benchmark dataset; For each historical data point in the benchmark dataset, the regression ingress flow is calculated using the pressure correction formula: Qh = , Where Qr is the inbound flow, Pr is the inbound pressure of this data, and P0 is the average pressure of the benchmark dataset; Establish valve opening K and return inlet flow rate The regression equation is: , Where A, B, C, and D are the undetermined coefficients of the regression equation, and K is the valve opening degree.

[0007] Optionally, the calculation of the theoretical inlet flow includes: Multiple sets of real-time data within the current evaluation period are collected, cleaned, and then used to form an evaluation dataset. Calculate the average valve opening of the evaluation dataset. Average inlet pressure and average inbound flow ; Substituting the average valve opening value into the regression equation yields the regression inlet flow rate.

[0008] The inflow rate can be calculated using the following formula: ,in, To calculate the inlet flow, In order to regain the traffic from the entry point, To evaluate the average ingress pressure of the dataset, The average inlet pressure is the central average pressure of the benchmark dataset.

[0009] Optionally, the deviation calculation and early warning judgment are as follows: First, calculate the evaluation parameters. Where ε is the evaluation parameter, Average inbound traffic of the actual evaluation dataset.

[0010] Secondly, an evaluation judgment is given. ,in, This is the alarm threshold.

[0011] Optionally, the preset alarm threshold is adaptively determined based on the statistical characteristics of the benchmark dataset, and the calculation formula is as follows: , where σ is the standard deviation of the regression residuals of the benchmark dataset, and k is the safety factor.

[0012] On the other hand, the present invention also provides a control valve fault state early warning device, comprising: The data acquisition module is used to acquire historical and real-time operating data of the control valve, including inlet pressure, inlet flow rate, and valve opening. The data cleaning module is used to remove outlier values ​​from the running data. The model building module is used to build a standardized regression model that performs any of the aforementioned control valve fault state early warning methods; The online diagnostic module is used to calculate the theoretical flow rate based on the standardized regression model and compare it with the measured flow rate. When the deviation exceeds the threshold, an early warning message is output.

[0013] Optionally, the online diagnostic module further includes: The dynamic threshold setting unit is used to adaptively adjust the alarm threshold based on the standard deviation of historical data residuals. The delayed confirmation unit is used to trigger an early warning only after the deviation has continuously exceeded the threshold for a preset time, so as to avoid false alarms caused by instantaneous interference.

[0014] Compared with the prior art, the present invention has at least the following beneficial effects: This invention constructs a pressure-standardized regression model to correct historical flow data under different inlet pressure conditions to a unified benchmark, effectively eliminating the interference of industrial pressure fluctuations on valve flow characteristic curves and enabling the regression equation to adapt to changing operating conditions. Simultaneously, by incorporating a cleaning mechanism for zero-value, negative-value, and anomalous data jumps, invalid data such as sensor malfunctions and signal transmission distortions can be effectively removed, ensuring the purity of the model training samples. This effectively solves the problems of high false alarm rates and delayed warnings caused by changes in operating conditions and data noise in traditional methods, thereby significantly improving the accuracy and reliability of fault diagnosis.

[0015] This invention systematically solves the complex dilemmas in industrial settings through the organic synergy of three major technical features: pressure standardization, adaptive threshold, and delayed confirmation. Pressure standardization eliminates the impact of operating condition fluctuations, adaptive threshold matches individual equipment differences, and delayed confirmation suppresses transient interference. All three are indispensable, which increases the early warning accuracy from 70% of existing technologies to over 95% and reduces the false alarm rate from 15% to below 3%. Attached Figure Description

[0016] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A flowchart of a method for early warning of fault status of a control valve provided by the present invention; Figure 2 A block diagram of a control valve fault status early warning device provided by the present invention. Detailed Implementation

[0017] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0018] In industrial automation control systems, control valves, as key actuators in fluid control systems, directly affect the safety and reliability of the entire industrial system. Especially in high-risk fields such as petrochemicals, power systems, and nuclear energy, control valve failures can lead to catastrophic accidents and huge economic losses. With the increasing level of industrial automation, real-time monitoring and early warning of control valve status has become one of the core requirements of modern industrial operation and maintenance. The development of valve status early warning technology at home and abroad has evolved from offline to online, and from simple to intelligent analysis. Common valve failure modes can be summarized into the following categories: jamming and sticking failures, leakage failures, and wear and corrosion failures. For jamming and sticking failures, some literature proposes judging based on actuator motor current fluctuations, deviations between valve position feedback signals and expected values, and abnormally prolonged action times. For leakage failures, judgment is made using changes in acoustic emission signal characteristics of the sealing part, abnormal fluctuations in pressure signals, and abnormal temperature distribution signals. For wear and corrosion failures, judgment is made by monitoring changes in the spectral characteristics of vibration signals, changes in the microstructure of the sealing surface, and gradual changes in flow characteristics. These judgments are made based on specific fault types and require the installation of corresponding monitoring instruments, especially for vibration and sound monitoring. However, these are often constrained by the on-site industrial environment, leading to increased judgment errors. Furthermore, the complexity of the judgment process and the large amount of computation make it difficult to output diagnostic conclusions in a timely manner. In reality, regardless of the type of fault, a mismatch between valve flow and its opening degree will occur. Employing multi-sensor information fusion technology to comprehensively assess the valve's health status is expected to improve the accuracy and timeliness of early warnings.

[0019] To address the issue of early warning for control valve malfunctions, this invention proposes a method for early warning of control valve malfunctions. This method utilizes the characteristic that the flow rate and valve opening are not coordinated when a valve malfunctions to provide early warning of valve malfunctions.

[0020] Specifically, please refer to Figure 1 The method provided in one embodiment of the present invention includes the following steps: S101 collects historical operating data of the control valve during normal operation, including time series of inlet pressure, inlet flow rate, and valve opening.

[0021] Specifically, a distributed control system (DCS) or a supervisory control system (SCADA) is used with a fixed sampling period Δt, exemplarily Δt being 10-60 seconds. Real-time data of the control valve's inlet pressure Pr, inlet flow rate Qr, and valve opening K are continuously collected, forming a timestamped triple {Pr(t), Qr(t), K(t)}. The normal operating period refers to the period during which the control valve operates continuously and stably for no less than 72 hours without any fault alarms or maintenance shutdowns. The collected data volume is no less than 3000 sets, covering 20% ​​to 100% of the valve's rated opening range. After collection, the data is stored in a historical database, using CSV or JSON structured storage format, with timestamp accuracy down to the second.

[0022] S102 performs data cleaning on the historical operating data, removing zero values, negative values, and abnormal jump data caused by factor sampling anomalies, to form a benchmark dataset.

[0023] Specifically, a three-level data cleaning rule is established: Level 1: Numerical range verification and removal. , The first level is for illegal data with K ≤ 0% or K > 100%; the second level is for physical logic verification, which removes jump data where the rate of change of flow rate |(Qr(t)-Qr(t-1)) / Δt| at adjacent time points exceeds 120% of the valve's rated flow rate change rate; the third level is for sliding window smoothing, which uses a median filter with a width of 5 to eliminate random noise on the data that has passed the first two levels of cleaning. The final benchmark dataset needs to have its statistical characteristics calculated, including the mean inlet pressure P0, the mean inlet flow rate Q0, and the standard deviation σ, for subsequent modeling.

[0024] S103 constructs a standardized regression model of pressure-opening-flow rate based on the benchmark dataset. The standardized regression model is used to eliminate the influence of inlet pressure changes and data fluctuations on the flow rate characteristic curve.

[0025] In this embodiment, the regression analysis of valve opening degree and flow rate includes: Calculate the average inlet pressure and average inlet flow rate in the regression dataset; The regression inflow is calculated using the following formula: in, In order to regain the traffic from the entry point, For inbound traffic, For inlet pressure, This represents the average inlet pressure.

[0026] The regression equation system can be established using the following formula: Where A, B, C, and D are the undetermined coefficients of the regression equation, and K is the valve opening degree.

[0027] The above system of equations is solved using a multiple linear regression algorithm to obtain the values ​​of A, B, C, and D, thereby obtaining the regression equation for the normal operation of the valve.

[0028] S104 collects the operating data of the regulating valve in real time, and calculates the theoretical inlet flow rate based on the standardized regression model and the current inlet pressure.

[0029] In this embodiment, the calculation of the inlet flow rate of the regulating valve includes: Multiple sets of real-time operating information of the control valve within a certain time period are collected, and data with inlet pressure, inlet flow, valve opening of 0 and negative values ​​are removed to form an evaluation dataset.

[0030] Calculate the average values ​​of inlet pressure, inlet flow rate, and valve opening in the evaluation dataset.

[0031] Substituting the average valve opening value into the regression equation yields the regression inlet flow rate.

[0032] The inflow rate can be calculated using the following formula: in, To calculate the inlet flow, In order to regain the traffic from the entry point, To evaluate the average ingress pressure of the dataset, The average ingress pressure is used for the benchmark dataset.

[0033] S105 compares the deviation between the theoretical inlet flow and the measured inlet flow with a dynamic threshold, and triggers an early warning when the deviation continues to exceed the dynamic threshold.

[0034] In this embodiment, the comparison between the calculated inbound flow and the actual inbound flow is performed using the following formula: Where ε is the evaluation parameter, This represents the average ingress traffic for the regression dataset.

[0035] Specifically, the relative deviation is first calculated. ×100%, of which To evaluate the average measured ingress traffic of the dataset, the evaluation criteria are then presented. ,in, The alarm threshold is defined as follows. The dynamic threshold ε0 is adaptively determined based on the standard deviation σ of the regression residuals of the benchmark dataset: ε0 = k·σ, where the safety factor k ∈ [3, 5]. "Continuously exceeding" means that ε is greater than ε0 for n consecutive evaluation periods, and the warning device outputs a warning message after the triggering condition is met.

[0036] In summary, the control valve fault state early warning method provided in this specification acquires historical operating information of the control valve during normal operation, including inlet pressure, inlet flow rate, and valve opening. Regression analysis is performed on the valve opening and flow rate to obtain the regression equation for normal valve operation. Real-time control valve operating information is collected, and the calculated inlet flow rate of the control valve is determined based on the regression equation and the real-time valve opening and inlet pressure. The valve operating state is judged by comparing the calculated inlet flow rate with the actual inlet flow rate, and an early warning is issued when the alarm threshold is exceeded. Thus, the fault state of the control valve can be predicted in a timely and accurate manner, laying the foundation for predictive maintenance of the control valve.

[0037] Furthermore, this invention systematically solves the complex challenges in industrial settings that cannot be addressed by a single technical approach through the organic synergy of three major technical features: pressure standardization, adaptive threshold, and delayed confirmation. Pressure standardization eliminates the interference of operating condition changes on the model baseline, ensuring that the flow characteristic curve remains stable under different pressure conditions, providing a reliable and unified benchmark for subsequent diagnosis. The adaptive threshold mechanism dynamically adjusts the alarm boundary based on individual equipment differences and performance degradation characteristics, avoiding the problems of being too lenient or too strict due to a fixed threshold. The delayed confirmation unit, as the last line of defense, effectively suppresses false alarms caused by transient interference. These three elements are indispensable and interconnected: without pressure standardization, operating condition fluctuations will cause the adaptive threshold to lose its stable calculation basis; without an adaptive threshold, even the most thorough data cleaning cannot adapt to the characteristic drift caused by equipment aging; without delayed confirmation, pressure fluctuations or occasional sensor jumps can still penetrate the first two lines of defense and trigger false alarms. This collaborative mechanism has increased the accuracy of early warning from 70% to over 95% of existing technologies, and reduced the false alarm rate from 15% to below 3%, truly achieving a leap from "regular maintenance" to "accurate prediction," and significantly reducing the risk of unplanned downtime and maintenance costs in high-risk industrial sectors.

[0038] To facilitate understanding and implementation by those skilled in the art, the following examples are provided: The data from the normal operation of a certain control valve, after removing data with valve inlet pressure, inlet flow, valve opening of 0 and negative values, forms a regression dataset, as shown in the table below.

[0039] The average inlet pressure and average inlet flow rate in the regression dataset are calculated to be 10.73738 kPa and 10.73738 kPa, respectively. .

[0040] The regression inflow is calculated using the following formula: The calculation results for each group of data are shown on the far right of the table above.

[0041] Using a multiple linear regression algorithm, the data from the above regression dataset are substituted into the following system of equations to obtain the values ​​of A, B, C, and D: Thus, the regression equation for the valve during normal operation is obtained, and the obtained regression equation is: Multiple sets of real-time operating data for the control valve within a certain time period are collected. Data with inlet pressure, inlet flow rate, and valve opening of 0 or negative values ​​are removed to form an evaluation dataset. The average values ​​of inlet pressure, inlet flow rate, and valve opening are calculated, as shown in the table below: Substituting the average valve opening value into the regression equation yields the regression inlet flow rate, i.e.: =-0.01786 * 19.428314043+ 0.341282 * 19.428314042 + 331.6073 *19.42831404 - 503.591 = 5936.8253 m The inflow rate can be calculated using the following formula: =6182.0820 m3 / h The inbound flow and the actual inbound flow are calculated using the following formula: =3.66755% Based on the standard deviation of the regression residuals and the range of values ​​for the safety factor, typically If we take 10%, then based on the above assessment, we can conclude that the valve is currently in normal operating condition.

[0042] Based on the same inventive concept, combined with Figure 2 As shown, this embodiment of the invention also provides a control valve fault state early warning device, comprising: The data acquisition module is used to acquire historical and real-time operating data of the control valve, including inlet pressure, inlet flow rate, and valve opening. The data cleaning module is used to remove outlier values ​​from the running data. The model building module is used to build a standardized regression model that performs any of the aforementioned control valve fault state early warning methods; The online diagnostic module is used to calculate the theoretical flow rate based on the standardized regression model and compare it with the measured flow rate. When the deviation exceeds the threshold, an early warning message is output.

[0043] In summary, the control valve fault state early warning device provided in this specification acquires historical operating information of the control valve during normal operation through an acquisition module, including the control valve inlet pressure, inlet flow rate, and valve opening; performs regression analysis on the valve opening and flow rate through a regression module to obtain the regression equation for normal valve operation; the acquisition module also collects control valve operating information in real time; the calculation module determines the calculated inlet flow rate of the control valve based on the regression equation, using the real-time valve opening and inlet pressure; and the diagnosis module judges the valve operating status by comparing the calculated inlet flow rate with the actual inlet flow rate, and issues an early warning when the alarm threshold is exceeded. Thus, an analytical function for a control valve fault state early warning method is realized, enabling timely and accurate early warning of control valve fault states, thereby laying the foundation for predictive maintenance of control valves.

[0044] This article uses specific examples to illustrate the inventive concept in detail. The description of the above embodiments is only for the purpose of helping to understand the core idea of ​​the present invention. It should be noted that any obvious modifications, equivalent substitutions or other improvements made by those skilled in the art without departing from the inventive concept should be included within the protection scope of the present invention.

[0045] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the foregoing claims.

[0046] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

[0047] It should be understood that "multiple" as used in this article refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0048] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0049] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method of regulating valve failure state early warning, characterized by, include: Collect historical operating data of the control valve during normal operation, including time series of inlet pressure, inlet flow, and valve opening. The historical operating data is cleaned to remove zero values, negative values, and abnormal jump data caused by sampling anomalies, forming a benchmark dataset; A standardized regression model of pressure-opening-flow is constructed based on the benchmark dataset. The standardized regression model is used to eliminate the influence of inlet pressure changes and data fluctuations on the flow characteristic curve. Real-time acquisition of the operating data of the control valve; calculation of the theoretical inlet flow rate based on the standardized regression model and the current inlet pressure. The deviation between the theoretical inlet flow and the measured inlet flow is compared with a dynamic threshold, and an early warning is triggered when the deviation exceeds the dynamic threshold.

2. The method of claim 1, wherein, The construction of the standardized regression model includes: Calculate the average inlet pressure P0 in the benchmark dataset; For each piece of historical data in the benchmark data set, the regression inlet flow rate is calculated by the pressure correction formula: , where Qr is the inlet flow rate, P0 is the average pressure of the P0 reference data set; A regression equation of valve opening K and regression inlet flow rate Q is established: ,​ Where A, B, C, and D are the undetermined coefficients of the regression equation, and K is the valve opening degree.

3. The method of claim 1, wherein, The calculated theoretical inflow includes: Multiple sets of real-time data within the current evaluation period are collected, cleaned, and then used to form an evaluation dataset. average valve opening of the evaluation dataset , average inlet pressure , and average inlet flow ; Substituting the average valve opening value into the regression equation yields the regression inlet flow rate; The inlet flow rate is calculated by the following equation : .

4. The method of claim 1, wherein, The deviation calculation and early warning judgment are as follows: First, the evaluation parameter is calculated where ε is the evaluation parameter, is the actual average inlet flow rate; Secondly, the evaluation judgment is given wherein, is an alarm threshold.

5. The method of claim 4, wherein, The preset alarm threshold is adaptively determined based on the statistical characteristics of the benchmark dataset, and the calculation formula is as follows: where σ is the regression residual standard deviation of the reference dataset and k is a safety factor.

6. A control valve fault state early warning device, characterized in that, include: The data acquisition module is used to acquire historical and real-time operating data of the control valve, including inlet pressure, inlet flow rate, and valve opening. The data cleaning module is used to remove outlier values ​​from the running data. A model building module for building a standardized regression model that performs the method described in any one of claims 1-3; The online diagnostic module is used to calculate the theoretical flow rate based on the standardized regression model and compare it with the measured flow rate. When the deviation exceeds the threshold, an early warning message is output.

7. The apparatus of claim 6, wherein, The online diagnostic module further includes: The dynamic threshold setting unit is used to adaptively adjust the alarm threshold based on the standard deviation of historical data residuals. The delayed confirmation unit is used to trigger an early warning only after the deviation has continuously exceeded the threshold for a preset time, so as to avoid false alarms caused by instantaneous interference.

Images