Liquid engine test bed valve fault monitoring method and system

Abstract

The invention discloses a liquid engine test bed valve fault monitoring method and system, and belongs to the technical field of liquid engine test bed valve monitoring, and the method comprises the following steps: collecting and processing complete data of a valve in all modes in previous test run, including a normal state and various fault cases, establishing quantitative correlation models of valve performance parameters, operation modes and environment variables, concluding safe and feasible parameter adjustment boundaries and correlation valve coordination strategies in different fault modes, and storing the models and the strategies into a decision knowledge base; through the combination of the steps, accurate judgment, layered compensation and early warning of the valve fault are achieved, the intelligence and precision level of fault monitoring of the test bed valve is improved, fault misjudgment is reduced, test bed safety is guaranteed, and then intelligent fault monitoring and risk prevention and control of the whole process of the liquid engine test bed valve are achieved.

Description

Technical Field

[0001] This invention relates to the field of valve monitoring technology for liquid engine test benches, and more specifically, to a method and system for monitoring valve faults on liquid engine test benches. Background Technology

[0002] Liquid engine test stands are widely used in commercial aerospace, and the valves on these stands are core control components of the propulsion system. Their operating conditions are complex, they come into contact with special media, and they operate in harsh environments. Their operational status directly determines the safety of the test process and the validity of the test data, making accurate fault monitoring of the valves crucial. Currently, valve fault monitoring technology for test stands still has many shortcomings: a lack of systematic integration and analysis of historical data from all valve modes and states; the absence of a quantitative correlation model between performance parameters and operating conditions / environment; and the lack of a unified fault decision knowledge base. This leads to blind operation in anomaly handling and fault diagnosis, resulting in low efficiency and accuracy. The system suffers from several problems: low accuracy; lack of quantitative assessment of valve health; rigid operating procedures; misjudgment of performance degradation caused by valve aging as a fault, leading to meaningless test interruptions; simplistic fault handling methods; failure to assess the remaining operational capacity of valves; and the inability to monitor and fail the entire test due to a single valve failure. Furthermore, it only addresses the faulty valve itself, lacking a health analysis and predictive warning mechanism for similar valve groups, which can easily trigger cascading failures and lacks accurate basis for preventative valve maintenance. These issues severely restrict the reliability of the test bench valve system, necessitating the development of a more intelligent and scientific valve fault monitoring method and system for liquid engine test benches. No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0003] To address the problems in related technologies, this invention proposes a valve fault monitoring method and system for liquid engine test benches, in order to overcome the problems of high valve misjudgment rate, inefficient fault handling, and lack of predictive early warning technology in existing related technologies. Therefore, the specific technical solution adopted by the present invention is as follows: A method and system for monitoring valve failures on a liquid engine test bench, the method comprising the following steps: S1. Collect and process complete data of valves under all modes during each commissioning, including normal state and various failure cases. Then, establish a quantitative correlation model between valve performance parameters, operating mode, and environmental variables. At the same time, summarize the safe and feasible parameter adjustment boundaries and related valve coordination strategies under different failure modes, and store these models and strategies in the decision knowledge base. S2. Classify valves according to their model, function, working medium, and severity level, extract multi-dimensional feature data of each category in the newly installed state, and quantify the health status of each valve based on historical data; S3. Based on the health status of each valve, optimize the predetermined operation plan, intelligently adjust the valve response time to increase the interval tolerance, and transform this plan into an intuitive and dynamically updated two-dimensional time sequence display diagram. S4. When the test bench starts, it collects sensor data in real time. When abnormal valve parameters are detected and verified as a real fault while remaining capacity is available, compensation is made by adjusting valve parameters. If the original target cannot be achieved, a compatible test plan is matched from the preset decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure that the test task is completed under the premise of safety. S5. For valves that have been identified as faulty, immediately initiate group health analysis and conduct key monitoring and analysis on other valves of the same category and with the same health level; when these valves begin to show the same characteristics, even if the fault threshold has not been reached, immediately issue a fault warning. As a preferred implementation, the process of collecting and processing complete data of the valve under all modes during each commissioning run, including normal conditions and various failure cases, and then establishing a quantitative correlation model between valve performance parameters, operating modes, and environmental variables, while summarizing safe and feasible parameter adjustment boundaries and related valve coordination strategies under different failure modes, and storing these models and strategies in a decision knowledge base includes the following steps: S11. Collect, clean, and synchronize the raw data from each test run to ensure that the time series data of each valve in each stage of startup, steady state, and shutdown, as well as under different thrust modes, are completely extracted from each test task, and the corresponding operating mode, environmental variables, and final state are labeled. The final state includes normal or specific fault type. S12. Analyze the statistical relationship between the key performance parameters of the valve, including action time, flow coefficient, sealing leakage rate, operating mode, and environmental variables, and train a regression model that can quantitatively describe the mapping between operating conditions and performance. S13. Perform reverse engineering and safety boundary simulation on the failure cases, and summarize the maximum adjustment range of valve parameters, namely opening degree and response speed, that are allowed to maintain system safety under each typical failure mode, as well as the list of related valves that need to be coordinated to compensate for their functional loss and the corresponding control strategies, including the increase in the opening degree of the upstream valve and the coordination sequence of the downstream valve. S14. The established model, parameter adjustment boundary rules, and associated valve coordination strategies are structured, coded, and verified, and stored in the decision knowledge base. As a preferred embodiment, the process of classifying valves according to their model, function, working medium, and severity level, extracting multi-dimensional feature data for each category in its newly installed state, and quantifying the health status of each valve based on historical data includes the following steps: S21. Establish a detailed classification and coding system based on the valve's model, specifications, core functions, working medium, and the severity level of the environment it is in. For each category, retrieve the standardized initial data acquisition program after its initial installation and start-up, and record its multi-dimensional characteristic data under calibrated working conditions, including complete opening and closing action time curves, dynamic waveforms of actuator drive current or air pressure, and benchmark sealing leakage rate obtained through precision testing, to form the original health fingerprint database of this type of valve. S22. Establish an individual profile for each independent valve, store its initial health fingerprint data as a baseline model; and based on the similar data continuously collected for the valve in each subsequent commissioning or test cycle, use time series analysis to track the long-term trend and short-term fluctuations of various performance parameters, including average operating time, peak current, and leakage rate, relative to their own baseline, thereby establishing their unique dynamic performance degradation model. S23. Based on the individual performance degradation model of each valve, focus on analyzing the data of its most recent complete operation, and compare it with the historical data of the valve, including the recent average, historical best and worst values, and the group data of valves of the same category from multiple perspectives. Through a weighted algorithm, integrate and calculate multiple normalized indicators such as action time delay rate, current curve distortion degree, and leakage rate growth rate, and finally output a comprehensive health index in the range of 0-100. This index directly reflects the degree of deviation of the valve's current state from its ideal state and its remaining capacity; In a preferred embodiment, a weighted algorithm is used to fuse multiple normalized indicators, including action time delay rate, current curve distortion, and leakage rate growth rate, for calculation. The calculation formula is as follows: ; in, To assess the number of indicators, For the first Normalized weights of each indicator, For the first The degree of degradation of each indicator after normalization. In a preferred embodiment, the predetermined operating plan is optimized by combining the health status of each valve, and the response time of the valves is intelligently adjusted to increase the interval tolerance. This plan is then transformed into an intuitive and dynamically updated two-dimensional time sequence display diagram. The specific steps are as follows: S31. Based on the range of the valve's comprehensive health index, set clear health level thresholds, defining a health index score ≥85 as excellent, a health index score between 65-79 as good, a health index score between 50-64 as slightly aging, and a health index score <50 as requiring close attention. S32. Configure specific tolerance adjustment coefficients for each health level. For valves of the excellent level, use 100% of the standard response time, that is, do not adjust the valve response time; for the good level, use 105%-110% of the standard response time according to its deterioration trend; for the slightly aging level, use 110%-135% of the quasi-response time according to its deterioration trend; for the key concern level, in addition to using 135%-160% of the quasi-response time, an additional status confirmation step will be forcibly added. S33. Before execution, each valve node in the operation sequence will be traversed to query its current health level and generate a new instruction sequence containing a flexible time window. S34. The optimized solution is transformed into an intuitive and dynamically updated two-dimensional time sequence display diagram, in which the health level of each valve node is intuitively displayed by the colors green, blue, yellow and orange. In a preferred embodiment, the test bench collects sensor data in real time upon startup. When abnormal valve parameters are detected and verified as a genuine fault while remaining capacity is available, compensation is achieved by adjusting the valve parameters. If the original target cannot be achieved, a compatible test plan is matched from a preset decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure the test task is completed safely. This includes the following steps: S41. After the test bench is started, it collects sensor data of all valves and related pipelines in real time at high frequency. The sensor data includes pressure, flow, valve position and temperature data, and continuously compares them with the dynamic expected values ​​generated based on the preset digital twin model. When any parameter deviates from its normal fluctuation range, an abnormal event is immediately triggered and the valve is marked as a state to be diagnosed. At the same time, it is evaluated to determine whether intervention control is required. S42. For valves marked as abnormal, the associated valve coordination strategy in the decision knowledge base is automatically invoked. Under the premise of ensuring overall safety, a preset active test sequence is executed. By observing the overall pressure and flow response pattern of the system and matching it with the simulation curves of various faults in the digital twin, it can be distinguished whether it is a pseudo-abnormality caused by sensor drift or transient interference, or a real fault of mechanical performance degradation of the valve body. S43. When the abnormal valve parameters are not caused by valve mechanical failure, but by sensor instantaneous drift, environmental interference or fluid transient effects, the adaptive compensation mechanism is also activated through the associated valve coordination strategy in the decision knowledge base to bring the parameters back to normal. S44. When the valve is confirmed to be faulty, continue to evaluate the remaining working capacity of the valve. If it still has remaining working capacity, the system will first try to adjust the control parameters of the faulty valve itself according to the coordination strategy preset for this type of fault in the decision knowledge base. If the effect is insufficient, it will automatically coordinate its associated valves to perform joint compensation so that the predetermined flow and pressure core test targets can be achieved within the original control sequence. S45. If the system determines that the original test plan's objective cannot be safely achieved, i.e., the rated thrust cannot be reached, it will immediately match a new test plan from the preset knowledge decision base that is fully compatible with the remaining capacity of the current faulty valve and the status of other valves. Then, it will dynamically adjust the control sequence, timing parameters, and engine setpoint of the global valves and execute the downgraded test task to ensure the completion of this test. In a preferred embodiment, when the verification indicates a valve failure, further evaluation of the valve's remaining operational capability includes the following steps: S441. After verifying that the valve is faulty, first diagnose and lock the specific fault mode, and then call the known performance degradation data and historical case data of the fault mode from the decision knowledge base as the initial benchmark for evaluating its remaining capacity. S442. To avoid secondary risks to the system, the system designs and executes a targeted micro-test sequence under strict safety boundary control. This includes sending a small adjustment command to the faulty valve or changing the local differential pressure by coordinating its associated valves. At the same time, the actual response data of the valve is collected at high frequency to directly measure its real-time performance residual under the current fault state. S443. Using the acquired real-time performance parameters as boundary conditions, establish a digital twin model of the valve, and input the real-time performance into the digital twin model of the valve to simulate and deduce the various operating conditions that it can withstand in the subsequent complete test task. S444. Based on the above data, the remaining capacity of the valve is quantitatively calculated to generate key indicators including the maximum reliable adjustment range, the maximum pressure difference that can be withstood, the action time delay coefficient, and the upper limit of the allowable leakage rate. Based on this, a clear and conservative safe operating window is defined, and its allowable operation and operating range are clearly defined. In a preferred embodiment, for valves already identified as faulty, immediately initiating group health analysis, and focusing on monitoring and analyzing other valves of the same category and with the same health level; when these valves begin to exhibit the same characteristics, even if the fault threshold has not been reached, immediately issuing a fault warning includes the following steps: S51. Extract a set of standardized multidimensional feature vectors from the real-time data of valves with confirmed faults. These include the core performance parameters that cause the fault, namely the specific vibration spectrum peak, the shape of the action time delay curve, the slope of the leakage rate increase and its evolution trajectory, forming a digital fingerprint of the fault mode. S52. Based on the valve classification system, select valves that belong to the same model, function, medium, and severity level as the valves confirmed to be faulty; further, from these similar valves, accurately match those individuals whose real-time health index is in a similar range to form a target group to be monitored. S53. Increase the monitoring frequency and data granularity of all valves in the target group, and start a dedicated feature comparison engine to calculate the similarity between the current performance data of each valve and the fault digital fingerprint extracted in the previous step in real time, and quantify whether its deviation from the healthy baseline pattern begins to match the known fault characteristics. S54. When the comparison engine finds that the feature similarity of a certain valve exceeds the preset warning threshold, it determines that the valve has shown the same early degradation trend, and then triggers a predictive fault warning and outputs the warning information. The warning threshold is set by statistical methods based on the analysis of the normal valve performance fluctuation range and early fault characteristic evolution pattern in historical data. A liquid engine test bench valve fault monitoring system, which adopts any of the above-mentioned liquid engine test bench valve fault monitoring methods, including a data acquisition and preprocessing module, a decision knowledge base establishment module, a valve health assessment module, an operation plan optimization module, a visualization module, a fault compensation module, and an early warning module. The data acquisition and preprocessing module collects raw data from all valve modes during each test run, multi-dimensional feature data of newly installed valves, and real-time sensor data when the test bench is started. It also performs preprocessing tasks such as data cleaning, time synchronization and alignment, working condition and status labeling, and feature data extraction. The decision knowledge base building module trains a quantitative correlation regression model between valve performance parameters, operating modes, and environmental variables based on preprocessed historical data; through reverse engineering and simulation of fault cases, it summarizes the parameter adjustment boundaries and associated valve coordination strategies under various fault modes; and stores all models, rules, and strategies after structured coding and verification. The valve health assessment module quantifies the health level of valves. First, it establishes a classification coding system and an original health fingerprint database based on valve model and function. Then, it creates an individual file for each valve and constructs a dynamic performance degradation model. Finally, it combines recent and historical data of valves and data of the same category group, and calculates a comprehensive health index of 0-100 by integrating normalized indicators such as action time delay rate through a weighted algorithm, and classifies the valve health level. The operation scheme optimization module optimizes the trial operation scheme based on the valve health assessment results, configures corresponding tolerance adjustment coefficients for different health levels, intelligently adjusts the valve response time, and generates a new instruction sequence with a flexible time window. The visualization module transforms the scheme into a dynamically updated two-dimensional time-series display diagram, visually displaying the health level of each valve in green, blue, yellow, and orange colors; The fault compensation module and early warning module collect sensor data at high frequency during commissioning and compare it with the dynamic expected values ​​of the digital twin model to identify and distinguish between pseudo-anomalies and real faults. For anomalies or faults, the valve's own parameters are adjusted or related valves are coordinated for compensation. For real faults, the remaining capacity is quantified first. If compensation is ineffective, a compatible test plan is matched from the decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure the safe completion of the test task. After a valve is diagnosed with a fault, its fault digital fingerprint is extracted. Valves of the same category and with similar health levels are selected for key monitoring. The monitoring frequency is increased, and the similarity between performance data and fault fingerprints is compared in real time. If the similarity exceeds the early warning threshold, a predictive fault warning is triggered to identify the early degradation trend of similar valves in advance. The beneficial effects of this invention are as follows: 1. This invention collects full-mode, full-state data of valves during all test runs and establishes a quantitative correlation model between performance parameters and operating conditions and environment. Simultaneously, it summarizes parameter adjustment boundaries and associated valve coordination strategies under fault modes and stores them in a decision knowledge base. This provides a scientific and practical theoretical and strategic basis for subsequent monitoring and adjustment of abnormal valve parameters and accurate fault diagnosis. This knowledge base allows the system to directly retrieve matching models and strategies to perform parameter adjustments and associated valve coordination when facing anomalies, without blindly debugging. This significantly improves the efficiency and accuracy of anomaly handling and fault diagnosis, and also provides pre-set strategy support for matching solutions under subsequent fault conditions, ensuring that the entire fault handling process is systematic and predictable. 2. This invention categorizes valves by model, function, and other dimensions, and establishes a health fingerprint database for newly installed valves. By combining this database with historical data to quantify the health status of each valve, it can accurately capture the actual performance status and degradation trend of the valves. Based on the health status, the optimization of the operation plan and the intelligent adjustment of the valve response time effectively increase the tolerance of the operation interval. It can be specifically adapted to the performance characteristics of slightly aged valves, avoiding the misjudgment of normal performance degradation phenomena caused by aging as faults. This reduces meaningless test interruptions and false alarms, while ensuring that the test operation plan is highly matched with the actual performance of the valves, thus improving the stability of the test bench and the accuracy of the test data. 3. When a valve malfunctions, its remaining operational capacity is first assessed. If remaining capacity is available, the parameters of the malfunctioning valve itself are adjusted first. If the parameter adjustment is insufficient, related valves are coordinated for compensation. If the original test objective still cannot be achieved, a compatible solution is matched from the decision knowledge base, and the global control sequence and parameters are dynamically adjusted. This layered approach maximizes the utilization of the malfunctioning valve's remaining operational capacity, ensuring continuous monitoring during the commissioning process and preventing test monitoring interruptions due to a single valve failure. Furthermore, even if the test cannot be completed according to the original plan, a different plan can be implemented under safe conditions, ensuring the successful completion of the entire commissioning test and preventing the entire test from failing due to a single valve failure. 4. Once a valve is identified as faulty, its digital fingerprint is immediately extracted. Group health analysis and focused monitoring are then conducted on other valves of the same category and with similar health levels. By comparing the similarity between performance data and the fault fingerprint in real time, a fault warning is triggered in advance before the valve reaches its fault threshold. This achieves predictive prevention and control of valve faults. This approach transforms passive post-fault handling into proactive pre-risk warning, enabling timely identification of early degradation trends in similar valves. This effectively reduces the possibility of cascading failures in valves of the same category, improves the overall operational reliability of the test bench valve system, and provides accurate data for preventative valve maintenance, reducing the risk of subsequent test failures from the source. Attached Figure Description To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 This is a flowchart of a valve fault monitoring method for a liquid engine test bench according to an embodiment of the present invention. Figure 2 This is a block diagram of a valve fault monitoring system for a liquid engine test bench according to an embodiment of the present invention. Detailed Implementation To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention. The components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components. According to embodiments of the present invention, a method and system for monitoring valve failures on a liquid engine test bench are provided. The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, the liquid engine test bench valve fault monitoring method according to an embodiment of the present invention includes the following steps: S1. Collect and process complete data of valves under all modes during each commissioning, including normal state and various failure cases. Then, establish a quantitative correlation model between valve performance parameters, operating mode, and environmental variables. At the same time, summarize the safe and feasible parameter adjustment boundaries and related valve coordination strategies under different failure modes, and store these models and strategies in the decision knowledge base. Furthermore, complete data on valves under all modes during each commissioning run are collected and processed, including normal conditions and various failure cases. A quantitative correlation model between valve performance parameters, operating modes, and environmental variables is then established. Simultaneously, safe and feasible parameter adjustment boundaries and associated valve coordination strategies under different failure modes are summarized. These models and strategies are then stored in a decision knowledge base, including the following steps: S11. Collect, clean, and synchronize the raw data from each test run to ensure that the time series data of each valve in each stage of startup, steady state, and shutdown, as well as under different thrust modes, are completely extracted from each test task, and the corresponding operating mode, environmental variables, and final state are labeled. The final state includes normal or specific fault type. S12. Analyze the statistical relationship between the key performance parameters of the valve, including action time, flow coefficient, sealing leakage rate, operating mode, and environmental variables, and train a regression model that can quantitatively describe the mapping between operating conditions and performance. S13. Perform reverse engineering and safety boundary simulation on the failure cases, and summarize the maximum adjustment range of valve parameters, namely opening degree and response speed, that are allowed to maintain system safety under each typical failure mode, as well as the list of related valves that need to be coordinated to compensate for their functional loss and the corresponding control strategies, including the increase in the opening degree of the upstream valve and the coordination sequence of the downstream valve. S14. The established model, parameter adjustment boundary rules, and associated valve coordination strategies are structured, coded, and verified, and stored in the decision knowledge base. S2. Classify valves according to their model, function, working medium, and severity level, extract multi-dimensional feature data of each category in the newly installed state, and quantify the health status of each valve based on historical data; Furthermore, based on the valve's model, function, working medium, and severity level, the valves are categorized, and multi-dimensional feature data of each category in its newly installed state are extracted. The health status of each valve is then quantified based on historical data, including the following steps: S21. Establish a detailed classification and coding system based on the valve's model, specifications, core functions, working medium, and the severity level of the environment it is in. For each category, retrieve the standardized initial data acquisition program after its initial installation and start-up, and record its multi-dimensional characteristic data under calibrated working conditions, including complete opening and closing action time curves, dynamic waveforms of actuator drive current or air pressure, and benchmark sealing leakage rate obtained through precision testing, to form the original health fingerprint database of this type of valve. S22. Establish an individual profile for each independent valve, store its initial health fingerprint data as a baseline model; and based on the similar data continuously collected for the valve in each subsequent commissioning or test cycle, use time series analysis to track the long-term trend and short-term fluctuations of various performance parameters, including average operating time, peak current, and leakage rate, relative to their own baseline, thereby establishing their unique dynamic performance degradation model. S23. Based on the individual performance degradation model of each valve, we focus on analyzing the data of its most recent complete operation and compare it with the historical data of the valve, including the recent average, historical best and worst values, and the group data of valves of the same category from multiple perspectives. Through a weighted algorithm, we integrate and calculate multiple normalized indicators such as action time delay rate, current curve distortion degree, and leakage rate growth rate, and finally output a comprehensive health index in the range of 0-100. Furthermore, a weighted algorithm is used to fuse multiple normalized indicators, including action time delay rate, current curve distortion, and leakage rate growth rate, for calculation. The formula is as follows: ; in, To assess the number of indicators, For the first Normalized weights of each indicator, For the first The degree of degradation of each indicator after normalization; It should be noted that the evaluation metrics include action time delay rate, current curve distortion, and leakage rate growth rate. The degradation level is 3. The value is between 0 and 1, and the degree of deterioration is obtained by analyzing the historical performance data of the valve. First, the typical value of each indicator in a brand new state is determined from the historical data as the health benchmark. Then, based on the analysis of a large number of operation and failure cases, the maximum safe limit of the indicator is determined as the failure threshold. The degree of deterioration d is calculated as (current measured value - health benchmark) / (failure threshold - health benchmark) and truncated between 0 and 1. The weighting coefficients of each indicator are determined by comprehensively analyzing the statistical contribution of each indicator in historical failure data and combining it with human experience in safety and reliability, so as to ensure that the weighting allocation objectively reflects the data pattern and conforms to the safety priority of the project. S3. Based on the health status of each valve, optimize the predetermined operation plan, intelligently adjust the valve response time to increase the interval tolerance, and transform this plan into an intuitive and dynamically updated two-dimensional time sequence display diagram. Furthermore, based on the health status of each valve, the predetermined operating plan is optimized, and the valve response time is intelligently adjusted to increase the interval tolerance. This plan is then transformed into an intuitive and dynamically updated two-dimensional time sequence diagram. The specific steps are as follows: S31. Based on the range of the valve's comprehensive health index, set clear health level thresholds, defining a health index score ≥85 as excellent, a health index score between 65-79 as good, a health index score between 50-64 as slightly aging, and a health index score <50 as requiring close attention. S32. Configure specific tolerance adjustment coefficients for each health level. For valves of the excellent level, use 100% of the standard response time, that is, do not adjust the valve response time; for the good level, use 105%-110% of the standard response time according to its deterioration trend; for the slightly aging level, use 110%-135% of the quasi-response time according to its deterioration trend; for the key concern level, in addition to using 135%-160% of the quasi-response time, an additional status confirmation step will be forcibly added. S33. Before execution, each valve node in the operation sequence will be traversed to query its current health level and generate a new instruction sequence containing a flexible time window. S34. The optimized solution is transformed into an intuitive and dynamically updated two-dimensional time sequence display diagram, in which the health level of each valve node is intuitively displayed by the colors green, blue, yellow and orange. S4. When the test bench starts, it collects sensor data in real time. When abnormal valve parameters are detected and verified as a real fault while remaining capacity is available, compensation is made by adjusting valve parameters. If the original target cannot be achieved, a compatible test plan is matched from the preset decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure that the test task is completed under the premise of safety. Furthermore, the test bench collects sensor data in real time during startup. When abnormal valve parameters are detected and verified as a genuine fault while remaining capacity is available, compensation is achieved by adjusting the valve parameters. If the original target cannot be achieved, a compatible test plan is matched from the preset decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure the test task is completed safely. This includes the following steps: S41. After the test bench is started, it collects sensor data of all valves and related pipelines in real time at high frequency. The sensor data includes pressure, flow, valve position and temperature data, and continuously compares them with the dynamic expected values ​​generated based on the preset digital twin model. When any parameter deviates from its normal fluctuation range, an abnormal event is immediately triggered and the valve is marked as a state to be diagnosed. At the same time, it is evaluated to determine whether intervention control is required. The pre-defined digital twin model is established by integrating the physical mechanism of the valve with massive historical operating data, and using system identification and machine learning methods for joint training and calibration. S42. For valves marked as abnormal, the associated valve coordination strategy in the decision knowledge base is automatically invoked. Under the premise of ensuring overall safety, a preset active test sequence is executed. By observing the overall pressure and flow response pattern of the system and matching it with the simulation curves of various faults in the digital twin, it can be distinguished whether it is a pseudo-abnormality caused by sensor drift or transient interference, or a real fault of mechanical performance degradation of the valve body. S43. When the abnormal valve parameters are not caused by valve mechanical failure, but by sensor instantaneous drift, environmental interference or fluid transient effects, the adaptive compensation mechanism is also activated through the associated valve coordination strategy in the decision knowledge base to bring the parameters back to normal. It should be noted that by fine-tuning the parameters of the associated valves to counteract abnormal disturbances, the key output parameters of the entire system can be restored to a stable state within the expected range, thereby ensuring the accuracy of the process and the validity of the data without interrupting the test. S44. When the valve is confirmed to be faulty, continue to evaluate the remaining working capacity of the valve. If it still has remaining working capacity, the system will first try to adjust the control parameters of the faulty valve itself according to the coordination strategy preset for this type of fault in the decision knowledge base. If the effect is insufficient, it will automatically coordinate its associated valves to perform joint compensation so that the predetermined flow and pressure core test targets can be achieved within the original control sequence. Furthermore, when a valve malfunction is confirmed, the remaining operational capability of the valve is further assessed by the following steps: S441. After verifying that the valve is faulty, first diagnose and lock the specific fault mode, and then call the known performance degradation data and historical case data of the fault mode from the decision knowledge base as the initial benchmark for evaluating its remaining capacity. S442. To avoid secondary risks to the system, the system designs and executes a targeted micro-test sequence under strict safety boundary control. This includes sending a small adjustment command to the faulty valve or changing the local differential pressure by coordinating its associated valves. At the same time, the actual response data of the valve is collected at high frequency to directly measure its real-time performance residual under the current fault state. S443. Using the acquired real-time performance parameters as boundary conditions, establish a digital twin model of the valve, and input the real-time performance into the digital twin model of the valve to simulate and deduce the various operating conditions that it can withstand in the subsequent complete test task. S444. Based on the above data, the remaining capacity of the valve is quantitatively calculated to generate key indicators including the maximum reliable adjustment range, the maximum pressure differential it can withstand, the action time delay coefficient, and the upper limit of the allowable leakage rate. A clear and conservative safe operating window is then defined, explicitly specifying its permissible operation and operating range. S45. If the system determines that the original test plan's objective cannot be safely achieved, i.e., the rated thrust cannot be reached, it immediately selects a new test plan from the preset knowledge decision base that is fully compatible with the remaining capacity of the currently faulty valve and the status of other valves. Then, it dynamically adjusts the control sequence, timing parameters, and engine setpoints of all global valves, executes the downgraded test task, and ensures the completion of this test run. S5. For valves that have been identified as faulty, immediately initiate group health analysis and conduct key monitoring and analysis on other valves of the same category and with the same health level; when these valves begin to show the same characteristics, even if the fault threshold has not been reached, immediately issue a fault warning. Furthermore, for valves already identified as faulty, a group health analysis is immediately initiated. Other valves of the same category and with similar health levels are then subject to focused monitoring and analysis. When these valves begin to exhibit the same characteristics, a fault warning is immediately issued, even if the fault threshold has not been reached, including the following steps: S51. Extract a set of standardized multidimensional feature vectors from the real-time data of valves with confirmed faults. These include the core performance parameters that cause the fault, namely the specific vibration spectrum peak, the shape of the action time delay curve, the slope of the leakage rate increase and its evolution trajectory, forming a digital fingerprint of the fault mode. S52. Based on the valve classification system, select valves that belong to the same model, function, medium, and severity level as the valves confirmed to be faulty; further, from these similar valves, accurately match those individuals whose real-time health index is in a similar range to form a target group to be monitored. S53. Increase the monitoring frequency and data granularity of all valves in the target group, and start a dedicated feature comparison engine to calculate the similarity between the current performance data of each valve and the fault digital fingerprint extracted in the previous step in real time, and quantify whether its deviation from the healthy baseline pattern begins to match the known fault characteristics. S54. When the comparison engine finds that the feature similarity of a certain valve exceeds the preset warning threshold, it determines that the valve has shown the same early degradation trend, and then triggers a predictive fault warning and outputs the warning information. A liquid engine test bench valve fault monitoring system, which adopts any of the above liquid engine test bench valve fault monitoring methods, including a data acquisition and preprocessing module, a decision knowledge base establishment module, a valve health assessment module, an operation plan optimization module, a visualization module, a fault compensation module, and an early warning module. The data acquisition and preprocessing module collects raw data from all valve modes during each test run, multi-dimensional feature data of newly installed valves, and real-time sensor data when the test bench is started. It also performs preprocessing tasks such as data cleaning, time synchronization and alignment, working condition and status labeling, and feature data extraction. The decision knowledge base building module trains a quantitative correlation regression model between valve performance parameters, operating modes, and environmental variables based on preprocessed historical data; through reverse engineering and simulation of fault cases, it summarizes the parameter adjustment boundaries and associated valve coordination strategies under various fault modes; and stores all models, rules, and strategies after structured coding and verification. The valve health assessment module quantifies the health status of valves. First, it establishes a classification coding system and an original health fingerprint database based on valve model and function. Then, it creates an individual file for each valve and constructs a dynamic performance degradation model. Finally, it combines recent and historical data of valves and data of the same category group, and uses a weighted algorithm to integrate normalized indicators such as action time delay rate to calculate a comprehensive health index of 0-100, and classifies the valve health level. The operation plan optimization module optimizes the commissioning operation plan based on the valve health assessment results, configures corresponding tolerance adjustment coefficients for different health levels, intelligently adjusts the valve response time, and generates a new instruction sequence with a flexible time window. The visualization module transforms the solution into a dynamically updated two-dimensional time-series display diagram, visually displaying the health level of each valve using green, blue, yellow, and orange colors; The fault compensation module and early warning module collect sensor data at high frequency during commissioning and compare it with the dynamic expected values ​​of the digital twin model to identify and distinguish between pseudo-anomalies and real faults. For anomalies or faults, the valve's own parameters are adjusted or related valves are coordinated for compensation. For real faults, the remaining capacity is quantified first. If compensation is ineffective, a compatible test plan is matched from the decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure the safe completion of the test task. After a valve is diagnosed with a fault, its fault digital fingerprint is extracted. Valves of the same category and with similar health levels are selected for key monitoring. The monitoring frequency is increased and the similarity between performance data and fault fingerprints is compared in real time. If the similarity exceeds the warning threshold, a predictive fault warning is triggered to identify the early degradation trend of similar valves in advance. The above are merely preferred embodiments of the present invention and are 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 within the protection scope of the present invention.

Claims

1. A method for monitoring valve faults on a liquid engine test bench, characterized in that, The method includes the following steps: S1. Collect and process complete data of valves under all modes during each commissioning, including normal state and various failure cases. Then, establish a quantitative correlation model between valve performance parameters, operating mode, and environmental variables. At the same time, summarize the safe and feasible parameter adjustment boundaries and related valve coordination strategies under different failure modes, and store these models and strategies in the decision knowledge base. S2. Classify valves according to their model, function, working medium, and severity level, extract multi-dimensional feature data of each category in the newly installed state, and quantify the health status of each valve based on historical data; S3. Based on the health status of each valve, optimize the predetermined operation plan, intelligently adjust the valve response time to increase the tolerance of the interval, and transform this plan into an intuitive and dynamically updated two-dimensional time sequence display diagram. S4. When the test bench starts, it collects sensor data in real time. When abnormal valve parameters are detected and verified as a real fault while remaining capacity is available, compensation is made by adjusting valve parameters. If the original target cannot be achieved, a compatible test plan is matched from the preset decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure that the test task is completed under the premise of safety. S5. For valves that have been identified as faulty, immediately initiate group health analysis and conduct key monitoring and analysis on other valves of the same category and with the same health level; when these valves begin to show the same characteristics, even if the fault threshold has not been reached, immediately issue a fault warning.

2. The method for monitoring valve faults on a liquid engine test bench according to claim 1, characterized in that, The process of collecting and processing complete valve data under all modes during each commissioning run, including normal conditions and various failure cases, and then establishing a quantitative correlation model between valve performance parameters, operating modes, and environmental variables, while summarizing safe and feasible parameter adjustment boundaries and related valve coordination strategies under different failure modes, and storing these models and strategies in a decision knowledge base includes the following steps: S11. Collect, clean, and synchronize the raw data from each test run to ensure that the time series data of each valve in each stage of startup, steady state, and shutdown, as well as under different thrust modes, are completely extracted from each test task, and the corresponding operating mode, environmental variables, and final state are labeled. The final state includes normal or specific fault type. S12. Analyze the statistical relationship between the key performance parameters of the valve, including action time, flow coefficient, sealing leakage rate, operating mode, and environmental variables, and train a regression model that can quantitatively describe the mapping between operating conditions and performance. S13. Perform reverse engineering and safety boundary simulation on the failure cases, and summarize the maximum adjustment range of valve parameters, namely opening degree and response speed, that are allowed to maintain system safety under each typical failure mode, as well as the list of related valves that need to be coordinated to compensate for their functional loss and the corresponding control strategies, including the increase in the opening degree of the upstream valve and the coordination sequence of the downstream valve. S14. The established model, parameter adjustment boundary rules, and associated valve coordination strategies are structured, coded, and verified, and stored in the decision knowledge base.

3. The method for monitoring valve faults on a liquid engine test bench according to claim 1, characterized in that, The process of classifying valves according to their model, function, working medium, and severity level, extracting multi-dimensional feature data for each category in its newly installed state, and quantifying the health status of each valve based on historical data includes the following steps: S21. Establish a detailed classification and coding system based on the valve's model, specifications, core functions, working medium, and the severity level of the environment it is in. For each category, retrieve the standardized initial data acquisition program after its initial installation and start-up, and record its multi-dimensional characteristic data under calibrated working conditions, including complete opening and closing action time curves, dynamic waveforms of actuator drive current or air pressure, and benchmark sealing leakage rate obtained through precision testing, to form the original health fingerprint database of this type of valve. S22. Establish an individual profile for each independent valve, store its initial health fingerprint data as a baseline model; and based on the similar data continuously collected for the valve in each subsequent commissioning or test cycle, use time series analysis to track the long-term trend and short-term fluctuations of various performance parameters, including average operating time, peak current, and leakage rate, relative to their own baseline, thereby establishing their unique dynamic performance degradation model. S23. Based on the individual performance degradation model of each valve, focus on analyzing the data of its most recent complete operation, and compare it with the historical data of the valve, including the recent average, historical best and worst values, and the group data of valves of the same category from multiple perspectives. Through a weighted algorithm, integrate and calculate multiple normalized indicators such as action time delay rate, current curve distortion degree, and leakage rate growth rate, and finally output a comprehensive health index in the range of 0-100.

4. The method for monitoring valve failure on a liquid engine test bench according to claim 3, characterized in that, The weighted algorithm integrates multiple normalized indicators, including action time delay rate, current curve distortion, and leakage rate growth rate, for calculation. The formula is as follows: ; in, To assess the number of indicators, For the first Normalized weights of each indicator, For the first The degree of degradation of each indicator after normalization.

5. The method for monitoring valve faults on a liquid engine test bench according to claim 1, characterized in that, The process involves optimizing a pre-determined operating plan by combining the health status of each valve, intelligently adjusting the valve response time to increase the interval tolerance, and transforming this plan into an intuitive and dynamically updated two-dimensional time-series display diagram. The specific steps are as follows: S31. Based on the range of the valve's comprehensive health index, set clear health level thresholds, defining a health index score ≥85 as excellent, a health index score between 65-79 as good, a health index score between 50-64 as slightly aging, and a health index score <50 as requiring close attention. S32. Configure specific tolerance adjustment coefficients for each health level. For valves of the excellent level, use 100% of the standard response time, that is, do not adjust the valve response time; for the good level, use 105%-110% of the standard response time according to its deterioration trend; for the slightly aging level, use 110%-135% of the quasi-response time according to its deterioration trend; for the key concern level, in addition to using 135%-160% of the quasi-response time, an additional status confirmation step will be forcibly added. S33. Before execution, each valve node in the operation sequence will be traversed to query its current health level and generate a new instruction sequence containing a flexible time window. S34. The optimized solution is transformed into an intuitive and dynamically updated two-dimensional time sequence display diagram, in which the health level of each valve node is intuitively displayed by the colors green, blue, yellow and orange.

6. The method for monitoring valve failure on a liquid engine test bench according to claim 1, characterized in that, When the test bench starts up, it collects sensor data in real time. When abnormal valve parameters are detected and verified as a real fault while remaining capacity is available, compensation is achieved by adjusting the valve parameters. If the original target cannot be achieved, a compatible test plan is matched from the preset decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure the test task is completed safely. This includes the following steps: S41. After the test bench is started, it collects sensor data of all valves and related pipelines in real time at high frequency. The sensor data includes pressure, flow, valve position and temperature data, and continuously compares them with the dynamic expected values ​​generated based on the preset digital twin model. When any parameter deviates from its normal fluctuation range, an abnormal event is immediately triggered and the valve is marked as a state to be diagnosed. At the same time, it is evaluated to determine whether intervention control is required. S42. For valves marked as abnormal, the associated valve coordination strategy in the decision knowledge base is automatically invoked. Under the premise of ensuring overall safety, a preset active test sequence is executed. By observing the overall pressure and flow response pattern of the system and matching it with the simulation curves of various faults in the digital twin, it can be distinguished whether it is a pseudo-abnormality caused by sensor drift or transient interference, or a real fault of mechanical performance degradation of the valve body. S43. When the abnormal valve parameters are not caused by valve mechanical failure, but by sensor instantaneous drift, environmental interference or fluid transient effects, the adaptive compensation mechanism is also activated through the associated valve coordination strategy in the decision knowledge base to bring the parameters back to normal. S44. When the valve is confirmed to be faulty, continue to evaluate the remaining working capacity of the valve. If it still has remaining working capacity, the system will first try to adjust the control parameters of the faulty valve itself according to the coordination strategy preset for this type of fault in the decision knowledge base. If the effect is insufficient, it will automatically coordinate its associated valves to perform joint compensation so that the predetermined flow and pressure core test targets can be achieved within the original control sequence. S45. If the system determines that the original test plan's objective cannot be safely achieved, i.e., the rated thrust cannot be reached, it will immediately match a new test plan from the preset knowledge decision base that is fully compatible with the remaining capacity of the current faulty valve and the status of other valves. Then, it will dynamically adjust the control sequence, timing parameters, and engine setpoint of the global valves and execute the downgraded test task to ensure the completion of this test.

7. The method for monitoring valve faults on a liquid engine test bench according to claim 6, characterized in that, When the verification indicates a valve malfunction, the remaining operational capability of the valve is further assessed through the following steps: S441. After verifying that the valve is faulty, first diagnose and lock the specific fault mode, and then call the known performance degradation data and historical case data of the fault mode from the decision knowledge base as the initial benchmark for evaluating its remaining capacity. S442. To avoid secondary risks to the system, the system designs and executes a targeted micro-test sequence under strict safety boundary control. This includes sending a small adjustment command to the faulty valve or changing the local differential pressure by coordinating its associated valves. At the same time, the actual response data of the valve is collected at high frequency to directly measure its real-time performance residual under the current fault state. S443. Using the acquired real-time performance parameters as boundary conditions, establish a digital twin model of the valve, and input the real-time performance into the digital twin model of the valve to simulate and deduce the various operating conditions that it can withstand in the subsequent complete test task. S444. Based on the above data, the remaining capacity of the valve is quantitatively calculated to generate key indicators including the maximum reliable adjustment range, the maximum pressure difference that can be withstood, the action time delay coefficient, and the upper limit of the allowable leakage rate. Based on this, a clear and conservative safe operating window is defined, and its allowable operation and operating range are clearly defined.

8. The method for monitoring valve faults on a liquid engine test bench according to claim 1, characterized in that, For valves already identified as faulty, a group health analysis is immediately initiated. Other valves of the same category and with similar health levels are then subject to focused monitoring and analysis. When these valves begin to exhibit the same characteristics, a fault warning is immediately issued even if the fault threshold has not been reached. This includes the following steps: S51. Extract a set of standardized multidimensional feature vectors from the real-time data of valves with confirmed faults. These include the core performance parameters that cause the fault, namely the specific vibration spectrum peak, the shape of the action time delay curve, the slope of the leakage rate increase and its evolution trajectory, forming a digital fingerprint of the fault mode. S52. Based on the valve classification system, select valves that belong to the same model, function, medium, and severity level as the valves confirmed to be faulty; further, from these similar valves, accurately match those individuals whose real-time health index is in a similar range to form a target group to be monitored. S53. Increase the monitoring frequency and data granularity of all valves in the target group, and start a dedicated feature comparison engine to calculate the similarity between the current performance data of each valve and the fault digital fingerprint extracted in the previous step in real time, and quantify whether its deviation from the healthy baseline pattern begins to match the known fault characteristics. S54. When the comparison engine finds that the feature similarity of a certain valve exceeds the preset warning threshold, it determines that the valve has shown the same early degradation trend, and then triggers a predictive fault warning and outputs the warning information.

9. A valve fault monitoring system for a liquid engine test bench, characterized in that, The system adopts the liquid engine test bench valve fault monitoring method as described in any one of claims 1-8, including a data acquisition and preprocessing module, a decision knowledge base establishment module, a valve health assessment module, an operation plan optimization module, a visualization module, a fault compensation module, and an early warning module as described below; The data acquisition and preprocessing module collects raw data from all valve modes during each test run, multi-dimensional feature data of newly installed valves, and real-time sensor data when the test bench is started. It also performs preprocessing tasks such as data cleaning, time synchronization and alignment, working condition and status labeling, and feature data extraction. The decision knowledge base building module trains a quantitative correlation regression model between valve performance parameters, operating modes, and environmental variables based on preprocessed historical data; through reverse engineering and simulation of fault cases, it summarizes the parameter adjustment boundaries and associated valve coordination strategies under various fault modes; and stores all models, rules, and strategies after structured coding and verification. The valve health assessment module quantifies the health level of valves. First, it establishes a classification coding system and an original health fingerprint database based on valve model and function. Then, it creates an individual file for each valve and constructs a dynamic performance degradation model. Finally, it combines recent and historical data of valves and data of the same category group, and calculates a comprehensive health index of 0-100 by integrating normalized indicators such as action time delay rate through a weighted algorithm, and classifies the valve health level. The operation scheme optimization module optimizes the trial operation scheme based on the valve health assessment results, configures corresponding tolerance adjustment coefficients for different health levels, intelligently adjusts the valve response time, and generates a new instruction sequence with a flexible time window. The visualization module transforms the scheme into a dynamically updated two-dimensional time-series display diagram, visually displaying the health level of each valve in green, blue, yellow, and orange colors; The fault compensation module and early warning module collect sensor data at high frequency during the test run and compare it with the dynamic expected value of the digital twin model to identify and distinguish between pseudo-anomalies and real faults. For anomalies or faults, the valve's own parameters are adjusted or related valves are coordinated for compensation. For real faults, the remaining capacity is quantified first. If compensation is ineffective, a compatible test plan is matched from the decision knowledge base, and the global control sequence and parameters are dynamically adjusted to ensure the safe completion of the test task. After a valve is diagnosed with a fault, its fault digital fingerprint is extracted. Valves of the same category and with similar health levels are selected for key monitoring. The monitoring frequency is increased and the similarity between performance data and fault fingerprints is compared in real time. If the similarity exceeds the early warning threshold, a predictive fault warning is triggered to identify the early degradation trend of similar valves in advance.

Images