An intelligent adjustment method for reverse flow valve based on adaptive control algorithm

Abstract

The present invention relates to an intelligent regulation method for a reverse flow valve based on an adaptive control algorithm, aiming to improve the accuracy and response efficiency of fluid regulation; the method obtains fluid flow, pressure, temperature and flow direction data collected by sensors installed at the inlet and outlet of the reverse flow valve, and combines the structural parameters of the reverse flow valve to calculate the initial target valve opening under the current fluid working condition using a working condition opening prediction model; based on the relationship between flow, pressure and valve opening, the initial target valve opening is adjusted using a fuzzy rule library to obtain the final target valve opening, thereby adapting to the dynamic changes of complex fluid working conditions; according to the final target valve opening, the actuator is controlled to adjust the operating state of the reverse flow valve to achieve precise regulation; the present invention combines working condition prediction with fuzzy control to improve the regulation accuracy and stability of the reverse flow valve, and is particularly suitable for fluid conveying systems with large flow and pressure fluctuations, and has broad application prospects in the energy, chemical, environmental protection and other industries.

Description

Technical Field

[0001] The present invention relates to the technical field of reverse flow valve control, and in particular to an intelligent adjustment method for a reverse flow valve based on an adaptive control algorithm. Background Art

[0002] Reverse flow valves are widely used in fluid piping systems to control the direction and flow rate of fluids, ensuring stable system operation. Traditional reverse flow valve adjustment methods typically rely on preset control logic or simple PID control strategies, which set fixed control rules based on flow rate, pressure, or opening. In industrial applications, some systems incorporate sensor data for closed-loop control to improve fluid regulation accuracy. Due to the complexity and real-time variability of fluid operating conditions, existing control methods have certain limitations when dealing with nonlinear characteristics, operating condition fluctuations, and external interference.

[0003] The main problem with existing technologies is that fixed rules or simple control algorithms are difficult to adapt to dynamic changes in fluid conditions, resulting in limited regulation accuracy. Furthermore, control methods lacking adaptive regulation capabilities can cause valves to overshoot or undershoot, affecting fluid delivery stability. Furthermore, in complex fluid environments, the lack of intelligent prediction mechanisms makes existing methods difficult to proactively adapt to flow fluctuations, which in turn affects the overall control response speed and energy efficiency.

[0004] Therefore, it is necessary to develop an intelligent adjustment method for reverse flow valves based on adaptive control algorithm. Summary of the Invention

[0005] The present application provides an intelligent regulation method for a reverse flow valve based on an adaptive control algorithm to improve the regulation accuracy and stability of the reverse flow valve.

[0006] The present application provides a reverse flow valve intelligent adjustment method based on an adaptive control algorithm, comprising:

[0007] Obtaining fluid flow, pressure, temperature and flow direction data collected by sensors installed at the inlet and outlet of the reverse flow valve;

[0008] Based on the acquired fluid data and the structural parameters of the reverse flow valve, the initial target valve opening under the current fluid working condition is predicted using the working condition opening prediction model;

[0009] Using a fuzzy rule base based on the relationship between flow, pressure and valve opening, the initial target valve opening is adjusted in real time to obtain the final target valve opening;

[0010] According to the final target valve opening, the actuator is controlled to adjust the operating state of the reverse flow valve.

[0011] The beneficial effects of the technical solution provided by this application include:

[0012] (1) By combining the operating condition opening prediction model with the structural parameters of the reverse flow valve, the target opening can be accurately predicted according to different fluid operating conditions. The fuzzy rule base can be used for real-time adjustment to keep the valve at the optimal opening in a dynamic environment, thereby improving the regulation accuracy and adaptability. (2) The fuzzy rule base is used for real-time adjustment, so that the system can maintain stable operation under large flow and pressure fluctuations, avoiding the problem that traditional PID control is easily affected by changes in operating conditions, and improving the robustness and anti-interference ability of the regulation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of a reverse flow valve intelligent adjustment method based on an adaptive control algorithm provided in the first embodiment of the present application. DETAILED DESCRIPTION

[0014] The following description sets forth many specific details to facilitate a thorough understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of the present application. Therefore, the present application is not limited to the specific implementations disclosed below.

[0015] The first embodiment of the present application provides a reverse flow valve intelligent adjustment method based on an adaptive control algorithm. Figure 1 , which is a schematic diagram of the first embodiment of this application. Figure 1 The first embodiment of the present application provides a detailed description of an intelligent adjustment method for a reverse flow valve based on an adaptive control algorithm.

[0016] Step S101: obtaining fluid flow, pressure, temperature and flow direction data collected by sensors installed at the inlet and outlet of the reverse flow valve.

[0017] In step S101, suitable sensors are first arranged at the inlet and outlet of the counter-flow valve to obtain key parameters of the fluid. Flow rate can be measured using a differential pressure flowmeter, electromagnetic flowmeter, or ultrasonic flowmeter. The specific selection can be determined based on the physical properties of the pipeline medium, the pipe diameter, and the measurement accuracy requirements. For example, for conductive fluids, an electromagnetic flowmeter can be used, while for non-conductive fluids such as gas or certain oil fluids, an ultrasonic flowmeter or mass flowmeter can be used. The pressure sensor can be a piezoresistive, strain gauge, or capacitive pressure sensor, and it should be ensured that it can withstand the maximum pressure within the operating range of the counter-flow valve to avoid sensor damage or excessive measurement errors. The temperature sensor can be a thermocouple or thermistor to ensure stable measurement under high or low temperature conditions. Flow direction data can be obtained through the bidirectional measurement function of the flowmeter, or by arranging multiple flow sensors at different locations and inferring the flow direction by comparing the flow velocity direction at the inlet and outlet.

[0018] The sensor must be installed coaxially with the pipeline to prevent bubbles or sediment from affecting the measurement accuracy. Flange, insert, or clamp installation methods can be used. The appropriate installation method should be selected based on the pipeline material and site environment. In addition, to ensure the reliability of data acquisition, the sensor needs to be calibrated regularly, especially after long-term operation, as measurement errors may increase due to factors such as temperature drift and mechanical wear. During the data acquisition process, data filtering and outlier rejection methods should be combined to avoid the impact of instantaneous pulsation, flow fluctuations, or pressure pulses on system control accuracy. Signal processing methods such as median filtering, mean filtering, or Kalman filtering can be used to improve data stability and credibility.

[0019] The frequency of data collection should be set based on the dynamic response requirements of the reverse flow valve. Generally, the update frequency of the fluid control system can be set between 50 ms and 1 s to ensure that the data can reflect changes in the fluid operating conditions in real time. The collected flow, pressure, temperature, and flow direction data can be transmitted to the host computer or controller via a fieldbus (such as Modbus, CAN, etc.) or wireless communication (such as LoRa, NB-IoT, etc.) for subsequent valve opening calculation and adjustment.

[0020] During data transmission, appropriate communication protocols and error detection mechanisms should be implemented to ensure data integrity and accuracy. For example, CRC (cyclic redundancy check) can be used to verify transmitted data to prevent erroneous control caused by transmission errors. Furthermore, to ensure real-time and stability of the system, edge computing can be used to perform preliminary data processing, such as filtering, data fusion, and feature extraction, at the data acquisition end close to the sensor. This reduces communication latency and computational burden, thereby improving data processing efficiency.

[0021] Through the above steps, the fluid flow, pressure, temperature and flow direction data of the reverse flow valve inlet and outlet can be obtained accurately and in real time, providing accurate and reliable basic data for subsequent valve opening prediction and intelligent adjustment.

[0022] Furthermore, the acquisition of fluid flow, pressure, temperature and flow direction data collected by sensors provided at the inlet and outlet of the reverse flow valve includes:

[0023] Dynamically adjust the sensor's data acquisition frequency according to the rate of change of flow and pressure. When the rate of change exceeds the set threshold, the acquisition frequency is increased; when the rate of change is lower than the set threshold, the acquisition frequency is reduced.

[0024] Store a predetermined amount of historical flow, pressure, temperature and flow direction data, and compare it with the historical data every time new data is collected. When the new data deviates from the set range of historical data, it will be marked as abnormal data and the data will be corrected;

[0025] Based on the structural parameters and historical data of the reverse flow valve, the flow direction data is compensated. When the flow direction data changes suddenly and is accompanied by a significant change in the valve opening, the compensation factor is calculated and the current flow direction data is adjusted to improve the accuracy of flow direction detection.

[0026] In the process of obtaining fluid flow, pressure, temperature and flow direction data collected by sensors at the inlet and outlet of the reverse flow valve, in order to ensure the accuracy, stability and real-time nature of the data, it is necessary to adopt methods such as dynamic acquisition frequency adjustment, historical data storage and comparison, abnormal data correction and flow direction data compensation to adapt to changes in fluid working conditions and provide reliable data input.

[0027] During data collection, the system first sets the sensor's initial sampling frequency. This frequency can be set based on the system's operational requirements, the characteristics of the fluid in the pipeline, and the response time of the reverse flow valve. Generally, the initial sampling frequency should ensure sufficient fluid state information under stable operating conditions while avoiding the data processing burden caused by oversampling. During system operation, the flow rate and pressure rate of change per unit time are calculated in real time. When the rate of change of either parameter exceeds a set upper threshold, such as when a sharp increase or decrease in flow rate is detected within a short period of time, or when there are large pressure fluctuations, the system automatically increases the sensor's sampling frequency, shortening the sampling interval to more precisely capture the rapid changes in fluid state. During this process, the system can use a graded adjustment method, determining the sampling frequency adjustment range based on the magnitude of the change rate. For example, when the flow or pressure rate of change exceeds the first set threshold but does not exceed the second set threshold, the sampling frequency is increased to 1.5 times the initial frequency. When the rate of change exceeds the second set threshold, the sampling frequency is increased to 2 times or more to ensure data integrity under unexpected operating conditions. On the contrary, when the rate of change of flow and pressure is lower than the preset lower threshold, it indicates that the system is in a stable working condition and the continuous data collected by the sensor changes little. At this time, the sampling frequency is reduced to reduce data redundancy, improve system computing efficiency, and reduce sensor energy consumption.

[0028] To ensure data integrity and continuity, each new data collection is compared with historical data, and a certain amount of historical data is stored to detect anomalies. Data storage utilizes a sliding window mechanism. The window size is determined by the number of historical data points and is dynamically adjusted based on changes in fluid conditions to ensure that historical data effectively reflects current operating trends. When new data is written to the storage window, the system calculates the mean and variance of all flow, pressure, temperature, and flow direction data within the current window and compares the mean deviation of the new data with that of the historical data. When the newly collected data value deviates from the historical mean by more than a set threshold, or when its trend is significantly inconsistent with the historical trend, such as when multiple consecutive data points fall outside the normal fluctuation range, the data is marked as an anomaly and anomaly correction is performed. The correction method for anomaly data is selected based on the anomaly type. If the data changes suddenly within a single sampling period but returns to normal in the next sampling period, linear interpolation or historical averaging can be used to correct the data without disrupting the overall trend during data processing. If abnormal data deviates from the normal range for multiple consecutive sampling periods, further investigation is needed to determine whether it is caused by sensor drift, environmental interference, or system failure. For example, when determining sensor drift, data from multiple sensors can be compared, such as the consistency between the measured value of a flow sensor and the theoretical flow calculated by a pressure sensor. Combined with other stored historical data, Kalman filtering or dynamic offset compensation can be used to correct the abnormal data to ensure data reliability and consistency.

[0029] During flow direction data processing, the fluid's motion can be affected by changes in the reverse flow valve's opening. Therefore, flow direction data collected solely by sensors may not accurately reflect the true flow direction. To improve the accuracy of flow direction detection, the system needs to compensate the flow direction data by combining the reverse flow valve's structural parameters and historical data. When the flow direction sensor detects a sudden change in the flow direction data and the system also records a significant adjustment in the valve opening within the same time period, a flow direction compensation factor is calculated and used to correct the current flow direction data. The flow direction compensation factor is calculated based on the magnitude of the valve opening change, fluid inertia, and historical flow direction data. Specifically, when a valve opens or closes rapidly, fluid inertia can cause short-term abnormal fluctuations in the flow direction data. For example, when a reverse flow valve closes, backflow in the pipeline may cause a brief reverse flow, which the flow direction sensor would detect as an abnormal reverse flow signal. In this case, the system needs to combine the flow direction change trends before and after the valve opening adjustment to calculate the influence of inertia during this short period and correct the flow direction data to reflect the true flow state. For different fluid media, due to different parameters such as viscosity and density, the calculation method of the compensation factor can be adjusted based on experimental data to adapt to different types of fluid working conditions.

[0030] Through the above method, the system can ensure the integrity and accuracy of the data while adjusting the sensor sampling frequency in real time, and effectively correct data anomalies caused by sudden working conditions, sensor drift or changes in valve opening, so that the acquired fluid parameters more accurately reflect the actual working conditions and provide reliable data support for subsequent valve adjustments.

[0031] Furthermore, the method stores a predetermined amount of historical flow, pressure, temperature and flow direction data, and compares the data with the historical data each time new data is collected. When the new data deviates from the set range of the historical data, it is marked as abnormal data and the data is corrected, including:

[0032] The data storage window is preset. During each data collection, the latest flow, pressure, temperature and flow direction data are stored in the window. The window length is set according to time or data capacity so that the data in the window can reflect the current working condition trend.

[0033] Calculate data distribution characteristics based on historical data within the storage window, including data mean, variance, and change trend. When the deviation of newly collected data values from the mean of historical data exceeds a set threshold, or the data change trend is significantly inconsistent with the historical trend, the data is marked as abnormal data.

[0034] Correct the data marked as abnormal. When the abnormal data is in a short-term mutation state and no continuous abnormalities occur, use interpolation or historical mean to correct it. When the abnormal data shows a continuous deviation trend, determine whether it is caused by sensor drift or environmental influence, and compensate based on the correlated data of multiple sensors to ensure that the corrected data can accurately reflect the actual fluid working conditions.

[0035] To ensure the system can accurately detect and effectively correct historical flow, pressure, temperature, and flow direction data during storage and processing, a data storage window must be set to retain historical data over a period of time. The size of this storage window can be set based on time or data capacity to accommodate a sufficient number of data points to provide a valid reference for current operating conditions. Each time new data is collected, the system stores the latest flow, pressure, temperature, and flow direction data in this storage window and deletes the oldest data to maintain a fixed storage window size, ensuring that the data within the window always reflects recent operating trends. The storage window length can be set based on the system's response speed, the stability of the fluid operating conditions, and data computational power. For example, in rapidly changing operating conditions, a shorter window length can be used to more quickly capture data trends, while in relatively stable conditions, a longer window length can be used to enhance the robustness of data analysis.

[0036] After data is stored, statistical analysis is performed on the data within the storage window to calculate its distribution characteristics. Based on these characteristics, the newly collected data is judged as anomalous. During each sampling period, the system calculates the mean, variance, and trend of all historical data within the storage window to serve as a benchmark for determining the plausibility of new data. The mean represents the central tendency of flow, pressure, temperature, and flow direction; the variance measures the degree of dispersion; and the trend identifies patterns of data fluctuation over short periods of time. If the deviation between the newly collected data value and the historical mean within the storage window exceeds a set threshold, or if the trend of the new data significantly deviates from the historical trend—for example, if there is an excessively large flow or pressure fluctuation within a short period of time, while historical data indicates that the operating conditions are generally stable—the data is flagged as anomalous to prevent misleading system control. The threshold for anomaly detection can be set based on the standard deviation of the historical data. For example, data deviating from the mean by more than a certain number of standard deviations is considered an anomaly. This approach adaptively adjusts the judgment criteria based on the fluctuations of different fluid operating conditions, improving the accuracy of anomaly detection.

[0037] Data marked as abnormal requires correction based on its characteristics to ensure data continuity and accuracy. When abnormal data exhibits short-term, sudden changes and similar anomalies have not occurred within multiple consecutive sampling periods, correction can be performed using interpolation or historical averaging. For example, if flow data within a sampling period exhibits a sudden, significant deviation while the preceding and following data remain stable, linear interpolation can be used to calculate a reasonable correction value based on the preceding and following data to replace the abnormal data and smooth the data series. If the abnormal data exhibits a persistent trend of deviation, such as pressure data gradually deviating from the normal range over multiple consecutive sampling periods, further analysis is required to determine whether the cause of the anomaly is sensor drift or environmental influences. In this case, correlation analysis can be performed by combining measurement data from multiple sensors. For example, comparing flow and pressure data collected by different sensors to examine whether there is a consistent pattern of deviation, or by using data from redundant sensors as a reference to identify the specific source of the anomaly.

[0038] If abnormal data is determined to be caused by sensor drift, dynamic compensation methods based on historical data can be used to correct it. For example, long-term trend analysis can be used to identify sensor drift deviations and automatically correct them in subsequent measurement data. If abnormal data is likely caused by environmental factors, such as a temperature sensor being disturbed by an external heat source, compensation can be combined with ambient temperature data to ensure the reliability of the measurement data. This method effectively filters the impact of sensor errors and environmental interference on the data, enabling the system to more accurately perceive fluid conditions and provide reliable data support for subsequent valve control.

[0039] Step S102: Based on the acquired fluid data and in combination with the structural parameters of the reverse flow valve, an initial target valve opening under the current fluid working condition is predicted using a working condition opening prediction model.

[0040] In step S102, the fluid flow rate, pressure, temperature, and flow direction data collected in step S101 are first used in conjunction with the structural parameters of the counterflow valve to establish a working condition opening prediction model. This working condition opening prediction model is based on the principles of fluid mechanics and the operating characteristics of the counterflow valve. It considers the fluid flow state within the pipeline, pressure loss, the effect of temperature on fluid viscosity, and the nonlinear relationship between valve opening and flow rate.

[0041] When calculating the initial target valve opening, fluid data must be preprocessed. Flow, pressure, and temperature data can be affected by sensor noise, short-term fluctuations, or environmental interference. Therefore, data filtering techniques, such as sliding average filtering, median filtering, or Kalman filtering, are required to improve data stability and reliability. Furthermore, flow direction data can be determined based on the flow sensor readings at the inlet and outlet of the reverse flow valve. If the inlet flow exceeds the outlet flow, there may be a risk of backflow, and the valve opening must be adjusted appropriately to prevent backflow.

[0042] After data preprocessing is complete, preliminary calculations are performed based on the reverse flow valve's structural parameters. These parameters include, but are not limited to, the valve's nominal diameter, body material, disc type, flow capacity coefficient (Cv or Kv value), and flow resistance coefficient. These parameters can be obtained from manufacturer datasheets or actual testing and used to correct the fluid flow characteristics during the calculation process. Based on fluid dynamics equations such as the Bernoulli equation, the continuity equation, and the flow resistance calculation formula, the relationship between the target flow rate and valve opening under specific fluid conditions can be calculated.

[0043] To improve prediction accuracy, valve opening prediction methods based on machine learning or data regression can be used. For example, regression models trained on historical operating data, such as polynomial regression, support vector machine regression (SVR), or neural network models, can accurately predict the optimal valve opening under different fluid conditions. Furthermore, if the system has strong nonlinear characteristics, an adaptive neural fuzzy inference system (ANFIS) can be used for modeling to enhance the model's generalization capabilities.

[0044] When calculating the initial target valve opening, first use flow and pressure data to determine the fluid's flow regime, such as laminar, transitional, or turbulent. If the flow is laminar, the target valve opening can be calculated based on the Poiseuille flow equation. If the flow is turbulent, corrections can be made based on empirical formulas or experimental data. For higher or lower temperature conditions, the effect of temperature on fluid viscosity must also be considered and the calculation formula adjusted.

[0045] After calculating the initial target valve opening, it needs to be verified for plausibility. For example, the calculated valve opening can be compared with the valve's mechanical limits. If it exceeds the allowable range, corrections are required. Furthermore, trends in the calculated results can be compared against historical operating data. If the current calculated opening deviates significantly from historical openings under similar operating conditions, the calculation parameters may need to be readjusted or the model updated.

[0046] Through the above steps, the initial target valve opening under the current fluid working conditions can be obtained, providing basic data for subsequent real-time adjustment and optimized control.

[0047] Furthermore, the operating condition opening prediction model includes a data preprocessing module, a feature extraction module, an opening calculation module and an adaptive correction module;

[0048] The data preprocessing module is used to receive flow rate, pressure, temperature and flow direction data, and normalize the data in combination with the structural parameters of the reverse flow valve, remove outliers, smooth the data through a filtering algorithm, and output optimized working condition data;

[0049] The feature extraction module receives the output of the data preprocessing module and calculates the flow-pressure change rate ratio, the fluid pressure drop gradient and the flow direction stability to construct a feature vector reflecting the current working condition, and outputs the feature vector to the opening calculation module;

[0050] The opening calculation module calculates a preliminary target opening based on the output of the feature extraction module using historical data regression analysis and real-time fluid simulation methods, wherein the historical data regression analysis is used for short-term trend prediction, and the real-time fluid simulation is combined with the reverse flow valve structural parameters to correct the predicted opening, and the calculated preliminary target opening is output to the adaptive correction module;

[0051] The adaptive correction module receives the output of the opening calculation module and calculates the prediction error based on real-time feedback data. When the error exceeds a set threshold, the parameters of the opening calculation module are adjusted and the preliminary target opening is corrected to obtain the final predicted initial target valve opening.

[0052] In the process of implementing the operating condition opening prediction model, the flow, pressure, temperature and flow direction data must first be processed through the data preprocessing module to ensure the stability and reliability of the input data. This module receives the raw data from the sensors at the inlet and outlet of the counterflow valve, and normalizes the data in combination with the structural parameters of the counterflow valve to make the numerical range of each physical quantity consistent, so as to avoid the influence of different dimensions on the calculation. In addition, in order to reduce the interference of measurement errors on subsequent calculations, the data needs to be filtered and outliers removed. When it is detected that the data deviates from the historical operating condition mean by more than the set threshold, and the deviation persists for multiple sampling cycles, the data is marked as an outlier and replaced by historical trend interpolation or predicted values calculated based on the physical model. In addition, in order to reduce the influence of high-frequency noise, an adaptive low-pass filtering method is used to smooth the data so that the data is more in line with the actual change trend of the fluid working condition, thereby improving the input data quality of the feature extraction module.

[0053] After data preprocessing is completed, the feature extraction module calculates the optimized operating data to extract key characteristic parameters that can accurately characterize the current fluid state. First, by calculating the ratio of the flow rate and pressure change rate within consecutive time steps, the flow-pressure change rate ratio is obtained. This ratio can reflect the dynamic changes in fluid resistance and is used to evaluate the sensitivity of valve regulation. Secondly, the fluid pressure drop gradient is calculated, that is, the rate of pressure drop between different measurement points. This parameter can quantify the throttling effect of the reverse flow valve on fluid flow and serves as an important basis for predicting valve opening. In addition, the changes in flow direction data within multiple time steps are analyzed and the flow direction stability is calculated to assess whether the fluid is in a steady state or an unstable state such as turbulence. The output of the feature extraction module is a feature vector containing multiple operating condition characteristic parameters. This feature vector serves as the input of the opening calculation module to ensure that the prediction model can calculate the target valve opening based on complete operating condition information.

[0054] After receiving the output from the feature extraction module, the opening calculation module combines historical data regression analysis with real-time fluid simulation methods to calculate a preliminary target opening. During this historical data regression analysis, a time series regression method is used to predict the most likely short-term opening trend based on stored past operating condition data. For example, a weighted moving average method is used to calculate the optimal opening after recent operating condition adjustments, and exponential smoothing methods are used to predict short-term trends, ensuring that the calculated opening value can adapt to historical operating condition changes. Furthermore, to prevent the predicted value from relying solely on historical trends and failing to adapt to the complex dynamic changes of current operating conditions, a real-time fluid simulation method is further incorporated for correction. This simulation method, based on the structural parameters of the reverse flow valve, the current fluid characteristics, and the characteristic operating condition parameters, uses numerical calculation methods to simulate the fluid state of the valve at different openings. The predicted preliminary target opening is then optimized by calculating physical quantities such as pressure loss and flow velocity distribution. Ultimately, the output of the opening calculation module is a preliminary target opening value, which is further optimized in the subsequent adaptive correction module.

[0055] The adaptive correction module receives the output of the opening calculation module and calculates the prediction error through real-time feedback data. When there is a significant deviation between the actual fluid operating conditions of the valve after adjustment and the predicted results, the module analyzes the source of the error and adjusts the parameters of the opening calculation module. For example, when the system detects that the prediction error is continuously too large or too small, it may mean that the historical data regression model fails to fully consider the changes in the current working conditions. Therefore, the regression weight is adjusted to reduce the influence of historical data and increase the weight of real-time fluid simulation. In addition, if the error mainly comes from parameter mismatch in the fluid simulation calculation, such as changes in fluid viscosity or pipe roughness leading to deviations in simulation results, the fluid characteristic parameters in the simulation model are updated through feedback data to improve the accuracy of the simulation calculation. Finally, the module calculates the final initial target valve opening based on the prediction results after error adjustment, and uses this value to control the actuator to adjust the operating state of the reverse flow valve.

[0056] The entire prediction model achieves precise prediction from sensor data to the final target valve opening through a cascade of four modules: data preprocessing, feature extraction, valve opening calculation, and adaptive correction. The output of each module serves as the input to the next module, ensuring that the prediction results not only consider historical trends but also enable real-time optimization based on current fluid conditions. This allows the reverse flow valve to adapt to different operating conditions and achieve precise control.

[0057] Furthermore, the feature extraction module includes a data conversion unit, a feature calculation unit and a dynamic feature optimization unit;

[0058] The data conversion unit is used to receive the output of the data preprocessing module and perform nonlinear transformation on the normalized flow, pressure, temperature and flow direction data to enhance the data differentiation. For flow and pressure data, a piecewise logarithmic mapping function is used to differentially scale data in different ranges so that it has higher resolution in the low value range. For temperature data, an exponential smoothing method is used to reduce the impact of transient changes, thereby obtaining converted data for feature calculation;

[0059] The feature calculation unit receives the output of the data conversion unit and calculates key features of the working condition, including the flow-pressure change rate ratio, the fluid pressure drop gradient and the flow direction stability, wherein the flow-pressure change rate ratio is obtained by calculating the ratio of the flow rate and the pressure change rate in continuous time steps to measure the dynamic response characteristics of the fluid working condition, the fluid pressure drop gradient is calculated based on the flow rate and the pipeline resistance parameter to quantify the influence of the valve opening on the fluid pressure, and the flow direction stability is obtained by calculating the variance of the flow direction data in multiple time steps to evaluate the stability of the fluid flow, and the calculated characteristic parameters are combined into a characteristic vector;

[0060] The dynamic feature optimization unit is used to receive the output of the feature calculation unit and adaptively adjust the feature weight according to the historical operating data. Specifically, based on the historical operating data of the reverse flow valve, the weighted sliding average method is used to calculate the long-term change trend of each feature, and the weight of each feature in the feature vector is adjusted according to the feature change trend to enhance the characterization ability of key influencing factors, so that the generated feature vector is more adapted to the current operating conditions, and the optimized feature vector is output to the opening calculation module.

[0061] In the specific implementation of the feature extraction module, the data conversion unit first receives the output of the data preprocessing module, including normalized flow, pressure, temperature, and flow direction data. Because the numerical ranges and changing trends of different physical quantities can vary significantly, directly using normalized data can result in certain features being underweighted during the calculation process, thus affecting the model's prediction accuracy. To enhance data discrimination, a nonlinear transformation is performed on the input data to ensure sufficient resolution of key operating parameters across different value ranges. During the conversion of flow and pressure data, a piecewise logarithmic mapping function is used to adjust the data, resulting in higher resolution in low-value ranges and smoother changes in high-value ranges. This improves the ability to perceive small flow and pressure fluctuations and avoids neglecting small values at low flow rates or low pressures. For temperature data, since transient temperature changes can introduce noise or short-term anomalies, exponential smoothing is used to reduce the interference of temperature data on the prediction model. This gives higher weight to newer data and gradually attenuates the influence of historical data, resulting in a smooth temperature input data set and improving the robustness of the overall feature. After being processed by the data conversion unit, the converted data is used as the input of the feature calculation unit to further calculate the key features.

[0062] The feature calculation unit receives the output data from the data conversion unit and extracts key operating condition features to construct a feature vector for predicting valve opening. First, the flow-pressure rate of change ratio is calculated to measure the dynamic response characteristics of the fluid system. The flow and pressure rates of change are calculated at each time step, and the ratio is calculated. A large change in the ratio indicates a fast system response and may require more precise valve control. A small change in the ratio indicates that the system is stable and the adjustment frequency can be appropriately reduced. Second, the fluid pressure drop gradient is calculated to quantify the impact of valve opening on fluid pressure. This parameter measures the pressure change at different locations in the pipeline and, combined with the current flow rate, calculates the pressure drop rate per unit length of pipeline, thereby characterizing the regulating effect of the reverse flow valve. A large pressure drop gradient may indicate insufficient valve opening and require an appropriate increase in opening. A small pressure drop gradient may indicate excessive valve opening, resulting in reduced system efficiency. Finally, the flow direction stability is calculated to assess the stability of the fluid flow. The changing trend of flow direction data can be measured by calculating the variance of the flow direction data over multiple consecutive time steps. A small flow direction variance indicates stable fluid flow, while a large flow direction variance may indicate fluid disturbance or transient reverse flow, requiring adjustment of the valve opening to improve flow conditions. Once these characteristic parameters are calculated, they are combined into a feature vector, which is then passed to the dynamic feature optimization unit to further optimize the feature weights and improve the adaptability of the prediction model.

[0063] The dynamic feature optimization unit receives the feature vector output by the feature calculation unit and adaptively adjusts the feature weights based on historical operating data. Because the impact of each feature on valve opening prediction varies under different operating conditions, using fixed weights may result in reduced prediction accuracy under specific conditions. Therefore, dynamic feature weight optimization based on historical operating data is required. First, a weighted moving average method is used to calculate the long-term trend of each feature to identify which features have the greatest impact on valve opening over different time periods. Then, the weights of each feature in the feature vector are adjusted based on their historical variations. For example, if flow direction stability fluctuates significantly over a period of time, while the flow-pressure rate of change ratio and fluid pressure drop gradient remain relatively stable, this indicates that flow direction fluctuations are the primary factor affecting the current operating conditions. Therefore, the weight of flow direction stability is appropriately increased, while the weights of other features are reduced, allowing the prediction model to prioritize the impact of flow direction fluctuations. Conversely, if the flow-pressure rate of change ratio experiences significant fluctuations while flow direction stability remains stable, the weight of the flow-pressure rate of change ratio should be increased, allowing the prediction model to prioritize the dynamic response characteristics of the system. The optimized eigenvectors more accurately represent the current operating conditions and can adapt to varying fluid dynamics, thereby improving the robustness of the prediction model. Ultimately, the optimized eigenvectors are output to the valve opening calculation module as key input for calculating the target valve opening, enabling the prediction model to maintain high-precision prediction capabilities under various operating conditions.

[0064] Furthermore, the opening calculation module includes a short-term trend prediction unit, a real-time fluid simulation unit and an opening correction unit;

[0065] The short-term trend prediction unit is used to analyze the changing trends of historical flow, pressure, temperature and flow direction data based on the output of the feature extraction module using a multi-order regression model, and calculate the target opening within a short time scale; by constructing a recursive linear regression or exponential smoothing regression method, using the data of multiple time steps in the past to fit the trend curve, the optimal opening at the current moment is predicted;

[0066] The real-time fluid simulation unit is used to receive the target opening calculated by the short-term trend prediction unit, and, in combination with the structural parameters of the reverse flow valve and real-time operating data, to correct the target opening through numerical simulation calculation methods. A numerical solution method based on the finite volume method or the finite difference method is used to simulate the fluid flow characteristics under different openings, and to calculate key parameters such as fluid flow rate, pressure drop, and turbulence intensity to determine whether the initially predicted opening meets the current operating conditions. If the simulation results indicate that the predicted target opening may cause flow overshoot or excessive pressure drop, the opening value is adjusted to meet the physical constraints to ensure the rationality of the prediction.

[0067] The opening correction unit is used to receive the corrected target opening output by the real-time fluid simulation unit, and fine-tune the opening in combination with the stability factor of the current working condition to improve the adjustment accuracy; by calculating the stability index of the working condition characteristic vector, the change amplitude of the current fluid state is judged. When the working condition changes drastically, the method of decreasing the adjustment step size is adopted to avoid the impact of the sudden change in the predicted opening on the system; when the working condition tends to be stable, the weighted smoothing adjustment strategy is adopted to make the final target opening more in line with the natural adjustment trend of the fluid; the output preliminary target opening is used for the subsequent adaptive correction module to achieve more precise valve control.

[0068] In the specific implementation of the valve opening calculation module, the short-term trend prediction unit first analyzes historical flow, pressure, temperature, and flow direction data to calculate the target valve opening over a short timescale. This unit receives the feature vectors output by the feature extraction module and uses a multi-order regression model to analyze the data's changing trends to ensure that the prediction results reflect the system's dynamic characteristics. Historical data regression analysis uses recursive linear regression or exponential smoothing regression methods. By fitting data over multiple time steps, a trend curve is constructed to predict the optimal valve opening value at the current moment. For example, in the recursive linear regression process, the influence coefficients of each variable on the valve opening are calculated based on the flow and pressure trends over several past time steps. A regression equation is then established to calculate the optimal valve opening setting at the current moment. In the exponential smoothing regression method, recent data points are given higher weights, allowing the prediction results to quickly adapt to short-term changes in fluid conditions and improve responsiveness to fluid dynamic characteristics. After determining the initial target valve opening, this value is passed as input to the real-time flow simulation unit to further optimize the prediction results.

[0069] The real-time flow simulation unit receives the target opening calculated by the short-term trend prediction unit and, based on the reverse flow valve's structural parameters and real-time operating data, modifies this target opening using numerical simulation methods. Using numerical calculation methods such as the finite volume method or the finite difference method, this unit simulates fluid flow characteristics at different openings and calculates key operating parameters, including flow rate, pressure drop, and turbulence intensity, to determine the rationality of the predicted opening. During the simulation, a discretized mathematical model of the fluid flow is first established. The fluid motion equations are then solved within the structural boundary conditions of the reverse flow valve to simulate the changes in fluid parameters at different valve openings. For example, for a given opening value, the flow velocity distribution per unit time across the pipe cross-section is calculated, and the corresponding pressure loss and turbulence level are evaluated. If the simulation results indicate that the predicted target opening may result in system flow overshoot, excessive pressure drop, or turbulence intensity exceeding a preset threshold, the system adjusts the target opening to ensure compliance with physical constraints. The correction strategy adjusts the opening range based on the fluid load conditions in the simulation results. For example, when the fluid pressure drop is too large, the opening is appropriately increased to reduce drag losses, while when the turbulence intensity is too high, the opening is reduced to improve flow stability. Ultimately, the corrected target opening calculated by this unit is passed as input to the opening correction unit to further optimize the accuracy of the opening adjustment.

[0070] The valve opening correction unit receives the corrected target opening calculated by the real-time fluid simulation unit and, based on the stability factor of the current operating conditions, fine-tunes the valve opening value to improve control accuracy and reduce shock effects during valve adjustment. This unit calculates the stability index of the operating condition characteristic vector to determine the magnitude of the current fluid state fluctuations and, accordingly, determines the valve opening adjustment strategy. When operating conditions fluctuate dramatically, such as when flow or pressure fluctuations exceed a set threshold within a short period of time, the system uses a decreasing adjustment step size to smooth the valve opening adjustment process, thereby avoiding system oscillation or overshoot caused by predicted sudden changes in the valve opening. When operating conditions stabilize, meaning the fluctuations in fluid parameters are within a normal range, the system uses a weighted smoothing adjustment strategy to gradually reduce the target opening adjustment amplitude, better matching the natural regulation trend of the fluid system. For example, during valve adjustment, if the fluid state is detected to be stable, the valve opening adjustment amplitude is exponentially decayed to smooth the final output target opening, reducing mechanical losses associated with frequent adjustments. After the above optimization, the preliminary target opening output by the unit is used as the final prediction value and passed to the adaptive correction module to further improve the prediction accuracy and achieve more precise valve control.

[0071] Furthermore, the adaptive correction module includes an error calculation unit, a weight optimization unit and a dynamic compensation unit;

[0072] The error calculation unit is used to receive the output of the opening calculation module, obtain the actual valve opening and fluid working condition feedback data, and calculate the prediction error; the error calculation includes opening deviation, flow error and pressure error, among which the opening deviation is calculated by comparing the target opening with the actual opening, and the flow error and pressure error are evaluated based on the deviation between the real-time collected data and the calculated value of the prediction model, and the error index is calculated by comprehensive weighting as the input for subsequent optimization;

[0073] The weight optimization unit is used to adaptively adjust the weights of various calculation methods in the opening calculation module based on the error index; when the flow error and pressure error are large, the weight of the real-time fluid simulation method is increased to enhance its adaptability to the current working conditions; when the opening deviation is mainly due to the instability of the short-term trend forecast, the weight of the historical data regression analysis is increased to optimize the accuracy of the trend forecast, thereby achieving dynamic adjustment of the weight based on the error distribution;

[0074] The dynamic compensation unit is used to correct the preliminary target opening according to the optimized weight and generate the final valve opening prediction value; when the error index continuously exceeds the set threshold, an adaptive step adjustment strategy is adopted to avoid large-scale adjustments that cause system oscillations; when the error index tends to stabilize, the adjustment amplitude is gradually converged to ensure a smooth change in the predicted opening.

[0075] During the implementation of the adaptive correction module, the error calculation unit first receives the preliminary target opening generated by the opening calculation module. It also obtains the actual valve opening adjusted by the actuator and real-time fluid condition feedback data, including flow and pressure. To accurately measure the prediction error, the unit calculates the opening deviation, flow error, and pressure error. The opening deviation is determined by comparing the numerical difference between the target opening and the actual executed opening and is used to assess the consistency between the predicted and executed values. The flow error and pressure error are calculated based on the deviation between the data collected by the real-time sensor and the predicted value calculated by the opening calculation module. Specifically, the difference between the measured flow rate and the predicted flow rate per unit time is calculated, and the deviation between the measured pressure and the predicted pressure is also calculated. To improve the integrity of the error assessment, an error index is comprehensively calculated, combining the opening deviation, flow error, and pressure error in a weighted manner. This allows the influence of different error factors on the overall correction to be dynamically adjusted based on the actual operating conditions. This error index serves as input for subsequent optimization.

[0076] After the error calculation is completed, the weight optimization unit adaptively adjusts the weights of each calculation method in the opening calculation module based on the size and distribution of the error index. When the flow error and pressure error are large, it means that the fluid operating conditions may undergo sudden or nonlinear changes. At this time, it is necessary to enhance the influence of the real-time fluid simulation method and increase the weight of the simulation model so that the calculation results are more dependent on the physical constraints of the current operating conditions to improve prediction accuracy and avoid error accumulation caused by changes in operating conditions. On the contrary, when the main error comes from the opening deviation and the flow and pressure errors are small, it means that there is great instability in the short-term trend prediction. At this time, the weight of the historical data regression analysis is increased, making the prediction more dependent on the past operating condition change trend to optimize short-term prediction accuracy. This weight optimization process is dynamic. The system will recalculate the error index in each calculation cycle and adjust the weight distribution according to the distribution of the error, so as to ensure that the prediction method can be adaptively optimized as the fluid operating conditions change and achieve gradual convergence of the prediction error.

[0077] After weight optimization is complete, the dynamic compensation unit adjusts the initial target opening based on the optimized weights to generate the final valve opening prediction. This correction process is based on the changing trend of the error index. When the error index exceeds the set threshold for multiple consecutive calculation cycles, it indicates that the current prediction model is unable to effectively adapt to changing operating conditions. In this case, an adaptive step-size adjustment strategy is implemented to increase the correction amplitude, accelerating the target opening adjustment and converging to a reasonable range as quickly as possible. This prevents the system from being in a state of long-term deviation, which could affect fluid control accuracy. Conversely, when the error index stabilizes, indicating that the prediction model has largely adapted to the current operating conditions, the adjustment amplitude is gradually narrowed, resulting in smoother changes in the opening correction, reducing system oscillation caused by frequent adjustments and improving valve control stability. Finally, the unit outputs the optimized target opening and transmits it to the valve actuator for precise adjustment of the reverse flow valve. This ensures that the valve opening adjustment process not only considers historical trends but also dynamically adapts to real-time operating conditions, thereby improving the overall control accuracy and stability of the system.

[0078] Step S103: using a fuzzy rule base based on the relationship between flow, pressure and valve opening, the initial target valve opening is adjusted in real time to obtain a final target valve opening.

[0079] In step S103, the initial target valve opening calculated in step S102 requires further real-time adjustment to accommodate dynamic changes in complex operating conditions and improve the accuracy and stability of valve regulation. This adjustment process relies on a fuzzy rule base, established based on the relationship between flow, pressure, and valve opening. This rule base comprehensively considers multiple factors and performs adaptive optimization adjustments. Fuzzy control methods are particularly effective in handling nonlinear and highly uncertain systems and are particularly suitable for reverse flow valve control scenarios where fluid conditions fluctuate significantly.

[0080] First, based on the real-time data input from the sensors, the collected fluid flow, pressure, and current valve opening are used as input variables for the fuzzy control system. To improve the system's response speed and robustness, all input variables need to be normalized to meet the computational requirements of fuzzy control. For example, a standardized interval can be defined for each of the flow, pressure, and opening, and the raw data can be mapped to the range required by the fuzzy control rule base through linear transformations or empirical formulas. In addition, to prevent the influence of data jitter or short-term outliers on control decisions, sliding mean filtering, median filtering, or exponentially weighted filtering can be used to smooth the input data, thereby enhancing control stability.

[0081] After input variable processing is complete, fuzzy membership functions need to be defined to represent the semantic relationships between different fluid parameters in the control system. For example, flow can be categorized into three levels: "low," "medium," and "high," pressure into "low pressure," "normal," and "high pressure," and valve opening into "closed," "small opening," "medium opening," and "large opening." The fuzzy membership of each input variable is determined by the specified membership function. Common membership function types include triangular, Gaussian, and trapezoidal functions. The specific selection can be optimized based on empirical data or experimental testing.

[0082] When building a fuzzy rule base, it's necessary to combine fluid dynamics characteristics with practical application experience to define a set of control rules. For example, when flow is high and pressure is within a normal range, the system may need to increase the valve opening to reduce flow resistance, while when flow is low and pressure is high, the opening may need to be reduced to prevent fluid backflow. The rule base can be constructed using an expert system approach, where experienced engineers set rules based on experimental data and practical experience, or using a data-driven approach, where a fuzzy inference system is trained using historical operating data to automatically optimize the fuzzy rule set.

[0083] During the fuzzy inference process, fuzzy logic operations are used to perform fuzzy inference calculations on the input variables to obtain the adjusted opening adjustment amount. Common inference methods include Mamdani-type reasoning and Sugeno-type reasoning. Mamdani-type reasoning obtains output results through fuzzy set operations and is more suitable for control systems that require strong interpretability, while Sugeno-type reasoning is suitable for scenarios with large computational load and high real-time requirements. The calculated opening adjustment amount is usually a fuzzy value and needs to be defuzzified to obtain the specific final target valve opening. Defuzzification methods include the maximum membership method, the weighted average method, and the central average method. Among them, the weighted average method is more commonly used and can comprehensively consider the influence of all fuzzy outputs, making the final opening adjustment smoother and more stable.

[0084] Once the final target valve opening is calculated, constraints must be placed on the adjustment results to ensure control system stability. For example, if the adjusted valve opening exceeds a preset safety range, a gradual adjustment approach can be employed to limit the maximum step size of a single adjustment to prevent system oscillation or overshoot. Furthermore, an adaptive adjustment mechanism can be introduced to dynamically optimize the fuzzy rule base based on historical control results to improve control accuracy and response speed.

[0085] The specific implementation steps of step S103 are described below by way of example:

[0086] In an industrial fluid pipeline system, the reverse flow valve is responsible for controlling the flow and preventing backflow. The system's goal is to maintain the outlet flow at 500 L / min while ensuring that the downstream pressure is stable at 0.8 MPa. To achieve this goal, the sensor collects the flow, pressure, temperature, and flow direction data of the fluid in real time, and calculates the initial target valve opening through the previous step. For example, under the current fluid working conditions, the prediction model calculates the initial target opening to be 60%. However, since the fluid system may be affected by factors such as upstream flow fluctuations, pipeline pressure changes, or temperature affecting viscosity, directly executing according to the initial target opening may cause flow or pressure overshoot, so further real-time adjustments are required to ensure optimal control of the valve.

[0087] When the system detects that the outlet flow rate is 480 L / min, lower than the target value, and the downstream pressure is 0.85 MPa, higher than the set value of 0.8 MPa, the valve opening needs to be corrected. According to the control strategy, when the flow rate is low but the pressure is high, it may mean that the current valve opening is too small and the opening needs to be increased to increase the flow rate. At the same time, fluctuations caused by excessive pressure should be avoided, so the adjustment range should not be too large. The fuzzy rule base will provide adjustment suggestions based on this situation. For example, under the current conditions, a moderate increase in the opening can improve the problem of insufficient flow without causing a sharp increase in downstream pressure. After calculation, the final decision was to increase the valve opening from 60% to 63%. After this adjustment is completed, the system continuously monitors changes in flow and pressure. If they still deviate from the target value, it will continue to optimize during the next adjustment so that the system ultimately stabilizes within the target flow and pressure range.

[0088] Throughout the adjustment process, the control system continuously calculates errors based on real-time sensor data and dynamically optimizes the valve opening using fuzzy rules, gradually aligning it with the optimal value. As flow and pressure change, the system continuously adjusts to ensure stable and precise control of the fluid delivery process, avoiding oscillations or delays caused by over-adjustment. In this way, the reverse flow valve can adapt to varying fluid conditions and automatically adjust to changes in the external environment, improving adjustment accuracy and ensuring reliable system operation.

[0089] Furthermore, the fuzzy rule base based on the relationship between flow, pressure and valve opening is used to adjust the initial target valve opening in real time to obtain the final target valve opening, including:

[0090] Build a fuzzy rule base, establish the corresponding relationship between flow, pressure and valve opening based on historical operating data, and set fuzzy input variables, including flow deviation, pressure deviation and opening change rate, to define control rules under different operating conditions;

[0091] Calculate the membership degree based on the current working condition data, perform fuzzy processing on the input variables through fuzzy reasoning method, and calculate the fuzzy membership degree of flow, pressure and opening to determine the matching degree of the input variables in the fuzzy rule base and calculate the fuzzy control output;

[0092] Perform real-time adjustments, calculate the adjusted target valve opening through the defuzzification method, and optimize the adjustment range in combination with the valve response characteristics. When the fluid operating conditions fluctuate slightly, a smooth adjustment strategy is adopted to ensure stability. When the operating conditions change drastically, adaptive gain adjustment is used to enhance the adjustment accuracy. Finally, the optimized final target valve opening is output to ensure that the valve adjustment effect is accurately matched with the fluid operating conditions.

[0093] To achieve the final target valve opening by adjusting the initial target valve opening in real time using a fuzzy rule base based on the relationship between flow, pressure, and valve opening, a comprehensive fuzzy control system must first be established to adapt to the dynamic changes in different fluid operating conditions. The core of this system is the fuzzy rule base, constructed based on historical operating and experimental data. By long-term monitoring of the valve opening response characteristics of the reverse flow valve under different flow, pressure, and temperature conditions, the nonlinear relationship between fluid characteristic changes and valve adjustment is analyzed. Based on this, fuzzy input variables are defined, including flow deviation, pressure deviation, and valve opening change rate. These input variables are divided into different fuzzy sets, such as "low," "medium," and "high," or "negative large," "negative small," "zero," "positive small," and "positive large," to more accurately describe the mapping between fluid operating conditions and valve adjustment. Subsequently, based on experimental analysis and empirical rules, fuzzy control rules linking the input variables to valve opening adjustment are developed to ensure that the control system can adapt to the dynamic changes in fluid operating conditions.

[0094] During the real-time adjustment process, the membership of the current operating condition data within the fuzzy rule base must first be calculated. This involves fuzzifying the flow deviation, pressure deviation, and valve opening change rate so that their corresponding membership values within the fuzzy set can be used for fuzzy inference calculations. Flow deviation is calculated by measuring the difference between the current flow rate and the target flow rate. If the deviation is large, the system should make corresponding adjustments to reduce the error between the valve opening and the target opening. Pressure deviation is calculated similarly, measuring the difference between the current pressure value and the predicted pressure value to determine whether the current operating condition is stable. The valve opening change rate is calculated based on the magnitude of the change in the current valve opening compared to the previous time step. If the rate of change is too rapid, the adjustment may need to be reduced to prevent system oscillation. However, if the rate of change is slow or the system response is sluggish, the adjustment may need to be increased to accelerate convergence. After the fuzzified results of these input variables are input into the fuzzy inference system, the system performs inference calculations based on the fuzzy rule base, comprehensively calculating the influence weights of the different input variables and outputting the fuzzy control adjustment amount, resulting in the fuzzy control output.

[0095] After obtaining the fuzzy control output, real-time adjustments are required to achieve the final target valve opening. Defuzzification is used to convert the fuzzy control output into a specific numerical adjustment value, and the adjustment strategy is optimized based on the valve's dynamic response characteristics. A weighted average defuzzification method can be used. This method calculates the target adjustment value based on the membership of different fuzzy rules to ensure that the adjusted opening changes meet the required fluid conditions. When performing adjustments, it is necessary to consider the optimal adjustment strategy for different operating conditions. For example, when the system detects minimal fluid fluctuations, a smoothing adjustment strategy is used to maintain valve stability, making the valve adjustment process more gradual and reducing unnecessary fluctuations. However, when the system detects drastic changes in fluid conditions, such as a sharp increase or decrease in flow or pressure within a short period of time, an adaptive gain adjustment strategy is used to dynamically adjust the valve's response rate to ensure that the control system can quickly adapt to sudden changes and improve control accuracy. In addition, during the real-time adjustment process, the control signal of the valve actuator also needs to be dynamically optimized to ensure the smoothness of the valve adjustment action, prevent overshoot or system oscillation, and ultimately achieve precise valve opening control, so that the valve adjustment effect is accurately matched with the fluid working conditions, and improve the overall stability and response speed of the system.

[0096] Step S104: controlling the actuator to adjust the operating state of the reverse flow valve according to the final target valve opening.

[0097] In step S104, the actuator needs to be precisely controlled based on the final target valve opening to adjust the operating state of the reverse flow valve so as to achieve the desired fluid regulation effect. The actuator serves as a driving device for the valve, and its type may include electric actuators, pneumatic actuators or hydraulic actuators. The specific selection depends on the application scenario and control requirements. For electric actuators, a stepper motor or servo motor is generally used to adjust the angle or linear displacement of the valve by receiving a control signal. Pneumatic actuators usually drive the piston through compressed air to achieve the opening and closing or precise adjustment of the valve, while hydraulic actuators use liquid pressure to control the valve opening, which is suitable for high-pressure working conditions or fine adjustment of large-diameter valves.

[0098] When controlling an actuator, the final target valve opening must first be converted into a control signal suitable for the actuator. For electric actuators, the control signal is typically a pulse signal or an analog voltage signal. The pulse signal is used to drive a stepper motor or servo motor, while the analog voltage signal is used to control the linear change of the valve opening. For pneumatic actuators, the control signal is generally adjusted using an electric proportional valve to control the air pressure and achieve precise valve opening control. Hydraulic actuators typically use proportional solenoid valves or servo valves to adjust the hydraulic pressure, thereby precisely controlling the valve opening.

[0099] When executing control signals, the actuator's response characteristics and the valve's inertia must be considered to avoid oscillation or overshoot during adjustment. For example, the stepper motor in an electric actuator may be affected by load fluctuations during operation, necessitating the use of acceleration and deceleration strategies during control to smoothly adjust the opening. For pneumatic actuators, hysteresis may occur during rapid response due to the compressibility of compressed air. Therefore, feedback control or pressure compensation mechanisms can be introduced to improve control accuracy. For hydraulic actuators, hydraulic shock must be avoided to prevent severe valve vibration during high-speed adjustment. Buffer circuits or flow control valves can be used to optimize hydraulic flow.

[0100] During valve adjustment, feedback control is a crucial tool for ensuring precision and stability. Actuators are typically equipped with position sensors or opening feedback devices, such as potentiometers, optical encoders, or LVDT displacement sensors, to monitor the actual valve opening in real time. The measured opening is compared with the target opening. If a deviation exists, closed-loop control adjusts the actuator output to gradually approach the target opening. PID control can be used as a closed-loop control strategy, and PID parameter optimization can be combined with adaptive adjustment algorithms to adapt to varying operating conditions, ensuring precise valve control in dynamic environments.

[0101] When actuators adjust valve openings, the impact of external environmental factors on control effectiveness must also be considered. For example, in high or low temperature environments, the mechanical components of the actuator may expand and contract due to heat, affecting control accuracy. In pneumatic systems, high humidity may cause condensation to accumulate, affecting movement sensitivity. In hydraulic systems, temperature fluctuations may cause changes in oil viscosity, affecting response speed. Therefore, in practical applications, environmental sensors can be used to monitor temperature and humidity, and compensation mechanisms can be incorporated into the control algorithm to optimize control effectiveness.

[0102] To further enhance valve regulation reliability, fault detection and protection mechanisms can be introduced. For example, upon detecting an actuator anomaly, such as a motor stall, insufficient air pressure, or a hydraulic system leak, the system can automatically adjust its control strategy and enter a safe mode to prevent valve malfunction or fluid system abnormalities caused by control failure. Furthermore, redundant control units or dual sensors can be integrated into the actuator to provide backup control paths, enhancing overall system reliability.

[0103] During the entire control process, by converting the final target valve opening into the control signal of the actuator, combined with closed-loop feedback control and fault detection protection mechanism, precise adjustment of the reverse flow valve can be achieved, ensuring the stable operation of the fluid system and meeting the adjustment requirements under different working conditions.

[0104] Furthermore, controlling the actuator to adjust the operating state of the reverse flow valve according to the final target valve opening includes:

[0105] Analyze the final target valve opening and calculate the actuator control signal based on the target opening value and the current valve state, including the valve opening change, execution time and execution rate, to ensure that the control signal meets the physical constraints of the valve and is optimally adjusted within the control range;

[0106] Dynamically compensate for execution errors. Based on real-time feedback of valve opening position, actuator operation response, and fluid working condition changes, the actual execution error is calculated. When the error exceeds the set threshold, the execution signal is adjusted, including compensating for execution time deviation and optimizing execution torque to ensure the accuracy of valve opening adjustment.

[0107] Optimize the control response strategy, combine the historical data of valve opening adjustment, and adaptively adjust the amplitude and response time of the control signal. When the fluid working condition is stable, adopt a smooth adjustment strategy to reduce the system impact. When the working condition fluctuates greatly, improve the execution response rate, enhance the system's rapid adjustment capability, and ensure the operating stability and adjustment accuracy of the reverse flow valve under different working conditions.

[0108] To control the actuator's adjustment of the reverse flow valve's operating state based on the target valve opening, the target valve opening must first be determined to generate a control signal suitable for the actuator. The control signal calculation must not only consider the target opening value but also the actual current valve state to ensure that the adjustment process complies with the valve's physical constraints, avoiding exceeding the valve's mechanical limits or causing control instability. Key control signal parameters include valve opening change, execution time, and execution rate. The valve opening change determines the current opening adjustment range, the execution time controls the actuator's duration, and the execution rate influences the speed of the valve's response. When the difference between the target opening and the current opening is small, a small-step, low-rate adjustment strategy should be adopted to gradually move the valve toward the target opening to maintain smooth fluid system operation. However, when the difference between the target opening and the current opening is large, the execution rate should be increased, taking into account operating conditions while maintaining system stability, to ensure that the fluid operating conditions adapt promptly to the adjusted valve state.

[0109] During valve actuation, execution errors may occur due to the nonlinear characteristics of the actuator and the dynamic changes in the fluid environment. This means that there is a deviation between the actual adjusted valve opening and the desired opening. Therefore, it is necessary to calculate and compensate for the execution error based on real-time feedback of the valve opening position, the actuator's operational response, and changes in fluid operating conditions. When the execution error is detected to exceed the set threshold, the system needs to adjust the control signal to compensate for the execution time deviation and optimize the execution torque so that the valve can more accurately reach the target opening. For example, if the actuator's opening adjustment lags due to inertia, the execution time can be increased or the signal amplitude can be adjusted to make the actuator more responsive. If the valve adjustment amplitude is too large due to fluid impact, the gain of the execution signal can be reduced and the execution torque can be optimized to prevent overshoot and ensure the accuracy of valve opening adjustment.

[0110] When controlling the actuator to adjust the operating state of a reverse flow valve, the control response strategy must be optimized to ensure smooth and efficient valve adjustment. The system integrates historical data on valve opening adjustments to adaptively adjust the control signal amplitude and response time based on different operating conditions. When fluid conditions are relatively stable, a smooth adjustment strategy should be employed to ensure the valve adjustment process is as gradual as possible, minimizing the impact of sudden changes on the system. When operating conditions fluctuate dramatically, the actuator response rate should be increased to enable the valve to quickly adapt to the new fluid conditions. For example, when flow or pressure changes suddenly, the system should accelerate valve adjustment to quickly restore fluid stability. When flow and pressure fluctuations are minimal, the adjustment rate can be appropriately reduced to reduce actuator load and extend the life of the equipment. Ultimately, by optimizing the control response strategy, the reverse flow valve can maintain stable operation under varying fluid conditions and possess strong adaptability, thereby improving overall adjustment accuracy and ensuring the operational stability of the fluid system.

[0111] Furthermore, the intelligent adjustment method for the reverse flow valve based on the adaptive control algorithm further includes:

[0112] Within a preset time period after the valve opening is adjusted, the flow, pressure, and temperature data at the inlet and outlet of the counterflow valve are continuously collected, and the actual operating condition changes are compared with the predicted operating condition data in real time to calculate the operating condition error. Based on the operating condition error, the actual control effect of the real-time adjustment of the valve opening is determined, and the rule entries corresponding to the current operating condition in the fuzzy rule library and their adjustment strategies are identified as to whether they are reasonable.

[0113] When the operating condition error exceeds the preset threshold, the corresponding control rules in the fuzzy rule base are modified to optimize the subsequent real-time adjustment strategy of the valve opening.

[0114] In this embodiment, after implementing the real-time adjustment of the reverse flow valve opening, it is necessary to further conduct refined tracking evaluation and dynamic optimization of the control effect to ensure long-term high-precision control performance. Specifically, after the actuator completes the real-time adjustment of the valve opening based on the final target valve opening, it enters a continuous data monitoring phase, which has a pre-set fixed time period, such as 5 seconds, 10 seconds or other cycle lengths suitable for specific application scenarios. During this preset time period, the system needs to continuously collect the actual flow, pressure and temperature data of the current fluid in real time through sensors installed at the inlet and outlet of the reverse flow valve, and record the collected data in a real-time data storage unit to ensure the continuity and integrity of data collection.

[0115] After collecting the real-time operating condition data, the system then performs a point-by-point, real-time comparison with the predicted operating condition data calculated by the aforementioned operating condition opening prediction model. Specifically, the system performs a differential calculation between the recorded actual flow, pressure, and temperature data and the initial predicted operating condition data (or the ideal operating condition data predicted by the previous adjustment). For example, the system calculates the magnitude of the flow error, pressure error, and temperature error. The magnitude of the error is used to assess the degree of agreement between the prediction and the actual condition, thereby determining the real-time error parameters for the current control condition. The specific calculation of the operating condition error can be performed using relative error, absolute error, or root mean square error (RMSE). For example, flow error can be expressed as (predicted flow - actual flow) / predicted flow × 100%. These error values provide an intuitive assessment of the effectiveness of the current real-time adjustment of the valve opening in improving the actual operating condition. Smaller errors indicate better control effectiveness.

[0116] After obtaining the above-mentioned operating error values, the system further evaluates the control effect, that is, through a pre-set evaluation index system, it clarifies the actual control effect of the real-time adjustment of the valve opening. The control effect evaluation index may include multi-dimensional parameters such as flow stability, pressure fluctuation amplitude, and system response speed. By comparing with the predicted data, the system can clarify whether the rule entries and adjustment strategies corresponding to this operating condition adjustment in the currently used fuzzy rule library are reasonable and effective. For example, if the actual measured flow deviates significantly from the target value and the pressure fluctuation increases significantly after the valve opening of a certain rule is increased, it can be inferred that the adjustment strategy corresponding to the rule may be defective or need to be corrected; on the contrary, if the flow and pressure quickly approach the target value and remain stable, the rule entry can remain unchanged and continue to be applied.

[0117] When the evaluation results of the above-mentioned working condition error show that the error exceeds the preset threshold set by the system (for example, the relative error exceeds the set 5% or other specifically set threshold), it indicates that the relevant rule entries in the fuzzy rule base can no longer meet the current working condition requirements, and there may be problems such as the adjustment range being too large or too small. In this case, the system automatically enters the rule base correction program, and makes appropriate corrections or optimization adjustments to the fuzzy sets, membership function shapes, rule inference weights, or adjustment range parameters of the rule output of the corresponding rule entries in the rule base to improve the adaptability of the rule base to the actual working conditions. The specific implementation of the rule base correction can adopt the method of adding, deleting, or adjusting the weight of fuzzy control rules. For example, for rules that lead to excessive opening adjustments, the rule output can be reduced; for rules with insignificant adjustment effects, the output can be appropriately increased or the coverage of the membership function can be adjusted to improve the overall adaptability and control accuracy of the system.

[0118] After the rule base is modified, the system automatically updates and reloads the modified fuzzy rule base for the next round of real-time adjustment of the valve opening, thereby continuously iterating and optimizing to achieve dynamic, long-term, stable, and precise control of the reverse flow valve's operating state. Through the closed-loop feedback mechanism of operating condition tracking, real-time error assessment, and dynamic rule modification, the present method continuously improves valve adjustment accuracy and adaptability, ultimately achieving more precise and reliable fluid control.

[0119] Furthermore, when the operating condition error exceeds a preset threshold, the corresponding control rule of the fuzzy rule base is modified to optimize the subsequent real-time adjustment strategy of the valve opening, including:

[0120] Based on historical operating condition data, a model for identifying operating condition change trends is established to calculate the correlation between operating condition change trends and rule adjustment effectiveness within a certain time window. A rule adjustment factor is dynamically generated based on this correlation. The rule adjustment factor is determined by the following formula 1:

[0121] ;

[0122] in, is the rule adjustment factor; The total number of data points for historical operating data analysis; Respectively represent the history The difference between the flow rate, pressure, and temperature of each data point and their respective target values; For the history The difference between the actual adjustment amount of the valve opening and the predicted adjustment amount corresponding to the data point; Respectively represent the weighted coefficients of flow, pressure, and temperature for the overall working condition, satisfying ; is a very small positive constant to prevent the denominator from being zero;

[0123] When the rule adjustment factor If the value is greater than the preset threshold, it indicates that the current fuzzy rule has obvious deviations in actual control. The output amplitude or rule confidence of the corresponding rule in the fuzzy rule base is updated to make the next valve opening adjustment closer to the actual working conditions.

[0124] In this embodiment, the control rules of the fuzzy rule base need to be optimized to adapt to the changes in real-time working conditions and improve the adjustment accuracy. Specifically, when the calculated working condition error exceeds the preset threshold, it is necessary to first build a working condition change trend identification model based on historical working condition data, and use mathematical analysis methods to calculate the correlation between the working condition change trend and the effectiveness of the rule adjustment within a certain time window. This correlation is used to generate the rule adjustment factor , thereby deciding whether the fuzzy rule base needs to be modified. The rule adjustment factor is calculated using formula (1):

[0125] ;

[0126] The formula is used to measure the influence of various working parameters in historical data on the door opening adjustment, and the calculated Evaluate the effectiveness of the current fuzzy rule base and decide whether to modify it.

[0127] In formula (1), the rule adjustment factor It reflects whether the trend of valve opening adjustment in historical data points is consistent with the changes in operating parameters. Significant deviation from 1 (e.g. If the value is greater than the preset threshold 2), it means that the valve opening adjustment strategy may not match the actual working conditions. Therefore, the rule base needs to be adjusted to optimize the subsequent control strategy.

[0128] parameter Represents the total number of data points for historical data analysis, usually set by the system as a fixed window, such as 100 or 500, to ensure the stability of statistical analysis. The choice of window size needs to balance real-time and data volume. Smaller window size is better. It can improve the response speed, but may cause greater fluctuations. This can increase robustness but may delay the triggering of adjustments.

[0129] variable Respectively represent The error change of flow, pressure and temperature at each historical data point is calculated as follows:

[0130] ;

[0131] in, Respectively The actual flow, pressure and temperature collected from each historical data point, is the target value for this operating condition and can be set based on empirical data or specific reverse flow valve adjustment requirements. The above calculations yield the deviation of each data point from the target value. The magnitude and direction of the deviation are used to measure the degree of deviation in the system operating state.

[0132] variable Representative The valve opening adjustment error for each data point is calculated as follows:

[0133] ;

[0134] in, is the actual valve opening of the data point, is the target opening given by the prediction model under this working condition. If it is too large, it means that the adjustment strategy of the control actuator fails to accurately match the system requirements, and the control rules may need to be optimized.

[0135] variable is the weighted coefficient of the importance of the working condition parameters, which is used to measure the influence of flow, pressure and temperature on the valve opening adjustment under the current working conditions. Their sum always satisfies:

[0136] ;

[0137] The specific value of the weighting coefficient can be set based on experience. For example, in a system where flow control is the main goal, It can be set to 0.5, while the weights of pressure and temperature can be set to 0.3 and 0.2 respectively. If the system is mainly affected by pressure fluctuations, you can increase value.

[0138] The sum of squares term in the denominator:

[0139] ;

[0140] Used to normalize the error margin to prevent a certain variable from having an extreme impact on the calculation result. Similarly, another square root term:

[0141] ;

[0142] The calculation denominator used to prevent valve adjustment errors is zero, where is a very small positive number (e.g. ) to avoid division by zero errors in mathematical calculations.

[0143] when The calculated value of is significantly greater than 1 (for example If the value is greater than the preset threshold 2), the adjusted valve opening deviates significantly from the target operating condition, and the control strategy fails to accurately match actual requirements. Therefore, the fuzzy rule base needs to be adjusted. This adjustment can be done by increasing or decreasing the output amplitude of the rule or adjusting the confidence level of the rule. For example, if adjusting a rule causes flow overshoot, the output amplitude of the rule can be reduced to reduce the aggressiveness of the opening adjustment. If adjusting a rule is insufficient to correct the deviation, the output weight of the rule can be increased to give it a greater role in the control strategy.

[0144] During the next operating condition adjustment, the system uses the updated fuzzy rule base for control, so that the valve opening adjustment is more in line with the actual needs of the fluid working conditions, thereby optimizing the regulation accuracy of flow, pressure and temperature, and improving the long-term stability of the system.

[0145] A second embodiment of the present application provides an electronic device, comprising:

[0146] processor;

[0147] The memory is used to store a program, which, when read and executed by the processor, executes an intelligent adjustment method for a reverse flow valve based on an adaptive control algorithm provided in the first embodiment of the present application.

[0148] The third embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the program executes an intelligent adjustment method for a reverse flow valve based on an adaptive control algorithm provided in the first embodiment of the present application.

[0149] Although the present application is disclosed as above with the preferred embodiments, it is not intended to limit the present application. Any person skilled in the art may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A reverse flow valve intelligent adjustment method based on an adaptive control algorithm, characterized in that: include: Obtaining fluid flow, pressure, temperature and flow direction data collected by sensors installed at the inlet and outlet of the reverse flow valve; Based on the acquired fluid data and the structural parameters of the reverse flow valve, the initial target valve opening under the current fluid working condition is predicted using the working condition opening prediction model; Using a fuzzy rule base based on the relationship between flow, pressure and valve opening, the initial target valve opening is adjusted in real time to obtain the final target valve opening; controlling the actuator to adjust the operating state of the reverse flow valve according to the final target valve opening; Within the preset time period after the valve opening is adjusted, the flow, pressure and temperature data at the inlet and outlet of the counter-flow valve are continuously collected, and the actual working condition changes are compared with the predicted working condition data in real time to calculate the working condition error; According to the working condition error, the actual control effect of the real-time adjustment of the valve opening is determined, and the rule entries corresponding to the current working condition in the fuzzy rule base and their adjustment strategies are identified as to whether they are reasonable; When the operating error exceeds the preset threshold, the corresponding control rules in the fuzzy rule base are modified to optimize the subsequent real-time adjustment strategy of the valve opening; When the operating condition error exceeds a preset threshold, the corresponding control rule of the fuzzy rule base is modified to optimize the subsequent real-time adjustment strategy of the valve opening, including: Based on historical operating condition data, a model for identifying operating condition change trends is established to calculate the correlation between operating condition change trends and rule adjustment effectiveness within a certain time window. A rule adjustment factor is dynamically generated based on this correlation. The rule adjustment factor is determined by the following formula 1: ； in, is the rule adjustment factor; The total number of data points for historical operating data analysis; Respectively represent the history The difference between the flow rate, pressure, and temperature of each data point and their respective target values; For the history The difference between the actual adjustment amount of the valve opening and the predicted adjustment amount corresponding to the data point; Respectively represent the weighted coefficients of flow, pressure, and temperature for the overall working condition, satisfying ; is a very small positive constant to prevent the denominator from being zero; When the rule adjustment factor If the value is greater than the preset threshold, it indicates that the current fuzzy rule has obvious deviations in actual control. The output amplitude or rule confidence of the corresponding rule in the fuzzy rule base is updated to make the next valve opening adjustment closer to the actual working conditions.

2. The intelligent adjustment method for reverse flow valve based on adaptive control algorithm according to claim 1 is characterized in that: The method of obtaining the fluid flow rate, pressure, temperature and flow direction data collected by sensors provided at the inlet and outlet of the reverse flow valve includes: Dynamically adjust the sensor's data acquisition frequency according to the rate of change of flow and pressure. When the rate of change exceeds the set threshold, the acquisition frequency is increased; when the rate of change is lower than the set threshold, the acquisition frequency is reduced. Store a predetermined amount of historical flow, pressure, temperature and flow direction data, and compare it with the historical data every time new data is collected. When the new data deviates from the set range of historical data, it will be marked as abnormal data and the data will be corrected; Based on the structural parameters and historical data of the reverse flow valve, the flow direction data is compensated. When the flow direction data changes suddenly and the valve opening changes significantly, the compensation factor is calculated and the current flow direction data is adjusted.

3. The intelligent adjustment method for reverse flow valve based on adaptive control algorithm according to claim 2 is characterized in that: The method stores a predetermined amount of historical flow, pressure, temperature and flow direction data, and compares the data with the historical data each time new data is collected. When the new data deviates from the set range of the historical data, it is marked as abnormal data and the data is corrected, including: The data storage window is preset. During each data collection, the latest flow, pressure, temperature and flow direction data are stored in the window. The window length is set according to time or data capacity so that the data in the window can reflect the current working condition trend. Calculate data distribution characteristics based on historical data within the storage window, including data mean, variance, and change trend. When the deviation of newly collected data values from the mean of historical data exceeds a set threshold, or the data change trend is significantly inconsistent with the historical trend, the data is marked as abnormal data. Correct the data marked as abnormal. When the abnormal data is in a short-term mutation state and no continuous abnormalities occur, use interpolation or historical mean to correct it. When the abnormal data shows a continuous deviation trend, determine whether it is caused by sensor drift or environmental influence, and compensate based on the related data of multiple sensors.

4. The intelligent adjustment method for reverse flow valve based on adaptive control algorithm according to claim 1, characterized in that: The operating condition opening prediction model includes a data preprocessing module, a feature extraction module, an opening calculation module and an adaptive correction module; The data preprocessing module is used to receive flow rate, pressure, temperature and flow direction data, and normalize the data in combination with the structural parameters of the reverse flow valve, remove outliers, smooth the data through a filtering algorithm, and output optimized working condition data; The feature extraction module receives the output of the data preprocessing module and calculates the flow-pressure change rate ratio, the fluid pressure drop gradient and the flow direction stability to construct a feature vector reflecting the current working condition, and outputs the feature vector to the opening calculation module; The opening calculation module calculates a preliminary target opening based on the output of the feature extraction module using historical data regression analysis and real-time fluid simulation methods, wherein the historical data regression analysis is used for short-term trend prediction, and the real-time fluid simulation is combined with the reverse flow valve structural parameters to correct the predicted opening, and the calculated preliminary target opening is output to the adaptive correction module; The adaptive correction module receives the output of the opening calculation module and calculates the prediction error based on real-time feedback data. When the error exceeds a set threshold, the parameters of the opening calculation module are adjusted and the preliminary target opening is corrected to obtain the final predicted initial target valve opening.

5. The intelligent adjustment method for reverse flow valve based on adaptive control algorithm according to claim 4 is characterized in that: The feature extraction module includes a data conversion unit, a feature calculation unit and a dynamic feature optimization unit; The data conversion unit is used to receive the output of the data preprocessing module and perform nonlinear transformation on the normalized flow, pressure, temperature and flow direction data to enhance the data differentiation. For flow and pressure data, a piecewise logarithmic mapping function is used to differentially scale data in different ranges so that it has higher resolution in the low value range. For temperature data, an exponential smoothing method is used to reduce the impact of transient changes, thereby obtaining converted data for feature calculation; The feature calculation unit receives the output of the data conversion unit and calculates key features of the working condition, including the flow-pressure change rate ratio, the fluid pressure drop gradient and the flow direction stability, wherein the flow-pressure change rate ratio is obtained by calculating the ratio of the flow rate and the pressure change rate in continuous time steps to measure the dynamic response characteristics of the fluid working condition, the fluid pressure drop gradient is calculated based on the flow rate and the pipeline resistance parameter to quantify the influence of the valve opening on the fluid pressure, and the flow direction stability is obtained by calculating the variance of the flow direction data in multiple time steps to evaluate the stability of the fluid flow, and the calculated characteristic parameters are combined into a characteristic vector; The dynamic feature optimization unit is used to receive the output of the feature calculation unit and adaptively adjust the feature weight according to the historical operating data. Specifically, based on the historical operating data of the reverse flow valve, the weighted sliding average method is used to calculate the long-term change trend of each feature, and the weight of each feature in the feature vector is adjusted according to the feature change trend to enhance the characterization ability of key influencing factors, so that the generated feature vector is more adapted to the current operating conditions, and the optimized feature vector is output to the opening calculation module.

6. The intelligent adjustment method for reverse flow valve based on adaptive control algorithm according to claim 4 is characterized in that: The opening calculation module includes a short-term trend prediction unit, a real-time fluid simulation unit and an opening correction unit; The short-term trend prediction unit is used to analyze the changing trends of historical flow, pressure, temperature and flow direction data based on the output of the feature extraction module using a multi-order regression model, and calculate the target opening within a short time scale; by constructing a recursive linear regression or exponential smoothing regression method, using the data of multiple time steps in the past to fit the trend curve, the optimal opening at the current moment is predicted; The real-time fluid simulation unit is used to receive the target opening calculated by the short-term trend prediction unit, and, in combination with the structural parameters of the reverse flow valve and real-time operating data, to correct the target opening through numerical simulation calculation methods. A numerical solution method based on the finite volume method or the finite difference method is used to simulate the fluid flow characteristics under different openings, and to calculate key parameters such as fluid flow rate, pressure drop, and turbulence intensity to determine whether the initially predicted opening meets the current operating conditions. If the simulation results indicate that the predicted target opening may result in flow overshoot or excessive pressure drop, the opening value is adjusted to meet the physical constraints. The opening correction unit is used to receive the corrected target opening output by the real-time fluid simulation unit, and fine-tune the opening in combination with the stability factor of the current working condition to improve the adjustment accuracy; by calculating the stability index of the working condition characteristic vector, the change amplitude of the current fluid state is judged. When the working condition changes drastically, the method of decreasing the adjustment step size is adopted to avoid the impact of the predicted opening mutation on the system; when the working condition tends to be stable, the weighted smoothing adjustment strategy is adopted to make the final target opening more in line with the natural adjustment trend of the fluid; the output preliminary target opening is used for the subsequent adaptive correction module.

7. The intelligent adjustment method for reverse flow valve based on adaptive control algorithm according to claim 4 is characterized in that: The adaptive correction module includes an error calculation unit, a weight optimization unit and a dynamic compensation unit; The error calculation unit is used to receive the output of the opening calculation module, obtain the actual valve opening and fluid working condition feedback data, and calculate the prediction error; the error calculation includes opening deviation, flow error and pressure error, among which the opening deviation is calculated by comparing the target opening with the actual opening, and the flow error and pressure error are evaluated based on the deviation between the real-time collected data and the calculated value of the prediction model, and the error index is calculated by comprehensive weighting as the input for subsequent optimization; The weight optimization unit is used to adaptively adjust the weights of various calculation methods in the opening calculation module based on the error index; when the flow error and pressure error are large, the weight of the real-time fluid simulation method is increased to enhance its adaptability to the current working conditions; when the opening deviation is mainly due to the instability of the short-term trend forecast, the weight of the historical data regression analysis is increased to optimize the accuracy of the trend forecast; The dynamic compensation unit is used to correct the preliminary target opening according to the optimized weight and generate the final valve opening prediction value; when the error index continuously exceeds the set threshold, an adaptive step adjustment strategy is adopted to avoid large-scale adjustments that cause system oscillations; when the error index tends to stabilize, the adjustment amplitude is gradually converged.

8. The intelligent adjustment method for reverse flow valve based on adaptive control algorithm according to claim 1, characterized in that: The method adopts a fuzzy rule base based on the relationship between flow, pressure and valve opening to adjust the initial target valve opening in real time to obtain the final target valve opening, including: Build a fuzzy rule base, establish the corresponding relationship between flow, pressure and valve opening based on historical operating data, and set fuzzy input variables, including flow deviation, pressure deviation and opening change rate, to define control rules under different operating conditions; Calculate the membership degree based on the current working condition data, perform fuzzy processing on the input variables through fuzzy reasoning method, and calculate the fuzzy membership degree of flow, pressure and opening to determine the matching degree of the input variables in the fuzzy rule base and calculate the fuzzy control output; Perform real-time adjustments, calculate the adjusted target opening through the defuzzification method, and optimize the adjustment range based on the valve response characteristics. When the fluid operating conditions fluctuate slightly, a smooth adjustment strategy is adopted to ensure stability. When the operating conditions change drastically, adaptive gain adjustment is used to enhance the adjustment accuracy, and finally output the optimized final target valve opening.

9. The intelligent adjustment method for reverse flow valve based on adaptive control algorithm according to claim 1, characterized in that: The controlling the actuator to adjust the operating state of the reverse flow valve according to the final target valve opening includes: Analyze the final target valve opening and calculate the actuator control signal based on the target opening value and the current valve state, including the valve opening change, execution time and execution rate, to ensure that the control signal meets the physical constraints of the valve and is optimally adjusted within the control range; Dynamically compensate for execution errors. Based on real-time feedback of valve opening position, actuator operation response, and fluid working condition changes, the actual execution error is calculated. When the error exceeds the set threshold, the execution signal is adjusted, including compensating for execution time deviation and optimizing execution torque. Optimize the control response strategy, combine the historical data of valve opening adjustment, and adaptively adjust the amplitude and response time of the control signal. When the fluid working condition is stable, adopt a smooth adjustment strategy to reduce the system impact. When the working condition fluctuates greatly, increase the execution response rate and enhance the system's rapid adjustment capability.

Images