Remote gate opening control method and system for pump station

Abstract

The application relates to the field of hydraulic engineering and automatic control technology, and discloses a remote water gate opening degree control method and system for a pump station. The method comprises the following steps: obtaining flow rate data and operation parameters to obtain a preliminary resistance estimation; combining real-time working conditions to deduce diffusion trajectories to determine a potential deviation range; obtaining a current operation damping ratio to obtain a deviation vector and determine a jamming probability; extracting stress characteristics and force characteristics according to the jamming probability and the current operation damping to calculate a risk value and generate a correction coefficient; inputting environment and communication parameters and the correction coefficient into a preset mapping model to obtain a preliminary control signal; comparing the predicted water level and the predicted flow rate with preset flood control constraints respectively to obtain an operation log to obtain a refined resistance value; and matching and adjusting the opening degree signal and the current change amount according to the refined resistance value to generate an adjustment command. The method can accurately identify dynamic mechanical resistance and perform feedforward compensation to avoid jamming failure.

Description

Technical Field

[0001] This invention relates to the field of water conservancy engineering and automation control technology, and in particular to a remote sluice gate opening control method and system for pumping stations. Background Technology

[0002] Currently, pumping stations and sluice gates serve as core hubs for water resource allocation and flood control and drainage, and their operational stability and control precision directly impact the safety of the river basin's water network. With the continuous development of water conservancy informatization, remote automatic control of sluice gates has become a key means to improve pumping station management efficiency and reduce the risks of manual intervention.

[0003] In existing technologies, traditional SCADA systems are typically used for remote adjustment and status monitoring of sluice gate openings. These solutions primarily utilize pre-set conventional PID control logic or fixed opening-time execution curves in the lower-level PLC to directly drive the motor based on single position commands issued by the operator or simple water level thresholds, thereby completing the gate's raising and lowering actions. However, during actual long-term service, the external fluid environment of a sluice gate (such as sudden changes in water flow velocity, uneven siltation, and surges in head pressure) and the physical state of the equipment itself (such as nonlinear wear of guide rails and fatigue aging of transmission components) are highly dynamic and complex. Existing technologies, when executing control commands, rely solely on static control parameters or simple end-position feedback, treating the actuator as an ideal linear model, lacking the ability to deeply identify transient changes in the gate's mechanical resistance and a feedforward compensation mechanism. This causes unpredictable and severe fluctuations in the mechanical resistance of the gate when encountering sudden and complex water conditions or localized equipment degradation. The original fixed control logic cannot detect and adjust the dynamic output torque and control gain in time, resulting in a serious disconnect and delay between the issuance of instructions and the actual physical execution.

[0004] Therefore, existing technologies have the problem that sluice gates are prone to serious inaccuracies in opening control, delayed response, and even physical jamming failures under complex operating conditions. Summary of the Invention

[0005] This invention provides a remote sluice gate opening control method and system for pumping stations, in order to solve the problem that sluice gates in the prior art are prone to serious inaccurate opening control, delayed response, or even physical jamming failure under complex operating conditions.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a method for remote sluice gate opening control in a pumping station, comprising: Acquire flow velocity data and operating parameters, use a preset classifier to divide and calculate distances, and obtain a preliminary resistance estimate; The system acquires real-time operating conditions, constructs an initial state vector by combining the preliminary resistance estimate with the preset aging model, calculates the prediction residual using the Kalman filter algorithm and adjusts the feedback gain to obtain the control deviation, and if the control deviation exceeds the preset accuracy threshold, it uses the preset transition probability to deduce the diffusion trajectory and determine the potential deviation range. If the potential deviation range exceeds the preset deviation threshold, the current operating damping is obtained and the resistance record is retrieved from the preset historical database. The deviation vector is calculated by comparison, and the probability distribution of the deviation vector is deduced using the Weibull distribution to determine the jamming probability. Based on the jamming probability and the current operating damping, the force characteristics are extracted. The force characteristics are input into a preset fatigue model to extract stress characteristics. The cumulative loss increment is calculated based on the stress characteristics. The cumulative loss increment is input into a preset degradation function to obtain a risk value. Based on the risk value, gain reconstruction is performed to generate correction coefficients. Obtain environmental and communication parameters, input the environmental and communication parameters and the correction coefficients into a preset mapping model for compensation calculation, and obtain a preliminary control signal; The predicted water level and predicted flow rate corresponding to the preliminary control signal are predicted using a preset prediction model. The predicted water level and predicted flow rate are compared with the preset flood control constraints to calculate the safety redundancy. If the safety redundancy is lower than the preset stability threshold, the operation log is obtained. The operation log is input into the preset identification model to analyze the balance relationship and correct the damping coefficient to obtain the refined resistance value. The opening signal and current change are acquired, and the opening signal and current change are matched and adjusted according to the refining resistance value using a preset drive mapping. An adjustment command is generated and closed-loop control is executed.

[0007] Secondly, the present invention provides a remote sluice gate opening control system for a pumping station, comprising: The resistance estimation module is used to acquire flow velocity data and operating parameters, and to perform segmentation and distance calculation using a preset classifier to obtain a preliminary resistance estimate. The deviation prediction module is used to obtain real-time operating conditions, construct an initial state vector by combining the preliminary resistance estimate with the preset aging model, calculate the prediction residual using the Kalman filter algorithm and adjust the feedback gain to obtain the control deviation. If the control deviation exceeds the preset accuracy threshold, the diffusion trajectory is deduced using the preset transition probability to determine the potential deviation range. The jamming analysis module is used to obtain the current operating damping and retrieve resistance records from the preset historical database if the potential deviation range exceeds the preset deviation threshold, compare and calculate the deviation vector, and use the Weibull distribution to deduce the probability distribution of the deviation vector to determine the jamming probability. The gain reconstruction module is used to extract the force characteristics based on the jamming probability and the current operating damping, input the force characteristics into a preset fatigue model to extract stress characteristics, calculate the cumulative loss increment based on the stress characteristics, input the cumulative loss increment into a preset degradation function to obtain the risk value, and perform gain reconstruction to generate correction coefficients based on the risk value. The compensation calculation module is used to acquire environmental and communication parameters, input the environmental and communication parameters and the correction coefficients into a preset mapping model for compensation calculation, and obtain a preliminary control signal; The resistance refining module is used to predict the predicted water level and predicted flow rate corresponding to the preliminary control signal using a preset prediction model. The predicted water level and predicted flow rate are compared with the preset flood control constraints to calculate the safety redundancy. If the safety redundancy is lower than the preset stability threshold, the operation log is obtained. The operation log is input into the preset identification model to analyze the balance relationship and correct the damping coefficient to obtain the refined resistance value. The closed-loop control module is used to acquire the opening signal and the current change, and to match and adjust the opening signal and the current change according to the refining resistance value using a preset drive mapping, generate adjustment commands and execute closed-loop control.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention acquires flow velocity data and operating parameters and constructs an initial state vector by combining a preset aging model. It uses the Kalman filter algorithm and transition probability to deduce the potential deviation range. Then, when the deviation exceeds the limit, it uses the Weibull distribution to determine the jamming probability. Finally, it combines the fatigue model and degradation function to perform gain reconstruction and generate correction coefficients. It overcomes the perception distortion and lag problems caused by treating the actuator as an ideal linear model and relying only on static parameters in the prior art. It realizes the deep nonlinear identification of the transient mechanical resistance and physical fatigue degradation state of the sluice gate and intervenes in advance to perform feedforward compensation before the actual physical jamming occurs. It significantly improves the resistance perception accuracy and control system robustness under the complex dynamic water flow and equipment aging intertwined environment, and effectively avoids the mechanical jamming and operational instability of the sluice gate.

[0009] (2) This invention calculates compensation by inputting environmental and communication parameters, including flow velocity, sediment, head pressure and communication delay, into a preset mapping model, and uses a preset prediction model to convert the preliminary control signal into predicted water level and predicted flow, and then compares it with the preset flood control constraints to calculate the safety redundancy; it decouples and synchronously feeds forward compensation between hydrodynamic environmental interference and network communication lag factors, and establishes a predictive verification barrier based on physical environmental red lines (flood warning and overflow limit), and completes the closed-loop verification from abstract electrical signals to specific physical hydraulic effects before the actual issuance of control commands; it completely eliminates command execution deviations caused by communication delays or sudden water conditions under complex working conditions, and greatly enhances the flood control safety of sluice gate scheduling and the physical feasibility of control commands.

[0010] (3) This invention obtains the refined resistance value by using the operation log and identification model when the safety redundancy is insufficient, and uses the preset drive mapping to match and adjust the opening signal and current change based on the refined resistance value, and finally generates the adjustment command and executes closed-loop control; it opens up the precise coordination path from the underlying hardware drive (voltage pulse width modulation delay compensation) to the mechanical end physical feedback (opening displacement and current change), so that the system can dynamically adjust the PID control parameters according to the real-time nonlinear damping change, and completes the seamless connection from the macro water resource scheduling command to the micro hardware motor execution; it realizes the high smoothness and high precision dynamic adjustment of the sluice gate opening, solves the problem of opening inaccuracy and overshoot oscillation caused by sudden mechanical force changes in the traditional SCADA control system, and extends the long-term safe service life of the core equipment of the pumping station. Attached Figure Description

[0011] Figure 1 This is a schematic flowchart of a remote sluice gate opening control method for pumping stations provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of the remote sluice gate opening control system for pumping stations provided in the second embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] Reference Figure 1 The first embodiment of the present invention provides a remote sluice gate opening control method for a pumping station, comprising the following steps: S11: Obtain flow velocity data and operating parameters, use a preset classifier to divide and calculate distances, and obtain a preliminary resistance estimate; S12, obtain real-time operating conditions, construct an initial state vector by combining the preliminary resistance estimate with the preset aging model, calculate the prediction residual using the Kalman filter algorithm and adjust the feedback gain to obtain the control deviation, if the control deviation exceeds the preset accuracy threshold, use the preset transition probability to deduce the diffusion trajectory and determine the potential deviation range. S13, if the potential deviation range exceeds the preset deviation threshold, obtain the current operating damping and retrieve the resistance record from the preset historical database, compare and calculate the deviation vector, use the Weibull distribution to deduce the probability distribution of the deviation vector, and determine the jamming probability. S14. Extract the force characteristics based on the jamming probability and the current operating damping, input the force characteristics into a preset fatigue model to extract stress characteristics, calculate the cumulative loss increment based on the stress characteristics, input the cumulative loss increment into a preset degradation function to obtain the risk value, and perform gain reconstruction to generate correction coefficients based on the risk value. S15, acquire environmental and communication parameters, input the environmental and communication parameters and the correction coefficient into a preset mapping model for compensation calculation, and obtain a preliminary control signal; S16, using a preset prediction model to predict the predicted water level and predicted flow rate corresponding to the preliminary control signal, comparing the predicted water level and predicted flow rate with the preset flood control constraints respectively, and calculating the safety redundancy. If the safety redundancy is lower than the preset stability threshold, the operation log is obtained, and the operation log is input into the preset identification model to analyze the balance relationship and correct the damping coefficient to obtain the refined resistance value. S17, acquire the opening signal and the current change, and adjust the opening signal and the current change according to the refining resistance value using a preset drive mapping, generate an adjustment command and execute closed-loop control.

[0014] In step S11, flow velocity data and operating parameters are acquired, and a preset classifier is used for segmentation and distance calculation to obtain a preliminary resistance estimate, including: When the flow rate data changes abruptly, the motor current, bearing vibration, lubrication status and transmission torque in the operating parameters are extracted, and the motor current, bearing vibration, lubrication status and transmission torque are jointly mapped to a high-dimensional feature space; In the high-dimensional feature space, a feature vector is constructed based on the mapped parameters. The classifier is used to perform nonlinear partitioning of the feature vector to determine the offset of the mapped feature data from the preset baseline feature distribution. The coordinate position of the feature vector in the high-dimensional feature space is corrected according to the offset. If the corrected coordinate position is in a preset abnormal range, the friction resistance feature component is extracted. The frictional resistance feature components are matched with the support vectors in the classifier, and the Euclidean distance between the support vectors and the corrected feature vectors is calculated to obtain the resistance level. By comparing the resistance level with preset historical operating records, the preliminary resistance estimate is obtained when the preset classification convergence condition is met.

[0015] In one implementation, this embodiment uses an ultrasonic Doppler velocimeter deployed at the pump station inlet to collect flow velocity data in real time in the form of a time series. This embodiment sets a sliding time window containing a fixed number of sampling points and calculates the time derivative of the flow velocity data within this sliding time window. This embodiment pre-collects historical flow velocity data from continuous operation under standard stable conditions, calculates the arithmetic mean and standard deviation of its time derivative series, and calculates the upper limit of the normal fluctuation distribution based on the 3-times-standard-deviation principle. This upper limit of the normal fluctuation distribution is set as the flow velocity mutation threshold. When the time derivative of the currently calculated flow velocity data within the sliding time window exceeds the flow velocity mutation threshold, a mutation in the flow velocity data is determined. At this time, this embodiment simultaneously extracts the motor current, bearing vibration, lubrication status, and transmission torque from the operating parameters recorded in real time in the programmable logic controller. This embodiment uses the minimum-maximum normalization method to uniformly linearly map the above four types of physical parameters with different dimensions to the dimensionless interval of 0 to 1. The radial basis function is used to check the normalized parameters and perform nonlinear spatial mapping operation to jointly map the motor current, the bearing vibration, the lubrication state and the transmission torque to a high-dimensional feature space.

[0016] It should be noted that, within the high-dimensional feature space, this embodiment concatenates the mapped motor current, bearing vibration, lubrication state, and transmission torque in a fixed dimensional index order, and constructs feature vectors based on the mapped parameters. This embodiment pre-calculates the mean vector and covariance matrix of a multivariate Gaussian distribution based on a historical normal operating condition dataset to construct a preset baseline feature distribution. The classifier is a preset support vector machine (SVM) classifier. The pre-training process of the SVM classifier involves acquiring a sample dataset containing labels of various known resistance states, constructing a Lagrange dual optimization function whose objective is to maximize the classification margin, iteratively solving for the optimal Lagrange multipliers using a sequential minimum optimization algorithm, and determining the optimal hyperplane and corresponding support vectors that distinguish different resistance states. The penalty parameters and kernel function parameters during model training are determined by using 5-fold cross-validation and a grid search algorithm to find the parameter combination that maximizes the classification accuracy. In this embodiment, the classifier is used to perform nonlinear partitioning of the feature vector, and the vertical geometric distance and relative spatial direction vector from the feature vector to the optimal hyperplane are calculated. The vertical geometric distance and the relative spatial direction vector are multiplied to determine the offset of the mapped feature data from the preset benchmark feature distribution.

[0017] It is worth noting that the training samples of the support vector machine classifier are obtained by synchronously recording operating parameters such as flow rate, motor current, bearing vibration, lubrication status, and transmission torque during the commissioning and normal operation of the pump station, and correspondingly recording the actual resistance status labels of the gate, such as normal, slight jamming, and severe friction, which are confirmed by maintenance personnel or directly measured by high-precision torque sensors. The classifier is trained using these labeled data sample sets.

[0018] In another implementation, this embodiment corrects the coordinate position of the feature vector in the high-dimensional feature space based on the offset. Specifically, the original coordinate sequence of the feature vector is subtracted from the spatial offset value corresponding to the offset to correct the coordinate position. This embodiment uses a local anomaly factor algorithm to perform density clustering analysis on the historical equipment mechanical jamming and fault dataset, extracting the outer contour coordinate set of the abnormal data clusters. The multidimensional closed geometric space formed by this outer contour coordinate set is set as a preset anomaly interval. If the corrected coordinate position falls within the anomaly interval, this embodiment performs principal component analysis on the covariance matrix corresponding to the corrected coordinate position, extracting the data dimension corresponding to the principal component whose cumulative contribution rate exceeds a preset contribution rate threshold, separating it, and determining it as a frictional resistance feature component. The preset contribution rate threshold is determined by the inflection point analysis method of the cumulative variance line graph, specifically using the number of principal components corresponding to a cumulative contribution rate of 85% as the selection criterion.

[0019] It should be noted that this embodiment utilizes the frictional resistance feature components to match the support vectors in the classifier. Specifically, the execution logic involves iterating through all the support vectors retained after the classifier training is completed, and selecting a specific support vector that has the highest cosine similarity value in the isomorphic dimension to the frictional resistance feature components. Subsequently, this embodiment calculates the Euclidean distance between the specific support vector and the corrected feature vector. The calculation process involves calculating the difference between the coordinate values ​​of each dimension of the specific support vector and the corresponding coordinate values ​​of the corrected feature vector, squaring each difference, summing the results, and finally taking the square root of the sum. This embodiment inputs the calculated Euclidean distance into a preset continuous piecewise mapping function to obtain a discretized classification numerical label, which is then determined as the resistance level.

[0020] It is worth noting that this embodiment compares the resistance level with a preset historical operating record. The preset historical operating record is a time series sequence of the device's periodic resistance levels over a preset period of time, stored in a relational database. The preset period is the most recent three years to ensure that the data includes the changing trend of the device's current wear stage, while avoiding interference from outdated data on real-time operating conditions. This embodiment compares the currently calculated resistance level with several recent resistance levels in the time series. The preset classification convergence condition in this embodiment is set by the logic that the variance of the currently calculated resistance level with the resistance levels recorded in the most recent five consecutive sampling periods is less than a preset convergence tolerance threshold. The convergence tolerance threshold is objectively determined by extracting the maximum value of the resistance level fluctuation variance within the historical stable wear cycle of the device; specifically, by collecting stable wear cycle data of the device that has been running continuously for the past five years without any jamming alarms, the resistance level is extracted on a monthly basis, the variance of the resistance level value in each cycle is calculated, and the maximum value of the variance of all cycles during the five-year period is taken as the convergence tolerance threshold. When the preset classification convergence condition is met, this embodiment uses the currently calculated resistance level as the final convergence output to obtain the preliminary resistance estimate.

[0021] In step S12, real-time operating conditions are acquired, and an initial state vector is constructed by combining the preliminary resistance estimate with a preset aging model. The Kalman filter algorithm is used to calculate the prediction residual, and the feedback gain is adjusted to obtain the control deviation. If the control deviation exceeds a preset accuracy threshold, the diffusion trajectory is deduced using a preset transition probability to determine the potential deviation range, including: The resistance level is extracted from the preliminary resistance estimate, and the resistance level is used to match the aging rate in the aging model to determine the wear increment. The initial state vector is constructed by combining the wear increment with the real-time operating conditions. The evolution of the initial state vector on the time axis is calculated using the Kalman filter algorithm to obtain the prediction residual. The control deviation is obtained by adjusting the feedback gain in the observation matrix using the predicted residual and performing recursive correction on the predicted residual in the state space. If the control deviation exceeds the accuracy threshold, the diffusion trajectory of the control deviation is calculated using the transfer probability to determine the initial deviation range; The initial deviation range is nonlinearly mapped to the aging model, and the initial state vector is updated in the iterative loop to determine the potential deviation range.

[0022] In one implementation, this embodiment extracts a resistance level containing discrete numerical features from the preliminary resistance estimate. This resistance level is then input into a preset aging model for matching and retrieval. It should be noted that the preset aging model is a mathematical mapping table constructed by collecting historical operation logs and periodic maintenance measurement data of similar sluice gate equipment throughout its entire lifecycle, and fitting this table using a multinomial regression algorithm. This mapping table records the material wear rate per unit time corresponding to different resistance levels. Specifically, periodically, such as quarterly, physical measurements are taken of the wear of key transmission components of the pump station or the same type of sluice gate, such as lead screws and nuts, and the corresponding cumulative operating time and average resistance level during that period are recorded. Based on multiple sets of cumulative operating time, wear amount, and resistance level data, a multinomial regression is used to fit the mapping relationship between the resistance level and the wear increment per unit time. This embodiment obtains the corresponding aging rate through matching and retrieval, and simultaneously acquires real-time operating conditions including the cumulative operating time of the equipment through a sensor network. In this embodiment, the cumulative running time is divided by a preset time base to obtain a time coefficient, and then the aging rate is multiplied by the time coefficient to calculate the wear increment in units of length. The time base is set according to the equipment design life and is taken as the expected total operating hours of the equipment. For example, for a sluice gate hoist with a design life of 20 years, the time base is 14,600 hours, calculated based on an average daily operation of two hours.

[0023] It is worth noting that this embodiment combines the wear increment with the real-time operating conditions to construct the initial state vector. Specifically, the real-time water flow velocity and real-time motor output torque in the real-time operating conditions are extracted, and the wear increment, the real-time water flow velocity, and the real-time motor output torque are arranged in a fixed dimensional order to construct a multi-dimensional column vector, which is the initial state vector. Subsequently, this embodiment uses the Kalman filter algorithm to calculate the evolution of the initial state vector on the time axis. This embodiment constructs a state transition matrix, multiplies the initial state vector with the state transition matrix, and obtains the prior state estimate for the current time step. This embodiment obtains the actual observation vector through physical sensors and maps the prior state estimate to the observation space using a preset observation matrix to obtain the prior observation estimate. This embodiment subtracts the prior observation estimate from the actual observation vector to calculate the prediction residual.

[0024] It should be noted that this embodiment utilizes the predicted residual to adjust the feedback gain in the observation matrix. The specific calculation logic is as follows: The prior error covariance matrix for the current time step is calculated based on the error covariance matrix of the previous time step and the system process noise covariance matrix; the prior error covariance matrix is ​​multiplied by the transpose of the observation matrix to obtain the numerator; the continuous product of the observation matrix, the prior error covariance matrix, and the transpose of the observation matrix, plus the measurement noise covariance matrix, is added to the denominator; the numerator is divided by the denominator to calculate the Kalman gain matrix, which is the feedback gain. This embodiment performs recursive correction on the predicted residual in the state space, specifically by multiplying the feedback gain by the predicted residual and superimposing the product onto the prior state estimate to obtain the posterior state estimate. This embodiment extracts the displacement feature component from the posterior state estimate, subtracts the ideal reference displacement value from this displacement feature component, and calculates the difference vector, which is the control deviation.

[0025] In another implementation, this embodiment determines whether the control deviation exceeds a preset accuracy threshold. The preset accuracy threshold is determined by statistically analyzing the finite element fatigue simulation results of the sluice gate's mechanical structure and extracting the maximum allowable deviation value that does not induce irreversible structural deformation. If the absolute value of the control deviation exceeds the accuracy threshold, this embodiment calculates the diffusion trajectory of the control deviation using a preset transition probability. It should be noted that this embodiment pre-divides the continuous numerical range of the control deviation into multiple discrete deviation state intervals according to a fixed numerical span. The preset transition probability is represented by a Markov state transition matrix, which is obtained by statistically analyzing historical mechanical instability diffusion case data using the maximum likelihood estimation method. This matrix records the probability value of the current deviation state interval transitioning to a more severe deviation state interval within the next very short time window. This embodiment performs continuous matrix multiplication iterations between the current state vector containing the control deviation and the Markov state transition matrix, with the number of operations corresponding to a preset number of future prediction time steps. This embodiment connects the deviation vectors output from each matrix multiplication iteration in chronological order to generate the diffusion trajectory. In this embodiment, the upper and lower limits of the diffusion trajectory at each time point are extracted to form an envelope interval, which is then determined as the initial deviation range.

[0026] It is worth noting that the probability values ​​of the Markov state transition matrix are obtained by analyzing historical monitoring data. Specifically, the historical control deviation data sequence is divided into several discrete states according to the percentage of the deviation threshold, such as 0-20% for state 1 and 20-40% for state 2. The frequency of state transitions in all adjacent sampling times is counted, and then this frequency is divided by the total frequency of transitions from a certain state to obtain the transition probability. If there is not enough historical data, it can be initialized as an identity matrix, which means that the state is assumed to be maintained in a high probability and is gradually updated during operation.

[0027] It should be noted that this embodiment performs a nonlinear mapping between the initial deviation range and the aging model. Specifically, the numerical boundary of the initial deviation range is used as an input parameter and substituted into the accelerated degradation polynomial function of the aging model to calculate the acceleration penalty factor. This embodiment multiplies the acceleration penalty factor by the wear increment element in the original initial state vector to update the initial state vector during iterative iteration. Using the updated initial state vector, this embodiment repeats the above steps of Kalman filter residual calculation, state deduction, and transition probability trajectory deduction. After each iteration, the absolute value of the difference between the deviation range boundary value obtained in this iteration and the deviation range boundary value obtained in the previous iteration is calculated. If the absolute value of this difference is less than a preset convergence scalar for three consecutive iterations (the nominal resolution of the selected absolute value photoelectric rotary encoder is 0.01 mm), the iterative iteration is terminated. This embodiment determines the deviation range output in the last iteration as the potential deviation range.

[0028] In step S13, if the potential deviation range exceeds a preset deviation threshold, the current operating damping is obtained and resistance records are retrieved from a preset historical database. A deviation vector is calculated by comparison, and the probability distribution of the deviation vector is derived using a Weibull distribution to determine the jamming probability, including: If the potential deviation range exceeds the deviation threshold, the resistance record that matches the flow velocity data is retrieved from the historical database, and the friction features of the corresponding gate opening are extracted. The friction features are matched with historical operating conditions to extract similar trajectories, and the historical reference damping corresponding to the current moment is extracted from the similar trajectories. Calculate the vector difference between the current operating damping and the historical reference damping to obtain the deviation vector; If the deviation vector exceeds a preset vector threshold, the Weibull distribution is used to calculate the blocking frequency of the deviation vector within a specific time window; The probability distribution is obtained by mapping the evolution trend of the blocking frequency in the state space, and the jamming probability is determined.

[0029] In one implementation, this embodiment extracts the upper limit of the potential deviation range and determines whether this upper limit exceeds a preset deviation threshold. The preset deviation threshold is determined by establishing a three-dimensional dynamic simulation model of the sluice gate actuator, applying a step disturbance load for offline simulation, and extracting the limit displacement deviation value that does not cause irreversible deformation or resonance of the mechanism. Specifically, it extracts the daily deviation vector magnitude data of the equipment under normal operating conditions over the past three years, removes outliers exceeding three times the standard deviation due to sensor malfunctions, calculates the standard deviation of the remaining data, and uses the three-times-standard-deviation value as the vector threshold. If the preset deviation threshold is exceeded, this embodiment reads the real-time output torque and speed of the actuator motor and converts them into linear force values ​​based on the mechanical transmission ratio to obtain the current operating damping. This embodiment constructs a preset historical database, which is a relational database built by collecting the sluice gate's monitoring and data acquisition system operation logs over the past five years, cleaning them according to the mapping relationship between flow velocity, gate opening, and resistance values. In this embodiment, the current flow velocity data is used as the search keyword to retrieve corresponding data entries in the historical database whose Euclidean distance error is within 5%, and these entries are identified as the resistance records that match the flow velocity data. This embodiment extracts the historical resistance statistical mean and variance corresponding to the current gate opening from these records and identifies them as friction characteristics.

[0030] It should be noted that this embodiment matches the friction features with historical operating conditions to extract similar trajectories. Specifically, the real-time operating damping within the current time window is extracted to form a query time series, and historical operating damping under the same operating condition dimension in the historical database is extracted to form multiple candidate time series. This embodiment uses a dynamic time warping algorithm to calculate the minimum cumulative distance path between the query time series and each candidate time series, and determines the candidate time series that generates the minimum cumulative distance path as the similar trajectory. In this embodiment, the resistance value at the path alignment point corresponding to the current moment is extracted from the similar trajectory and determined as the historical reference damping. This embodiment calculates the difference between the current operating damping and the historical reference damping to obtain the deviation vector. This embodiment determines whether the deviation vector exceeds a preset vector threshold. The preset vector threshold is determined by calculating the standard deviation of the deviation vector set under historical normal operating conditions and taking three times that standard deviation.

[0031] It is worth noting that if the deviation vector exceeds the preset vector threshold, this embodiment uses the Weibull distribution to calculate the congestion frequency of the deviation vector within a specific time window. The Weibull distribution includes shape parameters and scale parameters; it can be initialized based on reliability manual data of similar equipment. For example, the shape parameter can be initially set to 1, an exponential distribution, indicating a constant failure rate, and the scale parameter can be set to the characteristic life reference equipment overhaul cycle. After the system is running, the parameters are gradually corrected using a Bayesian update method based on actual monitored abnormal events, such as the interval between deviations exceeding the limit. This embodiment uses the maximum likelihood estimation method, iteratively calculating based on historically occurring jamming fault record time interval data, and fitting and solving for the corresponding shape parameters and scale parameters. This embodiment substitutes the magnitude of the deviation vector into the probability density function of the fitted Weibull distribution, performs integral calculation over the preset future 12-hour specific time window, and obtains the congestion frequency reflecting the expected number of mechanical blockages during that time period. This embodiment utilizes the evolution trend of the congestion frequency in the state space for mapping. Specifically, the congestion frequency at different time steps is input into the cumulative distribution function and mapped to a continuous probability value sequence between 0 and 1 to obtain the probability distribution. The probability value corresponding to the current moment is then extracted and determined as the jamming probability.

[0032] In step S14, force characteristics are extracted based on the jamming probability and the current operating damping. These force characteristics are then input into a preset fatigue model to extract stress characteristics. The cumulative loss increment is calculated based on the stress characteristics. This cumulative loss increment is input into a preset degradation function to obtain a risk value. Gain reconstruction is then performed based on the risk value to generate correction coefficients, including: Based on the jamming probability and the current operating damping, the force characteristics of the equipment under the current operating conditions are extracted; The stress characteristics are input into the fatigue model to extract the stress features; The cumulative loss increment of the equipment during its operating cycle is determined based on the stress characteristics. The cumulative loss increment is input into the degradation function to calculate the risk value that reflects the degree of performance degradation; Calculate the deviation vector of the risk value in the state space, use the deviation vector to perform the gain reconstruction, and generate the correction coefficient.

[0033] In one implementation, this embodiment extracts the stress characteristics of the equipment under the current operating condition based on the jamming probability and the current operating damping. Specifically, the calculation process involves multiplying the current operating damping by the jamming probability to obtain an additional risk damping equivalent, and then adding the current operating damping to the additional risk damping equivalent to obtain the stress characteristics reflecting the expected extreme stress. This embodiment inputs the stress characteristics into a preset fatigue model. The preset fatigue model is an SN curve model constructed by performing standard tensile-compressive fatigue material tests on key metal structural components of a sluice gate and fitting the correspondence between different stress levels and the number of failure cycles. This embodiment uses the rainflow counting method to extract features from the time-load sequence composed of the stress characteristics, identifying full-cycle and half-cycle features in the sequence, calculating the stress amplitude and average stress of each cycle, and combining them to determine the stress characteristics.

[0034] It should be noted that this embodiment determines the cumulative loss increment of the equipment within its operating cycle based on the stress characteristics. This embodiment employs Miner's linear cumulative damage theory, dividing the actual number of cycles corresponding to each extracted stress amplitude by the ultimate failure cycle number corresponding to that stress amplitude in the fatigue model to obtain the individual damage degree for each stress level. This embodiment sums the individual damage degrees of all levels within the current operating cycle to obtain the cumulative loss increment. This embodiment inputs the cumulative loss increment into a preset degradation function.

[0035] The preset degradation function is an exponential performance degradation mathematical function. Its degradation coefficient is objectively determined by least-squares fitting of wear thickness data from major overhauls of the equipment over the years. The specific calculation logic of this exponential performance degradation mathematical function is as follows: calculate the product of the degradation coefficient and the actual value of the cumulative loss increment, take the negative of the product as the exponent, calculate the exponent of the natural constant, and finally subtract the calculated result of the exponent from the number 1 to obtain the risk value between 0 and 1, which reflects the degree of performance degradation of the physical entity. Here, 1 represents complete equipment failure, and 0 represents the equipment in a brand-new state.

[0036] In another implementation, this embodiment calculates the deviation vector of the risk value in the state space. This embodiment obtains a baseline risk value of the device under ideal health conditions, subtracts the baseline risk value from the currently calculated risk value, and obtains the deviation vector reflecting the degree of deviation. This embodiment uses the deviation vector to perform the gain reconstruction. Specifically, this embodiment pre-establishes a mapping lookup table between the deviation vector and the PID control gain adjustment factor. In this embodiment, a linear interpolation algorithm is used in the mapping lookup table, with the current deviation vector as the input index, to calculate the corresponding feedforward gain compensation amount. This embodiment determines the feedforward gain compensation amount as the correction coefficient, which is directly used as a product factor in the opening adjustment control command issued in the next stage.

[0037] In step S15, environmental and communication parameters are acquired, and the environmental and communication parameters and the correction coefficients are input into a preset mapping model for compensation calculation to obtain a preliminary control signal, including: The environmental flow velocity and sediment content in the environmental and communication parameters are obtained, and the environmental flow velocity, sediment content and correction coefficient are input into the preset mapping model to determine the hydrodynamic force compensation value. The head pressure in the environmental and communication parameters is obtained, and the ideal control signal is calculated based on the head pressure and the hydrodynamic force compensation value. Extract the communication delay and signal strength from the environmental and communication parameters, and calculate the instruction transmission lag based on the communication delay and signal strength; The initial control signal is obtained by using the instruction transmission lag to perform time-axis feedforward compensation on the ideal control signal.

[0038] In one implementation, this embodiment uses an underwater acoustic current profiler and an optical turbidimeter deployed at the flow measurement section to acquire the environmental flow velocity and sediment content from the environmental and communication parameters, respectively. This embodiment inputs the environmental flow velocity, sediment content, and correction coefficient into a preset mapping model. The preset mapping model is a multivariate second-order polynomial response surface model. The mathematical coefficients of this response surface model are objectively determined by using least squares regression fitting to solve the multivariate data from scaled hydraulic model experiments under different combinations of sediment concentration and flow velocity. This embodiment uses the preset mapping model to calculate the foundation hydrodynamic pressure and multiplies the foundation hydrodynamic pressure by the correction coefficient to determine the hydrodynamic force compensation value.

[0039] Specifically, the construction of the multivariate second-order polynomial response surface model adopts a hydraulic physics model with a geometric scale of 1:20. Eleven working conditions are set with the environmental flow velocity in increments of 0.5 m / s, ranging from 0 m / s to 5 m / s. Eleven working conditions are set with the sediment content in increments of 0 kg / m³ to 10 kg / m³. These conditions are combined to form 121 working condition points for measurement. After model fitting, the coefficient of determination greater than 0.95 is used as the goodness-of-fit test criterion.

[0040] It should be noted that in this embodiment, the absolute elevation values ​​of the water levels upstream and downstream of the gate are obtained using a piezoresistive level gauge. The absolute elevation value of the upstream water level is subtracted from the absolute elevation value of the downstream water level, and the difference is multiplied by the water density constant and the gravitational acceleration constant to calculate the head pressure in the environmental and communication parameters. Based on the head pressure, this embodiment retrieves the basic opening command from the hydraulic scheduling surface lookup table. Simultaneously, the hydrodynamic force compensation value is divided by the pre-calibrated gate winch lifting stiffness coefficient to obtain the opening compensation amount. This embodiment adds the basic opening command and the opening compensation amount to calculate the ideal control signal. This ideal control signal is expressed as a percentage of the target gate's physical opening.

[0041] It is worth noting that this embodiment extracts the communication latency and signal strength from the environmental and communication parameters in the monitoring and data acquisition network status message. This embodiment calculates the instruction transmission lag based on the communication latency and signal strength. Specifically, this embodiment extracts the standard data packet size value for a single control instruction. This standard data packet size value is objectively set according to the fixed data frame length of the industrial Ethernet communication protocol. This embodiment divides the standard data packet size value by the effective real-time network bandwidth value corresponding to the signal strength mapping to calculate the dynamic transmission time. This embodiment adds the dynamic transmission time to the current communication latency, and the sum of the operations is determined as the instruction transmission lag.

[0042] In another implementation, this embodiment utilizes the instruction transmission lag to perform time-axis feedforward compensation on the ideal control signal. Specifically, the execution logic is as follows: this embodiment extracts the target execution timestamp pre-bound to the ideal control signal, subtracts the time value corresponding to the instruction transmission lag from the target execution timestamp, and generates a control instruction time sequence with an advance triggering time attribute. This embodiment outputs the instruction packet carrying the advance triggering time attribute to obtain the preliminary control signal.

[0043] In step S16, a preset prediction model is used to predict the predicted water level and predicted flow rate corresponding to the preliminary control signal. The predicted water level and predicted flow rate are compared with preset flood control constraints to calculate the safety redundancy. If the safety redundancy is lower than a preset stability threshold, the operation log is obtained. The operation log is input into a preset identification model to analyze the balance relationship and correct the damping coefficient to obtain the refined resistance value, including: Obtain the water level warning limit and overflow constraint conditions in the flood control constraints; The preset prediction model is used to predict the predicted water level and the predicted flow rate corresponding to the preliminary control signal under the current operating conditions. The predicted water level is compared with the water level warning limit, and the predicted flow rate is compared with the overflow constraint condition. The safety redundancy is calculated by combining the results. If the safety redundancy is lower than the stability threshold, then current fluctuation characteristics and opening displacement sequence are extracted from the operation log; The current fluctuation characteristics and the opening displacement sequence are input into the identification model to calculate the dynamic balance relationship between the frictional resistance torque and the load inertia torque, and to obtain preliminary identification results.

[0044] In one implementation, this embodiment reads and obtains the water level warning limit and excess flow constraint from the static relational database of the basin flood control command center. This embodiment uses a preset prediction model to predict the predicted water level and predicted flow corresponding to the preliminary control signal under the current operating conditions. The preset prediction model is a hydrodynamic numerical solution model based on a one-dimensional unsteady flow partial differential equation. This model uses a four-point implicit difference scheme to discretize the partial differential equation into a system of algebraic equations, and uses the pursuit method to solve for the hydrodynamic state of each spatial physical node in the time step. Specifically, the spatial step size of the one-dimensional unsteady flow partial differential equation model is set to fifty meters, the time step size is automatically adjusted using a Courant number less than or equal to 0.8, and the model calibration uses continuous hydrological observation data of the river section for the past ten years, with a root mean square error of less than 0.1 meters between the simulated water level and the measured water level as the calibration pass standard. In this embodiment, the opening input value represented by the preliminary control signal and the current upstream water flow obtained by the sensor are used as the boundary condition parameters input of the preset prediction model. After numerical calculation, the extreme value of water level and the arithmetic mean of flow rate of the downstream control section in the future prediction time window are output. These two values ​​are determined as the predicted water level and the predicted flow rate, respectively.

[0045] It should be noted that this embodiment compares the predicted water level with the water level warning limit. The calculation logic is as follows: subtract the predicted water level from the water level warning limit, and divide the difference by the water level warning limit to obtain the water level margin ratio. Simultaneously, this embodiment compares the predicted flow rate with the overflow constraint. Specifically, subtract the predicted flow rate from the overflow constraint, and divide the difference by the overflow constraint to obtain the flow margin ratio. This embodiment compares the water level margin ratio and the flow margin ratio, extracts the smaller value, and determines it as the safety redundancy obtained from the comprehensive calculation.

[0046] In another implementation, this embodiment determines whether the safety redundancy is lower than a preset stability threshold. The preset stability threshold is determined by analyzing extreme operating conditions recorded in the historical database of flood control drills over the past twenty years where no structural scour failure occurred, extracting the historical minimum safety redundancy data for the corresponding time point, and multiplying this value by an amplification factor of 1.5. If the safety redundancy is lower than the preset stability threshold, this embodiment extracts a time series containing the effective value of the stator current from the operating log of the device's local programmable logic controller as a current fluctuation feature, and simultaneously extracts the actual mechanical coordinate change sequence recorded by the photoelectric position encoder as an opening displacement sequence.

[0047] It is worth noting that in this embodiment, the current fluctuation characteristics and the opening displacement sequence are input into the identification model. The identification model is a pre-established dynamic parameter identification algorithm model based on the recursive least squares method. The solution principle of this model is to obtain the output electromagnetic torque sequence by multiplying the current fluctuation characteristics with the known motor torque constant, and to obtain the angular acceleration sequence by performing two time difference operations on the opening displacement sequence. In this embodiment, a torque equation is constructed within the identification model, setting the output electromagnetic torque sequence to be equal to the product of the load inertia moment and the angular acceleration sequence plus the frictional resistance moment. In this embodiment, the state parameter vector is iteratively updated using the recursive least squares method until the sum of squares of the predicted output residuals of the equation reaches a minimum extreme value, thereby calculating and separating the dynamic balance relationship between the frictional resistance moment and the load inertia moment. In this embodiment, the separated frictional resistance moment is extracted and determined as the resistance component output included in this preliminary identification result.

[0048] The process includes inputting the current fluctuation characteristics and the opening displacement sequence into the identification model, calculating the dynamic balance between the frictional resistance torque and the load inertia torque, and obtaining preliminary identification results. The process further includes: If the drag component in the preliminary identification result exceeds the preset design load range, the initial damping coefficient in the identification model is corrected. The nonlinear friction force distribution is recalculated using the corrected initial damping coefficient; The nonlinear friction force distribution is processed using a multi-round numerical approximation algorithm to obtain the refining resistance value.

[0049] In one implementation, this embodiment extracts the resistance component from the preliminary identification result and determines whether the resistance component exceeds a preset design load range. The preset design load range is a closed numerical interval determined by combining the upper limit of the rated mechanical torque parameter explicitly provided in the manufacturer's white paper for the sluice gate equipment with the tensile ultimate yield strength data of the structural steel through material mechanics derating calculations. If the resistance component exceeds this closed numerical interval, this embodiment corrects the initial damping coefficient in the identification model. The correction process involves dividing the current resistance component value exceeding the upper limit by the upper limit boundary value of the design load range to calculate the overload ratio scalar. In this embodiment, the initial damping coefficient configured in the identification model under the current physical state is directly multiplied by the overload ratio scalar, and the product of the two is used to replace and update the corrected initial damping coefficient.

[0050] It should be noted that this embodiment recalculates the nonlinear friction force distribution using the corrected initial damping coefficient. This embodiment pre-deploys a dynamic tribological physical simulation equation that incorporates low-speed nonlinear friction effects within the system. The corrected initial damping coefficient is substituted into the viscous friction sensitivity coefficient term of this physical simulation equation. Based on the aforementioned opening displacement sequence, this embodiment performs first-order time-difference differentiation to obtain a velocity time series. This velocity time series is then substituted into the physical simulation equation to independently calculate the static Coulomb friction force component unaffected by velocity, the nonlinear friction force component that decays exponentially with velocity, and the linear viscous friction force component proportional to the current velocity constant. This embodiment performs numerical superposition calculations on a unified time coordinate axis for the three types of discrete components obtained above, resulting in a numerical set that dynamically evolves with time and velocity. This numerical set is the recalculated nonlinear friction force distribution.

[0051] It is worth noting that this embodiment utilizes a multi-round numerical approximation algorithm to process the nonlinear friction force distribution. The multi-round numerical approximation algorithm used in this embodiment is the Newton-Raphson iterative algorithm. Specifically, the execution steps are as follows: This embodiment extracts the arithmetic mean of the time series corresponding to the nonlinear friction force distribution and sets it as the initial iteration starting point of the iterative optimization algorithm. In each iteration loop, this embodiment calculates the first derivative of the gradient of the current estimated value along the tangent direction of the target smoothing function and updates the estimated value along the direction of the steepest gradient descent. This embodiment continuously calculates the absolute value of the difference between the estimated values ​​output from two adjacent iterations. When the absolute value of this difference is less than a preset convergence tolerance, the multi-round iteration is actively terminated. The preset convergence tolerance is determined based on a direct mapping of the minimum resolution analog accuracy constant of the mechanical side torque sensor. In this embodiment, the convergence and stability estimate value output at the final iteration termination time is extracted separately and determined as the final output refined resistance value. The refined resistance value is an optimized resistance parameter obtained by reverse correction using an identification model based on the initial resistance estimate during the safety redundancy verification process. Its physical meaning is the same as the initial resistance estimate, both representing the mechanical resistance experienced by the gate under the current operating conditions, but the refined resistance value has higher identification accuracy. The ideal reference displacement value is the theoretical displacement value corresponding to the target opening command issued by the host computer, which is generated by the water conservancy dispatching system according to water level control requirements.

[0052] In step S17, the opening signal and current change are acquired, and the opening signal and current change are matched and adjusted according to the refining resistance value using a preset drive mapping. An adjustment command is generated and closed-loop control is executed, including: The driving torque reference is determined based on the refining resistance value, and the corresponding voltage pulse width modulation sequence is retrieved using the driving torque and pulse width modulation duty cycle mapping table in the driving mapping. The voltage pulse width modulation sequence is matched with a preset transmission efficiency mapping table to obtain an initial control pulse with time delay compensation characteristics, and the desired opening trajectory is obtained by parsing the initial control pulse. The positional deviation between the desired opening trajectory and the opening signal is compared. If the positional deviation exceeds a preset synchronization error range, a preset gain adjustment operator is used to perform PID adjustment to generate the adjustment command. The power unit is driven by the adjustment command, and the displacement control increment is determined according to the change in current. The displacement control increment is applied to the actuator and the end position data is fed back in real time to execute the closed-loop control.

[0053] In one implementation, this embodiment extracts the refined resistance value calculated and converged in the previous steps, divides the refined resistance value by the pre-extracted mechanical reduction ratio constant of the actuator reducer, and multiplies the quotient by the safety factor scalar specified in the equipment white paper to calculate the required expected value of the motor end-side torque, which is then determined as the driving torque reference. This embodiment calls a preset driving mapping in the local storage medium. This driving mapping internally encapsulates a pre-constructed driving torque and pulse width modulation duty cycle mapping table. The driving torque and pulse width modulation duty cycle mapping table is a two-dimensional lookup table generated by applying multi-step mechanical resistance loads to an offline motor test bench, recording the duty cycle values ​​required for the motor to maintain its rated speed under different steady-state output torques, and fitting the data using a cubic spline interpolation algorithm. This embodiment uses the calculated driving torque reference as the input index value into this mapping table, and extracts the corresponding discrete pulse control signal array containing the duty cycle extrema and chopping frequency through interpolation addressing, determining it as a voltage pulse width modulation sequence.

[0054] It should be noted that this embodiment utilizes the voltage pulse width modulation sequence to match a preset transmission efficiency mapping table. The preset transmission efficiency mapping table is a dataset established by comparing the backlash of the sluice gate drive screw and gearbox under no-load and full-load conditions, extracting the mechanical backlash and transmission hysteresis time distribution characteristics under different speed and torque ranges. In this embodiment, the duty cycle characteristics of the voltage pulse width modulation sequence are input into the transmission efficiency mapping table to extract the corresponding mechanical transmission hysteresis time constant. This embodiment performs a negative translation operation on the time axis of the voltage pulse width modulation sequence, with the translation time equal to the mechanical transmission hysteresis time constant, thereby generating an initial control pulse with time delay compensation characteristics. Subsequently, this embodiment extracts the effective voltage value and high-level duration parameter of the initial control pulse, combines it with the rated speed constant of the motor and the lead of the drive screw, and performs a numerical integration operation over time to obtain the absolute displacement numerical sequence over time in the theoretical dimension. This absolute displacement numerical sequence is then determined as the desired opening trajectory.

[0055] In another implementation, this embodiment uses an absolute photoelectric rotary encoder physically connected directly to the main shaft of the gate winch to collect real-time rotation angle data. This data is then converted into linear displacement values ​​according to a circumferential ratio and determined as the real-time opening signal. In this embodiment, under the same aligned timestamp slice, the theoretical displacement value of the desired opening trajectory at that moment is obtained. The actual displacement value of the opening signal is subtracted from this theoretical displacement value, and the absolute value of the difference is calculated and determined as the position deviation. This embodiment compares the position deviation with a preset synchronization error range. The preset synchronization error range is objectively determined by adding the factory physical measurement error constant of the rotary encoder to the static clearance value of the mechanical transmission chain under maximum load. If the position deviation exceeds the preset synchronization error range, this embodiment calculates the deviation change rate based on the position deviation of two adjacent sampling periods. The current position deviation and the deviation change rate are used as inputs to a preset gain adjustment operator. The preset gain adjustment operator is a dynamic polynomial equation of control coefficients pre-constructed based on the Ziegler-Nichols tuning rule. In this embodiment, the operator is used to output updated proportional coefficients, integral coefficients, and differential coefficients through algebraic calculations. The updated three coefficients are then used to perform discrete proportional-integral-differential calculations on the position deviation to generate an incremental correction instruction package for the voltage pulse duty cycle, which is then determined as the adjustment command.

[0056] It is worth noting that in this embodiment, the adjustment command drives the power unit to output the actual drive voltage to the stator winding of the drive motor. During the load operation of the actuator, this embodiment continuously collects the effective value of the armature current through a current sensor, calculates the difference between the effective values ​​of the current in two adjacent sampling time steps, and obtains the current change. This embodiment determines the displacement control increment based on the current change. Specifically, this embodiment divides the current change by a pre-calibrated motor torque current constant, and multiplies the result of this division by a preset elastic deformation compensation coefficient to calculate the displacement control increment. The motor torque current constant is obtained through the motor's factory test report or determined through on-site no-load experiments, measuring the no-load current under the motor's rated voltage, and back-calculating using the known no-load torque. The elastic deformation compensation coefficient is an objective physical constant derived and calculated in advance based on the Young's modulus, effective cross-sectional area, and force-bearing span of the transmission screw material using Hooke's law in mechanics of materials. This embodiment adds the displacement control increment to the target opening setpoint for the next control cycle and applies it to the actuator. In this embodiment, the terminal position data is continuously fed back for closed-loop iterative calculation, and the closed-loop control is executed until the position deviation is less than the synchronization error range for three consecutive sampling periods, at which point the adjustment loop is exited.

[0057] It should be noted that the displacement control increment does not directly participate in the PID adjustment calculation of the current control cycle, but is superimposed as a feedforward compensation amount into the expected opening trajectory of the next control cycle to correct the opening tracking deviation caused by mechanical elastic deformation. In specific implementation, the displacement control increment and the initial value of the expected opening trajectory at the beginning of the next cycle are algebraically added to obtain the corrected expected opening trajectory, and then the position deviation of the corrected trajectory is compared with the real-time opening signal.

[0058] In summary, this invention constructs a deep identification and feedforward gain reconstruction mechanism for transient fluctuations in mechanical resistance and physical jamming risks of sluice gates by integrating Kalman filter residual analysis, Weibull distribution derivation, and nonlinear degradation functions. It rigorously decouples hydrodynamic load interference and network communication lag, and introduces a pre-set prediction model for advance verification and resistance refinement under flood control safety red line constraints. Finally, through underlying drive delay compensation mapping and adaptive PID control closed loop based on current change feedback, it establishes a precise collaborative path from macroscopic water conservancy scheduling commands to microscopic electrical actuators. This achieves high-precision and smooth dynamic adjustment of sluice gate opening under complex dynamic water flow and equipment aging environments, completely solving the problems of response lag, overshoot oscillation, and mechanical jamming caused by the perception black box and fixed logic in traditional remote control systems. This significantly improves the absolute flood control safety of remote pump station scheduling and the long service life of core equipment.

[0059] Reference Figure 2 The second embodiment of the present invention provides a remote sluice gate opening control system for a pumping station, comprising: The resistance estimation module is used to acquire flow velocity data and operating parameters, and to perform segmentation and distance calculation using a preset classifier to obtain a preliminary resistance estimate. The deviation prediction module is used to obtain real-time operating conditions, construct an initial state vector by combining the preliminary resistance estimate with the preset aging model, calculate the prediction residual using the Kalman filter algorithm and adjust the feedback gain to obtain the control deviation. If the control deviation exceeds the preset accuracy threshold, the diffusion trajectory is deduced using the preset transition probability to determine the potential deviation range. The jamming analysis module is used to obtain the current operating damping and retrieve resistance records from the preset historical database if the potential deviation range exceeds the preset deviation threshold, compare and calculate the deviation vector, and use the Weibull distribution to deduce the probability distribution of the deviation vector to determine the jamming probability. The gain reconstruction module is used to extract the force characteristics based on the jamming probability and the current operating damping, input the force characteristics into a preset fatigue model to extract stress characteristics, calculate the cumulative loss increment based on the stress characteristics, input the cumulative loss increment into a preset degradation function to obtain the risk value, and perform gain reconstruction to generate correction coefficients based on the risk value. The compensation calculation module is used to acquire environmental and communication parameters, input the environmental and communication parameters and the correction coefficients into a preset mapping model for compensation calculation, and obtain a preliminary control signal; The resistance refining module is used to predict the predicted water level and predicted flow rate corresponding to the preliminary control signal using a preset prediction model. The predicted water level and predicted flow rate are compared with the preset flood control constraints to calculate the safety redundancy. If the safety redundancy is lower than the preset stability threshold, the operation log is obtained. The operation log is input into the preset identification model to analyze the balance relationship and correct the damping coefficient to obtain the refined resistance value. The closed-loop control module is used to acquire the opening signal and the current change, and to match and adjust the opening signal and the current change according to the refining resistance value using a preset drive mapping, generate adjustment commands and execute closed-loop control.

[0060] It should be noted that the remote sluice gate opening control system for pumping stations provided in this embodiment of the invention is used to execute all the process steps of the remote sluice gate opening control method for pumping stations described in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0061] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0062] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for remote sluice gate opening control in a pumping station, characterized in that, include: Acquire flow velocity data and operating parameters, use a preset classifier to divide and calculate distances, and obtain a preliminary resistance estimate; The system acquires real-time operating conditions, constructs an initial state vector by combining the preliminary resistance estimate with the preset aging model, calculates the prediction residual using the Kalman filter algorithm and adjusts the feedback gain to obtain the control deviation, and if the control deviation exceeds the preset accuracy threshold, it uses the preset transition probability to deduce the diffusion trajectory and determine the potential deviation range. If the potential deviation range exceeds the preset deviation threshold, the current operating damping is obtained and the resistance record is retrieved from the preset historical database. The deviation vector is calculated by comparison, and the probability distribution of the deviation vector is deduced using the Weibull distribution to determine the jamming probability. Based on the jamming probability and the current operating damping, the force characteristics are extracted. The force characteristics are input into a preset fatigue model to extract stress characteristics. The cumulative loss increment is calculated based on the stress characteristics. The cumulative loss increment is input into a preset degradation function to obtain a risk value. Based on the risk value, gain reconstruction is performed to generate correction coefficients. Obtain environmental and communication parameters, input the environmental and communication parameters and the correction coefficients into a preset mapping model for compensation calculation, and obtain a preliminary control signal; The predicted water level and predicted flow rate corresponding to the preliminary control signal are predicted using a preset prediction model. The predicted water level and predicted flow rate are compared with the preset flood control constraints to calculate the safety redundancy. If the safety redundancy is lower than the preset stability threshold, the operation log is obtained. The operation log is input into the preset identification model to analyze the balance relationship and correct the damping coefficient to obtain the refined resistance value. The opening signal and current change are acquired, and the opening signal and current change are matched and adjusted according to the refining resistance value using a preset drive mapping. An adjustment command is generated and closed-loop control is executed.

2. The remote sluice gate opening control method for a pumping station according to claim 1, characterized in that, The process of acquiring flow velocity data and operating parameters, dividing and calculating distances using a preset classifier, and obtaining a preliminary resistance estimate includes: When the flow rate data changes abruptly, the motor current, bearing vibration, lubrication status and transmission torque in the operating parameters are extracted, and the motor current, bearing vibration, lubrication status and transmission torque are jointly mapped to a high-dimensional feature space; In the high-dimensional feature space, a feature vector is constructed based on the mapped parameters. The classifier is used to perform nonlinear partitioning of the feature vector to determine the offset of the mapped feature data from the preset baseline feature distribution. The coordinate position of the feature vector in the high-dimensional feature space is corrected according to the offset. If the corrected coordinate position is in a preset abnormal range, the friction resistance feature component is extracted. The frictional resistance feature components are matched with the support vectors in the classifier, and the Euclidean distance between the support vectors and the corrected feature vectors is calculated to obtain the resistance level. By comparing the resistance level with preset historical operating records, the preliminary resistance estimate is obtained when the preset classification convergence condition is met.

3. The remote sluice gate opening control method for a pumping station according to claim 1, characterized in that, The process involves acquiring real-time operating conditions, constructing an initial state vector by combining the preliminary resistance estimate with a preset aging model, calculating the prediction residual using a Kalman filter algorithm, and adjusting the feedback gain to obtain the control deviation. If the control deviation exceeds a preset accuracy threshold, a diffusion trajectory is deduced using a preset transition probability to determine the potential deviation range, including: The resistance level is extracted from the preliminary resistance estimate, and the resistance level is used to match the aging rate in the aging model to determine the wear increment. The initial state vector is constructed by combining the wear increment with the real-time operating conditions. The evolution of the initial state vector on the time axis is calculated using the Kalman filter algorithm to obtain the prediction residual. The control deviation is obtained by adjusting the feedback gain in the observation matrix using the predicted residual and performing recursive correction on the predicted residual in the state space. If the control deviation exceeds the accuracy threshold, the diffusion trajectory of the control deviation is calculated using the transfer probability to determine the initial deviation range; The initial deviation range is nonlinearly mapped to the aging model, and the initial state vector is updated in the iterative loop to determine the potential deviation range.

4. The remote sluice gate opening control method for a pumping station according to claim 1, characterized in that, If the potential deviation range exceeds a preset deviation threshold, the current operating damping is obtained, and resistance records are retrieved from a preset historical database. A deviation vector is calculated by comparison, and the probability distribution of the deviation vector is derived using a Weibull distribution to determine the jamming probability, including: If the potential deviation range exceeds the deviation threshold, the resistance record that matches the flow velocity data is retrieved from the historical database, and the friction features of the corresponding gate opening are extracted. The friction features are matched with historical operating conditions to extract similar trajectories, and the historical reference damping corresponding to the current moment is extracted from the similar trajectories. Calculate the vector difference between the current operating damping and the historical reference damping to obtain the deviation vector; If the deviation vector exceeds a preset vector threshold, the Weibull distribution is used to calculate the blocking frequency of the deviation vector within a specific time window; The probability distribution is obtained by mapping the evolution trend of the blocking frequency in the state space, and the jamming probability is determined.

5. The remote sluice gate opening control method for a pumping station according to claim 1, characterized in that, The process involves extracting force characteristics based on the jamming probability and the current operating damping, inputting these force characteristics into a preset fatigue model to extract stress characteristics, calculating the cumulative loss increment based on the stress characteristics, inputting the cumulative loss increment into a preset degradation function to obtain a risk value, and performing gain reconstruction to generate correction coefficients based on the risk value. Based on the jamming probability and the current operating damping, the force characteristics of the equipment under the current operating conditions are extracted; The stress characteristics are input into the fatigue model to extract the stress features; The cumulative loss increment of the equipment during its operating cycle is determined based on the stress characteristics. The cumulative loss increment is input into the degradation function to calculate the risk value that reflects the degree of performance degradation; Calculate the deviation vector of the risk value in the state space, use the deviation vector to perform the gain reconstruction, and generate the correction coefficient.

6. The remote sluice gate opening control method for a pumping station according to claim 1, characterized in that, The process of acquiring environmental and communication parameters, inputting these parameters and the correction coefficients into a preset mapping model for compensation calculation, and obtaining a preliminary control signal includes: The environmental flow velocity and sediment content in the environmental and communication parameters are obtained, and the environmental flow velocity, sediment content and correction coefficient are input into the preset mapping model to determine the hydrodynamic force compensation value. The head pressure in the environmental and communication parameters is obtained, and the ideal control signal is calculated based on the head pressure and the hydrodynamic force compensation value. Extract the communication delay and signal strength from the environmental and communication parameters, and calculate the instruction transmission lag based on the communication delay and signal strength; The initial control signal is obtained by using the instruction transmission lag to perform time-axis feedforward compensation on the ideal control signal.

7. The remote sluice gate opening control method for a pumping station according to claim 1, characterized in that, The method involves using a preset prediction model to predict the predicted water level and predicted flow rate corresponding to the initial control signal, comparing the predicted water level and predicted flow rate with preset flood control constraints, calculating the safety redundancy, and if the safety redundancy is lower than a preset stability threshold, obtaining the operation log, inputting the operation log into a preset identification model to analyze the balance relationship and correct the damping coefficient to obtain the refined resistance value, including: Obtain the water level warning limit and overflow constraint conditions in the flood control constraints; The preset prediction model is used to predict the predicted water level and the predicted flow rate corresponding to the preliminary control signal under the current operating conditions. The predicted water level is compared with the water level warning limit, and the predicted flow rate is compared with the overflow constraint condition. The safety redundancy is calculated by combining the results. If the safety redundancy is lower than the stability threshold, then current fluctuation characteristics and opening displacement sequence are extracted from the operation log; The current fluctuation characteristics and the opening displacement sequence are input into the identification model to calculate the dynamic balance relationship between the frictional resistance torque and the load inertia torque, and to obtain preliminary identification results.

8. The remote sluice gate opening control method for a pumping station according to claim 7, characterized in that, After inputting the current fluctuation characteristics and the opening displacement sequence into the identification model to calculate the dynamic balance relationship between the frictional resistance torque and the load inertia torque, and obtaining the preliminary identification results, the method further includes: If the drag component in the preliminary identification result exceeds the preset design load range, the initial damping coefficient in the identification model is corrected. The nonlinear friction force distribution is recalculated using the corrected initial damping coefficient; The nonlinear friction force distribution is processed using a multi-round numerical approximation algorithm to obtain the refining resistance value.

9. The remote sluice gate opening control method for a pumping station according to claim 1, characterized in that, The process of acquiring the opening signal and current change, matching and adjusting the opening signal and current change according to the refining resistance value using a preset drive mapping, generating an adjustment command, and executing closed-loop control includes: The driving torque reference is determined based on the refining resistance value, and the corresponding voltage pulse width modulation sequence is retrieved using the driving torque and pulse width modulation duty cycle mapping table in the driving mapping. The voltage pulse width modulation sequence is matched with a preset transmission efficiency mapping table to obtain an initial control pulse with time delay compensation characteristics, and the desired opening trajectory is obtained by parsing the initial control pulse. The positional deviation between the desired opening trajectory and the opening signal is compared. If the positional deviation exceeds a preset synchronization error range, a preset gain adjustment operator is used to perform PID adjustment to generate the adjustment command. The power unit is driven by the adjustment command, and the displacement control increment is determined according to the change in current. The displacement control increment is applied to the actuator and the end position data is fed back in real time to execute the closed-loop control.

10. A remote sluice gate opening control system for a pumping station, characterized in that, include: The resistance estimation module is used to acquire flow velocity data and operating parameters, and to perform segmentation and distance calculation using a preset classifier to obtain a preliminary resistance estimate. The deviation prediction module is used to obtain real-time operating conditions, construct an initial state vector by combining the preliminary resistance estimate with the preset aging model, calculate the prediction residual using the Kalman filter algorithm and adjust the feedback gain to obtain the control deviation. If the control deviation exceeds the preset accuracy threshold, the diffusion trajectory is deduced using the preset transition probability to determine the potential deviation range. The jamming analysis module is used to obtain the current operating damping and retrieve resistance records from the preset historical database if the potential deviation range exceeds the preset deviation threshold, compare and calculate the deviation vector, and use the Weibull distribution to deduce the probability distribution of the deviation vector to determine the jamming probability. The gain reconstruction module is used to extract the force characteristics based on the jamming probability and the current operating damping, input the force characteristics into a preset fatigue model to extract stress characteristics, calculate the cumulative loss increment based on the stress characteristics, input the cumulative loss increment into a preset degradation function to obtain the risk value, and perform gain reconstruction to generate correction coefficients based on the risk value. The compensation calculation module is used to acquire environmental and communication parameters, input the environmental and communication parameters and the correction coefficients into a preset mapping model for compensation calculation, and obtain a preliminary control signal; The resistance refining module is used to predict the predicted water level and predicted flow rate corresponding to the preliminary control signal using a preset prediction model. The predicted water level and predicted flow rate are compared with the preset flood control constraints to calculate the safety redundancy. If the safety redundancy is lower than the preset stability threshold, the operation log is obtained. The operation log is input into the preset identification model to analyze the balance relationship and correct the damping coefficient to obtain the refined resistance value. The closed-loop control module is used to acquire the opening signal and the current change, and to match and adjust the opening signal and the current change according to the refining resistance value using a preset drive mapping, generate adjustment commands and execute closed-loop control.

Images