A regulation method for open-channel gates based on a nonlinear model predictive control algorithm

Abstract

The present invention discloses a regulation method for open channel gates based on a non-linear model predictive control algorithm, which relates to the technical field of water conveyance projects and includes: using the Saint-Venant equations as the prediction model of the non-linear model predictive control system, and constructing a water diversion outlet model and a gate control model; establishing a real-time state correction framework; substituting the corrected state variables into the real-time state correction framework, and combining with the regulation strategy to conduct water level prediction; establishing an optimization model with the minimization of the sum of squares of the measured water level deviation and the number of gate regulations as the objective, and using the zebra optimization algorithm to solve the regulation strategy; adopting a rolling optimization closed-loop control strategy to dynamically correct the water level prediction result according to the difference between the measured value and the predicted value, and execute the regulation strategy. The present invention realizes real-time tracking of flow rate changes, thereby improving the water level prediction accuracy of the model, reducing the number of control times of the canal pool gates at the same time, improving the gate control efficiency, and ensuring the stability of the canal pool water level and flow rate in the open channel irrigation system.

Description

Technical Field

[0001] The present invention relates to the technical field of water conveyance projects, and specifically to a regulating method for open channel gates based on a non - linear model predictive control algorithm. Background Technique

[0002] With the increase in population and the development of agriculture, the demand for agricultural water is constantly growing. Especially in areas where the temporal and spatial distribution of water resources is uneven, ensuring reliable and stable water supply is crucial for promoting the sustainable development of agriculture. However, outdated equipment and improperly designed irrigation water supply systems pose challenges to maintaining the stability of the water level and flow in canal pools, resulting in a decline in the water supply efficiency of the system. Therefore, it is necessary to modernize irrigation canal pools to ensure more stable water levels and flows and improve the safety of agricultural water use. Generally speaking, in order to avoid potential safety hazards caused by large fluctuations in water levels in open - channel irrigation systems (OCIS), manual operation is mostly used to regulate the water level upstream of the gates. However, when complex water conditions occur in the canal pool water network, the performance of manual operation is not good, which makes the automatic control of the canal pool water level the focus of current research. Currently, a variety of automatic control algorithms have been developed, starting from the distributed feedback control algorithm focusing on the proportional - integral - delay model (PID) at the beginning; then developing to the centralized feedback control algorithm focusing on model predictive control (MPC) and linear quadratic regulator (LQR); and later gradually evolving into intelligent control. Among them, MPC has been proven to be suitable for the real - time water level control of irrigation canal pools under complex water diversion changes due to its feedback mechanism and real - time rolling prediction ability.

[0003] MPC refers to a large class of algorithms, usually consisting of an internal prediction model, objective function optimization, constraints, and a rolling optimization strategy. The prediction accuracy of the internal prediction model greatly affects the performance of the control algorithm. In 1871, the Saint-Venant equations (SVEs) were proposed and applied to describe one-dimensional flow in open channels. However, it is difficult to directly couple the SVEs with a feedback mechanism and perform real-time result correction, so they cannot be directly used as the internal prediction model of the control algorithm. Moreover, the numerical model for solving the SVEs using the discrete difference method has a relatively large load, so various simplified models have emerged. There are usually two methods for constructing an internal prediction model suitable for irrigation channels and pools: one is to discretize and linearize the SVEs to construct a linear state vector space, such as the single-input single-output (SISO) model derived from the linearized SVEs; the other modeling method is completely different from the SVEs. It directly uses the dynamic characteristics of the water level and flow rate in the channel and pool to construct a calculation model for the water level-flow rate relationship, such as the integral delay (ID) model and the integral zero-delay (IDZ) model. If the ID model is used as the internal prediction model of MPC and control interval constraints and water level soft constraints are added, the gate adjustment time can be shortened without affecting water level regulation. Although this model has been widely used, it makes many linear assumptions about the physical characteristics of the channel and pool. When the non-linear characteristics of the water level are obvious, it cannot accurately reflect the real physical process. On the other hand, the control actions generated by the adjustment instructions generated by the model controller do not strictly follow the linear relationship with the flow rate, which will cause obvious deviations between the prediction results of the internal prediction model and the future actual situation, resulting in a significant decline in the performance of the predictive control algorithm. Summary of the Invention

[0004] In view of the above problems, the present invention is proposed.

[0005] Therefore, the technical problem to be solved by the present invention is: how to solve the problem that the linear assumptions of the existing internal predictive control algorithm for water level and flow rate lead to a reduction in the accuracy of the prediction model.

[0006] To solve the above technical problem, the present invention provides the following technical solution: An open-channel gate regulation method based on a non-linear model predictive control algorithm, comprising: using the Saint-Venant equations as the prediction model of the non-linear model predictive control system, and constructing a water diversion port model and a gate control model; establishing a real-time state correction framework, and using Kalman filtering to feedback the change in the water diversion port flow rate into the prediction model; substituting the corrected state quantity into the real-time state correction framework, and combining with the regulation strategy to perform water level prediction; establishing an optimization model with the minimization of the sum of the squares of the measured water level deviation and the number of gate regulations as the objective, and using the zebra optimization algorithm to solve the regulation strategy; adopting a rolling optimization closed-loop control strategy to dynamically correct the water level prediction result according to the difference between the measured value and the predicted value, and execute the regulation strategy.

[0007] As a preferred solution of the open channel gate regulation method based on the non - linear model predictive control algorithm of the present invention, wherein: the water diversion port model is obtained based on the flow balance calculation, and the specific formula is as follows,

[0008] Q e =Q f +Q i ;

[0009] Wherein, Q e is the flow rate of the channel section before the water diversion port; Q f is the flow rate of the channel section after the water diversion port; Q i is the flow rate of the water diversion port.

[0010] The gate control model is based on the calculation of the flow - through capacity, specifically obtained according to the continuity equation and the water level - flow rate relationship. The specific formula is:

[0011]

[0012] Wherein, Q is the flow - through capacity of the gate; e is the opening of the gate; b is the width of the gate; Z u is the water level immediately upstream of the gate; Z d is the water level immediately downstream of the gate; C d is the flow - through coefficient; G is the comprehensive flow - through coefficient, which includes the influence of gate operation and hydraulic conditions.

[0013] As a preferred solution of the open channel gate regulation method based on the non - linear model predictive control algorithm of the present invention, wherein: the establishment process of the real - time state correction framework includes the following steps. Obtain the discretized format of the Saint - Venant equation through discretization processing; couple the discretized Saint - Venant equation with the Kalman filter framework. Among them, the coupling includes a prediction process and an update process; use the measured data for state correction. Specifically, assume that there is an uncertain change q in the water diversion flow rate at each water diversion port and correct the state vector.

[0014] Among them, the Saint - Venant equation is the basic equation describing the one - dimensional flow of an open channel, which consists of a continuity equation and a momentum equation. The specific formula is:

[0015]

[0016]

[0017] Wherein, A is the cross - sectional area of the flowing water; Q is the flow velocity; Z is the water level; g is the acceleration due to gravity; C is the Chezy friction coefficient; R is the hydraulic radius; t is the time; x is the position along the flow direction.

[0018] The specific formula of the discretized Saint - Venant equation is:

[0019] α1Qi-1 / 2 +β1Z i +γ1Q i+1 / 2 =δ1;

[0020] α2Z i +β2Q i+1 / 2 +γ2Z i+1 =δ2;

[0021] Wherein, Q and Z respectively represent the flow rate and water level value to be solved at time step t+1, and the subscripts i, i+1 / 2 and i-1 / 2 respectively represent the variables at the staggered grid points i, i+1 / 2 and i-1 / 2; α1, β1, γ1, δ1, α2, β2, γ2, δ2 are variables calculated from the variables Q and Z at time step t.

[0022] Combining the discretized Saint-Venant equations to obtain an equation describing the relationship between the water levels of node i and its adjacent nodes:

[0023] α i Z i-1 +β i Z i +γ i Z i+1 =δ i ;

[0024] Wherein, α i , β i , γ i , δ i are variables calculated from the variables Q and Z at time step t.

[0025] As a preferred scheme of the open channel gate regulation method based on the nonlinear model predictive control algorithm of the present invention, wherein: the formula of the prediction process is,

[0026] X t+1|t =M t X t|t +ω t , ω t ~N(0,W t );

[0027] Wherein, X t+1|t represents the prediction stage vector at time step t+1; M t represents the system model matrix at time step t; X t\t represents the update stage vector at time step t; ω t is the system model error, N(0, W t ) is a normal distribution, 0 is the mean value, and W t is the diagonal covariance matrix, representing the distribution characteristics of the error.

[0028] The formula for the update process is

[0029]

[0030] where X t+1|t+1 represents the stage vector updated by Kalman filtering at time step t+1; K represents the Kalman gain matrix; Y t+1 obs represents the observation vector measured at time step t+1; H represents the observation matrix that maps the observation position to the state vector X.

[0031] After coupling the discretized Saint-Venant equation with the Kalman filtering framework, it is expressed in matrix form as:

[0032]

[0033] where the vector Z t and Z t+1 represent the water level changes at time steps t and t+1 respectively, Z = [Z1, Z2,..., Z Nc T , where N c represents the total number of staggered grid points, and M t Z is the transformed system model matrix.

[0034] As a preferred solution of the open channel gate regulation method based on the nonlinear model predictive control algorithm described in the present invention, wherein: the correction process of the state vector includes the following steps: taking the state vector considering the uncertain diversion flow rate change q as an additional state variable and adding it to the state variables of data assimilation to form the corrected state variables:

[0035] X * = [X, Q] T ;

[0036] where Q = [q1, q2,..., q Nq T , N q is the number of diversion outlets; X * is the corrected state variable.

[0037] Bringing the corrected state quantity into the real-time state correction framework and combining the regulation strategy for water level prediction includes the following steps: bringing the corrected state variable X * into the numerical model to obtain the EnKF framework for water level prediction under uncertain diversion disturbances:

[0038]

[0039] ​​

[0040] Among them,

[0041]

[0042]

[0043] In the formula, X * t+1|t is the predicted phase vector at time step t + 1 after correction; X t+1|t represents the predicted phase vector at time step t + 1; M * t is the new system model matrix calculated using the corrected task; X t\t represents the updated phase vector at time step t; Q * t|t is the process noise covariance formula at time step t after correction; Y t+1 obs represents the observed vector measured at time step t + 1; Y t+1 obs* represents the observed vector measured at time step t + 1 after correction; H represents the observation matrix that maps the observed position to the state vector X; H * represents the corrected observation matrix; K * t+1|t is the new Kalman gain matrix, calculated using the new system predicted state variables, observation variable error covariance, and observation error variance; Q * t|t is the additional state vector at time step t; Q * t+1|t is the additional state vector at time step t + 1.

[0044] Input the regulation strategy to be carried out for water level prediction;

[0045] Correct the water level prediction result. Specifically, when predicting the future water level of the observation point at prediction step t + i based on the water level prediction value, do not directly use the water level prediction value Z t+i|t , but use the difference between the measured water level at step t and the assimilated water level Z t|t to correct the calculated water level. The correction formula is:

[0046]

[0047] Among them, Z f,t+1|t represents the final predicted value of the observation point at time step t + i generated by the internal model according to the calculation result at time step t; Z t+i|tis the predicted water level of the observation point at time step t+i; Z t|t is the assimilated water level of the observation point at time step t; is the measured water level at the observation point at time step t.

[0048] As a preferred solution of the open channel gate control method based on the nonlinear model predictive control algorithm described in the present invention, the optimization model adopts a multi-objective optimization algorithm to solve the optimal strategy, and the objective function is:

[0049]

[0050] Among them, h it and h' it are the calculated water level value and the measured water level value of the i-th water level station at time t, respectively; r and s are the number of regulating gates and the total number of observation points, respectively; n is the number of control gates; A1 and A2 are the sum of weighted water level deviations and the weight of gate control time, respectively.

[0051] The optimization constraint conditions of the optimization model are as follows: the gate flow rate shall not be greater than the maximum allowable gate flow rate; the gate opening shall not be greater than the maximum opening and closing height of the gate.

[0052] The population matrix of the zebra optimization algorithm is:

[0053]

[0054] Where X is the population of gate control schemes; Xi is the i-th gate control scheme; Xi,j is the value of the j-th problem variable proposed by the i-th problem; N is the number of population members; m is the number of decision variables; the objective function is evaluated according to each proposed value of the problem variable; the value obtained by the objective function is represented by a vector:

[0055]

[0056] Where F is the vector of objective function values; F i is the i-th objective function value obtained; when the objective function value is the smallest, it is the optimal candidate solution.

[0057] As a preferred solution of the open channel gate control method based on the nonlinear model predictive control algorithm described in the present invention, a closed-loop rolling optimization strategy is adopted, that is, a neural network performs single-step prediction, uses the data at time t to predict the data at time t+1, and rolls in new data and eliminates old data.

[0058] To solve the above technical problems, the present invention further provides the following technical solutions: An open-channel gate regulation system based on a non-linear model predictive control algorithm, comprising a data acquisition module for collecting real-time water level and flow rate data; a model construction module for constructing a non-linear internal prediction model based on the Saint-Venant equations; a state correction module for feeding the measured data back into the prediction model, using the Kalman filter to perform real-time correction on the state variables, and predicting the water level in combination with the corrected state; an optimization solution module for establishing an optimization model with the goal of minimizing the sum of squares of the measured water level deviation and the number of gate regulations, and using the zebra optimization algorithm to solve for the optimal regulation strategy; and a control execution module for adopting a rolling optimization closed-loop control strategy, dynamically correcting the water level prediction result according to the difference between the measured value and the predicted value, and executing the regulation strategy to control the gate.

[0059] A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that when the processor executes the computer program, the steps of the open-channel gate regulation method based on the non-linear model predictive control algorithm as described above are implemented.

[0060] A computer-readable storage medium, having stored thereon a computer program, characterized in that when the computer program is executed by a processor, the steps of the open-channel gate regulation method based on the non-linear model predictive control algorithm as described above are implemented.

[0061] The beneficial effects of the present invention: The non-linear model predictive controller NM-MPC proposed by the present invention can obtain more accurate water level predictions under flexible water diversion disturbances and can significantly reduce the gate adjustment frequency. In an open-channel irrigation system, when significant non-linear changes occur in the water level of a large irrigation canal pond, whether the water diversion changes are predictable or not, the present invention shows excellent water level prediction advantages, enabling the irrigation water supply system to maintain a stable flow rate and water level, and greatly improving the water supply efficiency of the system. The present invention realizes real-time tracking of flow rate changes, thereby improving the water level prediction accuracy of the model, reducing the number of times of controlling the canal pond gates at the same time, improving the gate control efficiency, and ensuring the stability of the canal pond water level and flow rate in the open-channel irrigation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings without creative efforts based on these drawings.

[0063] Figure 1 FIG. 18 is an overall flowchart of an open-channel gate regulation method based on a non-linear model predictive control algorithm provided by an embodiment of the present invention.

[0064] Figure 2 Schematic diagram of a test channel pool for a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0065] Figure 3 Flow rate change diagram of the 6th water diversion outlet in the test scheme of a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0066] Figure 4 NM-MPC water level prediction result diagram under predictable water diversion for a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0067] Figure 5 NM-MPC water level prediction result diagram under unpredictable water diversion for a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0068] Figure 6 NM-MPC water level prediction result diagram for a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0069] Figure 7 Diagram of a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0070] Figure 8 Diagram of a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0071] Figure 9 Diagram of a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0072] Figure 10 Diagram of a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0073] Figure 11 Diagram of a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0074] Figure 12 Diagram of a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0075] Figure 13 Diagram of a canal gate regulation method based on a nonlinear model predictive control algorithm provided by an embodiment of the present invention.

[0076] Figure 14 This is a diagram of a regulation method for a open channel gate based on a non - linear model predictive control algorithm provided by an embodiment of the present invention. Detailed implementation manners

[0077] To make the above - mentioned objects, features and advantages of the present invention more obvious and understandable, the following will describe the detailed implementation manners of the present invention with reference to the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0078] Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0079] Embodiment 1

[0080] Referring to Figure 1 , an embodiment of the present invention provides a regulation method for an open - channel gate based on a non - linear model predictive control algorithm. When the water demand changes in an open - channel irrigation system, a non - linear predictive controller NM - MPC with a one - dimensional hydrodynamic model as the internal model can more accurately predict the water - level change. Based on the obtained predicted water level, the ZOA algorithm is used to optimize and solve the optimal gate control strategy, and the rolling prediction regulation is used to execute the control strategy, which can greatly improve the gate control efficiency and ensure the stability of the water level and flow rate in the irrigation canal pond.

[0081] The present invention can be summarized into the following three parts:

[0082] 1. Based on the theory of hydraulic numerical simulation, on the basis of analyzing the flow - passing characteristics of the channel, gate and water - dividing outlet, each building is generalized and coupled with the Saint - Venant equation to form a hydrodynamic simulation model that can simulate different incoming water and different gate scheduling strategies.

[0083] 2. Couple the discretized Saint - Venant equation with the Kalman filter framework, and use the measured data to correct the state vector to establish a real - time state correction framework considering the change of the water - dividing flow rate. Then, use the difference between the measured water level and the assimilated water level to correct the predicted water level to avoid the influence of cumulative error on the prediction accuracy.

[0084] 3. Based on the generated regulation prediction model, establish an optimization model, use the ZOA algorithm for optimization and solution, and generate and execute the optimal gate control strategy.

[0085] A method for regulating a sluice gate of an open channel based on a nonlinear model predictive control algorithm provided in this embodiment specifically includes the following steps:

[0086] S1: Take the Saint-Venant equation as the prediction model of the nonlinear model predictive control system, and construct a water diversion outlet model and a gate control model.

[0087] S1.1: The basic equation describing one-dimensional flow in an open channel is the Saint-Venant equation (SVEs), which consists of a continuity equation and a momentum equation, and is expressed as:

[0088]

[0089]

[0090] In the formula, A is the cross-sectional area of the water flow, m 3 ; Q is the flow velocity, m 3 / s; Z is the water level, m; g is the acceleration due to gravity, m / s 2 ; C is the Chezy friction coefficient, m 1 / 2 / s; R is the hydraulic radius, m; t is the time; x is the position along the flow direction.

[0091] S1.2: Construct a water diversion outlet model. The channel pool that diverts water outward is called a water diversion outlet. The water diversion outlet is the channel for the river to supply water to the outside world. The change in the flow rate of the water diversion outlet is an important factor leading to the change in the water level of the channel pool. The modeling of the water diversion outlet model is mainly based on the flow balance calculation formula:

[0092] Q e =Q f +Q i ;

[0093] Among them, Q e is the flow rate of the channel pool section in front of the water diversion outlet, m 3 / s; Q f is the flow rate of the channel pool section behind the water diversion outlet, m 3 / s; Q i is the flow rate of the water diversion outlet, m 3 / s.

[0094] S1.3: Construct a gate control model. In a water conveyance project, a check gate is often arranged in the river channel to regulate the water level and flow rate. The check gate generally adjusts the opening of the gate to regulate the flow rate in the open channel. Based on the flow rate calculation formula, a gate model is established according to the continuity equation and the water level-flow rate relationship:

[0095]

[0096] Among them, Q is the flow rate through the gate, m 3 / s; e is the gate opening, in m; b is the gate width, in m; Z u is the water level immediately upstream of the gate, in m; Z d is the water level immediately downstream of the gate, in m; C d is the flow coefficient, in m 1 / 2 / s. G is the comprehensive flow coefficient, which includes the effects of gate operation and hydraulic conditions.

[0097] Preferably, the one-dimensional hydrodynamic numerical model of the Saint-Venant equations (SVEs) is used as the internal prediction model of the non-linear model prediction control system, which can accurately reflect the non-linear characteristics of the open-channel hydrodynamic process, improve the accuracy of water level prediction, while the traditional linear model often has a large prediction deviation when the water level changes greatly.

[0098] S2: Establish a real-time state correction framework, and use the Kalman filter to feedback the change of the water diversion flow to the prediction model.

[0099] S2.1: Discretize the numerical model.

[0100] First, use the staggered conservative scheme to discretize the hydrodynamic calculation model to obtain the discretized form of the Saint-Venant equations:

[0101] α1Q i-1 / 2 +β1Z i +γ1Q i+1 / 2 =δ1;

[0102] α2Z i +β2Q i+1 / 2 +γ2Z i+1 =δ2;

[0103] where Q and Z respectively represent the flow rate and water level values to be solved at time step t + 1, and the subscripts i, i + 1 / 2, and i - 1 / 2 represent the variables at the staggered grid points i, i + 1 / 2, and i - 1 / 2 respectively. α1, β1, γ1, δ1, α2, β2, γ2, δ2 are variables calculated from the variables Q and Z at time step t.

[0104] Combining the above two equations gives an equation describing the relationship between the water levels of node i and adjacent nodes:

[0105] α i Z i-1 +β i Z i +γ i Z i+1 =δ i ;

[0106] where α i , β i , γ i , δi A variable calculated from variables Q and Z at time step t.

[0107] S2.2: The discrete model coupled Kalman filter framework (i.e., the standard EnKF framework) has two main processes, namely the prediction process and the update process. The prediction process is described by the following formula:

[0108] X t+1|t = M t X t|t + ω t , ω t ~ N(0, W t );

[0109] where, X t+1|t represents the prediction stage vector at time step t + 1; M t represents the system model matrix at time step t; X t\t represents the update stage vector at time step t; ω t is the system model error. N(0, W t ) is a normal distribution, 0 is the mean, and W t is the diagonal covariance matrix, representing the distribution characteristics of the error.

[0110] The update process can be described by the state update equation:

[0111]

[0112] where, X t+1|t+1 represents the update stage vector at time step t + 1 through Kalman filtering; K represents the Kalman gain matrix; Y t+1 obs represents the observed vector measured at time step t + 1; H represents the observation matrix that maps the observation position to the state vector X.

[0113] The discretized Saint-Venant equation is coupled with the standard EnKF and represented in matrix form:

[0114]

[0115] where, the vectors Z t and Z t+1 represent the water level changes at time steps t and t + 1 respectively, Z = [Z1, Z2,..., Z Nc T , where N c represents the total number of staggered grid points, and M t Z is the transformed system model matrix.

[0116] S2.3: Use the measured data to correct the state.​

[0117] Specifically, in the irrigation canal pool, there is uncertainty in the flow process of water intake. The randomness of the water consumption change at the water intake will lead to a deterioration in the prediction effect of the internal prediction model. To cope with sudden uncertain water diversion disturbances, it is assumed that each water diversion point has an uncertain water diversion flow change q, and the state vector is corrected.

[0118] Preferably, by establishing a real-time state correction framework considering the uncertain water diversion flow change, the actual flow change at the water diversion point is fed back into the prediction model through Kalman filtering, improving the response ability to sudden flow disturbances.

[0119] S3: Substitute the corrected state quantity into the numerical model and combine the regulation strategy to predict the water level.

[0120] S3.1: Substitute the corrected state quantity into the numerical model.

[0121] Take the above state vector considering the uncertain water diversion flow change q as an additional state variable and add it to the state variables of data assimilation to form the corrected state variable:

[0122] X * =[X,Q] T ;

[0123] where, Q = [q1, q2, …, q Nq T , N q is the number of water diversion points; X * is the corrected state variable.

[0124] Substitute the corrected state variable X * into the numerical model to obtain the EnKF framework for water level prediction under uncertain water diversion disturbances:

[0125]

[0126]

[0127] where,

[0128]

[0129]

[0130] where, M * t is the new system model matrix calculated using the corrected task. K* t+1|t is the new Kalman gain matrix, which is calculated using the new system predicted state variable, the observation variable error covariance, and the observation error variance. ​

[0131] S3.2: Input the regulation strategy to be carried out and perform water level prediction.

[0132] Specifically, based on the real-time regulation model, corresponding regulation strategies are specified according to the requirements of each canal pond, and the regulation strategies are input into the prediction model to predict the water level changes of the canal pond under different diversion flow rate changes.

[0133] S3.3: Correct the water level prediction results.

[0134] In the present invention, both the control interval and the model update interval under data feedback are set to 10 minutes, and a relatively large spatial interval is adopted to reduce the computational complexity and calculation time. Since there will be modeling errors when using a numerical model with a relatively large spatial interval for prediction, there is a certain degree of error between the assimilation value and the true value. Although the error at each step when calculating the water level state is relatively small, the cumulative effect of the generated errors has a greater impact.

[0135] When predicting the future water level of the observation point at the prediction step t + i based on the water level prediction value of formula , the water level prediction value Z t+i|t is not directly used, but the difference between the measured water level at step t and the assimilated water level Z t|t is used to correct the calculated water level. The correction formula is:

[0136]

[0137] Among them, Z f,t+1|t represents the final prediction value of the observation point at time step t + i generated by the internal model according to the calculation result at time step t; Z t+i|t is the predicted water level of the observation point at time step t + i; Z t|t is the assimilated water level of the observation point at time step t; is the measured water level of the observation point at time step t.

[0138] Preferably, through this gate optimization regulation strategy, the sum of squares of the measured water level deviation and the number of gate regulations are used as a dual-objective optimization, taking into account both the water level control accuracy and the gate regulation efficiency.

[0139] S4: Establish an optimization model with the goal of minimizing the sum of squares of the measured water level deviation and the number of gate regulations, and use the zebra optimization algorithm to solve the optimal regulation strategy.

[0140] S4.1: Establish an optimization model: optimization objective, optimization variables, optimization constraints.

[0141] Specifically, to balance the safety requirements of water level control and the economic considerations of gate control, a multi-objective optimization algorithm is adopted to solve the optimal strategy, with the sum of the squares of the measured water level deviations at each observation point and the minimum total number of gate regulation times as the optimization objectives. The expression of the objective function is:

[0142]

[0143] where h it and h' it are the calculated water level value and the measured water level value at the i-th water level station at time t, respectively; r and s are the number of check gates and the total number of observation points, respectively; n is the number of controlled gates; A1 and A2 are the weights of the sum of weighted water level deviations and the gate control time, respectively. The larger A1 and A2 are, the better the optimization effect is.

[0144] The application area of the present invention is an open-channel pool divided by check gates. In m pool sections, there are 2 variables to be optimized, namely the water level deviation between the measured water level and the predicted water level and the number of gate regulation times.

[0145] The specific optimization constraints are as follows:

[0146] The flow rate through the gate shall not be greater than the maximum allowable flow rate of the gate:

[0147] Q i,j ≤Q i,max ;

[0148] where Q i,j is the flow rate of the i-th check gate in the j-th time period, m 3 / s; Q i,max is the maximum flow rate through the check gate, m 3 / s.

[0149] The opening of the gate shall not be greater than the maximum opening and closing height of the gate:

[0150] e≤h max ;

[0151] where e is the opening of the gate, m; h max is the maximum opening and closing height of the gate, m.

[0152] S4.2: Conduct optimization and solution. When solving the above model, there are multiple decision variables and constraint conditions. There are many large search spaces that do not meet the feasible solutions in the optimization algorithm's search process. Therefore, the selection of the algorithm, the handling method of constraint conditions, etc. play a key role in finding the optimal gate control scheme. The present invention selects a new bionic meta-heuristic algorithm - Zebra Optimization Algorithm (ZOA) to effectively solve this problem.

[0153] In the ZOA algorithm, the position of each zebra in the search space determines the value of the decision variable, and the initial position of the zebra in the search space is randomly assigned. The population matrix of ZOA is as follows:

[0154]

[0155] Among them, X is the population of gate control schemes; Xi is the i-th gate control scheme; Xi,j is the value of the j-th problem variable proposed by the i-th one; N is the number of population members; m is the number of decision variables. The objective function is evaluated according to each proposed value of the problem variable. The value obtained by the objective function is represented by a vector:

[0156]

[0157] Among them, F is the vector of objective function values; F i is the objective function value obtained by the i-th one. By comparing the objective function values, the quality of their corresponding candidate solutions is effectively analyzed, and the best candidate solution for the given problem is identified. This invention is a minimization problem, so the scheme with the smallest objective function value is the optimal candidate solution.

[0158] The optimization process is divided into a foraging stage and a defense stage. The position update formula in the foraging stage is as follows:

[0159]

[0160]

[0161] Among them, is the value of the j-th iteration; is the latest status of the i-th zebra based on the first stage; is the objective function value; PZ is the best member pioneer zebra in the population; PZ j is PZ in the j-th iteration; r is a random number between [0,1]; I = round(1 + rand), rand is a random number between [0,1], so I ∈ [1,2], and the larger I is, the greater the change to the population.

[0162] The position update formula in the defense stage is as follows:

[0163]

[0164]

[0165] Among them, is the j-th dimension value; is the latest status of the i-th zebra based on the second stage; is the objective function value; t is the iteration profile; T is the maximum number of iterations; R = 0.01; Ps represents the probability of randomly selecting one of two solutions generated within the interval [0, 1]; AZ is the state of the attacked zebra, and AZ j is its value after the j-th iteration.

[0166] In each iteration of the ZOA algorithm, the population members are updated based on the foraging stage and the defense stage. During consecutive iterations, the best candidate solutions are updated and saved until the algorithm completely ends. ZOA will use the best candidate solution as the optimal solution to the given problem. The total computational complexity of ZOA is O(N·m·(2c·T)).

[0167] S5: Adopt a closed-loop control strategy of rolling optimization, dynamically correct the water level prediction result according to the difference between the measured value and the predicted value, and execute the optimal regulation strategy.

[0168] The present invention adopts a closed-loop rolling optimization strategy, that is, uses a neural network for single-step prediction, predicts the data at time t + 1 with the data at time t, rolls in new data and eliminates old data. Under this control mode, after each iteration, execute the regulation actions at adjacent times in the strategy generated in step four, and at the next regulation time, repeat the water level prediction process in step two and the optimization process in step five until the optimal water level prediction result and the optimal gate regulation scheme are obtained.

[0169] Preferably, by introducing the Zebra Optimization Algorithm (ZOA), the above multi-objective optimization problem is effectively solved, avoiding falling into local optimal solutions. In addition, a closed-loop control strategy of rolling optimization is adopted, and the prediction result is dynamically corrected according to the difference between the measured water level and the predicted water level, improving the accuracy of long-term prediction.

[0170] In summary, the Nonlinear Model Predictive Controller NM-MPC proposed by the present invention can obtain more accurate water level predictions under flexible water diversion disturbances and can significantly reduce the gate adjustment frequency. In an open-channel irrigation system, when significant nonlinear changes occur in the water level of a large irrigation canal pond, whether the water diversion change is predictable or not, the present invention shows excellent water level prediction advantages, enabling the irrigation water supply system to maintain stable flow and water level, and greatly improving the water supply efficiency of the system.

[0171] Embodiment 2

[0172] Refer to Figures 2 - 3 , which is an embodiment of the present invention, and provides an open-channel gate regulation system based on a nonlinear model predictive control algorithm, including:

[0173] A data acquisition module for collecting real-time water level and flow data;

[0174] A model construction module for constructing a nonlinear internal prediction model based on the Saint-Venant equation;

[0175] A state correction module, which is used to feedback the measured data into the prediction model, perform real-time correction on the state quantity by using Kalman filtering, and predict the water level in combination with the corrected state;

[0176] An optimization solving module, which is used to establish an optimization model with the goal of minimizing the sum of squares of the deviation between the measured water level and the number of gate regulation times, and solve the optimal regulation strategy by using the zebra optimization algorithm;

[0177] A control execution module, which is used to adopt a rolling optimization closed-loop control strategy, dynamically correct the water level prediction result according to the difference between the measured value and the predicted value, and execute the regulation strategy to control the gate.

[0178] Embodiment 3

[0179] This is an embodiment of the present invention. What is different from the previous embodiment is that:

[0180] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that makes contributions to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0181] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0182] More specific examples (nonexhaustive list) of computer-readable media include the following: electrical connections (electronic devices) having one or more wirings, portable computer disk cartridges (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other suitable processing as necessary, and then stored in a computer memory.

[0183] It should be understood that the various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0184] Example 4

[0185] Referring to Figures 2 - 14 , an embodiment of the present invention provides a regulation method for open channel gates based on a nonlinear model predictive control algorithm. To verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0186] As Figure 1 shown, the No. 2 test reach developed by the Canal Automation Algorithm Committee of the American Society of Civil Engineers is selected as the test reach of the present invention. The No. 2 test reach is relatively large, with a total of eight canal pools. Each canal pool is separated by a check gate, and there is a water diversion port at the downstream section of each canal pool. The end of the reach is closed, and the downstream section of the last canal pool is assumed to be a check gate. The flow coefficient of each gate is assumed to be 0.8, and the dead zone is 5 mm. The remaining relevant parameter indicators are shown in Table 1:

[0187] Table 1 Parameters of the ASCE No. 2 Test Canal in the Study

[0188]

[0189]

[0190] The implementation process of this experiment is as follows:

[0191] Step 1: Set the control variables and research scenarios of the research channel pool.

[0192] The initial flow rate at the most downstream is 3 m 3 / s, and the initial flow rate at the water inlet is 11 m 3 / s. The initial flow rate at the water diversion outlet of each channel pool is 1 m 3 / s. At the 5h moment, the flow rate at the water diversion outlet of channel pool 6 increases from 1 m 3 / s to 3 m 3 / s, and the flow rates at the other water diversion outlets always remain 1 m 3 / s. The specific changes in the water conveyance flow rate and the flow rate at the water diversion outlet of the channel pool are as shown in Figure 3 shown.

[0193] The research scenarios are set as follows: the case where water diversion is predictable (i.e., the control model can collect information on changes in downstream water demand in advance) and the case where water diversion is unpredictable (i.e., the control model cannot collect information on changes in downstream water demand), and the ability of the proposed control method to cope with various downstream water demand changes is evaluated by comparison.

[0194] Step 2: Configure the controller and establish the optimization algorithm.

[0195] Take NM-MPC as the control algorithm of the research case simulation model, and obtain the model characteristic parameters related to the research case through the hydrodynamic simulation model in advance. The specific parameter values are shown in Table 2.

[0196] Table 2 Model characteristic parameters

[0197]

[0198]

[0199] Set the calculation time interval of the internal prediction model in the NM-MPC model to 3 min, the number of calculation nodes to 45, the regulation interval and the model update interval under data feedback to 10 min, and the optimization solution algorithm to the E-WOA algorithm.

[0200] The implementation results of this experiment are as follows: The case tests the prediction ability of NM-MPC for the water level change at the upstream observation point of the gate when the flow rate at the downstream water diversion outlet changes. In addition, the influence of gate regulation on the water level change is analyzed to obtain the optimal gate regulation scheme through the MPC controller.

[0201] 1. Analysis of NM-MPC prediction performance.

[0202] (1) Prediction performance when the flow rate at the water diversion outlet changes.

[0203] The predicted start times are 300 min, 330 min, 360 min, 420 min, and 480 min respectively. The water level prediction results of the water level observation points downstream of the 6th pool under the two scenarios are as Figure 4 and Figure 5 shown, where Figure 4 is the regulation result under the condition that the water diversion disturbance is predictable. When the change of the water diversion flow is predictable, no matter when the prediction starts, the water level prediction result of the NM-MPC model for the observation point is always close to the measured value, showing good prediction performance. Figure 5 is the regulation result under the condition that the water diversion disturbance is unpredictable. When the change of the water diversion flow is unpredictable, at 300 min, the actual water level of the canal pool remains unchanged, and the NM-MPC cannot predict the change of the water level; at 330 min, the change of the flow rate at the water diversion outlet causes a sudden change in the water level, and the NM-MPC starts to predict the change of the water level of the observation point, but the NM-MPC needs to spend a certain amount of time for assimilation and correction, so the water level prediction accuracy is relatively low at this time; at 360 min, the NM-MPC has completed the water level correction, and the water level prediction result of the observation point is close to the measured value.

[0204] (2) Prediction performance when the flow rate at the water diversion outlet changes.

[0205] Manually adjust the opening of the 7th gate to reduce the outlet flow rate of the 6th canal pool by 2 m 3 / s at 300 min. The adjustment amplitude is calculated inversely according to the overflow formula of the 7th gate, and the opening of the other check gates remains unchanged. Figure 6 shows the water level prediction of the water level observation point downstream of the 6th canal pool. The results show that no matter when the starting time of the water level prediction is, the water level prediction effect of the NM-MPC is excellent. When the opening of the check gate changes, the flow rate of the canal pool does not change stably due to the change of the gate opening. The change of the gate water level and the flow rate change show a non-linear relationship, but the NM-MPC is a non-linear model and shows excellent performance when the water level changes.

[0206] 2. Control results of the MPC controller.

[0207] Use the deviation value obtained by subtracting the initial value from the simulated values of the water level and the gate opening to represent the change of the water level in each canal pool and the control process of the check gate. As Figures 7 - 10 shown, when the water diversion flow is predictable, the maximum water level deviations of the 5th and 6th canal pools predicted by the NM-MPC are 0.035 m and 0.034 m respectively, and there is almost no gate callback. This shows that the NM-MPC can maintain a more stable water level trend during the control process, thus obtaining a more accurate predicted water level and generating a more reasonable gate control strategy.

[0208] As Figures 11 - 14As shown, when the water diversion flow rate is unpredictable, the maximum water level deviations of Canal Basin 5 and Canal Basin 6 predicted by NM-MPC are 0.083 m and 0.092 m respectively, and the number of gate adjustments is 30 times. Although the gate opening adjustment of NM-MPC is relatively conservative when the downstream water diversion changes, over time, the internal prediction model of NM-MPC gradually adjusts the model-calculated water level using the measured data. With the completion of the flow rate correction, the generated gate control strategy gradually becomes reasonable.

[0209] From the results of this experimental case, in NM-MPC, the prediction accuracy of NM-MPC is greatly improved by adding flow rate correction during the assimilation process, and it can well solve the water level control problem caused by sudden or unpredictable changes in water demand. Under the same water diversion disturbance conditions, NM-MPC can obtain a better gate control strategy, greatly reducing the number of gate control times and significantly improving the gate control efficiency.

[0210] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An open channel gate control method based on a nonlinear model predictive control algorithm, characterized in that: include: The Saint-Venant equation is used as the prediction model of the nonlinear model predictive control system, and the water diversion model and gate control model are constructed; Establish a real-time state correction framework and use Kalman filtering to feed back the flow changes of the water diversion outlet into the prediction model; Bring the corrected state quantity into the real-time state correction framework, and perform water level prediction in combination with the control strategy; An optimization model is established with the goal of minimizing the sum of squares of measured water level deviations and the number of gate regulation times, and the zebra optimization algorithm is used to solve the regulation strategy. Adopting the closed-loop control strategy of rolling optimization, the water level prediction results are dynamically corrected according to the difference between the measured value and the predicted value, and the control strategy is implemented; The process of establishing the real-time status correction framework includes the following steps: The discretization format of Saint-Venant equation is obtained through discretization processing; The discretized Saint-Venant equation is coupled with the Kalman filter framework; wherein the coupling includes a prediction process and an update process; The measured data is used to correct the state, specifically, assuming that each water diversion outlet has an uncertain water diversion flow change q, and the state vector is corrected; The discretized Saint-Venant equation is coupled with the Kalman filter framework and expressed in matrix form as follows: Among them, the vector Z t and Z t+1 They represent the water level changes at time steps t and t+1, respectively, Z = [Z1, Z2, …, Z Nc ] T , where N c represents the total number of staggered grid points, M t Z is the transformed system model matrix.

2. The open channel gate control method based on the nonlinear model predictive control algorithm according to claim 1, characterized in that: The water diversion model is calculated based on flow balance, and the specific formula is as follows: Q e =Q f +Q i ; Among them, Q e Q is the flow rate of the channel section before the water diversion outlet; f is the flow rate of the channel section after the diversion outlet; Q i is the diversion outlet flow rate; The gate control model is based on flow calculation, which is specifically obtained according to the continuity equation and the water level-flow relationship. The specific formula is: Among them, Q is the gate flow rate; e is the gate opening; b is the gate width; Z u Z is the water level just upstream of the gate; d is the water level immediately downstream of the gate; C d is the flow coefficient; G is the comprehensive flow coefficient, which includes the influence of gate operation and hydraulic conditions.

3. The open channel gate control method based on the nonlinear model predictive control algorithm according to claim 2, characterized in that: The Saint-Venant equation is a basic equation for describing one-dimensional flow in an open channel, which consists of the continuity equation and the momentum equation. The specific formula is: Among them, A is the cross-sectional area of ​​the water flow; Q is the flow velocity; Z is the water level; g is the acceleration of gravity; C is the friction coefficient of the friction; R is the hydraulic radius; t is the time; x is the position along the flow direction; The specific formula of the discretized Saint-Venant equation is: a1Q i-1 / 2 +β1Z i +γ1Q i+1 / 2 =δ1; a2Z i +β2Q i+1 / 2 +γ2Z i+1 =δ2; Where Q and Z represent the flow and water level values ​​to be solved at time step t+1, respectively; subscripts i, i+1 / 2 and i-1 / 2 represent the variables at staggered grid points i, i+1 / 2 and i-1 / 2, respectively; α1, β1, γ1, δ1, α2, β2, γ2, δ2 are the variables calculated from variables Q and Z at time step t; Combining the discretized Saint-Venant equations, we get the equation describing the relationship between the water levels of node i and its neighboring nodes: α i WITH i-1 +β i WITH i +γ i WITH i+1 =δ i ; Among them, α i , β i , γ i , δ i is the variable calculated from variables Q and Z at time step t.

4. The open channel gate control method based on the nonlinear model predictive control algorithm according to claim 3, characterized in that: The formula for the prediction process is, X t+1|t =M t X t|t +oh t ,oh t ~N(0,W t ); Among them, X t+1|t represents the prediction phase vector when the time step is t+1; M t represents the system model matrix when the time step is t; X t\t represents the update phase vector when the time step is t; ω t is the system model error; N(0, W t ) is a normal distribution, 0 is the mean, W t is the diagonal covariance matrix, which represents the distribution characteristics of the error; The formula for the update process is, Among them, X t+1|t+1 represents the phase vector updated by the Kalman filter at time step t+1; K represents the Kalman gain matrix; Y t+1 obs represents the observation vector measured at time step t+1; H represents the observation matrix that maps the observation position to the state vector X.

5. The open channel gate control method based on the nonlinear model predictive control algorithm according to claim 4, characterized in that: The correction process of the state vector comprises the following steps: The state vector considering the uncertain water diversion flow change q is added as an additional state variable to the state variable of data assimilation to form the modified state variable: X * =[X,Q] T ; Where Q = [q1,q2,…,q Nq ] T , N q is the number of water outlets; X is the state variable; X * is the modified state variable; The method of bringing the corrected state quantity into the real-time state correction framework and performing water level prediction in combination with the control strategy includes the following steps: The corrected state variable X * Bringing it into the numerical model, we get the EnKF framework for water level prediction under uncertain water diversion disturbance: in, Where, X * t+1|t is the prediction phase vector at the corrected time step t+1; X t+1|t represents the prediction phase vector when the time step is t+1; M * t is the new system model matrix calculated using the revised task; X t\t represents the update phase vector at time step t; Q * t|t is the process noise covariance formula when the time step is t after correction; Y t+1 obs represents the observation vector measured at time step t+1; Y t+1 obs* represents the observation vector measured at the corrected time step t+1; H represents the observation matrix that maps the observation position to the state vector X; H * represents the corrected observation matrix; K * t+1|t is the new Kalman gain matrix, which is calculated using the new system prediction state variables, observation variable error covariance, and observation error variance; Q * t|t is the additional state vector at time step t; Q * t+1|t is the additional state vector at time step t+1; Input the control strategy to be implemented and make water level prediction; The water level prediction results are corrected. When the water level prediction value Z is used to predict the future water level of the observation point with step length t+i, the water level prediction value Z is not directly used. t+i|t , but uses the measured water level at step length t Assimilation water level Z t|t The difference between them is used to correct the calculated water level. The correction formula is: Among them, Z f,t+1|t It represents the final prediction value of the observation point at time step t+i generated by the internal model based on the calculation result of time step t; Z t+i|t is the predicted water level of the observation point at time step t+i; Z t|t is the assimilated water level of the observation point at time step t; is the measured water level at the observation point at time step t.

6. The open channel gate control method based on the nonlinear model predictive control algorithm according to claim 5, characterized in that: The optimization model uses a multi-objective optimization algorithm to solve the optimal strategy, and the objective function is: Among them, h it and h' it are the calculated water level value and the measured water level value of the i-th water level station at time t; r and s are the number of control gates and the total number of observation points, respectively; n is the number of control gates; A1 and A2 are the sum of weighted water level deviations and the weight of gate control time, respectively; The optimization constraints of the optimization model are: The gate flow rate shall not be greater than the gate's maximum allowable flow rate; The gate opening shall not be greater than the maximum opening and closing height of the gate; The population matrix of the zebra optimization algorithm is: Where X is the population of gate control schemes; Xi is the i-th gate control scheme; Xi,j is the value of the j-th problem variable proposed by the i-th problem; N is the number of population members; m is the number of decision variables; the objective function is evaluated according to each proposed value of the problem variable; the value obtained by the objective function is represented by a vector: Where F is the vector of objective function values; F i is the i-th objective function value obtained; when the objective function value is the smallest, it is the optimal candidate solution.

7. The open channel gate control method based on the nonlinear model predictive control algorithm according to claim 6, characterized in that: A closed-loop rolling optimization strategy is adopted, that is, the neural network performs single-step prediction, uses the data at time t to predict the data at time t+1, and rolls in new data and eliminates old data.

8. A system using the open channel gate control method based on the nonlinear model predictive control algorithm as claimed in any one of claims 1 to 7, characterized in that: include: Data acquisition module, used to collect real-time water level and flow data; A model building module for building a nonlinear internal prediction model based on the Saint-Venant equation; The state correction module is used to feed back the measured data to the prediction model, use Kalman filtering to correct the state quantity in real time, and predict the water level based on the corrected state; The optimization solution module is used to establish an optimization model with the goal of minimizing the sum of squares of measured water level deviations and the number of gate regulation times, and uses the zebra optimization algorithm to solve the optimal regulation strategy; The control execution module is used to adopt a rolling optimization closed-loop control strategy, dynamically correct the water level prediction result according to the difference between the measured value and the predicted value, and execute the regulation strategy to control the gate.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the open channel gate control method based on the nonlinear model predictive control algorithm described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the open channel gate control method based on a nonlinear model predictive control algorithm according to any one of claims 1 to 7 are implemented.

Images