Intelligent water network real-time decision control system and instruction generation method

Abstract

The application discloses a real-time decision control system and a command generation method for an intelligent water network, and the command generation method comprises the following steps: acquiring real-time monitoring data of the water network system; providing a data processing system for the monitoring station, removing interference and noise from the real-time monitoring data, and obtaining current water network state monitoring data sets consistent with the data by fuzzy processing to form instant feedback and establish a prediction control model of the water network flow state; calculating the deviation between the real-time monitoring information and the pre-set target, inputting the deviation into the prediction control model, and obtaining a joint automatic fuzzy control command of the gate group. The application can acquire the state of the water network project in real time and accurately, solve the problem of automatic control of strong coupling, nonlinearity, large time delay and other obstacles in the water supply process, improve the timeliness of the gate group control, and improve the stability of the water network control system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of water conservancy automation and smart water conservancy, in particular to an intelligent water network real-time decision control system and a command generation method. BACKGROUND

[0002] Generally, the complex water supply network artificially built on the basis of natural river and lake system, which has the functions of water diversion, water distribution, water use and stable operation, and serves multiple objects, is called water network, and the gate is the most important hydraulic facility for regulating the operation state of the water network. The so-called intelligent water network is to realize the unattended automatic operation of the water network engineering group through the joint automatic control of various types of gates. However, in the process of regulating the flow of the water network channel, the upper and lower gates are linked and restricted, which makes it difficult to realize real-time and collaborative control of the water network gate group.

[0003] Since the 1950s, some countries in Europe and the United States have taken the lead in carrying out automatic research on canal systems, and some gate-controlled water network channel systems have achieved autonomous operation without human management, for example, the California water transfer project and the Colorado River water transfer project in the United States use new technologies such as remote transmission and remote control to control the regulating gates of the water supply network. The automatic construction of computer monitoring, water regime measurement and safety monitoring of the Luan-Tian water diversion project in China began in the 1980s. After decades of development, the intelligent management level of water network engineering is still not high. Not only is the data quality poor, but the measurement accuracy and reliability also need to be improved. The traditional point monitoring method cannot meet the real-time control of water network flow state, especially for complex water networks with multiple water supply objects and multiple water supply targets that change over time. The time lag in the water transfer process has a significant impact on control timeliness. At present, the mechanism model driven control decision model not only has low precision and poor stability, but also lacks guidance control for the characteristics of complex water network engineering such as complex hydraulic fluctuations, strong coupling and large time lag. Moreover, the measurement error, simulation error and control error in the control process will form accumulation and superposition, which will continuously affect the control of the non-steady-state system, cause deviation in the formulation of steady state, and make it impossible to perform stable control. SUMMARY

[0004] In view of the above shortcomings in the prior art, the intelligent water network real-time decision control system and the command generation method provided by the present application solve the problems of low control precision, lack of guidance control and inability to perform stable control for the characteristics of complex water network engineering such as complex hydraulic fluctuations, strong coupling and large time lag.

[0005] In order to achieve the above-mentioned application purposes, the technical scheme adopted by the present application is as follows: an intelligent water network real-time decision control system, comprising a hydrological monitoring module, a data processing module, a prediction control module, a prediction-fuzzy control module and a fuzzy control module.

[0006] hydrological monitoring module, for obtaining real-time monitoring data of the water network system according to existing water network engineering layout and various types of hydrological monitoring sites arranged;

[0007] The data processing module is used for removing interference and noise from the real-time monitoring data and obtaining current water network state monitoring data set consistent with the data by fuzzy processing, so as to form instant feedback.

[0008] The prediction control module is used for establishing a prediction control mode of water flow state of the water network according to the instant feedback result, the set water network water supply, the water volume target of the water receiving side and the expected water supply process path.

[0009] The prediction-fuzzy control module is used for obtaining an adjusted prediction-fuzzy control model by using a control effect fuzzy evaluation and feedback correction method according to the established prediction control mode.

[0010] The fuzzy control module is used for completing joint automatic fuzzy control by using a neural network based on a state space model according to the water network prediction fuzzy control model.

[0011] The intelligent water network real-time decision control system instruction generation method comprises the following steps:

[0012] S1, obtaining real-time monitoring data of the water network system according to existing water network engineering layout and various types of hydrological monitoring sites arranged;

[0013] S2, removing interference and noise from the real-time monitoring data by using a system equipped with data processing at the monitoring sites, and obtaining current water network state monitoring data set consistent with the data by fuzzy processing, so as to form instant feedback.

[0014] S3, establishing a prediction control mode of water flow state of the water network according to the instant feedback result, the set water network water supply, the water volume target of the water receiving side and the expected water supply process path.

[0015] S4, obtaining a water network prediction fuzzy control model by using a control effect fuzzy evaluation and feedback correction method according to the established prediction control mode.

[0016] S5, completing joint automatic fuzzy control by using a neural network based on a state space model according to the water network prediction fuzzy control model.

[0017] The intelligent water network real-time decision control system instruction generation method comprises the following steps: The present application adds the processes of fuzzy sampling control of measured data and self-disturbance filtering, can ensure real-time and accurate acquisition of the state of the water network engineering, processes the problems of strong coupling, nonlinearity, large time lag and other disturbances in automatic control of the water supply process, realizes more automatic mechanism control, ensures system stability and response speed, and improves the timeliness and refinement degree of control. Attached Figure Description

[0018] Figure 1 This is a flowchart of the present invention;

[0019] Figure 2 The application process of fuzzy sampling for real-time monitoring data;

[0020] Figure 3 The mechanism for removing disturbances using the Kalman-Active Disturbance Rejection Filtering method. Detailed Implementation

[0021] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0022] like Figure 1 As shown, the intelligent water network real-time decision control system and command generation method include the following steps:

[0023] S1. Based on the existing water network engineering layout and the various hydrological monitoring stations deployed, obtain real-time monitoring data of the water network system;

[0024] S2. Equip the monitoring stations with a data processing system to remove interference and noise from the real-time monitoring data and perform fuzzing processing to obtain a current water network status monitoring dataset that is consistent with the data, thereby generating immediate feedback.

[0025] S3. Based on the results of real-time feedback, the set water supply network, the water volume target on the receiving side, and the expected water supply process path, establish a predictive control mode for the water flow status of the water network.

[0026] S4. Based on the established predictive control model, the fuzzy evaluation and feedback correction method of control effect is adopted to obtain the water network predictive fuzzy control model.

[0027] S5. Based on the water network predictive fuzzy control model, use a state-space model-based neural network to complete joint automatic fuzzy control.

[0028] The specific implementation method of step S1 is as follows:

[0029] S1-1. Divide the water network into two levels: water network channel river sections and nodes;

[0030] S1-2. Using the discrete Saint-Venant equations, 2(N-1) algebraic equations for the water network channels are obtained. Based on the boundary conditions of the water network nodes, two more algebraic equations for the boundary conditions are added, resulting in a total of 2N nonlinear algebraic equations. The discrete Saint-Venant equations are shown below:

[0031]

[0032]

[0033] In the formula, a 1i =1, c 1i =1,

[0034]

[0035]

[0036]

[0037]

[0038] After eliminating variables from the discretized equations, a double-pursuit equation is obtained. The linear equation system is then solved using the pursuit method to calculate the h′ water level, Q′ flow rate, and gate opening e of the existing water network project. 1i The relationship is as follows: j represents the cycle step, i represents the node, and a, b, c, and d represent the conversion factors for water level and flow rate to opening degree; θ is the angle between the tangent of the lower edge of the arc gate and the horizontal direction; g is the acceleration due to gravity; B is the corresponding gate; A is the cross-sectional area of ​​the water passage; M represents the design condition; Δt represents the measurement time; Δx i This represents the change in spatial coordinates;

[0039] S1-3. Optimize the spatial layout of monitoring stations based on the existing hydrodynamic element relationships of the water network. The existing hydrodynamic element relationships of the water network include the storage capacity of each channel of the water network, the design water level and flow rate, the top elevation of the dike, the operating mode adopted by each channel of the water network, the upstream normal water level, the downstream normal water level, equal volume operation and storage balance control.

[0040] S1-4. Based on the water network operation mode, determine the water level, flow rate, sensor type, layout method and installation location, and obtain real-time monitoring data of the water network system.

[0041] The specific implementation method of step S2 is as follows:

[0042] S2-1. The continuous information from the monitoring station is instantaneously sampled with a period TH, and the sampled value is used as the external input signal W;

[0043] S2-2. Convert the external input signal W into an input signal Y suitable for the sampler S, and input it into the sampler S to obtain discrete state values ​​Y with a periodic pattern. d ;

[0044] S2-3, The periodic discrete state values ​​Y d Input fuzzy controller K d The discrete signal U is obtained. d ;

[0045] S2-4, Discrete signal U d The control input U is formed by a zero-order hold;

[0046] S2-5. According to the formula:

[0047]

[0048] C n-1 (t)=S(t)

[0049]

[0050]

[0051] The output value of the state variables C(t) of the discretized system model after removing disturbances and noise is obtained; where V represents the number of nodes that cause coupling disturbances to the water flow; li represents the li-th node; Indicates the parameters of the Karman filter. and Indicates the numerical value of the filter factor; C n-1 (t), C n (t), C n+1 (t) represents the state variable obtained during the application of the Active Disturbance Rejection Algorithm, T ab tj represents the inertial time constant of the water flow in the water network; x ,tj g ,tj y The constant m represents the self-regulating coefficient for eliminating fluctuations in the water flow of a small disturbance network. df This indicates large fluctuations. The control input U is converted into a real-time input signal S(t), where S1(t-τ) is the approximate input extracted by the differential tracker, and S2(t) is the differential signal. and Used to measure the relative deviation between the output current water level and flow rate of the water network channel and the actual value; t represents the measurement time point; e qy and e qh Both represent the self-adjustment coefficient; E li F represents the energy fluctuation caused by the coupling disturbance effect of water flow; li T represents the dynamic force that generates the coupling disturbance effect of water flow. we represents the water flow time delay constant of the water network channel; h T is the self-adjusting coefficient for the sampling period; y e is the time delay constant of the self-adjustment coefficient; y This is the self-adjusting coefficient; For the state values ​​and differential signals in the time interval from n-1 to n; For the state values ​​and differential signals in the time interval from n to n+1; For the state values ​​and differential signals in the time interval from n-1 to n+1;

[0052] S2-6. The output value C(t) of the state variable of the discretized system model after removing disturbances and noise is used as the real-time status of the station detection to form instant feedback. The instant feedback includes all parameters related to water.

[0053] The specific implementation method of step S3 is as follows:

[0054] S3-1. According to the formula:

[0055]

[0056]

[0057]

[0058] The fuzzy control plan P is obtained JC ;in, This represents the weight of the water supply plan in group i2, and satisfies... Sim ya (i2, j2) represents the simulation plans for various prediction methods in historical data, Q sj (j2) represents the corresponding actual water supply; This represents the prediction of the water supply simulation for the i2th group; m indicates that there are m prediction simulation methods; S represents the sliding surface;

[0059] S3-2, According to the fuzzy control plan P JC This creates fuzzy control.

[0060] The specific implementation method of step S4 is as follows:

[0061] S4-1. According to the formula:

[0062]

[0063]

[0064] The water network channel control model, and the functions of upstream and downstream water depth relative to flow rate are obtained. in, The deviation between the water level at the control point at time t and the steady-state water level e′ in the river segment i3 (i3=1,2,3) corresponding to the water diversion channel; This represents the area of ​​the return water zone of the i3th water intake channel; These represent the deviations of the inflow, outflow, and intake flow rates of the i3th water intake channel from the steady state, respectively. The flow time delay corresponds to the i-th water diversion channel; L is the weir length or gate width associated with the gate; k a k b h is the reduction factor. m h1 and h2 represent the water depths upstream and downstream of the gate, respectively; ω represents the gate opening.

[0065] S4-2. Based on the water network channel control model, the current state of the water network is obtained as a function of upstream water depth and downstream water depth relative to the flow rate.

[0066] S4-3, According to the formula:

[0067]

[0068] The steady-state control conditions s of the water network system are obtained. e (t); where W -1 ω1 represents the inverse matrix representing the state space of the water supply network project; ω1 represents the time variable that measures the current state input of the water network. Represents the state variables of the water network system; ∈1, ∈2, and ku are the setting parameters of the sliding surface; This indicates the bounded range of the external disturbance; sat(·) represents the state of the sliding surface; This is a matrix representing the state space of the water supply network project, obtained from the time variables input from the state of the water network.

[0069] S4-4. According to the formula:

[0070]

[0071] The output u of the control decision for the steady-state water network is obtained. i (t); where, It represents the bounded range of the external disturbance; sign(·) is the sign function, which indicates the sign of (·);

[0072] S4-5. According to the formula:

[0073]

[0074] The adjusted predictive fuzzy control model is obtained; where z(t) is the final state of the time-delay control. This is the output of time-delay control; All are system matrices of the i5th subsystem; Represents the membership degree of the antecedent variable in set N; ζ(t) = [ζ1(t), ... ζ p [(t)] represents the antecedent variable; N is the fuzzy set; r is the prediction simulation method; x(t-τ) is the approximate input extracted by the differential tracker corresponding to the original data obtained from the monitoring station; x(t) represents the original data obtained from the monitoring station; u(t) represents the output of the control decision of the steady-state water network; ω(t) is the time variable that measures the current state input of the water network.

[0075] The specific implementation method of step S5 is as follows:

[0076] S5-1. According to the formula:

[0077]

[0078] The water level and flow rate constraint equations are obtained; where Q g For the flow rate through the gate, C d Δh is the gate orifice flow coefficient; u' is the gate opening; b1 is the gate orifice width; Δh is the water level difference between upstream and downstream of the gate.

[0079] S5-2, According to the formula:

[0080] L 2 δx(k+1)=R 2 δx(k)+Wδq(k)

[0081]

[0082]

[0083]

[0084]

[0085] The matrix solution space of the water level and flow rate constraint equations is obtained; where, L 2 R represents the coefficient matrix representing the influence of predicted conditions on water level and flow rate based on the control and operation of water network channels. 2 The matrix represents the coefficients that influence water level and flow rate under the current water network channel control operation. W′ represents the comprehensive influence coefficient matrix of lateral outflow on the water network state under time-delay control. [·] T Z represents the transpose matrix; δq(k) represents the relative steady-state outflow deviation of water intake on both sides of the water network channel; δx(k) represents the deviation of the raw data obtained by the monitoring station at time k; δx(k+1) represents the deviation of the raw data obtained by the monitoring station at time k+1; j+2 Z represents the water level at the (j+2)th cycle step. j+1Let ε be the water level at the (j+1)th cycle step; ε be the differential error. For the incoming traffic to the node; δQ p This represents the current traffic value of the node.

[0086] S5-3, According to the formula:

[0087]

[0088] Obtain a predictive model x representing the state of the water network. m (k+i6); where J represents the number of basis functions; This represents the linear weighting coefficients obtained from the basis function optimization calculation; The value of the basis function at a given time point within the sampling period is represented by j3; j3 represents the j3rd basis function; i6 represents the number of responses; G m Represents the prediction function; This represents the form of the prediction function in the i6-1 time period; This represents the form of the prediction function in the i6-2 time period; H m The control function represents the prediction function;

[0089] S5-4. According to the formula:

[0090]

[0091]

[0092]

[0093] This involves decoupling the control gates of the water network channel from the upstream gates controlling the incoming water, decoupling the upstream gates controlling the incoming water from the channel / pipe head gates, and decoupling the output for flow rate; among these, This indicates the opening degree of the i7th control gate in the upstream section of the river / pipeline; This indicates the water distribution volume at the corresponding i7th gate outlet. This indicates the deviation between the water level in front of the gate at water distribution point i7 in the previous time period and the corresponding water supply target setting, where k represents the time period; KD0 and KD1 represent the proportional and integral coefficients of the PID controller for the upstream gate of the i7th river segment, respectively, with 0 representing the coefficient of the water diversion gate at the water diversion channel; These are the gate opening and the decoupling parameters for decoupling the flow output, respectively. This indicates the deviation between the water level at the diversion point i7 at time k and the corresponding water supply target setting; KP0 and KI0 represent the proportional coefficient and integral coefficient of the PID controller at the water intake gate of the water intake channel.

[0094] S5-5, According to the formula:

[0095] x'(t)=e At x(0)+∫0 t e A(t-τ) Bu(τ)dτ

[0096] y'(t)=Ce At x(0)+C∫0 t e A(t-τ) Bu(τ)dτ+Du(t)

[0097] The solutions to the state equations and the output equations in the state-space representation are obtained; where e At x(0) is a homogeneous solution in the matrix solution space of the water level and flow limit equations; Ce is the product of the Laplace transform of the matrix solution space of the water level and flow rate constraint equations and the Laplace transform of the input; At x(0) is the zero-input response; Du(t) is the zero-state response;

[0098] S5-6. Based on the output of the adjusted predictive fuzzy control model, the solution of the state equation, and the solution of the output equation, the control information of the opening degree of each gate, the water level control information of the upstream water network, and the flow control information of the downstream water network are used to control the opening degree of each gate, thereby completing the joint automatic fuzzy control.

[0099] like Figure 2 As shown, the sampling process for the current water level and flow signals of the system collects data over a weighted time period TH, obtaining instantaneous values ​​of continuous information from the monitoring station. This initial monitoring data, including interference, noise, and water network channel system operation commands, is used as the external input signal W. Then, a regulating system, along with corresponding samplers S and fuzzy controllers K, are added. d A zero-order hold H forms a fuzzy system for data acquisition and preprocessing. It converts the external input signal W into a sampling input signal Y suitable for sampling operation control. Then, through the combined use of water level and flow monitoring equipment and a sampler, it obtains periodic discrete state values ​​Y from continuous real-time changes in water level and flow parameters. d The discrete state values ​​are then further processed by a discrete-time fuzzy controller to form a discrete signal U. d The input is then passed through a zero-order hold to form the control input U, and then through an automatic processing system to obtain the corresponding output Z.

[0100] like Figure 3As shown, the accurate water level information is obtained by removing disturbances and noise using a Kalman-active disturbance rejection filter mechanism. S is the real-time input signal, S1 is the approximate input extracted by the differential tracker, and S2 is the differential signal; δe1 and δe2 are both error signals; KS1, KS2, and KS3 are the state observation signals of the extended state controller; u0 is the input control gain, u is the controller input, and bc is the feedback compensation factor; m df This indicates large fluctuations in load; C is the relative deviation between the current water level and flow rate of the water network channel and the steady state.

[0101] In one embodiment of the present invention, the sampling period TH is calculated as follows:

[0102]

[0103] Where ω and β represent variables that measure the frequency of water level and flow fluctuations within the system, ρ(β,t) represents the periodicity of the influence of water flow fluctuations on the system state, and Γ represents the fractional integral operator.

[0104] Based on the inherent uncertainty of water flow, for a specific random variable Z and confidence level α, the expected maximum negative benefit YQ(X) is obtained; where YQ represents the moment matching value of the distribution of accumulated future rewards for each action. σ represents the α quantile of the standard Gaussian distribution. h μ represents the dynamic model benefit reduction factor at the current time step. h This represents the numerical value of the Gaussian probability density function;

[0105] Make judgments and verifications based on the direction and trend of variables.

[0106]

[0107] Among them, Q(L) dq ,t) represents flow rate, L dq H(L) represents the length of the currently demonstrated channel / pipe segment in the overall water supply network system. dq ,t) represents the water pressure generated by the turbulent water flow, and a represents the water hammer wave velocity corresponding to the control effect.

[0108] Based on the simulation results, the parameters of the sliding surface and its controller are adjusted, and the control decision output for the gate opening is obtained through feedback and adjustment.

[0109] u = -(B T P) -1 (B T PAx(t)+B T Pf(x,t)+||B T PD||×||B TPEs(t)||+μsign(s))

[0110] According to the formula:

[0111]

[0112] To obtain the maximum benefit J during the state change process η , where r is the number of fuzzy rules; argmax represents finding the maximum independent variable; a1 represents the water hammer wave speed corresponding to the control action.

[0113] Decoupling calculations are performed on the prediction model representing the state of the water network system and the prediction model representing the opening degree of each gate within the water network. This reduces the mutual influence of hydrodynamic disturbances generated during the operation of each gate on water flow and gate control disturbances, and further eliminates the influence of these complex disturbances in the process of formulating control methods. Decoupled control algorithms are applied to individual gates, decoupling the gate opening degree and flow rate from the upstream and downstream directions of the water network channel. For control gates located in various river sections / pipelines, decoupling upstream transmits the current gate operation amount to the upstream gate controlling the incoming water, preventing flow disturbances generated by closed-loop control from propagating upstream towards water intake. For gates controlling downstream incoming water, decoupling downstream transmits a portion of the current gate operation amount to the downstream canal / pipe gates, preventing the propagation of control disturbances downstream and isolating the impact of gate opening and closing on subsequent water diversion and distribution. The flow rate is then... Output decoupling and opening output decoupling are integrated into practical applications. Flow control is used as an indirect decoupling mechanism, and the actual gate is used as the control structure. The flow controller converts the flow output into the gate opening output. In the control process under decoupling calculation, when the flow in the water intake channel changes and the water level in the regulating pool deviates from the target value, the water level error is input to the feedback controller to obtain the flow error for the water intake channel. At the same time, it is added to the output flow obtained from the decoupling calculation in the upstream direction and then input to the flow controller. The flow controller calculates the opening adjustment value of the water intake gate and controls the gate actuator to operate.

[0114] This invention incorporates fuzzy sampling control and active disturbance rejection filtering based on measured data, ensuring real-time and accurate acquisition of the water network project's status. It addresses the challenges of automatic control of disturbances such as strong coupling, nonlinearity, and large time delays in water supply processes, achieving more automated mechanism-based control. This ensures system stability and response speed, improving the timeliness and precision of control.

Claims

1. A method for generating instructions for a real-time decision-making control system for an intelligent water network, characterized in that, Includes the following steps: S1. Based on the existing water network engineering layout and the various hydrological monitoring stations deployed, obtain real-time monitoring data of the water network system; S2. Equip the monitoring stations with a data processing system to remove interference and noise from the real-time monitoring data and perform fuzzing processing to obtain a current water network status monitoring dataset that is consistent with the data, thereby generating immediate feedback. S3. Based on the real-time feedback results, the set water supply network, the water volume target on the receiving side, and the expected water supply process path, establish a predictive control mode for the water flow status of the water network. The specific implementation method is as follows: S3-1. According to the formula: Obtain fuzzy control plan ;in, Indicates the first The weight of the water supply plan is determined, and it meets the requirements. ; This represents simulation plans for various prediction methods based on historical data. This indicates the corresponding actual water supply. Indicates the first Predictions from group water supply rehearsals; m Indicates that there is m Predictive simulation methods; Indicates the sliding surface; S3-2, According to the fuzzy control plan Establish a predictive control model for water flow status in the water network; S4. Based on the established predictive control model, the fuzzy evaluation and feedback correction method of control effect is adopted to obtain the adjusted predictive fuzzy control model. S5. Based on the water network predictive fuzzy control model, use a state-space model-based neural network to complete joint automatic fuzzy control.

2. The method for generating instructions for a real-time decision-making control system for an intelligent water network according to claim 1, characterized in that, The specific implementation method of step S1 is as follows: S1-1. Divide the water network into two levels: water network channel river sections and nodes; S1-2, Using the discrete Saint-Venant equations, we obtain the 2(...) of the water network channel. N -1) algebraic equations, and supplemented with 2 algebraic equations for boundary conditions based on the water network node boundaries, forming a total of 2 N The discrete Saint-Venant equations are shown below, representing a system of nonlinear algebraic equations: In the formula, , , , ， ， The discretized equations are eliminated to obtain the double-chasing equations. The linear equation system is solved using the chasing method, and the results of the existing water network project are calculated. water level Flow rate, gate opening The relationship; among them, j Indicates the cycle step, i Represents a node. a , b , c , d All of these represent the conversion factors for water level and flow rate to opening degree; θ The angle between the tangent at the lower edge of the arc-shaped gate and the horizontal direction; It is the acceleration due to gravity; B For the corresponding gate; A This refers to the cross-sectional area of ​​the water passage. M Indicates the design conditions; Indicates the measurement time; This represents the change in spatial coordinates; S1-3. Optimize the spatial layout of monitoring stations based on the existing hydrodynamic element relationships of the water network. The existing hydrodynamic element relationships of the water network include the storage capacity of each channel of the water network, the design water level and flow rate, the top elevation of the dike, the operating mode adopted by each channel of the water network, the upstream normal water level, the downstream normal water level, equal volume operation and storage balance control. S1-4. Based on the water network operation mode, determine the water level, flow rate, sensor type, layout method and installation location, and obtain real-time monitoring data of the water network system.

3. The method for generating instructions for a real-time decision-making control system for an intelligent water network according to claim 2, characterized in that, The specific implementation method of step S2 is as follows: S2-1, with a periodicity TH Instantaneous values ​​are acquired from continuous information from monitoring stations, and these values ​​are used as external input signals. W ; S2-2, Input external signal W Transformed to be suitable for samplers S input signal Y and input sampler S This yields discrete state values ​​exhibiting a periodic pattern. Y d ; S2-3, Discrete state values ​​exhibiting periodic patterns Y d Input fuzzy controller K d To obtain discrete signals U d ; S2-4, Discrete signal U d The control input is formed by a zero-order hold. U ; S2-5. According to the formula: Obtain the state variable output values ​​of the discretized system model after removing disturbances and noise. ;in, V This indicates the number of nodes that cause coupling disturbances to the water flow. Indicates the first One node; Indicates the parameters of the Karman filter. and Indicates the numerical value of the filter factor; , , This represents the state variables obtained during the application of the Active Disturbance Rejection Algorithm. This represents the inertial time constant of the water flow in the water network; , , This represents the self-regulating coefficient constant for eliminating fluctuations in the water flow of a water network under small disturbances. Indicates significant fluctuations, control input U Converted into real-time input signal , The approximate input extracted by the differential tracker It is a differential signal; and Used to measure the relative deviation between the current water level and flow rate of the water network channel and the actual value; t Indicates the measurement time point; and Both represent self-adjustment coefficients; This indicates the energy fluctuations caused by the coupling disturbance effect of water flow; This indicates the driving force behind the coupling disturbance effect of water flow; This represents the water flow time delay constant of the water network channel; The self-adjusting coefficient for the sampling period; The time delay constant is the self-adjustment coefficient; This is the self-adjusting coefficient; for n -1 to n Time period state values ​​and state quantities of differential signals; for n arrive n +1 time period state value and state quantity of differential signal; for n- 1 to n +1 time period state value and state quantity of differential signal; S2-6. Remove disturbances and noise from the output values ​​of the state variables of the discretized system model. The station monitors the real-time status and generates immediate feedback, which includes all parameters related to water.

4. The method for generating instructions for a real-time decision-making control system for an intelligent water network according to claim 3, characterized in that, The specific implementation method of step S4 is as follows: S4-1. According to the formula: The water network channel control model, and the functions of upstream and downstream water depth relative to flow rate are obtained. ;in, The river section corresponding to the water diversion channel ( Water level and steady state at control point t in (=1,2,3) Deviation between water levels; Indicates the first The area of ​​the return water zone of each water diversion channel; 、 、 They represent the first The deviations of the inflow, outflow and intake flow rates of each water diversion channel from the steady state; Indicates the first The water flow time delay corresponding to each water diversion channel; L The weir length or gate width associated with the gate; 、 This is the reduction factor; 、 These are the water depths upstream and downstream of the sluice gate, respectively. This refers to the gate opening degree; S4-2. Based on the water network channel control model, the current state of the water network is obtained as a function of upstream water depth and downstream water depth relative to the flow rate. S4-3, According to the formula: Obtain the steady-state control conditions of the water network system ;in, The inverse matrix representing the state space of a water supply network project; This represents a time variable that measures the current state of the water network. Represents the state variables of the water network system; , , ku These are the parameters for setting the sliding surface; Indicates the bounded range of external disturbances; sat(·) The state of the sliding surface; This is a matrix representing the state space of the water supply network project, obtained from the time variables input from the state of the water network. S4-4. According to the formula: The output of the control decision for the steady-state water network is obtained. ;in, Indicates the bounded range of external disturbances; sign (·) is a sign function, indicating that the sign of (·) is taken; S4-5. According to the formula: The adjusted predictive fuzzy control model is obtained; where This represents the final state of time-delay control. This is the output of time-delay control; , , , , All are the first The system matrix of each subsystem; Indicates that the predecessor variable is in the set N Membership degree in; For the preceding variable; N It is a fuzzy set; For predictive simulation methods; The approximate input extracted by the differential tracker corresponding to the raw data obtained from the monitoring station; This represents the raw data obtained from the monitoring stations; This represents the output of the control decision for a steady-state water network; The time variable is used to measure the current state of the water network.

5. The method for generating instructions for a real-time decision-making control system for an intelligent water network according to claim 4, characterized in that, The specific implementation method of step S5 is as follows: S5-1. According to the formula: The water level and flow rate constraint equations are obtained; where, For the flow rate through the gate, The gate orifice flow coefficient; For the gate opening, The width of the gate opening; The difference in water level between the upstream and downstream sides of the sluice gate; S5-2, According to the formula: The matrix solution space of the water level and flow rate constraint equations is obtained; where, This represents the coefficient matrix indicating the impact of predicted conditions on water level and flow rate based on the control and operation of water network channels. This represents a coefficient matrix indicating the impact of current water network channel control and operation on water level and flow. This represents the matrix of comprehensive influence coefficients on the water network state under time-delay control. Represents the transpose matrix. This indicates the relative steady-state outflow deviation of water intake on both sides of the water network channel; express k Constantly monitor deviations in the raw data obtained from the monitoring stations; express k+ Deviation of raw data obtained from monitoring stations at time 1; For the first j+2 Water level during the circulation step; For the first j +1 cycle water level; This is the differential error; For the traffic flowing into the node; This represents the current traffic value of the node. S5-3, According to the formula: A predictive model representing the state of the water network is obtained. ;in, J Indicates the number of basis functions; This represents the linear weighting coefficients obtained from the basis function optimization calculation; This represents the value of the basis function at a given time point within the sampling period; Indicates the first One basis function; Indicates the number of responses; Represents the prediction function; Indicates the prediction function in The format of the time period; Indicates the prediction function in The format of the time period; The control function represents the prediction function; S5-4. According to the formula: This involves decoupling the control gates of the water network channel from the upstream gates controlling the incoming water, decoupling the upstream gates controlling the incoming water from the channel / pipe head gates, and decoupling the output for flow rate; among these, Indicates the first control gate in the upstream section of the river / pipeline The opening degree of each; Indicates the corresponding first The water distribution volume of each sluice gate's water distribution point Indicates water distribution point The deviation between the water level upstream of the sluice gate in the previous time period and the corresponding water supply target setting. k Indicates a time period; , They represent the first The proportional and integral coefficients of the PID controller for the upstream gate of each river section, where 0 represents the coefficient of the water diversion gate at the water diversion channel; , These are the gate opening and the decoupling parameters for decoupling the flow output, respectively. Indicates water distribution point exist k The deviation between the water level in front of the gate at any given time and the corresponding water supply target setting; , This represents the proportional and integral coefficients of the PID controller for the water intake gate at the water intake channel. S5-5, According to the formula: The solution to the state equation in the state-space representation is obtained. and the solution of the output equation ;in, This represents the homogeneous solution in the matrix solution space of the water level and flow rate constraint equations; The product of the Laplace transform of the matrix solution space of the water level and flow rate constraint equations and the Laplace transform of the input quantities; A response with zero input; Zero-state response; Zero-state response; S5-6. Based on the output of the adjusted predictive fuzzy control model, the solution of the state equation, and the solution of the output equation, the control information of the opening degree of each gate, the water level control information of the upstream water network, and the flow control information of the downstream water network are used to control the opening degree of each gate, thereby completing the joint automatic fuzzy control.

6. A control system based on the instruction generation method for a real-time decision-making control system for an intelligent water network according to any one of claims 1-5, characterized in that, It includes a hydrological monitoring module, a data processing module, a predictive control module, a predictive fuzzy control module, and a fuzzy control module; The hydrological monitoring module is used to obtain real-time monitoring data of the water network system based on the existing water network engineering layout and various hydrological monitoring stations. The data processing module is used to equip the monitoring stations with data processing systems to remove interference and noise from real-time monitoring data and perform fuzzing processing to obtain a current water network status monitoring dataset that is consistent with the data, thus forming real-time feedback. The predictive control module is used to establish a predictive control mode for the water flow status of the water network based on the results of real-time feedback, the set water supply network, the water volume target on the receiving side, and the expected water supply process path. The predictive fuzzy control module is used to obtain the adjusted predictive fuzzy control model by adopting fuzzy evaluation and feedback correction methods for control effect based on the established predictive control mode. The fuzzy control module is used to perform joint automatic fuzzy control based on the water network prediction fuzzy control model and a neural network based on the state space model.

Images