CN-121979161-A - Water supply network leakage cooperative control method and system

CN121979161ACN 121979161 ACN121979161 ACN 121979161ACN-121979161-A

Abstract

A water supply network leakage cooperative control method and system comprises the steps of uniformly organizing multi-source operation information of an operation state of an execution mechanism based on a water supply network topological structure, constructing a motion condition leakage confidence field capable of being updated in a recursive mode along with control motions and monitoring responses, characterizing a self-adaptive determination control mode by taking the motion condition leakage confidence field as a control state under the conditions of water supply service level and operation safety constraint, restraining control motion strength of an active identification stage through excitation budget parameters, dividing the network into a plurality of cooperative control units and setting control main bodies, generating candidate control motions based on local observables and confidence abstract information exchanged by adjacent units by each control main body, projecting the candidate control motions to a safe movable working set through a motion mapping operator, executing consistency coordination correction at a partition boundary or a shared execution mechanism, and obtaining final control motions to be executed in a concurrent mode.

Inventors

ZHOU LI
SONG DONGPAN
FU MINGLEI
XU SHAOQING
LI JIA
MENG FANYU

Assignees

杭州莱宸科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260327

Claims (11)

1. A water supply network leakage cooperative control method comprises the following steps: S1, performing topology modeling on nodes and pipe sections of a water supply network, determining partition boundaries and key nodes, and configuring static properties and constraint parameters for the nodes, the pipe sections and an executing mechanism; s2, collecting water supply network operation data and actuating mechanism action data at preset time intervals, and constructing a control state sequence according to a sliding time window and the topological structure; S3, preprocessing the control state sequence, and fusing the control state sequence with the static attribute to obtain a control input characteristic sequence; S4, constructing a hydraulic consistency evidence and an action-response consistency evidence based on a control input characteristic sequence, fusing to obtain consistency indexes, and recursively updating leakage risk indexes and uncertainty indexes according to the consistency indexes to form and update an action condition leakage confidence field; s5, generating a safe actionable action set according to the water supply service constraint and the operation safety constraint, defining an action mapping operator, and defining a constraint boundary and a projection mode of projecting the candidate control action to the safe actionable action set; S6, taking the leakage confidence field of the action condition as a control state representation, judging a control mode according to an uncertainty index, and adaptively adjusting a control target weight according to the leakage risk index and the water supply service level requirement; S7, projecting candidate control actions into a safe movable action set according to an action mapping operator to obtain feasible actions, detecting the consistency of the actions of a plurality of control subjects at a partition boundary or a shared executing mechanism, and carrying out coordination correction on conflict actions according to boundary consistency constraint, water supply service priority and high risk object confidence level to output final control actions; And S8, issuing a final control action, collecting execution feedback data, updating action condition leakage confidence field and control strategy parameters according to the execution feedback data, and outputting a control effect evaluation result and alarm information.
2. The water supply network leakage cooperative control method as claimed in claim 1, wherein the step S1 comprises: S1-1, modeling a water supply network node as a node set, and modeling a pipeline for connecting the node and a hydraulic element as a pipe segment set to obtain a water supply network topological structure; s1-2, endowing static properties for the nodes and the pipe sections, wherein the static properties comprise elevation, pipe diameter, length, roughness coefficient, partition index and key node identification; s1-3, determining an actuating mechanism set of a pump station, a valve and a pressure regulating valve, and setting constraint parameters of an upper limit and a lower limit of actions, an upper limit of action change rate and a limit of start-stop frequency.
3. The water supply network leakage cooperative control method as claimed in claim 1, wherein the step S2 comprises: S2-1, collecting operation data, wherein the operation data comprise node pressure, node water head, pipe section flow, partition water inlet and outlet amount, operation state of an executing mechanism and water demand information; S2-2, organizing and collecting data according to a sliding time window, constructing a node measurement sequence, a pipe section measurement sequence, a partition metering sequence and an actuating mechanism action sequence, and establishing a corresponding relation with a topological structure to obtain a control state sequence.
4. The water supply network leakage cooperative control method as claimed in claim 1, wherein the step S3 comprises: S3-1, carrying out missing value processing, abnormality identification and rejection, unit unification and numerical value normalization processing on the control state sequence; s3-2, respectively forming a dynamic characteristic vector by the node measurement quantity, the pipe section measurement quantity and the actuating mechanism action quantity at each time step, and fusing the dynamic characteristic vector with the corresponding static attribute to form a control input characteristic; And S3-3, stacking the control input features of all time steps in the time window in time sequence to form a control input feature sequence.
5. The water supply network leakage cooperative control method as claimed in claim 1, wherein the step S4 comprises: S4-1, constructing partition water quantity balance deviation based on partition water inlet and outlet quantity and partition water demand, and constructing boundary flow residual error based on boundary measurement and adjacent partition exchange quantity; s4-2, constructing pressure deviation based on the pressure safety interval of the key node; s4-3, constructing action-response deviation based on the control action increment, the key node pressure after execution and the variation of boundary flow, and representing the consistency degree between the action and the hydraulic response; S4-4, integrating water balance deviation, boundary flow residual error, pressure deviation and action response deviation to form a consistency index, and generating leakage risk index and uncertainty index by the consistency index to form a high risk object set and a high uncertainty object set; S4-5, recursively updating the leakage risk index and the uncertainty index by using the latest acquired data to obtain an action condition leakage confidence field; s4-6, generating active identification excitation budget parameters according to the uncertainty index, wherein the excitation budget parameters comprise an upper limit of an action amplitude, an upper limit of an action change rate and an upper limit of duration.
6. The water supply network leakage cooperative control method as claimed in claim 1, wherein the step S5 comprises: s5-1, determining the minimum service pressure of a key node and the constraint of a node pressure safety interval, and determining the pressure feasible range of each partition or the key node based on the current moment and the water demand estimated value in a preset control period; S5-2, determining the action feasible range of each actuating mechanism according to the action upper and lower limits, the action change rate upper limit, the start-stop frequency limit and the allowed working range of the pump station, the valve and the pressure regulating valve; s5-3, determining forbidden action combination and linkage constraint rules under the switching scene of the shared executing mechanism, the serial-parallel valve group and the pump group according to the pipe network topological relation and the equipment linkage relation to obtain an operable action combination range; S5-4, synthesizing the constraints of the steps S5-1 to S5-3, generating a safe and feasible action set for limiting the value range of the candidate control action, and providing the safe and feasible action set for the subsequent step S7 for action mapping and checking.
7. The method for collaborative control of leakage of a water supply network according to claim 1, wherein in step S6, candidate control actions are generated or selected with the goal of reducing uncertainty index under the limitation of excitation budget parameters in an active identification mode, and in an economic control leakage mode, candidate control actions are generated or selected with the goal of reducing leakage risk and taking both operational economy and action smoothness into account.
8. The water supply network leakage cooperative control method as claimed in claim 1, wherein the step S6 includes: s6-1, judging a control mode according to an uncertainty index, and adaptively adjusting a control target weight according to a leakage risk index and a water supply service level requirement; s6-2, dividing a pipe network into a plurality of cooperative control units and setting a control main body, wherein the control main body is a strategy controller, the strategy controller is a reinforcement learning intelligent body or a rule-based optimization controller, and a local observation set and an action set are configured for each control main body; S6-3, the adjacent control main bodies exchange confidence abstract information, wherein the confidence abstract information comprises boundary node pressure statistics, boundary flow residual statistics, high risk object sets and corresponding confidence indexes; S6-4, under the active identification mode, each control main body outputs a candidate excitation action sequence under the limitation of excitation budget parameters, and selects with the uncertainty index reduction as a target; And S6-5, under the economic leakage control mode, each control main body outputs a candidate control action, and selects the target of reducing leakage risk indexes and considering running economy and action smoothness.
9. The water supply network leakage cooperative control method as claimed in claim 1, wherein the step S7 includes: S7-1, performing action projection on the candidate control actions according to boundary constraint of a safe feasible action set, wherein the action projection comprises constraint processing on upper and lower limits of actions, upper limit of action change rate and start-stop frequency limit, and outputting feasible actions meeting the constraint; And S7-2, detecting consistency of actions of the plurality of control subjects at partition boundaries or sharing execution mechanisms, and when conflicts exist, carrying out coordination correction on the conflict actions according to boundary consistency constraints and combining the priority of water supply service constraints and the confidence of high-risk objects, and outputting consistent final control actions.
10. The water supply network leakage cooperative control method as claimed in claim 1, wherein the step S8 comprises: S8-1, issuing a final control action to an executing mechanism for implementation, and collecting pressure, flow, partition metering and executing mechanism state feedback data after execution; s8-2, updating action condition leakage confidence fields, uncertainty indexes and control strategy parameters based on feedback data to realize online closed loop self-adaptive iteration; and S8-3, outputting a control effect evaluation result and alarm information, wherein the alarm information comprises a pressure out-of-limit alarm, an abnormal fluctuation alarm and a high risk area prompt.
11. The water supply network leakage cooperative control system is characterized by comprising a data acquisition module, a topology and parameter management module, a control state sequence construction module, an input characteristic construction module, an action condition leakage confidence field construction and updating module, a safe and feasible action set generation module, a mode judgment and cooperative decision module, a safety check and consistency correction module, an execution issuing and feedback updating module and a man-machine interaction and alarm module, wherein the whole water supply network leakage control process is cooperatively realized among the modules; The data acquisition module is used for carrying out data interaction with the existing information system of the water supply enterprise to acquire basic information, operation monitoring data and actuating mechanism action data of the water supply network; The topology and parameter management module is used for organizing and managing nodes, pipe sections and execution mechanisms of the water supply network, constructing a topological structure of the water supply network, determining partition boundaries and key nodes, and maintaining static properties of the nodes and the pipe sections and constraint parameters of the execution mechanisms; The control state sequence construction module is used for collecting operation data and action data at preset time intervals and constructing a control state sequence comprising node measurement, pipe section measurement, partition metering and actuating mechanism action sequences according to the sliding time window and the topological structure; The input characteristic construction module is used for preprocessing the control state sequence and fusing the control state sequence with the static attribute to form a control input characteristic sequence; The action condition leakage confidence field construction and updating module is used for constructing and recursively updating the action condition leakage confidence field based on the control input characteristics, including leakage risk indexes and uncertainty indexes, and generating an active identification excitation budget parameter according to the uncertainty indexes so as to limit the amplitude, the change rate and the duration of the control action in an active identification mode; the safe and feasible action set generation module is used for generating a safe and feasible action set based on the water supply service constraint and the operation safety constraint so as to limit the feasible control range of the execution mechanism; The mode judging and collaborative decision-making module is used for judging a control mode according to the uncertainty index and adaptively adjusting the control target weight; the system comprises a module, a control module and a control module, wherein the module divides a pipe network into a plurality of cooperative control units and sets a control main body, the control main body is a strategy controller, the strategy controller is an reinforcement learning intelligent body or a rule-based optimization controller, the control main body generates candidate control actions based on local observed quantity and confidence abstract information exchanged by adjacent main bodies; When a plurality of control main bodies are inconsistent at the partition boundary or the shared executing mechanism, the related control actions are coordinated and corrected according to the boundary consistency constraint, the water supply service priority and the high risk object priority, and the consistent final control actions are output; The execution issuing and feedback updating module is used for issuing a final control action and acquiring execution feedback data, updating action condition leakage confidence field, uncertainty index and control strategy parameters based on the feedback data, and realizing online closed loop self-adaptive iteration; and the man-machine interaction and alarm module is used for displaying the action condition leakage confidence field, the control mode, the control action and control effect evaluation result and outputting alarm information and treatment prompt when the pressure is out of limit, abnormal fluctuation or a high risk object appears.

Description

Water supply network leakage cooperative control method and system Technical Field The invention relates to the technical field of urban water supply and intelligent water service, in particular to a cooperative control method and system for leakage of a water supply network. Background Water leakage in water supply networks is an important problem in urban water resource management. On one hand, the leakage loss causes great loss of raw water and finished water, increases the energy consumption of water taking, water production and pressurization, and on the other hand, the leakage loss can also cause secondary risks such as pollutant invasion, ground subsidence and the like. Because the water supply network is buried underground more and the service range is wide, the tiny leakage is difficult to discover and dispose in time in a short time, so that the leakage management and operation safety management and control difficulty is increased. The existing leakage control adopts a mode of combining partition metering and manual experience in engineering, and is generally realized by establishing water distribution partitions and judging by combining indexes such as minimum night flow, partition water balance and the like, and meanwhile, the existing leakage control is interfered by means of cooperation of static partition pressure control, empirical scheduling, periodic inspection and the like. The mode has lower implementation threshold, but generally has the problems of long updating period, lag response to short-time working condition fluctuation and sudden leakage, dependence on manual experience on control actions, difficulty in forming closed loop optimization and the like, and in addition, the coarse granularity index of the partition layer is difficult to support fine cooperative regulation and control on various execution mechanisms such as valves, pump stations and the like. With the deployment of a monitoring and data acquisition system, a remote water meter and an online pressure and flow sensor, an automatic method for the operation regulation and control of a water supply network is gradually increased, but the existing control strategy still has defects on a leakage control target. On the one hand, the pressure control commonly used in engineering application is more developed around pressure stabilization and water supply guarantee, and the leakage inhibition is difficult to be used as a core target and continuously optimized while the pressure safety interval and the water supply service level constraint are met. On the other hand, the water supply network has the characteristics of multi-source water supply, multi-partition coupling and coexistence of multiple execution mechanisms, different areas have differences in the aspects of demand change, leakage sensitivity, pressure constraint, regulation and control cost and the like, the single controller or single-point optimization mode is difficult to realize the cooperative control of cross-partition and cross-equipment, and the fluctuation of control effect and insufficient robustness are easy to occur under the uncertain disturbance and equipment constraint conditions. Patent document CN118520806a discloses a water supply network pressure regulation method, which is constructed and corrected in real time through a hydraulic model, and a dynamic pressure regulation strategy is generated by combining an objective function and an optimization algorithm, so that the opening of a valve and the running state of a pump station are regulated. The method can realize automation of pressure regulation, but the control target is mainly pressure regulation, a target modeling and effect measuring mechanism for leakage control is lacked, an effective scheme is not provided for the collaborative decision and conflict resolution problems required by multi-partition and multi-actuator concurrent regulation, and the self-adaptive capacity under the complex conditions of demand disturbance and operation constraint still has room for improvement. Patent document CN119294550a discloses a method and system for controlling a water supply network using reinforcement learning by constructing a historical state database and training agents based on a reward function to generate a control strategy for control set points such as pump speed and introducing a mix set point for dynamic recommendation. The method is heuristic to control of a water supply network, but rewards design and control objects are more biased to running indexes such as energy consumption, water level and the like, feedback quantity and constraint expression which can directly drive control optimization are not constructed around leakage inhibition, and a control organization mode of multi-agent for carrying out division cooperation, local autonomy and global coordination on multi-class execution mechanisms in a multi-partition scene is not reflected, so that the long-term self-adaptive optimization