A plant-grid collaborative optimal operation system for integrated intelligent water plant

Abstract

The application is suitable for the field of intelligent water technology, and discloses a plant-network collaborative optimization operation system for integrated intelligent water plant, which comprises: an abnormal point module, which constructs a water supply pipe network topology graph and collects pipe network data in real time, detects pipe explosion events and locates pipe explosion nodes; a region demarcation module, which demarcates an affected area based on the pipe explosion nodes, extracts a candidate valve set, a boundary node set and a minimum connected subgraph in the affected area; and an instruction generation module, which obtains an optimal valve closing set based on the candidate valve set and the minimum connected subgraph, and generates a collaborative scheduling instruction. After the scheduling instruction is executed, the application monitors the pressure residual of the boundary nodes in real time, dynamically calculates a residual threshold value according to the measured pressure change rate, and performs hierarchical processing by fitting a straight line slope of the residual sequence. The application can dynamically adjust its model and parameters according to the actual state of the pipe network, improve the long-term prediction accuracy of the industrial data management system, and ensure the efficiency and reliability of the emergency response of the intelligent water plant.

Description

Technical Field

[0001] This invention relates to the field of smart water technology, specifically to a plant-network collaborative optimization operation system for integrated smart water plants. Background Technology

[0002] Water supply networks are a crucial component of urban infrastructure, and their safe and stable operation directly impacts public safety and residents' quality of life. With the advancement of smart city construction, integrated smart water plants are gradually becoming the development direction of the water industry. Their core feature is incorporating the water plant's production process, the network's transmission and distribution process, and the pumping station's pressure regulation process into a unified data collection and monitoring system, achieving optimized collaborative operation between the plant and the network. Among various emergencies in water supply networks, pipe bursts are the most common and most damaging. After a pipe burst, significant water resources are lost, localized water pressure drops sharply, and secondary disasters such as road collapses may even occur. Therefore, how to quickly and accurately identify the event, analyze the affected area, formulate the optimal valve closure and isolation plan, and coordinate the scheduling of water plants and pumping stations in the event of a pipe burst has become a key technical challenge in the construction of smart water plants.

[0003] In existing technologies, when a pipe bursts, dispatchers need to manually answer citizen complaints or wait for reports from patrol personnel, and then rely on experience to consult paper or electronic pipeline maps to determine which valves need to be closed. This manual decision-making model has a response delay; it usually takes a long time from the pipe burst to the valve closure, during which a large amount of water resources are lost, and the water outage area is often unnecessarily expanded. Summary of the Invention

[0004] The purpose of this invention is to provide a plant-network collaborative optimization operation system for integrated smart water plants, in order to solve the problem that existing technologies are unable to quickly analyze the affected area and provide the optimal valve shut-off scheme and scheduling instructions in the event of a pipe burst, so as to minimize the impact of the disaster.

[0005] The objective of this invention can be achieved through the following technical solution: a plant-network collaborative optimization operation system for integrated smart water plants, comprising:

[0006] The abnormal point module constructs a water supply network topology map and collects network data in real time, detects pipe burst events and locates the burst node.

[0007] The region delineation module, based on the burst pipe node, delineates the affected region and extracts the candidate valve set, boundary node set, and minimum connected subgraph within the affected region;

[0008] The instruction generation module obtains the optimal set of valves to be closed based on the candidate valve set and the minimum connected subgraph, and generates collaborative scheduling instructions.

[0009] The adaptive adjustment module executes collaborative scheduling instructions, monitors the pressure residuals of boundary nodes in real time, and makes adaptive adjustments based on the magnitude of the residuals.

[0010] Preferably, the pipeline network data specifically includes node pressure, pipe segment flow rate, valve opening, pipe length, and node adjacency relationships. The node refers to a valve, pipe intersection, pipe diameter change point, water source point, or key water use point identified from the pipeline network geographic information system and assigned a unique number. The node pressure refers to the hydrostatic pressure borne by the water body at a certain node location in the water supply network topology map. The node adjacency relationship refers to the fact that for any two nodes in the water supply network topology map, if there is a pipe connecting the two nodes, then the two nodes are called adjacent nodes.

[0011] Preferably, the specific process of detecting the pipe burst event and locating the burst node is as follows:

[0012] Obtain the pressure gradient between the target node and each of its adjacent nodes at the same time. The pressure gradient is equal to the difference between the pressure value of the target node and the pressure value of the adjacent node, divided by the length of the pipe connecting the target node and the adjacent node.

[0013] After obtaining the pressure gradients corresponding to all neighboring nodes of the target node, the arithmetic mean of all pressure gradients is calculated. The deviation between the pressure gradient of each neighboring node and the arithmetic mean is calculated. Each deviation is squared and then summed to obtain the sum of squared deviations. The sum of squared deviations is divided by the total number of neighboring nodes to obtain the variance of the pressure gradient of the target node.

[0014] If the pressure gradient variance of the target node exceeds the preset variance threshold for three consecutive time periods, and the node pressure decrease rate of the target node is lower than the preset decrease rate threshold, then it is determined that a pipe burst event has occurred near the target node, and the target node is a pipe burst node.

[0015] Preferably, the specific process of delineating the affected area based on the burst pipe node is as follows:

[0016] Mark all adjacent nodes directly connected to the burst pipe node as the main affected area;

[0017] Nodes that are first-order connected, second-order connected, and third-order connected to any node in the main influence area are selected respectively to form a candidate node set, and different weights are assigned to each candidate node.

[0018] In the topology graph, query the water source path of each candidate node and calculate the water source dependence of the candidate node. The water source dependence of the candidate node is equal to the ratio of the number of paths in the water source path of the candidate node that pass through any node in the main influence area to the total number of paths.

[0019] The weighted dependency of each candidate node is obtained by weighting and fusing the water source dependency of each candidate node with its corresponding weight.

[0020] A preset dependency threshold is set, and it is compared with the weighted dependency of each candidate node. If the weighted dependency of a candidate node is not less than the dependency threshold, the candidate node is marked as a secondary influence area. Otherwise, the node is considered not to be directly affected by the pipe burst and is not included in the affected area.

[0021] The primary and secondary affected areas are designated as the affected areas.

[0022] Preferably, the specific process of extracting the candidate valve set, boundary node set, and minimum connected subgraph within the affected area is as follows:

[0023] Candidate valve set: Identify all valves located within the affected area. If a valve has a pipe that is directly connected to a node outside the affected area, mark the valve as a candidate valve. All candidate valves constitute the candidate valve set.

[0024] Boundary node set: Find all nodes located inside the affected region. If a node has any edge connecting it to a node outside the affected region, then mark that node as a boundary node. All boundary nodes constitute the boundary node set.

[0025] Minimum Connected Subgraph: Query the shortest path from the burst pipe node to each boundary node in the set of boundary nodes from the water supply network topology graph. Record the number of edges traversed by each shortest path. Obtain the maximum number of edges corresponding to all boundary nodes. Obtain all nodes that can be reached from the burst pipe node through the edge with the maximum number of edges, as well as the edges connecting these nodes, from the water supply network topology graph. Construct the minimum connected subgraph from these nodes and edges.

[0026] Preferably, the specific process of obtaining the optimal set of closed valves is as follows:

[0027] Establish a pre-defined correspondence between user levels and weights, label each user with a level type and assign the corresponding weight;

[0028] Find all the minimum valve combinations that can disconnect the burst pipe from all water sources on the minimum connected subgraph. Each minimum valve combination constitutes a candidate valve shut-off set.

[0029] The user impact level, leakage water volume, and water hammer impact value are obtained for each candidate valve shut-off set, wherein the user impact level is determined by the maximum value among the pressure reduction coefficient, water quality degradation coefficient, and insufficient water supply coefficient;

[0030] The user impact level of each candidate valve set is weighted and fused with the user's corresponding weight to obtain the weighted total impact. The first preset number of candidate valve sets with the smallest weighted total impact value are selected as the first-level screening valve sets.

[0031] The primary screening valve sets are sorted in ascending order of leakage volume, and the first two preset number of primary screening valve sets are retained as secondary screening valve sets.

[0032] For each candidate valve set in the secondary screening valve set, the optimal closing sequence is solved using dynamic programming, and the minimum water hammer impact value under this sequence is calculated. The candidate valve set with the minimum water hammer impact value is selected as the optimal valve set. If the water hammer impact values ​​of multiple candidate valve sets are the same, the one with the smallest leakage is selected as the optimal valve set.

[0033] Preferably, the specific process of monitoring the pressure residual at the boundary node is as follows:

[0034] During the execution of the coordinated scheduling instructions, the measured pressure of the boundary nodes is obtained by pressure sensors installed on the boundary point set at a preset sampling frequency;

[0035] The predicted pressure is calculated for the same boundary node using a hydraulic model and then aligned with the measured pressure in the time dimension.

[0036] The absolute value of the difference between the measured pressure and the predicted pressure at the boundary node is the boundary node pressure residual.

[0037] Preferably, the specific process of performing closed-loop feedback is as follows:

[0038] Obtain the residual threshold, compare the pressure residual of each boundary node with the residual threshold, and if the pressure residual of all boundary nodes is less than the residual threshold, maintain the current cooperative scheduling instruction.

[0039] If the pressure residual of a boundary node is not less than the residual threshold and the duration exceeds the preset time threshold, then the residual data of the boundary node at the most recent sampling time are extracted, and a straight line is obtained by fitting the least squares method. The slope of the straight line reflects the trend of the residual changing with time.

[0040] Based on the comparison results of the absolute value of the slope with the first threshold and the second threshold, a graded adjustment is performed.

[0041] Preferably, the specific process of performing the hierarchical adjustment is as follows:

[0042] If the absolute value of the slope is less than the first threshold, and the residual falls back to within the residual threshold within a preset time threshold, it is determined to be an instantaneous disturbance and no adjustment is made.

[0043] If the absolute value of the slope is between the first threshold and the second threshold, it is determined to be a slow drift, the hydraulic model parameters are corrected, and the corrected hydraulic model is used for subsequent pressure prediction. If the corrected residual converges within a specified time, the original coordinated scheduling instruction is continued to be executed; otherwise, it is upgraded to continuous deterioration processing.

[0044] If the absolute value of the slope is greater than or equal to the second threshold, it is determined to be a continuous deterioration, and a return instruction is executed.

[0045] Preferably, the specific process of executing the return instruction is as follows:

[0046] If the residual of a single boundary node continues to increase while other boundary nodes remain normal, return to the instruction generation module and resolve for the optimal valve set while keeping the affected area unchanged.

[0047] If the residuals of multiple boundary nodes increase simultaneously and have the same trend, return to the region delineation module, redefine the affected region and update the minimum connected subgraph, and re-execute the instruction generation module.

[0048] The beneficial effects of this invention are:

[0049] 1. This invention calculates the pressure gradient variance between each node and its adjacent nodes in real time, reflecting the degree of disruption to the spatial consistency of local pressure distribution caused by a pipe burst. Combined with pressure drop characteristics, it effectively distinguishes pipe bursts from non-leakage disturbances such as valve regulation and pump switching. This system transforms the traditional passive detection mode relying on manual reporting into an active early warning mode, reducing false alarm and missed alarm rates, and enabling rapid and accurate location of pipe burst events, providing a reliable data source for emergency management of industrial water supply systems.

[0050] 2. This invention delineates the affected area by weighting and fusing the weights of multi-order candidate nodes with water source dependence. It assigns decreasing weights to first-, second-, and third-order candidate nodes and calculates the weighted dependence based on the proportion of water source paths passing through the marked affected area. This weighted dependence is then compared with a preset threshold to accurately delineate the affected area. This invention comprehensively considers the physical law of the attenuation of pipe burst impact with distance and the influence of the circuitous water supply capacity in a ring network on node dependence, avoiding over-expansion of the affected area or omission of key nodes, thus improving the processing efficiency and reliability of the plant-network collaborative optimization operation system at the industrial data level.

[0051] 3. This invention uses the maximum value among three coefficients—pressure reduction coefficient, water quality degradation coefficient, and insufficient water supply coefficient—to represent the degree of impact on users. Different weights are assigned based on the social importance of each user. A weighted fusion method is used to obtain a primary set of valves to be shut off. This set is then further filtered according to leakage volume to obtain a secondary set of valves to be shut off. A dynamic programming method is then used to solve for the optimal shut-off sequence and corresponding water hammer impact value for each candidate valve in the secondary set, ultimately determining the optimal set of valves to be shut off. This system comprehensively considers multiple objectives in risk management, including social equity, resource conservation, and infrastructure safety, providing a scientific basis for emergency dispatch decisions.

[0052] 4. This invention monitors the pressure residuals at boundary nodes in real time after executing scheduling commands and dynamically calculates the residual threshold based on the measured pressure change rate, avoiding erroneous adjustments triggered by sudden pressure changes. By classifying the slope of the fitted straight line of the residual sequence, instantaneous disturbances are only logged, and hydraulic model parameters are slowly corrected through drift. If the situation continues to deteriorate, the valve shut-off sequence is re-solved or the affected area is redefined based on the range of abnormal nodes. This invention can dynamically adjust its model and parameters according to the actual state of the pipeline network, improving the long-term prediction accuracy of the industrial data management system and ensuring the efficiency and reliability of emergency response in smart water plants. Attached Figure Description

[0053] The present invention will now be further described with reference to the accompanying drawings.

[0054] Figure 1 This is a framework diagram of the system provided in the embodiments of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Please see Figure 1 As shown, this invention is a plant-network collaborative optimization operation system for integrated smart water plants, comprising:

[0057] Anomaly Point Module S100: Constructs a water supply network topology map and collects network data in real time, detects pipe burst events and locates burst nodes.

[0058] In this embodiment of the invention, the specific process of constructing the water supply network topology is as follows:

[0059] Valves, pipe intersections, pipe diameter change points, water source points, and key water use points are set as nodes in the pipeline geographic information system and assigned unique numbers. Among them, pipe diameter change points refer to the locations where the pipe diameter changes.

[0060] Establish an edge between the two ends of each actual pipe and record the pipe length. All edges constitute an edge set.

[0061] Based on the edge set, generate an adjacent node set for each node to form a water supply network topology graph. Specifically, for any two nodes in the water supply network topology graph, if there is a pipeline connecting the two nodes, then the two nodes are called adjacent nodes.

[0062] Pipeline network data refers to a collection of information reflecting the hydraulic and equipment status of the pipeline network, specifically including node pressure, pipe segment flow, valve opening, pipe length, and node adjacency relationships;

[0063] Node pressure refers to the hydrostatic pressure exerted on the water body at a certain node location in the water supply network topology diagram. It is obtained by installing pressure sensors at the network nodes.

[0064] Pipe flow rate refers to the volume of water passing through the cross-section of a pipe section per unit time, which is obtained by installing an ultrasonic flow meter on the pipe section.

[0065] Valve opening degree refers to the extent to which a valve is opened, and is represented by a normalization coefficient of 0 to 1, where 0 represents fully closed and 1 represents fully open.

[0066] Pipe length refers to the actual physical length of the pipe connecting two adjacent nodes in a water supply network, while node adjacency refers to the connection relationship between nodes in the water supply network topology diagram. Both are obtained through the water supply network topology diagram.

[0067] The specific process for detecting pipe burst events and locating the burst point is as follows:

[0068] Obtain the pressure gradient between target node i and each of its neighboring nodes j at the same time. The formula is: ,in, This represents the node pressure value of target node i. This represents the node pressure value of node j. This represents the length of the pipe connecting target node i and node j;

[0069] After obtaining the pressure gradients corresponding to all neighboring nodes j, calculate the arithmetic mean of all pressure gradients, calculate the deviation of the pressure gradient of each neighboring node from the arithmetic mean of all pressure gradients, square each deviation and add them together to get the sum of squared deviations, and then divide the sum of squared deviations by the total number of neighboring nodes to get the pressure gradient variance of the target node i.

[0070] If the pressure gradient variance of target node i exceeds the preset variance threshold for three consecutive time periods, and the node pressure decrease rate of target node i is also lower than the preset decrease rate threshold, then it is determined that a pipe burst event has occurred near target node i, and target node i is a pipe burst node.

[0071] Because pipe bursts are continuous, transient disturbances can be eliminated by monitoring the variance over three consecutive time periods. At the same time, an increase in pressure gradient variance may also be caused by operations such as valve adjustment and pump switching, but these operations will not cause a rapid drop in pressure at the same time. The essential characteristic of a pipe burst is that the leakage from the rupture causes an instantaneous drop in pressure. By comparing the pressure drop rate with a preset drop rate threshold, pipe bursts can be effectively distinguished from other non-leakage disturbances. Only when both conditions are met can a pipe burst event be judged, thus reducing the false judgment rate.

[0072] The variance threshold is calculated by pre-collecting historical data of the pressure gradient variance of target node i under normal operating conditions, then calculating the arithmetic mean and standard deviation of all collected variance values, and adding three times the standard deviation to the mean as the variance threshold. This ensures that the variance value of target node i is below this variance threshold under normal operating conditions, and may exceed this variance threshold only when an anomaly occurs. At the same time, the variance threshold of each node is calculated independently and does not affect each other.

[0073] The pressure drop rate threshold is determined by collecting pressure drop rate data from multiple actual pipe burst events, statistically analyzing the minimum absolute value of the pressure drop rate across all burst events, and then taking the negative of this minimum absolute value as the threshold value. If no historical data is available, a commonly used empirical value in the field of water supply network engineering, namely -0.05 MPa / s, is adopted.

[0074] Specifically, under normal and stable operating conditions, the pressure gradient between each node in the pipeline network and its adjacent nodes is spatially consistent. When a pipe bursts, the pressure at the burst point drops instantly, causing the pressure gradient between the burst node and its upstream adjacent nodes to increase sharply, while the pressure gradient between the burst node and its downstream adjacent nodes may decrease, thus disrupting the original spatial consistency. The pressure gradient variance is a statistical indicator that reflects the degree of disruption of spatial consistency. The larger the pressure gradient variance, the more significant the difference in pressure gradients on the pipelines around the corresponding node, the more uneven the local pressure distribution, and the higher the probability of a pipe burst.

[0075] Furthermore, the aforementioned plant-network collaborative optimization operation system for integrated smart water plants also includes:

[0076] Region delineation module S200: Based on the burst pipe node, delineate the affected region and extract the candidate valve set, boundary node set and minimum connected subgraph within the affected region.

[0077] In this embodiment of the invention, the specific process of delineating the affected area based on the burst pipe node is as follows:

[0078] Mark all adjacent nodes directly connected to the burst pipe node as the main affected area;

[0079] Nodes that are first-order connected, second-order connected, and third-order connected to any node in the main influence area are selected respectively to form a candidate node set. Different weights are assigned to each candidate node, and the weights of the first-order connected, second-order connected, and third-order connected nodes decrease in that order.

[0080] In the topology graph, query the water source path for each candidate node, count the number m of paths m that pass through any node in the main influence area, and the total number of paths M. Calculate the water source dependency of unmarked nodes. The calculation formula is as follows: ;

[0081] The weighted dependency of each candidate node is obtained by weighting and fusing the water source dependency of each candidate node with its corresponding weight.

[0082] A preset dependency threshold is set, and it is compared with the weighted dependency of each candidate node. If the weighted dependency of a candidate node is not less than the dependency threshold, the candidate node is marked as a secondary influence area. Otherwise, the node is considered not to be directly affected by the pipe burst and is not included in the affected area.

[0083] The primary and secondary affected areas are designated as the affected areas.

[0084] It should be noted that the weights of nodes connected in the first, second, and third order decrease sequentially because the impact of a pipe burst gradually weakens as the propagation distance increases. The dependency threshold is the optimal threshold determined by collecting data on the measured impact of nodes in multiple actual pipe burst events and their corresponding theoretical water source dependency, using an accuracy maximization method.

[0085] The process of obtaining the candidate valve set is as follows:

[0086] Identify all valves located within the affected area and check if each valve has a pipe directly connected to an external node of the affected area. If so, mark the valve as a candidate valve. The set of all candidate valves is the candidate valve set. These valves are the control points for isolating the burst pipe node. Closing these valves can isolate the burst pipe node from the water source.

[0087] The process of obtaining the boundary node set is as follows:

[0088] Find all nodes located inside the affected region, and check if each node has any edge connecting it to nodes outside the affected region. If so, mark the node as a boundary node. The set of all boundary nodes is the boundary node set.

[0089] A minimum connected subgraph is a subgraph consisting of all nodes within an affected region and the edges connecting these nodes, such that the subgraph covers all nodes within the affected region but excludes any nodes outside the affected region. The process for obtaining this subgraph is as follows:

[0090] Find the shortest path from the burst pipe node to every node in the set of boundary nodes from the water supply network topology map;

[0091] Record the number of edges traversed in each shortest path, and obtain the maximum number of edges corresponding to all boundary nodes;

[0092] From the water supply network topology graph, obtain all nodes that can be reached from the burst pipe node by the maximum number of edges, as well as the edges connecting these nodes. Combine these nodes and edges to form a minimum connected subgraph.

[0093] Specifically, by assigning decreasing weights to candidate nodes of different orders, the physical law that the impact of a pipe burst gradually weakens as the propagation distance increases is reflected, avoiding the wrong inclusion of distant nodes and improving the rationality of area delineation. By obtaining the weighted dependence of each candidate node through its water source dependence and corresponding weight, it is possible to effectively distinguish between nodes in the ring network that can be supplied by other water sources and nodes that are truly dependent on the affected area, reducing misjudgments and improving accuracy.

[0094] Furthermore, the aforementioned plant-network collaborative optimization operation system for integrated smart water plants also includes:

[0095] Instruction generation module S300: Based on the candidate valve set and the minimum connected subgraph, obtain the optimal valve closing set and generate collaborative scheduling instructions.

[0096] In this embodiment of the invention, the optimal set of valves to be closed is determined by the degree of user impact, leakage volume, and water hammer impact value in the candidate set of valves to be closed. This is because the degree of user impact reflects the impact of valve closure on social water supply services, leakage volume reflects water waste and potential secondary disaster risks, and water hammer impact reflects the peak pressure fluctuation of the pipeline during the valve closure process. The optimal combination of the three factors needs to be selected.

[0097] The degree to which users are affected is determined by the maximum value among the pressure reduction coefficient, water quality degradation coefficient, and insufficient water supply coefficient. This is because users' water experience is constrained by the worst factor. If any one of the pressure, water quality, or water supply deteriorates significantly, users will not be able to use water normally even if the other two are normal. Therefore, taking the maximum value of the three values ​​best reflects the actual degree of service degradation.

[0098] The pressure drop is due to a pipe rupture causing a decrease in pressure near the breach. The resulting pressure wave propagates and causes a subsequent drop in water pressure at the user's location. The pressure reduction coefficient ranges from [value missing]. The larger the value, the lower the pressure. The process for obtaining the pressure reduction coefficient is as follows:

[0099] First, obtain the normal water pressure value at the user node. After closing each candidate valve set, calculate the actual pressure value at the user node using the hydraulic model. If the actual pressure is not lower than the normal water pressure value, the pressure reduction coefficient is 0. If the actual pressure value is 0, the pressure reduction coefficient is 1. If the actual pressure value is between 0 and the normal water pressure value, first calculate the ratio of the actual pressure value to the normal water pressure value. The difference between 1 and this ratio is the pressure reduction coefficient.

[0100] Water quality degradation occurs when a burst pipe causes negative pressure that draws in contaminants, or when a change in water flow direction washes away deposits from the pipe walls, resulting in turbidity and a decline in water quality at the user's location. The range of values ​​for the water quality degradation coefficient is as follows: The larger the value, the lower the pressure. The process for obtaining the water quality degradation coefficient is as follows:

[0101] First, obtain the turbidity threshold under normal operating conditions at the user node. After closing each candidate valve set, simulate the actual turbidity value at the user node through the water quality model. If the actual turbidity is not higher than the normal turbidity threshold, the water quality degradation coefficient is 0. If the actual turbidity is not lower than the upper limit of unacceptable water quality, the water quality degradation coefficient is 1. If the actual turbidity is between the normal threshold and the upper limit of unacceptable water quality, first calculate the difference between the actual turbidity and the normal threshold, and then divide it by the difference between the upper limit of unacceptable water quality and the normal threshold. The resulting ratio is the water quality degradation coefficient.

[0102] Insufficient water supply is caused by a pipe burst leading to a decrease in the total water supply of the pipeline network, or by valve closure causing a drop in pressure in some areas, resulting in a slower flow rate. The actual flow rate received by users is insufficient to meet their water demand. The range of values ​​for the insufficient water supply coefficient is as follows: The larger the value, the more insufficient the water supply. The process of obtaining the water supply insufficiency coefficient is as follows:

[0103] First, obtain the normal water demand at the user node. After closing each candidate valve set, calculate the actual supply flow at the user node using the hydraulic model. If the actual supply flow is not lower than the normal water demand, the water supply insufficiency coefficient is 0. If the actual supply flow is 0, the water supply insufficiency coefficient is 1. If the actual supply flow is between 0 and the normal water demand, first calculate the ratio of the actual supply flow to the normal water demand. The difference between 1 and this ratio is the water supply insufficiency coefficient.

[0104] Water hammer is a pressure fluctuation caused by the sudden change in water flow velocity in a pipeline when a valve is closed rapidly. Excessive water hammer can damage the pipeline. The range of water hammer values ​​is as follows: The higher the value, the higher the risk of water hammer damage;

[0105] The process of obtaining the water hammer impact value is as follows: First, the pipeline flow velocity before the valve is closed is obtained through the hydraulic model. Then, it is multiplied with the water density and pressure wave velocity to obtain the single valve water hammer impact index. For multiple valves that need to be closed in sequence, the pressure wave superposition effect under different closing sequences is simulated through dynamic programming. The maximum water hammer pressure value that occurs in the whole process is taken as the water hammer impact value.

[0106] It should be noted that the calculation of node pressure values, actual supply flow rate, pipeline flow velocity before valve closure, and actual turbidity value by water quality model are conventional technical methods in the field of water supply network, and will not be elaborated here.

[0107] Leakage refers to the total amount of water lost from the time a pipe burst occurs until all valves in the corresponding candidate valve set close. Specifically, it is obtained by first determining the closing time required for each candidate valve to fully close, and then identifying the maximum value among the required closing times. Next, the initial flow rate q ejected from the burst point is estimated. This initial flow rate is obtained by measuring the rate of pressure drop near the burst point. Since the burst flow rate decays over time, ... ,in, This represents the instantaneous leakage flow rate at time t2 after the pipe burst. To reflect the rate at which the flow rate of a burst pipe decreases over time, an attenuation coefficient is used. Then, the integral value of the instantaneous flow rate is calculated over the entire time period from the moment the pipe bursts until the maximum value of the required shut-off time is determined. This integral result is the leakage water volume under the corresponding candidate valve shut-off set.

[0108] The specific process for obtaining the optimal set of valves to be closed is as follows:

[0109] A user level and weight table is pre-established, each user is labeled with a level type, and users are assigned corresponding weights according to the level type.

[0110] Find all the minimum valve combinations that can disconnect the burst pipe from all water sources on the minimum connected subgraph. Each minimum valve combination constitutes a candidate valve shut-off set.

[0111] Obtain the user impact level, leakage water volume, and water hammer impact value for each candidate valve closure set;

[0112] The user impact level of each candidate valve set is weighted and fused with the user's corresponding weight to obtain the weighted total impact. The first preset number of candidate valve sets with the smallest weighted total impact value are selected as the first-level screening valve sets.

[0113] The primary screening valve sets are sorted in order of increasing leakage volume, and the first two preset number of primary screening valve sets are retained as secondary screening valve sets.

[0114] For each candidate valve set in the secondary screening valve set, the optimal closing sequence is solved using dynamic programming, and the minimum water hammer impact value under this sequence is calculated. The candidate valve set with the minimum water hammer impact value is selected as the final optimal valve set, and its corresponding closing sequence is the optimal closing sequence. If multiple candidate valve sets have the same water hammer impact value, the candidate valve set with the smallest leakage is selected as the final optimal valve set.

[0115] It should be noted that the user rating level is determined based on the user's social importance and the nature of water use. Users who undertake life-saving, safety protection and public education functions are rated as Level 1 users, such as hospitals, fire departments and schools. Users whose water outages would cause significant social impact or economic losses are rated as Level 2 users, such as large residential areas and industrial parks. Ordinary residents are rated as Level 3 users, and greening and sanitation are rated as Level 4 users.

[0116] Users are assigned corresponding weights based on their level type. For example, a level 1 user has a weight of 100, a level 2 user has a weight of 10, a level 3 user has a weight of 1, and a level 4 user has a weight of 0.1.

[0117] The first preset number is the number of preferred valve sets to be selected and retained based on the degree of user impact. It is dynamically adjusted according to the pipeline network scale and real-time requirements, and the value is usually 5 to 10. If the value is too small, potential optimal valve sets may be missed, and if it is too large, the computational burden will be increased. If the total number of selected preferred valve sets is less than the first preset number, all of them will be retained.

[0118] The second preset number is the number of optimal valve sets to retain after sorting by leakage volume. It is dynamically adjusted according to the network scale and real-time requirements, and the value is usually 3 to 5. The second preset number of schemes with the smallest leakage volume are retained to facilitate subsequent optimization. Similarly, if the total number of optimal valve sets to retain is less than the second preset number, all of them are retained.

[0119] Using dynamic programming to find the optimal closing sequence is a common technique in combinatorial optimization. The basic idea is to represent the set of closed valves using a binary bitmask, record the maximum water hammer peak value under the current closing sequence, and then enumerate the next valve to be closed and check for upstream water constraints to perform state transitions. Finally, the complete sequence that minimizes the water hammer peak value is obtained recursively. The specific implementation method is common knowledge in this field and will not be elaborated here.

[0120] The collaborative scheduling instructions specifically include valve control instructions, water plant scheduling instructions, and information release instructions. Specifically, the valve control instructions send closing commands to each valve in the optimal valve set according to the optimal closing sequence. The water plant scheduling instructions are based on the hydraulic balance calculation results of the remaining pipeline network after isolation, and issue pressure reduction or production reduction instructions to the water plant. The information release instructions generate a list of affected users and the estimated recovery time, and push them to the customer service center.

[0121] Specifically, degradation coefficients are calculated from three dimensions: pressure, water quality, and water supply, and the maximum value is taken as the user impact coefficient to comprehensively reflect the degree of damage to water supply services caused by pipe bursts. Then, different users are assigned corresponding level weights to achieve priority protection for key users. Finally, the optimal valve set and optimal closing sequence are determined according to three objectives: weighted total impact, leakage water volume, and water hammer impact value. This ensures optimization quality while reducing computational complexity and meeting the data management requirements of real-time emergency response.

[0122] Furthermore, the aforementioned plant-network collaborative optimization operation system for integrated smart water plants also includes:

[0123] Adaptive adjustment module S400: Executes collaborative scheduling instructions, monitors the pressure residuals of boundary nodes in real time, and makes adaptive adjustments based on the magnitude of the residuals.

[0124] In this embodiment of the invention, the boundary node pressure residual refers to the absolute value of the difference between the measured pressure and the predicted pressure at the boundary node at the same moment;

[0125] The measured pressure was acquired at a preset sampling frequency by pressure sensors installed on the boundary point set during the execution of the coordinated scheduling instructions.

[0126] The predicted pressure is the predicted value calculated by the hydraulic model for the same boundary node and aligned with the measured pressure in the time dimension.

[0127] Obtain the residual threshold, which is used to determine whether the residual between the actual pressure and the hydraulic model prediction is within an acceptable range:

[0128] During the valve closing process, the node pressure will fluctuate rapidly. At this time, the instantaneous deviation between the measured pressure and the predicted pressure will naturally increase. If a fixed residual threshold is used, the instantaneous deviation will be greater than the residual threshold, causing unnecessary back-reaction calculations and affecting stability.

[0129] The formula for calculating the residual threshold is: ;

[0130] in, Indicates the residual threshold;

[0131] This represents the basic threshold, indicating that the residual must be less than this basic threshold even when the pressure is stable to be considered acceptable. It is obtained from engineering experience and is usually set to 0.01 MPa.

[0132] This represents the proportionality coefficient, obtained by fitting historical data.

[0133] This represents the measured pressure at the current time t3. The absolute value of the rate of change;

[0134] This indicates the pressure change rate limit. It means that the residual threshold will only increase when the pressure change rate exceeds this limit. It is statistically derived from the maximum change rate of pressure fluctuation under normal operating conditions. Changes below this value are considered normal disturbances, and it is usually taken as 0.02 MPa.

[0135] The pressure residuals of the boundary nodes are compared with the residual threshold. If the pressure residuals of all boundary nodes are less than the residual threshold, the scheduling instruction is determined to meet the expected effect, and the current collaborative scheduling instruction is maintained.

[0136] If the pressure residual at a boundary node is not less than the residual threshold and the duration exceeds the preset time threshold, then the residual data of the most recent sampling times of the corresponding boundary node are extracted, and a straight line is obtained by fitting using the least squares method. The slope of the straight line reflects the trend of the residual changing with time.

[0137] It should be noted that the time threshold is set based on the sampling period of the pressure sensor and the duration of hydraulic transients in the pipeline network. It is usually set to 3 to 5 times the sampling period, which is an inherent parameter of the pressure sensor.

[0138] If the absolute value of the slope is less than the first threshold, and the residual falls back to within the residual threshold within the preset time threshold, it is determined to be an instantaneous disturbance, and only the log is recorded without triggering any adjustment.

[0139] If the absolute value of the slope is between the first threshold and the second threshold, it is determined to be a slow drift, triggering the correction of hydraulic model parameters. Specifically, the Kalman filter is used to correct the uncertain parameters in the hydraulic model, such as the pipe roughness coefficient and the node water demand. The corrected hydraulic model is immediately used for subsequent pressure prediction. If the corrected residual converges within a specified time, the original coordinated scheduling command is continued to be executed. If the corrected residual still does not converge, it is upgraded to continuous deterioration.

[0140] It should be noted that the specified time is an empirical threshold, usually set to 5 seconds;

[0141] If the absolute value of the slope is greater than or equal to the second threshold, it is determined to be a continuous deterioration, and a return instruction is executed. Specifically, if only the residual of a single boundary node continues to increase, while other boundary nodes are normal, it is determined to be a local valve execution delay or sensor failure, and the system returns to the instruction generation module. Under the premise of keeping the affected area unchanged, the optimal valve closing set is resolved. If the residuals of multiple boundary nodes increase simultaneously and the trend is consistent, it is determined to be a hydraulic model deviation or an expansion of the pipe burst area, and the system returns to the region delineation module. Using the current measured pressure distribution as the initial condition, the dynamically affected region is re-delineated and the minimum connected subgraph is updated. Then, the instruction generation module is re-executed.

[0142] It should be noted that the first threshold and the second threshold are calculated by collecting multiple pipe burst times and boundary node pressure residual sequences under normal operating conditions, calculating the slope distribution of residual changes, setting the upper limit of the absolute value of the slope that may be reached under normal operating conditions as the first threshold, and setting the minimum value of the absolute value of the slope that can be reached during the deterioration of the pipe burst as the second threshold.

[0143] Specifically, the residual threshold is dynamically calculated based on the measured pressure change rate. When the pressure change rate is lower than the set limit, the residual threshold remains unchanged. When the change rate exceeds the limit, the residual threshold increases linearly with the change rate, enabling the residual threshold to adapt to the severity of pressure fluctuations. The slope obtained by fitting is used to distinguish between instantaneous disturbances, slow drifts, and continuous deterioration. Three strategies are adopted respectively: logging, parameter correction, and recalculation, to avoid unnecessary global recalculation and improve the efficiency of industrial data management.

[0144] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0145] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A plant-network collaborative optimization operation system for integrated smart water plants, characterized in that, include: The abnormal point module constructs a water supply network topology map and collects network data in real time, detects pipe burst events and locates the burst node. The region delineation module, based on the burst pipe node, delineates the affected region and extracts the candidate valve set, boundary node set, and minimum connected subgraph within the affected region; The instruction generation module obtains the optimal set of valves to be closed based on the candidate valve set and the minimum connected subgraph, and generates collaborative scheduling instructions. The adaptive adjustment module executes collaborative scheduling instructions, monitors the pressure residuals of boundary nodes in real time, and makes adaptive adjustments based on the magnitude of the residuals.

2. The plant-network collaborative optimization operation system for integrated smart water plants according to claim 1, characterized in that, The pipeline network data includes node pressure, pipe segment flow rate, valve opening degree, pipe length, and node adjacency relationships.

3. The plant-network collaborative optimization operation system for integrated smart water plants according to claim 2, characterized in that, The specific steps for detecting pipe burst events and locating burst nodes are as follows: Obtain the pressure gradient between the target node and each of its neighboring nodes at the same time. After obtaining the pressure gradients corresponding to all neighboring nodes of the target node, calculate the arithmetic mean of all pressure gradients. Calculate the deviation between the pressure gradient of each neighboring node and the arithmetic mean. Square each deviation and sum them to obtain the sum of squared deviations. Divide the sum of squared deviations by the total number of neighboring nodes to obtain the pressure gradient variance of the target node. If the pressure gradient variance of the target node exceeds a preset variance threshold for three consecutive time points, and the node pressure decrease rate of the target node is lower than a preset decrease rate threshold, then it is determined that a pipe burst event has occurred near the target node, and the target node is marked as a pipe burst node.

4. The plant-network collaborative optimization operation system for integrated smart water plants according to claim 3, characterized in that, The specific process for delineating the affected area based on the aforementioned burst pipe node is as follows: Mark all adjacent nodes directly connected to the burst node as the main influence area; select nodes that are first-order connected, second-order connected, and third-order connected to any node in the main influence area to form a candidate node set, and assign different weights to each candidate node. In the topology graph, query the water source path of each candidate node and calculate the water source dependency of the candidate node; then, weight and fuse the water source dependency of each candidate node with its corresponding weight to obtain the weighted dependency of each candidate node. A preset dependency threshold is set, and it is compared with the weighted dependency of each candidate node. If the weighted dependency of a candidate node is not less than the dependency threshold, the candidate node is marked as a secondary impact area. Otherwise, the node is considered not to be directly affected by the pipe burst and is not included in the affected area. The primary impact area and the secondary impact area are defined as the affected areas.

5. A plant-network collaborative optimization operation system for integrated smart water plants according to claim 4, characterized in that, The specific process for extracting the candidate valve set, boundary node set, and minimum connected subgraph within the affected area is as follows: Candidate valve set: Identify all valves located within the affected area. If a valve has a pipe directly connected to a node outside the affected area, mark that valve as a candidate valve. All candidate valves constitute the candidate valve set. Boundary node set: Find all nodes located inside the affected region. If a node has any edge connecting it to a node outside the affected region, then mark that node as a boundary node. All boundary nodes constitute the boundary node set. Minimum Connected Subgraph: Query the shortest path from the burst pipe node to each boundary node in the set of boundary nodes from the water supply network topology graph. Record the number of edges traversed by each shortest path. Obtain the maximum number of edges corresponding to all boundary nodes. Obtain all nodes that can be reached from the burst pipe node through the edge with the maximum number of edges, as well as the edges connecting these nodes, from the water supply network topology graph. Construct the minimum connected subgraph from these nodes and edges.

6. The plant-network collaborative optimization operation system for integrated smart water plants according to claim 5, characterized in that, The specific process for obtaining the optimal set of valves to be closed is as follows: Establish a pre-defined correspondence between user levels and weights, label each user with a level type and assign the corresponding weight; Find all the minimum valve combinations that can disconnect the burst pipe from all water sources on the minimum connected subgraph. Each minimum valve combination constitutes a candidate valve shut-off set. Obtain the user impact level, leakage water volume, and water hammer impact value for each candidate valve closure set; The user impact level of each candidate valve set is weighted and fused with the user's corresponding weight to obtain the weighted total impact. The first preset number of candidate valve sets with the smallest weighted total impact value are selected as the first-level screening valve sets. The first-level screening valve sets are sorted in order of increasing leakage volume, and the first second preset number of first-level screening valve sets are retained as second-level screening valve sets. For each candidate valve set in the secondary screening valve set, the optimal closing sequence is solved using dynamic programming, and the minimum water hammer impact value under this sequence is calculated. The candidate valve set with the minimum water hammer impact value is selected as the optimal valve set. If the water hammer impact values ​​of multiple candidate valve sets are the same, the one with the smallest leakage is selected as the optimal valve set.

7. A plant-network collaborative optimization operation system for integrated smart water plants according to claim 6, characterized in that, The specific process for monitoring the pressure residual at the boundary node is as follows: During the execution of the coordinated scheduling instructions, the measured pressure of the boundary nodes is obtained by pressure sensors installed on the boundary point set at a preset sampling frequency; The predicted pressure is calculated for the same boundary node using a hydraulic model and then aligned with the measured pressure in the time dimension. The absolute value of the difference between the measured pressure and the predicted pressure at the boundary node is the boundary node pressure residual.

8. A plant-network collaborative optimization operation system for integrated smart water plants according to claim 7, characterized in that, The specific process of performing the closed-loop feedback is as follows: Obtain the residual threshold, compare the pressure residual of each boundary node with the residual threshold, and if the pressure residual of all boundary nodes is less than the residual threshold, maintain the current cooperative scheduling instruction. If the pressure residual of a boundary node is not less than the residual threshold and the duration exceeds the preset time threshold, then the residual data of the boundary node at the most recent sampling time are extracted, and a straight line is obtained by fitting the least squares method. The slope of the straight line reflects the trend of the residual changing with time. Based on the comparison results of the absolute value of the slope with the first threshold and the second threshold, a graded adjustment is performed.

9. A plant-network collaborative optimization operation system for integrated smart water plants according to claim 8, characterized in that, The specific process for performing the hierarchical adjustment is as follows: If the absolute value of the slope is less than the first threshold, and the residual falls back to within the residual threshold within a preset time threshold, it is determined to be an instantaneous disturbance and no adjustment is made. If the absolute value of the slope is between the first threshold and the second threshold, it is determined to be a slow drift, and the hydraulic model parameters are corrected; if the corrected residual converges within a specified time, the original cooperative scheduling instruction is continued to be executed; otherwise, it is upgraded to continuous deterioration processing. If the absolute value of the slope is greater than or equal to the second threshold, it is determined to be a continuous deterioration, and a return instruction is executed.

10. A plant-network collaborative optimization operation system for integrated smart water plants according to claim 9, characterized in that, The specific process of executing the return instruction is as follows: If the residual of a single boundary node continues to increase while other boundary nodes remain normal, return to the instruction generation module and resolve for the optimal valve set while keeping the affected area unchanged. If the residuals of multiple boundary nodes increase simultaneously and have the same trend, return to the region delineation module, redefine the affected region and update the minimum connected subgraph, and re-execute the instruction generation module.

Images