A water supply network optimization scheduling method and system based on multi-model coupling

Abstract

The present invention provides a water supply network optimization scheduling method and system based on multi-model coupling, comprising: collecting basic data on regional pipe networks and establishing a water supply network hydraulic model; obtaining water supply network operation scheduling targets and establishing a water supply network scheduling scenario model; obtaining historical monitoring data of the water supply network and using a neural network algorithm to construct a node water demand prediction model and a node water demand time-varying curve prediction model, which respectively outputs the node predicted water demand and the node water demand predicted time-varying curve; constructing a water supply network optimization scheduling model based on multi-model coupling, using an improved particle swarm algorithm for optimization, outputting a water supply network optimization scheduling plan, introducing optimization scheduling evaluation indicators to evaluate the scheduling plan; and constructing a water supply network optimization scheduling plan screening model to obtain the optimal water supply network optimization scheduling plan. The present invention adopts a multi-model coupling approach, effectively reducing prediction errors, improving optimization efficiency, and realizing intelligent and precise scheduling of the water supply network.

Description

Technical Field

[0001] The present invention belongs to the field of water supply network optimization scheduling, and specifically relates to a water supply network optimization scheduling method and system based on multi-model coupling. Background Art

[0002] With the continuous expansion of urban scale and the continuous increase of urban population, the scale and topological complexity of urban water supply network systems are also gradually increasing. Traditional experience-based manual scheduling methods cannot achieve intelligent and precise control, which easily leads to a large amount of energy waste. In addition, problems such as imbalance between supply and demand, insufficient water pressure during peak water consumption, excessive fluctuation of pipe network pressure, and increased probability of pipe burst often occur. It is difficult to adapt to the current water supply network scheduling needs. Therefore, it is very necessary to provide an intelligent and precise water supply network scheduling solution.

[0003] Chinese patent CN118644015A discloses a pipe network optimization scheduling system and method based on a hydraulic model and a neural network algorithm. This system constructs a pipe network hydraulic model based on collected water supply network structural data to simulate and analyze the dynamic behavior of water flow in the pipe network. A scheduling model is constructed based on historical pipe network water supply data and a neural network algorithm to learn and predict future water demand and water supply patterns, thereby improving scheduling accuracy and efficiency. Chinese patent CN115860192A discloses a water supply network optimization scheduling method based on a fuzzy neural network and a genetic algorithm. The water supply flow and pressure parameters of each water plant's water supply pump station are input into a trained neural network model to predict the flow and pressure values at each measuring point in the water supply network. The genetic algorithm is then used to optimize the water supply flow and pressure parameters at each water plant's water supply pump station. These methods utilize only a single neural network-based prediction model, resulting in large prediction errors and requiring a high sampling frequency, making it difficult to ensure the accuracy of the scheduling plan. Summary of the Invention

[0004] In response to the problems existing in the above-mentioned urban water supply network system and the limitations of the current method of obtaining the optimal scheduling plan for the water supply network, the present invention provides a water supply network optimal scheduling method and system based on multi-model coupling.

[0005] Based on the above purpose, the technical solution adopted by the present invention is as follows:

[0006] A water supply network optimization scheduling method based on multi-model coupling includes the following steps:

[0007] S1. Collect basic data of regional pipe network and establish hydraulic model of water supply network;

[0008] S2. Obtain the water supply network operation scheduling target and establish a water supply network scheduling scenario model;

[0009] S3. Obtain historical monitoring data of the water supply network, randomly divide the historical monitoring data of the water supply network into a training set and a validation set in proportion, use a neural network algorithm to construct a node water demand prediction model and a node water demand time-varying curve prediction model, which respectively output the node predicted water demand and the node water demand prediction time-varying curve, and use the training set to train the node water demand prediction model and the node water demand time-varying curve prediction model, respectively, to obtain a trained node water demand prediction model and a trained node water demand time-varying curve prediction model, respectively;

[0010] S4. Input the validation set into the trained node water demand prediction model and the trained node water demand time-varying curve prediction model, respectively, to generate node water demand prediction results and node water demand time-varying curve prediction results, respectively, set water demand prediction evaluation indicators and water demand time-varying curve evaluation indicators, and if the prediction results meet the evaluation criteria, respectively obtain the trained node water demand prediction model and the trained node water demand time-varying curve prediction model, and proceed to S5; if they do not meet the evaluation criteria, return to S3;

[0011] S5. Couple the water supply network hydraulic model, the water supply network scenario scheduling model, the node water demand prediction model, and the node water demand time-varying curve prediction model to construct a water supply network optimization scheduling model based on multi-model coupling, optimize the water supply network optimization scheduling model based on multi-model coupling using an improved particle swarm algorithm, and output a water supply network optimization scheduling plan;

[0012] The improved particle swarm algorithm is based on the standard particle swarm algorithm, including improving the inertia weight, adding a penalty function and limiting the particle speed and particle position range;

[0013] The calculation formula of the improved inertia weight is:

[0014] ;

[0015] Where, oh op represents the improved inertia weight, oh max represents the initial inertia weight, oh min represents the inertia weight at the maximum number of iterations, s represents the density of feasible solutions in the particle swarm;

[0016] The expression of the penalty function is:

[0017] ;

[0018] Where, fpe ( x ) represents the penalty function, where the penalty coefficient , m and c Indicates a non-negative parameter that needs to be debugged. α and β is a value not less than 1;

[0019] The particle velocity and particle position ranges are specifically limited as follows:

[0020] ;

[0021] Where, v max and v min Represent the maximum and minimum speeds of the particles, respectively. x max and x min represent the maximum position and minimum position of the particle respectively;

[0022] S6. Introduce the optimization scheduling evaluation index to evaluate the water supply network optimization scheduling plan. If it meets the optimization scheduling evaluation criteria, save the water supply network optimization scheduling plan and continue iterative calculation. If it does not meet the optimization scheduling evaluation criteria, proceed to step S7.

[0023] S7. Construct a water supply network optimization scheduling scheme screening model to obtain the optimal water supply network optimization scheduling scheme.

[0024] Preferably, step S1 also includes: the regional pipeline network basic data includes pipeline network GIS attributes, user revenue meter reading data, pipeline network flow monitoring data, pipeline network pressure monitoring data, pool water level monitoring data, water pump characteristic data, water pump operation data, month, weather, temperature and holiday distribution; the pipeline network GIS attributes include pipeline network topology information, pipeline material, pipeline size, node number and DMA partition number.

[0025] Preferably, step S2 further includes: the water supply network operation scheduling objectives include meeting water supply pressure requirements, reducing system operation energy consumption, reducing pressure fluctuations within the network, and reducing network leakage rate;

[0026] The water supply network scheduling scenario model includes an objective function and constraints. The objective function includes minimizing the total energy consumption. The calculation formula for the total energy consumption is:

[0027] ;

[0028] Where, E ( n) represents the total energy consumption of all pumps operating for 24 hours a day, t Any time of day, r represents the water density, g represents the acceleration due to gravity, Q i,t Indicates the i The water pump is t Traffic flow during the period, H i,t Indicates the i The water pump is t The lift during the period, or i,t Indicates the i The water pump is t Work efficiency during each period;

[0029] The constraint condition is expressed as:

[0030] ;

[0031] Where, n i,t Indicates the i The water pump is t The revolution ratio of the time period, P j,t Indicates the j The control point is t The pressure of time, V k,t Indicates the k The pool is in t The volume of water during a period.

[0032] Preferably, step S3 also includes: the water supply network historical monitoring data includes node daily meter reading water volume historical monitoring data and DMA partition main flow historical monitoring data, and the GA-BP neural network is used to respectively construct the node water demand prediction model and the node water demand time-varying curve prediction model.

[0033] The node daily meter reading water volume historical monitoring data is randomly divided into a training set 1 and a validation set 1 in a ratio of 8:2, the training set 1 is used to train the node water demand prediction model to obtain the trained node water demand prediction model;

[0034] The DMA partition main flow historical monitoring data is randomly divided into training set 2 and validation set 2 in a ratio of 8:2. The training set 2 is used to train the node water demand time-varying curve prediction model to obtain the trained node water demand time-varying curve prediction model.

[0035] Preferably, step S4 further includes: inputting the validation set 1 into the trained node water demand prediction model for verification to generate the node water demand prediction result; inputting the validation set 2 into the trained node water demand time-varying curve prediction model for verification to obtain a DMA partition main flow prediction time-varying curve, normalizing the DMA partition main flow prediction time-varying curve, and combining it with the node predicted water demand to generate the node water demand time-varying curve prediction result;

[0036] The water demand prediction evaluation indicators include NSE wd , and its calculation formula is:

[0037] ;

[0038] Where, q s Indicates the s The measured water demand of each node, Indicates the s The predicted water demand of each node, Indicates the s The average measured water demand of each node, the water demand forecast evaluation standard is NSE wd ≥0.8;

[0039] The water demand time-varying curve prediction evaluation index includes NSE tc , and its calculation formula is:

[0040] ;

[0041] Where, q l Indicates the l The measured flow rate of each pipe section, Indicates the l The predicted flow rate of each pipe section, Indicates the l The average measured flow rate of each pipe section, the water demand time change curve prediction evaluation standard is the NSE tc ≥0.8.

[0042] Preferably, step S6 also includes: the optimization scheduling evaluation index includes the number of iterations T. When T≤150, it meets the evaluation criteria, the water supply network optimization scheduling plan is saved, and the iterative calculation is continued. Otherwise, step S7 is entered.

[0043] Preferably, step S7 further includes: the water supply network optimization scheduling scheme screening model includes pump station operating costs, and the expression of the pump station operating costs is:

[0044] ;

[0045] Where, F cost represents the operating cost of the pump station, δ represents the electricity price, N t represents the number of pumps running in period t, It represents the motor efficiency of the i-th water pump in the t-th time period. The optimal water supply network optimization scheduling plan includes the lowest operating cost of the pump station.

[0046] The present invention also provides a water supply network optimization scheduling system based on multi-model coupling, which includes the following modules:

[0047] Data collection module: used to collect basic data of regional pipe network, operation and scheduling targets of water supply network and historical monitoring data of water supply network;

[0048] Model construction module: used to build a water supply network hydraulic model, a water supply network scheduling scenario model, a node water demand prediction model, a node water demand time-varying curve prediction model, a water supply network optimization scheduling model based on multi-model coupling, and a water supply network optimization scheduling scheme screening model;

[0049] Model training module: used to train and verify the node water demand prediction model and the node water demand time-varying curve prediction model;

[0050] Scheduling evaluation module: used to evaluate the water supply network optimization scheduling scheme output by the water supply network optimization scheduling model based on multi-model coupling;

[0051] Scheme screening module: used to obtain the optimal water supply network optimization scheduling scheme.

[0052] The advantages of the present invention compared to existing methods are:

[0053] (1) The present invention adopts a first-level optimization scheduling method to directly and uniformly schedule the water pumps of the water plant and pumping stations at all levels. Compared with the second-level optimization scheduling method of the traditional water supply network, it can effectively deal with the problems of high energy consumption of water supply network scheduling, insufficient water supply pressure, large pressure fluctuations in the network, and high leakage, thereby achieving a reasonable allocation of water supply resources and improving the overall operating efficiency and stability of the water supply system;

[0054] (2) The present invention supports the deployment of an overall solution with the existing or newly added infrastructure such as flow and pressure monitoring points and variable frequency water pumps, significantly reducing the renovation and operation and maintenance costs of the water supply network system;

[0055] (3) In the prior art, only a single neural network model is used to predict water demand. This method has problems such as large prediction error and high sampling frequency requirements. The present invention, based on the GA-BP neural network algorithm, constructs a node water demand prediction model and a node water demand time-varying curve prediction model. Through the coupling effect of the two prediction models, the total amount of water demand and time-varying characteristics are considered at the same time, and the accurate prediction of the node water demand timely changing curve is achieved, which is more in line with the application scenario of actual water supply network operation and scheduling;

[0056] (4) The present invention makes a number of improvements based on the standard particle swarm algorithm, including improving the inertia weight, adding a penalty function, and limiting the particle speed and position range, which greatly reduces the number of parameters that need to be adjusted. It constructs a water supply network optimization scheduling model based on multi-model coupling, and optimizes the model based on the improved particle swarm algorithm. It supports automatic adjustment and optimization based on the optimization results, which significantly improves the optimization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0058] Figure 1 This is a flow chart of a method for optimizing scheduling of a water supply network based on multi-model coupling according to an embodiment of the present invention;

[0059] Figure 2 Schematic diagram of the architecture of a water supply network optimization scheduling method based on multi-model coupling according to an embodiment of the present invention;

[0060] Figure 3 This is a flowchart of an optimization model for water supply network optimization based on multi-model coupling according to an embodiment of the present invention;

[0061] Figure 4 It is a structural diagram of a water supply network optimization scheduling system based on multi-model coupling according to an embodiment of the present invention. DETAILED DESCRIPTION

[0062] To make the objectives, technical solutions, and advantages of the present invention more clearly understood, the technical solutions of the present invention are described clearly and completely below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments described are only a portion of the embodiments of the present invention, not all of them. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0063] See also Figure 1 and Figure 2 As shown, the present invention provides a water supply network optimization scheduling method based on multi-model coupling, which includes the following steps:

[0064] S1. Collect basic data of regional pipe network and establish hydraulic model of water supply pipe network.

[0065] Step S1 in the embodiment of the present invention specifically includes:

[0066] Collect basic regional network data such as pipeline network GIS attribute data, user revenue meter reading data, pipeline network flow monitoring data, pipeline network pressure monitoring data, pool water level monitoring data, water pump characteristic data, water pump operation data, month, weather, temperature and holiday distribution. The pipeline network GIS attribute data includes pipeline network topology information, pipe material, pipe size, node number and DMA partition number. Based on the basic regional pipeline network data, use EPANET software to establish a water supply network hydraulic model of "water plant-water transmission pipeline network-boosting pump stations at all levels-water distribution pipeline network-user".

[0067] S2. Obtain the water supply network operation scheduling target and establish a water supply network scheduling scenario model.

[0068] Step S2 in the embodiment of the present invention specifically includes:

[0069] Obtain the water supply network operation scheduling objectives, including meeting water supply pressure requirements, reducing system operation energy consumption, reducing internal pressure fluctuations in the network, and reducing network leakage rates. Establish a water supply network scheduling scenario model consisting of an objective function and constraints. The objective function is to minimize the total operation energy consumption. The calculation formula for the total operation energy consumption is:

[0070] ;

[0071] Where, E ( n ) represents the total energy consumption of all pumps operating for 24 hours a day, t Any time of day, r represents the water density, g represents the acceleration due to gravity, Q i,t Indicates the i The water pump is t Traffic flow during the period, H i,t Indicates the i The water pump is t The lift during the period, or i,t Indicates the i The water pump ist Efficiency of time period.

[0072] The constraint condition is expressed as:

[0073] ;

[0074] Where, n i,t Indicates the i The water pump is t The revolution ratio of the time period, P j,t Indicates the j The control point is t The pressure of time, V k,t Indicates the k The pool is in t The volume of water during a period.

[0075] S3. Obtain historical monitoring data of the water supply network, randomly divide the historical monitoring data of the water supply network into a training set and a validation set in proportion, use a neural network algorithm to construct a node water demand prediction model and a node water demand time-varying curve prediction model, which respectively output the node predicted water demand and the node water demand prediction time-varying curve, use the training set to train the node water demand prediction model and the node water demand time-varying curve prediction model, respectively, and obtain a trained node water demand prediction model and a trained node water demand time-varying curve prediction model, respectively.

[0076] Step S3 in the embodiment of the present invention specifically includes:

[0077] The node daily meter reading water volume historical monitoring data and the DMA partition main flow historical monitoring data are obtained, and the node daily meter reading water volume historical monitoring data and the DMA partition main flow historical monitoring data are randomly divided into a training set and a validation set at a ratio of 8:2. In order to effectively distinguish the two groups of data, the training set and the validation set corresponding to the node daily meter reading water volume historical monitoring data are named training set 1 and validation set 1, respectively, and the training set and the validation set corresponding to the DMA partition main flow historical monitoring data are named training set 2 and validation set 2, respectively.

[0078] A GA-BP neural network is used to construct a node water demand prediction model and a node water demand time-varying curve prediction model respectively. The training set 1 is input into the node water demand prediction model for training to obtain a trained node water demand prediction model. The training set 2 is input into the node water demand time-varying curve prediction model for training to obtain a trained node water demand time-varying curve prediction model.

[0079] S4. Input the verification set into the trained node water demand prediction model and the trained node water demand time-varying curve prediction model respectively, generate node water demand prediction results and node water demand time-varying curve prediction results respectively, set water demand prediction evaluation indicators and water demand time-varying curve evaluation indicators, if the prediction results meet the evaluation criteria, obtain the trained node water demand prediction model and the trained node water demand time-varying curve prediction model respectively, and enter S5. If it does not meet the evaluation criteria, return to S3.

[0080] Step S4 in the embodiment of the present invention specifically includes:

[0081] Input the validation set 1 into the trained node water demand prediction model, output the node predicted water demand, and set the water demand prediction evaluation index NSE wd Evaluate the predicted water demand of the node, NSE wd The expression is as follows:

[0082] ;

[0083] Where, q s Indicates the s The measured water demand of each node, Indicates the s The predicted water demand of each node, Indicates the s The average measured water demand of each node, when the NSE wd ≥0.8, then the trained node water demand prediction model is obtained and the process goes to S5. NSE wd <0.8, the training and verification of the node water demand prediction model are repeated.

[0084] Input the validation set 2 into the trained node water demand time-varying curve prediction model to obtain the DMA partition main flow prediction time-varying curve, normalize the DMA partition main flow prediction time-varying curve, combine it with the node predicted water demand, output the node water demand prediction time-varying curve, and set the water demand time-varying curve prediction evaluation index. NSE tc Evaluate the change curve of the node water demand prediction, NSE tc The calculation formula is as follows:

[0085] ;

[0086] Where, ql Indicates the l The measured flow rate of each pipe section, Indicates the l The predicted flow rate of each pipe section, Indicates the l The average measured flow rate of each pipe section, when the NSE tc ≥0.8, then the trained node water demand time-varying curve prediction model is obtained, and the process goes to S5. NSE tc <0.8, the training and verification of the node water demand time-varying curve prediction model are repeated.

[0087] The daily meter reading water volume data of the node and the DMA partition main flow data are respectively input into the trained node water demand prediction model and the trained node water demand time-varying curve prediction model to obtain the node water demand for the next day and the node water demand time-varying curve for the next day.

[0088] S5. Couple the water supply network hydraulic model, the water supply network scenario scheduling model, the node water demand prediction model and the node water demand time-varying curve prediction model to construct a water supply network optimization scheduling model based on multi-model coupling, use an improved particle swarm algorithm to optimize the water supply network optimization scheduling model based on multi-model coupling, and output a water supply network optimization scheduling plan.

[0089] Step S5 in the embodiment of the present invention specifically includes:

[0090] The improved particle swarm algorithm is based on the standard particle swarm algorithm, including improving the inertia weight, adding a penalty function, and limiting the particle speed and position range, as follows:

[0091] The improved inertia weight includes:

[0092] The calculation formula of the particle velocity and position of the standard particle swarm algorithm in D-dimensional space is:

[0093] ;

[0094] Where, d =1,2,…, D represents the spatial dimension, m represents the number of population iterations, and Respectively represent i Particles passing through m After the first iteration d The velocity and displacement in dimension, oh Represents the inertia weight, which determines the proportion of the particle to maintain the speed of the previous iteration, and the value is non-negative.c 1 and c 2 represents the learning factor. The former determines the tendency of the particle to search for the historical optimal position and represents the memory cognition of the particle. The latter determines the tendency of the particle to search for the optimal position of the current population and represents the social cognition of the particle. Both usually take values of 0 to 2. and Respectively represent i Particles passing through m After the first iteration d The historical optimal position and the population optimal position in dimension.

[0095] The calculation formula of the improved inertia weight is:

[0096] ;

[0097] Where, oh op represents the improved inertia weight, s is the density of feasible solutions in the particle swarm. In the early stage of the optimization process, s The value is smaller, oh op A larger value is beneficial for searching for the best solution in the entire search space, quickly approaching the global optimal position, and avoiding falling into the local optimal solution. As the iterative process gradually advances, the s The value gradually increases, oh op The value gradually decreases, which is conducive to the particles concentrating on the global optimal position to find the global optimal solution.

[0098] The multi-objective constrained optimization problem is expressed as:

[0099] ;

[0100] Where, f ( x ) represents the multi-objective constrained optimization objective function, u ( x )and v ( x ) represents a constraint.

[0101] In the multi-objective constrained optimization objective function f ( x ) adopts the method of adding the penalty function to convert the multi-objective constrained optimization problem into:

[0102] ;

[0103] Where, fit ( x) represents the fitness function, f pe ( x ) represents the penalty function, where the penalty coefficient , m and c is a non-negative parameter that needs to be debugged. α and β is a value not less than 1, and the constraints of the converted multi-objective constrained optimization problem are still u ( x )and v ( x ).

[0104] In an embodiment of the present invention, the multi-objective constraint optimization objective function f ( x ) is the total energy consumption of the operation E ( n ), so the fitness function fit ( x ) is rewritten as:

[0105] ;

[0106] when x If the value is within the feasible region, then the fit ( x ) is 0, the fit ( x ) and the E ( n ), that is, the fitness function is the objective function.

[0107] The penalty coefficient A In the early stages of the optimization process, the σ value is small. A A larger value is conducive to prompting the particles to fly to the feasible region for searching. As the optimization process continues to iterate, the σ value gradually increases. A The value gradually decreases, which is conducive to prompting the particles to search for the optimal value of the objective function in the feasible region.

[0108] The particle's trajectory is often random during the search and flight process. Therefore, it is necessary to impose certain restrictions on the particle's speed and position range to ensure that when the particle flies beyond the boundary, the next flight direction must be within the feasible solution area and will not wander around the boundary. Specifically:

[0109] ;

[0110] Where, v max and v minRepresent the maximum and minimum speeds of the particles, respectively. x max and x min Respectively represent the maximum position and minimum position of the particle. During the iteration process, if the particle speed and position exceed the limit range, the boundary value is taken. If the x ≤ x min When v The value is | v |, when the x ≥ x min When v The value is -| v |.

[0111] See also Figure 3 As shown in FIG, the specific optimization process of the water supply network optimization scheduling model based on multi-model coupling is as follows:

[0112] The water demand of the node the next day and the change curve of the water demand of the node the next day are input into the hydraulic model of the water supply network to generate the hydraulic model of the water supply network the next day, and obtain t =0 time corresponding to the n i,t , input it into the hydraulic model of the next day's water supply network for simulation, and output the simulation results. P j,t 、 Q i,t 、 H i,t and stated V k,t , construct the E ( n ) and the fit ( x ), for the fit ( x ) to evaluate the quality of the individual and the population to update the optimal position and the fit ( x ), conduct continuous iterative optimization, and when the set time is met, output the optimized scheduling plan of the water supply network at each time of the next day.

[0113] S6. Introduce the optimization scheduling evaluation index to evaluate the water supply network optimization scheduling plan. If it meets the optimization scheduling evaluation criteria, save the water supply network optimization scheduling plan and continue iterative calculation. If it does not meet the optimization scheduling evaluation criteria, enter step S7.

[0114] Step S6 in the embodiment of the present invention specifically includes:

[0115] See also Figure 3 As shown, the optimization scheduling evaluation index includes the number of iterations. The optimization scheduling evaluation standard includes that when the number of iterations is ≤150, it is considered to meet the optimization scheduling evaluation standard, and the water supply network optimization scheduling plan corresponding to time t is saved, and the iterative optimization is continued. When the number of iterations is greater than 150, it is considered that it does not meet the optimization scheduling evaluation standard, and the 150th iteration at time t and the five lowest iterations are selected. E ( n ) and the corresponding water supply network optimization scheduling plan, and enter step S7.

[0116] S7. Construct a water supply network optimization scheduling scheme screening model to obtain the optimal water supply network optimization scheduling scheme.

[0117] Step S7 in the embodiment of the present invention specifically includes:

[0118] A water supply network optimization scheduling scheme screening model is constructed, including the pump station operation cost. The expression of the pump station operation cost is:

[0119] ;

[0120] Where, F cost represents the operating cost of the pump station, δ represents the electricity price, N t represents the number of pumps running in period t, It represents the motor efficiency of the i-th water pump in the t-th time period. The optimal water supply network optimization scheduling plan includes the lowest operating cost of the pump station.

[0121] Using the pump station operating costs, the multiple groups of water supply network optimization scheduling plans corresponding to each time of the next day saved in step S6 are screened, and the optimal water supply optimization scheduling plan for each time of the next day is output, and the optimal water supply optimization scheduling plan for each time is formed into the optimal water supply network optimization scheduling plan for the next day.

[0122] This embodiment obtains basic information on the regional pipe network and the operation and scheduling targets of the water supply network system, establishes a hydraulic model of the water supply network and a water supply network scheduling scenario model respectively, uses a neural network algorithm to construct a node water demand prediction model and a node water demand time-varying curve prediction model, obtains the node water demand for the next day and the node water demand time-varying curve for the next day, and uses a method of coupling the water supply network hydraulic model, the water supply network scheduling scenario model, the node water demand prediction model, and the node water demand prediction time-varying curve prediction model to construct a water supply network optimization scheduling model based on multi-model coupling. The improved particle swarm algorithm is used for optimization, and multiple groups of water supply network optimization scheduling plans for each time of the next day are output. A water supply network optimization scheduling plan screening model is constructed to obtain the optimal water supply network optimization scheduling plan for the next day.

[0123] This method adopts a multi-model coupling approach and uses a neural network algorithm to construct a node water demand prediction model and a node water demand time-varying curve prediction model, which effectively reduces the prediction error. It uses an improved particle swarm algorithm to greatly improve the optimization efficiency and realize intelligent and precise scheduling of the water supply network. It has the advantages of simple deployment and solves the problems of high energy consumption in the existing water supply network scheduling, insufficient water supply pressure, large pressure fluctuations in the network, and many leakages.

[0124] The following describes a water supply network optimization scheduling system based on multi-model coupling provided by the present invention. The water supply network optimization scheduling system based on multi-model coupling described below and the water supply network optimization scheduling method based on multi-model coupling described above can be compared with each other.

[0125] See also Figure 4 As shown, it includes: a data collection module 41, a model construction module 42, a model training module 43, a scheduling evaluation module 44, and a solution screening module 45, wherein:

[0126] Data collection module: used to collect basic data of regional pipe network, operation and scheduling targets of water supply network and historical monitoring data of water supply network;

[0127] Model construction module: used to build a water supply network hydraulic model, a water supply network scheduling scenario model, a node water demand prediction model, a node water demand time-varying curve prediction model, a water supply network optimization scheduling model based on multi-model coupling, and a water supply network optimization scheduling scheme screening model;

[0128] Model training module: used to train and verify the node water demand prediction model and the node water demand time-varying curve prediction model;

[0129] Scheduling evaluation module: used to evaluate the water supply network optimization scheduling scheme output by the water supply network optimization scheduling model based on multi-model coupling;

[0130] Scheme screening module: used to obtain the optimal water supply network optimization scheduling scheme.

[0131] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of the present invention.

Claims

1. A water supply network optimization scheduling method based on multi-model coupling, characterized in that: The steps include: S1. Collect basic data of regional pipe network and establish hydraulic model of water supply pipe network; S2. Obtain the water supply network operation scheduling target and establish a water supply network scheduling scenario model; S3. Obtain historical monitoring data of the water supply network, randomly divide the historical monitoring data of the water supply network into a training set and a validation set in proportion, use a neural network algorithm to construct a node water demand prediction model and a node water demand time-varying curve prediction model, which respectively output the node predicted water demand and the node water demand prediction time-varying curve, and use the training set to train the node water demand prediction model and the node water demand time-varying curve prediction model, respectively, to obtain a trained node water demand prediction model and a trained node water demand time-varying curve prediction model, respectively; S4. Input the validation set into the trained node water demand prediction model and the trained node water demand time-varying curve prediction model, respectively, to generate node water demand prediction results and node water demand time-varying curve prediction results, respectively, set water demand prediction evaluation indicators and water demand time-varying curve prediction evaluation indicators, and if the prediction results meet the evaluation criteria, respectively obtain the trained node water demand prediction model and the trained node water demand time-varying curve prediction model, and proceed to S5; if they do not meet the evaluation criteria, return to S3; S5. Couple the water supply network hydraulic model, the water supply network scheduling scenario model, the node water demand prediction model, and the node water demand time-varying curve prediction model to construct a water supply network optimization scheduling model based on multi-model coupling, optimize the water supply network optimization scheduling model based on multi-model coupling using an improved particle swarm algorithm, and output a water supply network optimization scheduling plan; The improved particle swarm algorithm is based on the standard particle swarm algorithm, including improving the inertia weight, adding a penalty function and limiting the particle speed and particle position range; The calculation formula of the improved inertia weight is: oh op =ω max -(oh max -oh min )s; Where, ω op represents the improved inertia weight, ω max represents the initial inertia weight, ω min represents the inertia weight at the maximum number of iterations, and σ represents the density of feasible solutions in the particle swarm; The expression of the penalty function is: Where, f pe (x) represents the penalty function, where the penalty coefficient μ and γ represent non-negative parameters that need to be adjusted, and α and β are values not less than 1; The particle velocity and particle position ranges are specifically limited as follows: Where, v max and v min Respectively represent the maximum speed and minimum speed of the particle, x max and x min represent the maximum position and minimum position of the particle respectively; S6. Introduce the optimization scheduling evaluation index to evaluate the water supply network optimization scheduling plan. If it meets the optimization scheduling evaluation criteria, save the water supply network optimization scheduling plan and continue iterative calculation. If it does not meet the optimization scheduling evaluation criteria, proceed to step S7. S7. Construct a water supply network optimization scheduling scheme screening model to obtain the optimal water supply network optimization scheduling scheme.

2. The water supply network optimization scheduling method based on multi-model coupling according to claim 1 is characterized in that: Step S1 also includes: the regional pipeline network basic data includes pipeline network GIS attributes, user revenue meter reading data, pipeline network flow monitoring data, pipeline network pressure monitoring data, pool water level monitoring data, water pump characteristic data, water pump operation data, month, weather, temperature and holiday distribution; the pipeline network GIS attributes include pipeline network topology information, pipeline material, pipeline size, node number and DMA partition number.

3. The water supply network optimization scheduling method based on multi-model coupling according to claim 1 is characterized in that: Step S2 also includes: the water supply network operation scheduling objectives include meeting water supply pressure requirements, reducing system operation energy consumption, reducing pressure fluctuations within the network, and reducing network leakage rate; The water supply network scheduling scenario model includes an objective function and constraints. The objective function includes minimizing the total energy consumption. The calculation formula for the total energy consumption is: Where E(n) represents the total energy consumption of all pumps for 24 hours a day, t represents any time period in a day, ρ represents the water density, g represents the acceleration of gravity, and Q i,t represents the flow rate of the i-th water pump in the t-th period, H i,t represents the head of the i-th water pump in the t-th period, η i,t represents the working efficiency of the i-th water pump in the t-th period; The constraint condition is expressed as: Where n i,t represents the speed ratio of the i-th water pump in the t-th period, P j,t represents the pressure of the jth control point at the tth time period, V k,t Represents the volume of water in the kth pool at the tth time period.

4. The water supply network optimization scheduling method based on multi-model coupling according to claim 1 is characterized in that: Step S3 further includes: using the water supply network historical monitoring data including the node daily meter reading water volume historical monitoring data and the DMA zone main flow historical monitoring data, and using the GA-BP neural network to respectively construct the node water demand prediction model and the node water demand time-varying curve prediction model; The node daily meter reading water volume historical monitoring data is randomly divided into a training set 1 and a validation set 1 in a ratio of 8:2, the training set 1 is used to train the node water demand prediction model to obtain the trained node water demand prediction model; The DMA partition main flow historical monitoring data is randomly divided into training set 2 and validation set 2 in a ratio of 8:

2. The training set 2 is used to train the node water demand time-varying curve prediction model to obtain the trained node water demand time-varying curve prediction model.

5. The water supply network optimization scheduling method based on multi-model coupling according to claim 4 is characterized in that: Step S4 also includes: inputting the validation set 1 into the trained node water demand prediction model for verification to generate the node water demand prediction result; inputting the validation set 2 into the trained node water demand time-varying curve prediction model for verification to obtain a DMA partition main flow prediction time-varying curve, normalizing the DMA partition main flow prediction time-varying curve, and combining it with the node predicted water demand to generate the node water demand time-varying curve prediction result; The water demand forecast evaluation index includes NSE wd , and its calculation formula is: Where q s represents the measured water demand of the sth node, q' s represents the predicted water demand of the sth node, Represents the average measured water demand of the sth node. The water demand prediction evaluation standard is the NSE wd ≥0.8; The water demand time-varying curve prediction evaluation index includes NSE tc , and its calculation formula is: Where q l represents the measured flow rate of the lth pipe section, q l ' represents the predicted flow of the lth pipe section, It represents the mean value of the measured flow rate of the lth pipe section. The water demand time variation curve prediction evaluation standard is the NSE tc ≥0.

8.

6. The water supply network optimization scheduling method based on multi-model coupling according to claim 1 is characterized in that: Step S6 also includes: the optimization scheduling evaluation index includes the number of iterations T. When T≤150, it meets the evaluation criteria, the water supply network optimization scheduling plan is saved, and the iterative calculation is continued. Otherwise, step S7 is entered.

7. The water supply network optimization scheduling method based on multi-model coupling according to claim 3 is characterized in that: Step S7 also includes: the water supply network optimization scheduling scheme screening model includes pump station operating costs, and the expression of the pump station operating costs is: Where, F cost represents the operating cost of the pump station, δ represents the electricity price, N t represents the number of pumps running in period t, η i ' ,t It represents the motor efficiency of the i-th water pump in the t-th time period. The optimal water supply network optimization scheduling plan includes the lowest operating cost of the pump station.

8. A water supply network optimization scheduling system based on multi-model coupling, used to implement the water supply network optimization scheduling method based on multi-model coupling as claimed in claim 1, characterized in that: Includes the following modules: Data collection module: used to collect basic data of regional pipe network, water supply network operation scheduling targets and historical monitoring data of water supply network; Model construction module: used to build a water supply network hydraulic model, a water supply network scheduling scenario model, a node water demand prediction model, a node water demand time-varying curve prediction model, a water supply network optimization scheduling model based on multi-model coupling, and a water supply network optimization scheduling scheme screening model; Model training module: used to train and verify the node water demand prediction model and the node water demand time-varying curve prediction model; Scheduling evaluation module: used to evaluate the water supply network optimization scheduling scheme output by the water supply network optimization scheduling model based on multi-model coupling; Scheme screening module: used to obtain the optimal water supply network optimization scheduling scheme.

Images