Optimal dispatching method of regional water supply system considering saltwater intrusion

Abstract

The application discloses a kind of regional water supply system optimization scheduling methods considering salt tide influence, including S1: the multi-source-multiple waterworks-multiple water area layered water supply topology structure covering target area affected by salt tide is constructed;S2: by the real-time water quality monitoring data of reservoir in the area affected by salt tide, determine salt tide period, dynamically adjust multi-objective multi-source optimization scheduling model objective function and constraint condition, preferentially schedule water quality safe water source;S3: based on the topology structure model constructed in step S1, take the salt tide state in step S2 as dynamic parameter, construct the three-dimensional objective function of multi-objective multi-source optimization scheduling model "water shortage, salt tide influence, cost difference", combined with constraint condition, solve Pareto optimal solution set by improved NSGA-III algorithm.The application is applicable to the collaborative scheduling optimization of multi-source system affected by salt tide, has wide application value in improving water resources allocation efficiency, guaranteeing water supply safety and the like.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of water resource management and urban water supply scheduling, and particularly relates to a regional water supply system optimal scheduling method considering the influence of salt tide. BACKGROUND

[0002] Reservoir scheduling is the core link of water resource system management. The traditional scheduling needs to coordinate multiple objectives such as flood control, power generation, ecology and water supply. In the salt tide affected areas with dense population and developed economy, water supply safety has become the primary task of reservoir scheduling. Salt tide intrusion is a unique hydrological phenomenon in coastal and estuarine areas, which refers to the intrusion of seawater along the river under the action of tides, resulting in the increase of chloride concentration in river water. When the concentration exceeds the limit value of 250 mg / L in the Standards for Drinking Water Health, it will directly threaten the safety of drinking water, and also affect industrial water, agricultural irrigation and the stability of estuarine ecosystems.

[0003] There are three major water supply challenges in the salt tide affected areas: 1) complex hydrological conditions: affected by both river runoff and ocean tides, when runoff decreases in dry season, salt tide intrusion distance can extend to tens to hundreds of kilometers inland, and water quality in water source area fluctuates dramatically; 2) prominent supply-demand contradiction: such areas are usually densely populated and economically active, with daily water demand often exceeding 5 million m³, while water supply capacity of water source area is limited during salt tide period, which is prone to double risks of "water supply gap" and "water quality exceeding standard"; 3) great scheduling difficulty: existing scheduling models mostly focus on "water balance and cost minimization", without quantifying water quality risks caused by salt tide into the objective system, and only rely on passive response of "salt avoidance and fresh water storage" engineering, which is prone to the contradiction of "protecting water quantity while water quality exceeds standard, and protecting water quality while water supply is insufficient" during extreme salt tide period; and mainstream multi-objective optimization algorithms have insufficient diversity of solution set and slow convergence speed in the three-dimensional objective space of "water quantity-water quality-cost", which is difficult to meet the flexibility demand of scheduling decision.

[0004] Taking the Yangtze River Estuary as an example, as a typical salt tide affected area, the permanent population in this area reached 24.7589 million in 2022, and the daily water demand exceeded 8 million m³. When the Yangtze River runoff is less than 12000 m³ / s in dry season, the salt tide intrusion distance can reach more than 100 km, and the chloride concentration at the water intake is often more than 300 mg / L. The existing scheduling strategy has obvious limitations: excessive reliance on experience rules, lack of mechanism to dynamically respond to salt tide changes; insufficient multi-objective coordination, unable to achieve global trade-off of water shortage, cost and water quality; poor algorithm adaptability, difficult to cope with the trend of intensifying salt tide caused by climate change.

[0005] Therefore, it is urgent to build a general optimal scheduling method that is not dependent on specific regional exclusive conditions and can adapt to different salt tide affected scenarios, to solve the technical bottlenecks of "water quality-water quantity-cost" coordinated optimization and dynamic response to salt tide changes. Summary of the Invention

[0006] The present invention aims to provide an optimized scheduling method for regional water supply systems that takes into account the impact of salinity intrusion, thereby solving the aforementioned technical problems existing in the prior art.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following solution:

[0008] The method for optimizing the scheduling of regional water supply systems that takes into account the impact of saltwater intrusion includes the following steps:

[0009] Step S1: Construct a multi-source, multi-water treatment plant, and multi-demand area stratified water supply topology covering the target area affected by saltwater intrusion. Based on graph theory, the reservoirs, water treatment plants, and demand areas in this region are abstracted as directed graph nodes, and the connections between nodes are abstracted as directed edges. The flow rate on each edge is defined as a decision variable. Specifically, this includes:

[0010] Step S1.1: Abstract the core reservoirs, water plant clusters, and water demand areas in the saltwater intrusion-affected areas into directed graph nodes to form a three-layer network structure of "water source area - water plant cluster - water demand area"; wherein, the water source area is the set of reservoirs in the area that undertake the main water supply task; the water plant cluster is a combination of several water plants in the area divided according to the water supply range or function; and the water demand area is a number of regional units in the area divided according to the water demand characteristics.

[0011] Step S1.2: Collect daily water demand, reservoir water supply capacity, water level-storage capacity relationship and water quality monitoring data in the area affected by saltwater intrusion, and determine the decision variables as the allocation from the water source to the water plant group and the allocation from the water plant group to the water demand area, as shown in Equations (1) and (2) respectively:

[0012] (1)

[0013] In the above formula, R This is a collection of core reservoirs in the area affected by saltwater intrusion. P This refers to the cluster of water plants in the area. T For the set of scheduling periods; X r,p,t Reservoir r During the period t water plant group p The amount of water supplied.

[0014] (2)

[0015] In the above formula, Z This refers to a collection of water-demanding areas affected by saltwater intrusion. Y p,z,t Indicates a group of water plants p During the period t To water-demand areas zThe amount of water delivered.

[0016] Step S2: Determine the salt tide period through real-time water quality monitoring data of reservoirs in the salt tide affected area, dynamically adjust the objective function and constraint conditions of the multi-objective multi-water source optimal scheduling model, and preferentially schedule water sources with safe water quality.

[0017] Step S3: Based on the topological structure model constructed in step S1, taking the salt tide state in step S2 as a dynamic parameter, a multi-objective multi-water source optimal scheduling model is constructed, and a three-dimensional objective function of "water shortage, salt tide influence, and cost difference" is constructed. Combined with the constraint conditions, the Pareto optimal solution set is solved by the improved NSGA-III algorithm; specifically including:

[0018] Step S3.1: Construct the objective function of the multi-objective multi-water source optimal scheduling model, as shown in formula (3), formula (4), and formula (5):

[0019] Objective 1: Minimize water shortage:

[0020] (3)

[0021] In the formula, is the water shortage objective function;

[0022] is the water shortage of the water demand area z in the time period t , D z,t is the water demand of the water demand area z in the time period t .

[0023] Objective 2: Minimize the salt tide affected water supply:

[0024] (4)

[0025] In the formula, f 2 ( X ) is the salt tide affected water supply objective function; , is the chloride concentration monitored by the water intake of the reservoir r in the first t time period, is the salt tide influence threshold value; is the reservoir affected by salt tide invasion.

[0026] Objective 3: Minimize the water supply cost difference:

[0027] (5)

[0028] In the formula, f3 ( X Let be the objective function for the difference in water supply costs; For reservoir r During the period t unit cost For reservoir r During the period t Total water supply K 0 This represents the actual cost of water supply.

[0029] Step S3.2: Set constraints, as shown in equations (6), (7), (8), (9), (10), (11), (12), and (13):

[0030] 1) Reservoir water supply capacity constraints:

[0031] (6)

[0032] In the above formula, For reservoir r During the period t The maximum water supply capacity.

[0033] 2) Water plant supply capacity constraints:

[0034] (7)

[0035] In the above formula, For water plant p During the period t The maximum water supply capacity.

[0036] 3) Pipeline transmission and distribution capacity constraints:

[0037] (8)

[0038] (9)

[0039] In the above formula, For reservoir r To the water plant p The maximum water delivery capacity of the corresponding water supply pipeline. For water plant p To water-demand areas z The maximum water delivery capacity of the corresponding water supply pipeline.

[0040] 4) Water plant water balance constraints:

[0041] (10)

[0042] 5) Water supply demand meets constraints:

[0043] (11)

[0044] 6) Non-negative constraint:

[0045] (12)

[0046] (13).

[0047] Step S3.3: solving by NSGA-III algorithm: encoding decision variables into multi-dimensional vectors, generating initial population by diversification strategy; introducing penalty function to deal with constraint violation problem in fitness evaluation; using adaptive crossover, mutation operators and feasibility repair operator based on engineering heuristic rules to optimize search process; dynamically evaluating convergence MIGD and distribution MHV of solution set, and outputting Pareto optimal solution set.

[0048] Further optimization, in step S1.1, the division of water plant groups is based on the layout of water supply projects and the distribution of water transmission networks in the salt tide affected area, including northern water plant group, central water plant group, eastern water plant group, southern water plant group, western water plant group, suburban water plant group, and suburban water plant group, each water plant group containing at least one water plant with conventional water treatment capacity.

[0049] The division of the water demand area is based on the administrative division, population density and water load characteristics of the area, including central urban area, northern area, eastern area, western area, southern area type, each water demand area containing at least one town or functional area with stable water demand.

[0050] Further optimization, in step S2, the salt tide period determination condition is that the chloride concentration at any reservoir water intake in the salt tide affected area exceeds 250 mg / L; during the salt tide period, the priority coefficient of water quality safety water source scheduling is increased by 1.2-1.5 times, and the upper limit of water supply quantity of salt tide affected water source is reduced by 20%-30%.

[0051] Further optimization, in step S3.3, the specific process of solving by NSGA-III algorithm includes:

[0052] Step S3.3.1: decision variable coding: converting the core decision variables of water supply scheduling into a mathematical form that can be processed by the algorithm, i.e., encoding the water supply quantity set of reservoirs to water plant groups and the water transmission quantity set of water plant groups to water demand areas into one-dimensional real chromosome, and the length of chromosome is R |×| P |×| T |+| P |×| Z |×|T |, the value range of each gene locus is ([0, Q max ], Q max To ensure that the coding results meet the engineering capacity boundaries, the maximum water supply of the corresponding reservoir or the maximum water transmission capacity of the water plant is determined.

[0053] Step S3.3.2: Diversified initial population generation: Seven engineering-oriented strategies are weighted and fused to generate an initial population of 100-300. The strategy group includes demand-oriented strategy, water supply capacity balancing strategy, historical scheduling fitting strategy, random focusing strategy, regional priority strategy, reservoir balancing strategy and mixed strategy. The weight of each sub-strategy in the mixed strategy is dynamically adjusted according to the scheduling deviation rate of the same period in the previous three years of the region.

[0054] Step S3.3.3: Fitness assessment and constraint handling, specifically including:

[0055] S3.3.3.1: Calculate multi-objective fitness: For each individual in the population, i.e. a complete set of water supply scheduling schemes, simultaneously calculate the water shortage target f1, the water supply target f2 affected by salinity intrusion, and the water supply cost difference target f3. Integrate the calculation results of the three targets into a three-dimensional fitness vector to intuitively reflect the individual's comprehensive performance in the dimensions of "water supply reliability-water quality safety-economy".

[0056] S3.3.3.2: Constraint Violation Penalty: A penalty function is introduced to modify fitness. The modified formula is: Modified Fitness = Original Fitness + λ1max(0, Water Shortage Limit) + λ2max(0, Water Supply Exceeds Limit); where λ1 is the supply and demand balance penalty coefficient. When the water shortage in the water demand area exceeds 5% of its total water demand, the value of λ1 is doubled to strengthen the penalty for severe water supply gaps; λ2 is the capacity constraint penalty coefficient. When the water supply of the reservoir exceeds 10% of its maximum water supply capacity, the value of λ2 is doubled to strictly limit overcapacity water supply.

[0057] S3.3.3.3: For individuals who still violate constraints after being penalized, repair them according to the project priority rules.

[0058] Step S3.3.4: Drive population iteration through the synergistic operation of crossover and mutation, balancing the capabilities of "global exploration" and "local optimization," specifically including:

[0059] Crossover operation: Simulated binary crossover SBX is used, with crossover probability... p c The initial value is 0.85~0.95, and it decreases linearly by 0.05 every 50 generations, with a lower limit of 0.6, gradually shifting towards local optimization; the crossover distribution index is fixed at 20 to ensure the diversity and stability of the crossover offspring.

[0060] Mutation operation: polynomial mutation strategy, mutation probability p m Initial value: 0.05~0.15, when the population diversity index is less than 0.3, p m Increase 0.02 to avoid premature convergence of the algorithm; the mutation distribution index is fixed at 20 to control the rationality of the mutation range.

[0061] Step S3.3.5: environmental selection and population update: to maintain the non-dominance and uniformity of the solution set, the population is updated by the following three steps, which include:

[0062] Step S3.3.5.1: combine the current parent population and the offspring population, and perform fast non-dominated sorting on the combined population. The non-dominated front of different levels is divided according to the dominance relationship of individuals. The first front is the current optimal candidate solution set.

[0063] Step S3.3.5.2: in the three-dimensional target space of "water shortage amount-salt tide influence-cost difference", preset the reference points with uniform distribution, the number is 1.2~1.5 times the size of the target population, calculate the correlation degree of non-dominated solutions and each reference point, and normalize the target value to eliminate the dimension difference, to ensure the fair competition of different targets.

[0064] Step S3.3.5.3: select individuals according to the correlation degree from high to low, and finally retain 100~150 individuals to form a new parent population, which ensures that the new population contains non-dominated solutions and can uniformly cover different trade-off areas in the target space.

[0065] Step S3.3.6: determine the iteration termination time and output the optimal scheme by dynamically monitoring the convergence and solution set distribution of the algorithm:

[0066] Step S3.3.6.1: dynamically calculate the average inverse generation distance MIGD and the average hyper volume MHV of the solution set;

[0067] Step S3.3.6.2: terminate evolution when any of the following conditions is met:

[0068] 1) the evolution number reaches the preset maximum value;

[0069] 2) the average inverse generation distance MIGD of 50 consecutive generations is less than 0.001, and the hyper volume variation coefficient is less than 5%.

[0070] Step S3.3.6.3: after the termination of evolution, output the final non-dominated solution set, which contains 30~50 Pareto optimal solutions, each solution contains complete and The allocation scheme comprehensively covers the multi-objective trade-off space of "water shortage-salt tide impact-cost difference", providing diversified alternative schemes for scheduling decisions.

[0071] Further optimization, in the step S3.3.2, the strategy includes:

[0072] Demand-oriented strategy: predict water demand in water-demanding areas and inversely allocate water supply from water sources;

[0073] Supply capacity balancing strategy: allocate water supply according to the proportion of the maximum water supply capacity of each reservoir ;

[0074] Historical scheduling fitting strategy: fit the scheduling data of the same period in the previous three years in the salt tide affected areas to generate an initial solution;

[0075] Random focus strategy: randomly select 1-2 water-demanding areas (such as core areas) to prioritize water supply;

[0076] Regional priority strategy: allocate water supply according to the priority of water-demanding areas , the priority of core areas is higher than that of non-core areas;

[0077] Reservoir balancing strategy: make the difference rate of water supply of each reservoir less than 15%;

[0078] Mixed strategy: weighted fusion of the above six strategies, the weight is dynamically adjusted according to the scheduling deviation rate in the same period in the previous three years, if the scheduling deviation rate of a certain strategy is higher than 10%, the weight is reduced by 5%-10%, to balance the diversity of the initial population and the engineering rationality.

[0079] Further optimization, in the step S3.3.3.3, the engineering priority rules of the feasibility repair operator include:

[0080] If the water supply of the reservoir is over the limit, the salt tide water source with the highest chloride concentration is preferentially reduced; for every 50 mg / L increase in concentration, the over-supply of the water source is reduced by 10%.

[0081] If the water plant water imbalance, that is, the deviation of the inflow and outflow is over the limit, the water supply of the water plant closest to the water-demanding area is preferentially adjusted, and the single adjustment amplitude does not exceed 15% of the design water supply capacity of the water plant.

[0082] If the water demand of the water-demanding area is over the limit, the core area with the highest population density is preferentially guaranteed, the water shortage of the core area is controlled to be less than 5%, and the water shortage of the non-core area is proportionally allocated according to the population density of each area, balancing the livelihood guarantee and water supply fairness.

[0083] Further optimization, in the step S3.3.6.1, the average reverse generation distance MIGD for evaluating the convergence of the solution set and the average hyper volume MHV for evaluating the distribution of the solution set, specifically includes:

[0084] Step S3.3.6.1.1: Calculate the mean reverse generation distance (MIGD) and convergence criteria:

[0085] First, let the current optimal solution set be... S ={ x 1, x 2,..., x n};in, x i For the solution set of the first i Each Pareto candidate solution corresponds to a set of solutions including the water supply from the reservoir to the water treatment plant group. Water supply from water plant clusters to water demand areas A complete water supply scheduling plan n This represents the total number of solutions in the current solution set.

[0086] Secondly, let the theoretically optimal solution set be... S *={ x 1*, x 2*,..., x k *};in, x j * is the first j Theoretically optimal solution.

[0087] Third, calculate the first i candidate solutions x i Reverse generation distance IGD ( x i The mean reverse generation distance (MIGD) is as follows:

[0088] (14)

[0089] (15)

[0090] In the above formula, IGD( x i )for x i To the theoretical optimal solution set S The three-dimensional Euclidean distance to the nearest solution in the *. f 1( x i ), f 2( x i ), f 3( x i ) are candidate solutions x iThe target values ​​for water shortage, water supply affected by saltwater intrusion, and water supply cost differences; f 1( x j *) f 2( x j *) f 3( x j *) represent the theoretical optimal solutions. x j The target value corresponding to *.

[0091] Fourth, convergence judgment rule: When the calculated MIGD < 0.001, it is determined that the current solution set has converged to the theoretical optimal region and meets the algorithm convergence requirement.

[0092] Step S3.3.6.1.2: Calculate the average hypervolume MHV and determine distribution equilibrium:

[0093] First, with the "water shortage target" f 1. Saltwater intrusion affects water supply targets f 2. Target for difference in water supply costs f Using "3" as the coordinate axis, a three-dimensional Euclidean space is formed. O - f 1 f 2 f 3. Spatial origin O (0,0,0) represents the ideal optimal state of "zero water shortage, zero saltwater intrusion impact, and zero cost difference".

[0094] Secondly, the effective hypervolume elements in the solution set are selected to form the effective hypervolume element set. V ={ V 1, V 2,..., V m};in, V j For the first j The volume of an effective hypervolume element, which is defined as an element that is not dominated by other candidate solutions and can independently contribute to the target space coverage. m MHV represents the total number of effective hypervolume units, where m ≤ n; MHV is the average hypervolume, which is the sum of the volumes of all effective hypervolume units.

[0095] Third, calculate the first j Volume of an effective hypervolume unit V j Coefficient of variation of mean supervolume MHV CV Specifically:

[0096] (16)

[0097] (17)

[0098] (18)

[0099] In the formula, x j is the target value of the candidate solution corresponding to the effective hyper-volume unit, and f 1( x j ), f 2( x j ), f 3( x j ) are respectively the water shortage, the water supply affected by salt tide and the difference in water supply cost of the solution; μ is the mean value of the MHV, and is the standard deviation of the effective hyper-volume unit volume.

[0100] Fourth, the distribution balance determination rule: when the calculated CV <5%, it is determined that the current solution set is distributed evenly in the three-dimensional target space without obvious aggregation or fault, and meets the distribution requirement of the algorithm.

[0101] Compared with the prior art, the beneficial effects of the present application are:

[0102] 1. The method has strong universality, and the core models of topological modeling, salt tide discrimination and NSGA-III algorithm do not depend on exclusive parameters of a specific region, and only need to input the number of reservoirs, water plant layout, water quality data and water demand characteristics of the target region to adapt, and can be widely applied to all regions affected by salt tide.

[0103] 2. Dynamic response is efficient, the salt tide period mode is triggered based on the universal water quality threshold, the response is fast, the salt tide period can be automatically identified according to real-time water quality data, compared with the traditional experience scheduling, the salt tide invasion can be responded to in advance, and the weight coefficient can be adaptively adjusted according to the salt tide intensity of the region, the salt tide characteristics difference of different regions is adapted, and the system resilience is enhanced.

[0104] 3. The multi-objective collaborative optimization effect is remarkable, the three target collaborative water supply scheduling of "water quantity-water quality-cost" is realized, and the water quality safety guarantee capability is significantly improved.

[0105] 4. The present application adopts the NSGA-III multi-objective evolutionary algorithm, solves the problems of uneven distribution of Pareto solution set under multi-objective and easy to fall into local optimum, improves the diversity and globality of the solution set, and through multi-source initialization and intelligent repair, the convergence speed and the proportion of feasible solutions of the algorithm are improved, and the actual complex constraints are adapted. BRIEF DESCRIPTION OF DRAWINGS

[0106] Figure 1 A flow chart of the calculation of the method for optimal scheduling of a regional water supply system considering the influence of salt tide;

[0107] Figure 2 A topological relationship diagram of the current water supply system in Shanghai;

[0108] Figure 3 A Pareto frontier diagram of multi-objective optimization in 2023;

[0109] Figure 4 A comparison diagram of the water supply process of the multi-objective optimization result, the traditional scheduling result and the actual scheduling scheme in 2023. DETAILED DESCRIPTION

[0110] The technical solutions of the present application will be described in detail below with reference to the embodiments, but the scope of protection of the present application is not limited to the embodiments.

[0111] In this embodiment, the water supply system in the Yangtze River Estuary is taken as an example for illustration. The Yangtze River Estuary is an important economic center and a sensitive area of water resources in China, which has long been affected by the serious influence of salt tide invasion on water sources. Important water sources such as Chenhang and Qingcaosha are limited by reservoir capacity, and their water supply guarantee capacity is insufficient during the continuous salt tide period. In addition, there is a lack of efficient two-way interconnection mechanism between the raw water systems, which leads to weak emergency scheduling capacity and seriously affects the efficiency of regional water resource supply. The current scheduling mode mainly adopts single economic guidance or segmented strategy, which is difficult to realize the coordinated optimization of water supply reliability, economic benefit and water quality safety.

[0112] In this embodiment, as shown in Figure 1 , the method for optimal scheduling of a regional water supply system considering the influence of salt tide takes the water supply scheduling of the Yangtze River Estuary in 2023 as an application scenario, which specifically includes the following steps:

[0113] Step S1: A multi-source-multi-plant-multi-demand area layered water supply topological structure covering the salt tide-affected Yangtze River Estuary region is constructed, as shown in Figure 2 , the decision variables are defined, which specifically include:

[0114] Step S1.1: In this embodiment, the water sources include Chenhang Reservoir, Qingcaosha Reservoir and Jinze Reservoir, a total of 3; the water plant groups include the north, center, Pudong, south, west, Songjin and suburban groups, a total of 7 groups, including 28 water plants, with a total design water supply capacity of 9 million m³ / day; the main water demand areas include the central urban area, the northern region, the western region and the southern region, with a daily water demand of 7.5-8.5 million m³. The connection relationship and water transmission capacity of the water source-water plant group-water demand area are defined.

[0115] Step S1.2: Collect 2023 annual operation data, including actual daily water demand in the affected area of the Yangtze River Estuary region, reservoir and water plant supply capacity, reservoir water level and effective storage capacity relationship, reservoir water intake chlorides concentration data, as well as the actual water supply scheme in 2023 and the traditional scheduling scheme. Among them, the annual daily water demand, reservoir water supply capacity, and chlorides concentration data come from the routine monitoring platform of Shanghai Water Affairs Bureau, the historical scheduling data come from the water supply annual report from 2020 to 2022, and they are all types of data that can be routinely obtained in the salt tide affected area.

[0116] The decision variables are determined as the allocation amount from the water source to the water plant group and the allocation amount from the water plant group to the water demand area, as follows:

[0117] Allocation amount from water source to water plant: (19).

[0118] In the formula, R ={Chenhang Reservoir, Qingcaosha Reservoir, Jinze Reservoir}, P={North Water Plant Group, Center Water Plant Group, Pudong Water Plant Group, South Water Plant Group, West Water Plant Group, Songjin Water Plant Group, Suburban Water Plant Group}, T is a set of scheduling periods.

[0119] Allocation amount from water plant to water demand area:

[0120] (20)

[0121] In the formula, Z={Central Urban Area, Northern Region, Western Region, Southern Region}.

[0122] Step S2: Determine the salt tide period through real-time water quality monitoring data of the reservoir in the salt tide affected area, dynamically adjust the objective function and constraint conditions of the multi-objective multi-water source optimization scheduling model, and preferentially schedule the water source with safe water quality.

[0123] In this embodiment, the chlorides concentration is monitored every 2 hours, and any reservoir with chlorides concentration exceeding 250 mg / L is determined as a salt tide period.

[0124] From January to March and November to December 2023, the salt tide in the Yangtze River Estuary region is high, the chlorides concentration in Qingcaosha Reservoir reaches 420 mg / L, the chlorides concentration in Chenhang Reservoir reaches 380 mg / L, and the chlorides concentration in Jinze Reservoir reaches 280 mg / L; on January 5, 2023, the chlorides concentration in Chenhang Reservoir reaches 260 mg / L, triggering the salt tide period mode: adjusting the salt tide affected water supply weight coefficient to 0.6, as the chlorides concentration in Chenhang Reservoir is close to 350 mg / L, the upper limit of the weight is taken; the upper limit of the water supply amount of Chenhang Reservoir is reduced to 1.44 million m³ / day (80% of the original 1.8 million m³ / day), and Jinze Reservoir is preferentially scheduled (the daily water supply amount is 1.1 million m³, and the proportion is increased to 15%).

[0125] On April 10, 2023, the chloride concentration of each reservoir was less than 200 mg / L, and the salt tide period mode was exited: the weight coefficient was restored to 0.3, the upper limit of the reservoir water supply was restored to the design value, and the priority of the dispatch was adjusted according to the "cost optimization".

[0126] Step S3: Based on the topological structure model constructed in step S1, the salt tide state in step S2 is taken as a dynamic parameter, a multi-objective multi-source optimal scheduling model is constructed, and a three-dimensional objective function of "water shortage, salt tide influence, and cost difference" is constructed. Combined with the constraint condition, the Pareto optimal solution set is solved by the improved NSGA-III algorithm. Specifically, it includes:

[0127] Step S3.1: Construct the objective function of the multi-objective multi-source optimal scheduling model, as shown in formula (21), formula (22), and formula (23):

[0128] Objective 1: Minimize water shortage

[0129] (21)

[0130] In the formula, is the water shortage objective function;

[0131] is the water demand area z The water shortage in the time period t , D z,t is the water demand area z The water demand in the time period t .

[0132] Objective 2: Minimize salt tide influence water supply:

[0133] (22)

[0134] In the formula, f 2 ( X ) is the salt tide influence water supply objective function; , is the chloride concentration monitored by the water intake of the reservoir r in the first t time period, is the salt tide influence threshold; is the reservoir affected by the salt tide invasion.

[0135] Objective 3: Minimize water supply cost difference:

[0136] (23)

[0137] In the formula, f 3( X ) is the water supply cost difference objective function; is the reservoir r in the time period t unit cost, is the reservoir r in the time period t total water supply, K 0 is the actual water supply cost.

[0138] Step S3.2: Set the constraint conditions, as shown in formula (24), formula (25), formula (26), formula (27), formula (28), formula (29), formula (30), formula (31):

[0139] 1) Reservoir water supply capacity constraint:

[0140] (24)

[0141] In the above formula, is the maximum water supply capacity of the reservoir r in the time period t , unit: m³.

[0142] 2) Water plant water supply capacity constraint:

[0143] (25)

[0144] In the above formula, is the maximum water supply capacity of the water plant p in the time period t , unit: m³.

[0145] 3) Pipeline transportation and distribution capacity constraint:

[0146] (26)

[0147] (27)

[0148] In the above formula, is the maximum water supply capacity of the reservoir r to the water plant p , unit: m³. is the maximum water supply capacity of the water plant p to the water demand area z , unit: m³.

[0149] 4) Water plant water balance constraint:

[0150] (28)

[0151] 5) Water supply demand satisfaction constraint:

[0152] (29)

[0153] 6) Non-negative constraint:

[0154] (30)

[0155] (31)

[0156] Step S3.3: Solve using NSGA-III algorithm: encode decision variables as a multi-dimensional vector, generate initial population through diversification strategy; introduce penalty function in fitness evaluation to handle constraint violation problem; use adaptive crossover, mutation operators and feasibility repair operators based on engineering heuristic rules to optimize search process; dynamically evaluate the convergence MIGD and distribution MHV of the solution set, and output the Pareto optimal solution set.

[0157] In this embodiment, the parameters are set: according to the complexity of the water supply system in the Yangtze River estuary region, the population size is 120, the maximum evolution generation is 200, the crossover probability is 0.9, which is reduced by 0.05 every 50 generations, the mutation probability is 0.1, and the diversity index is lower than 0.3. When it is raised to 0.12; parameter selection only depends on the complexity of the regional water supply system.

[0158] Initial population generation: based on the Yangtze River estuary scheduling data from January 2020 to January 2022, the demand-oriented strategy weight is 22%. Because the demand deviation rate in January 2022 is 12%, according to the general rule "high deviation rate reduces 5%-10% of weight", the weight is reduced by 3% compared with the benchmark of 25%, the water supply capacity balanced strategy weight is 18%, and the rest of the strategy is allocated according to the benchmark weight.

[0159] Feasibility repair: on January 10, 2023, the Qingcaosha Reservoir water supply exceeds the upper limit by 150,000 m³. According to the general repair rule, the water supply to the southern water plant group is preferentially reduced by 100,000 m³, accounting for 67% of the excess, and the remaining 50,000 m³ is reduced to the water plant group in the suburbs to ensure water balance in the water plant.

[0160] Convergence determination: when evolution reaches 200 generations, MIGD=0.0008<0.001, and MHV variation coefficient=4.2%<5%, stop evolution, and output 42 Pareto optimal schemes.

[0161] Table 1 is the simulation results of daily optimized scheduling water supply scheme in 2023, and Table 2 is the simulation results of daily optimized scheduling of each reservoir in 2023. Due to the limited space, only part of the data is provided here.

[0162] Table 1 Simulation results of daily optimized scheduling water supply scheme in 2023 in the Yangtze River estuary region affected by salt tide

[0163]

[0164] Table 2 daily optimized scheduling simulation results of each reservoir in the salt tide affected region of the Yangtze River Estuary in 2023

[0165]

[0166] Figure 3 For the method described in the present application, the Pareto front of multi-objective optimization is simulated for the salt tide affected region of the Yangtze River Estuary in each quarter of 2023. Among them, in the first quarter, the runoff is less and the salt tide intensity is high in the dry season of the Yangtze River Estuary, and the Pareto front is significantly dispersed in the "salt tide influence-water shortage" dimension, and the solution set extends along the direction of "high water quality safety (low salt tide influence) → high water supply reliability (low water shortage)", which is consistent with the dynamic target of "water quality safety priority in salt tide period" in the scheme. In the second and third quarters, the Yangtze River Estuary enters the flood season, and the salt tide threat is weakened, and the front is contracted and concentrated in the "water shortage-cost difference" two-dimensional dominant area, and the salt tide influence target value tends to be stable, which reflects the logic of "water supply cost / stability priority in non-salt tide period" in the scheme. In the fourth quarter, it is the transition period of salt tide, because the dry season restarts in October-December, the salt tide appears initially and is accompanied by dry season precursors, the solution set form is between winter and summer, the multi-objective game complexity rises (salt tide initial appearance + dry season precursor), and the ability of the scheme to "dynamically adapt to seasonal transition period" is verified.

[0167] Figure 4 For the comparison chart of the multi-objective optimization results, the traditional scheduling results and the actual scheduling scheme of water supply process in 2023, Figure 4 Fig. (a) is the comparison result of the multi-objective optimization results, the traditional scheduling results and the actual scheduling scheme of water supply process in Qingcaosha Reservoir in 2023; Figure 4 Fig. (b) is the comparison result of the multi-objective optimization results, the traditional scheduling results and the actual scheduling scheme of water supply process in Jinze Reservoir in 2023; Figure 4 Fig. (c) is the comparison result of the multi-objective optimization results, the traditional scheduling results and the actual scheduling scheme of water supply process in Chenhang Reservoir in 2023. As can be seen from the figure, the water supply curve of the optimization scheme is highly consistent with the actual water supply curve, the average water supply amount of the optimization scheme described in the present application deviates from the actual average water supply amount by less than 4.5%, and the average water supply amount of the traditional scheduling water supply scheme deviates from the actual average water supply amount by between 24-45%.

[0168] In summary, the verification of the Yangtze River Estuary example shows that the method described in the present application can significantly improve the water supply safety and scheduling economy in the salt tide period. In addition, the method is also applicable to other regions affected by salt tide, and has wide application prospects, providing general technical support for intelligent scheduling of urban water resources in all salt tide affected regions under the background of climate change.

[0169] As above, although the present application has been shown and expressed in connection with a certain embodiments thereof, it is to be understood that the application is not limited to the details of the foregoing description, but rather can be practiced with the exclusion of the features and aspects. It is the intention, therefore, to encompass within the scope of the application, all variations and equivalents that fall within the spirit and scope of the claims following.

Claims

1. A method for optimizing the scheduling of regional water supply systems considering the impact of saltwater intrusion, characterized in that, Includes the following steps: Step S1: Construct a multi-source, multi-water plant, multi-demand area stratified water supply topology covering the target area affected by saltwater intrusion. Based on graph theory, the reservoirs, water plants and demand areas in this area are abstracted as directed graph nodes, the connection relationships between nodes are abstracted as directed edges, and the flow on the edges is defined as a decision variable. Specifically, it includes: Step S1.1: Abstract the core reservoirs, water plant clusters, and water demand areas in the saltwater intrusion-affected region into directed graph nodes, forming a three-layer network structure of "water source area - water plant cluster - water demand area"; wherein, the water source area is the set of reservoirs in the region that undertake the main water supply task; the water plant cluster is a combination of several water plants in the region divided according to water supply range or function; and the water demand area is a number of regional units in the region divided according to water demand characteristics. Step S1.2: Collect daily water demand, reservoir water supply capacity, water level-storage capacity relationship and water quality monitoring data in the area affected by saltwater intrusion, and determine the decision variables as the allocation from the water source to the water plant group and the allocation from the water plant group to the water demand area, as shown in Equations (1) and (2) respectively: （1） In the above formula, R This is a collection of core reservoirs in the area affected by saltwater intrusion. P This refers to the cluster of water plants in the area. T For the set of scheduling periods; X r,p,t Reservoir r During the period t water plant group p Water supply; （2） In the above formula, Z This refers to a collection of water-demanding areas affected by saltwater intrusion. Y p,z,t Indicates a group of water plants p During the period t To water-demand areas z The amount of water delivered; Step S2: Determine the saltwater intrusion period using real-time water quality monitoring data of reservoirs in the affected area, dynamically adjust the objective function and constraints of the multi-objective multi-source optimization scheduling model, and prioritize the scheduling of water sources with safe water quality. Step S3: Based on the topology model constructed in Step S1, and using the salinity intrusion state in Step S2 as dynamic parameters, construct a multi-objective, multi-source water optimization scheduling model with a three-dimensional objective function of "water shortage, salinity intrusion impact, and cost difference." Combined with constraints, solve for the Pareto optimal solution set using the improved NSGA-III algorithm; specifically including: Step S3.1: Construct the objective function of the multi-objective, multi-water-source optimization scheduling model, as shown in equations (3), (4), and (5): Objective 1: Minimize water shortage: （3） In the above formula, Let the objective function be the amount of water shortage; For water-demand areas z During the period t The amount of water shortage, D z,t For water-demand areas z During the period t Water demand; Objective 2: Minimize the impact of saltwater intrusion on water supply. （4） In the above formula, f 2 ( X ) represents the objective function for the impact of saltwater intrusion on water supply; , For reservoir r No. t Chloride concentration was monitored at the water intake during specific time periods. The threshold for the influence of saltwater intrusion; The reservoir whose water supply is affected by saltwater intrusion; Objective 3: Minimize the difference in water supply costs: （5） In the above formula, f 3 ( X Let be the objective function for the difference in water supply costs; For reservoir r During the period t unit cost For reservoir r During the period t Total water supply K 0 This represents the actual cost of water supply. Step S3.2: Set constraints, as shown in equations (6), (7), (8), (9), (10), (11), (12), and (13): 1) Reservoir water supply capacity constraints: （6） In the above formula, For reservoir r During the period t Maximum water supply capacity; 2) Water plant supply capacity constraints: （7） In the above formula, For water plant p During the period t Maximum water supply capacity; 3) Pipeline transmission and distribution capacity constraints: （8） （9） In the above formula, For reservoir r To the water plant p The maximum water delivery capacity of the corresponding water supply pipeline. For water plant p To water-demand areas z The maximum water delivery capacity of the corresponding water supply pipeline; 4) Water plant water balance constraints: （10） 5) Water supply demand meets constraints: （11） 6) Nonnegativity constraint: （12） （13） Step S3.3: Solve using the NSGA-III algorithm: Encode decision variables into multidimensional vectors and generate an initial population through diversified strategies; introduce a penalty function in fitness evaluation to handle constraint violation problems; optimize the search process using adaptive crossover, mutation operators, and feasibility repair operators based on engineering heuristics rules; dynamically evaluate the convergence (MIGD) and distribution (MHV) of the solution set, and output the Pareto optimal solution set.

2. The method for optimizing the scheduling of a regional water supply system considering the impact of saltwater intrusion as described in claim 1, characterized in that, In step S1.1, the division of water plant groups is based on the layout of water supply projects and the distribution of water transmission pipelines in the areas affected by saltwater intrusion. The groups include northern water plant groups, central water plant groups, eastern water plant groups, southern water plant groups, western water plant groups, suburban water plant groups, and remote suburban water plant groups. Each water plant group contains at least one water plant with conventional water treatment capacity. The division of water demand areas is based on the administrative division, population density and water load characteristics of the area, including the central urban area, northern area, eastern area, western area and southern area types. Each water demand area includes at least one town or functional area with stable water demand.

3. The method for optimizing the scheduling of a regional water supply system considering the impact of saltwater intrusion according to claim 2, characterized in that, In step S2, the criteria for determining the saline tide period are: the chloride concentration at the intake of any reservoir in the area affected by the saline tide exceeds 250 mg / L; during the saline tide period, the priority coefficient for scheduling water sources with safe water quality is increased by 1.2-1.5 times, and the upper limit of water supply to water sources affected by the saline tide is reduced by 20%-30%.

4. The method for optimizing the scheduling of a regional water supply system considering the impact of saltwater intrusion as described in claim 3, characterized in that, The specific process of solving the problem using the NSGA-III algorithm in step S3.3 includes: Step S3.3.1: Encoding Decision Variables: Transforming the core decision variables of water supply scheduling into a mathematical form that the algorithm can process, that is, encoding the set of water supply volumes from the reservoir to the water plant group. The collection of water supply from water treatment plants to water demand areas The encoding is a one-dimensional real chromosome with a chromosome length of | R |×| P |×| T |+| P |×| Z |×| T |, the value range of each gene locus is ([0, Q max ], Q max To ensure that the coding results meet the engineering capacity boundaries, the maximum water supply capacity of the corresponding reservoir or the maximum water transmission capacity of the water plant is determined. Step S3.3.2: Diversified initial population generation: Seven engineering-oriented strategies are weighted and fused to generate an initial population of 100-300. The strategy group includes demand-oriented strategy, water supply capacity balancing strategy, historical scheduling fitting strategy, random focusing strategy, regional priority strategy, reservoir balancing strategy and mixed strategy. The weight of each sub-strategy in the mixed strategy is dynamically adjusted according to the scheduling deviation rate of the same period in the previous three years of the region. Step S3.3.3: Fitness assessment and constraint handling, specifically including: S3.3.3.1: Calculate multi-objective fitness: For each individual in the population, i.e. a complete set of water supply scheduling schemes, simultaneously calculate the water shortage target f1, the water supply target f2 affected by salinity intrusion, and the water supply cost difference target f3, and integrate the calculation results of the three targets into a three-dimensional fitness vector; S3.3.3.2: Constraint violation penalty: Introduce a penalty function to modify fitness. The modification formula is: Modified fitness = Original fitness + λ1max(0, Water shortage limit) + λ2max(0, Water supply exceeds the limit). Among them, λ1 is the supply and demand balance penalty coefficient. When the water shortage in the water demand area exceeds 5% of its total water demand, the value of λ1 is doubled to strengthen the penalty for severe water supply gaps; λ2 is the capacity constraint penalty coefficient. When the water supply of the reservoir exceeds 10% of its maximum water supply capacity, the value of λ2 is doubled to strictly limit the overcapacity water supply. S3.3.3.3: For individuals that still violate constraints after being penalized, repair them according to the project priority rules; Step S3.3.4: Drive population iteration through the synergistic operation of crossover and mutation, specifically including: Crossover operation: Simulated binary crossover SBX is used, with crossover probability... p c The initial value is 0.85~0.95, and it decreases linearly by 0.05 every 50 generations, gradually shifting towards local optimization; the crossover distribution index is fixed at 20. Mutation operation: Employs a polynomial mutation strategy, with mutation probability... p m Initially, the value is between 0.05 and 0.

15. When the population diversity index is less than 0.3, p m Increase by 0.02; the variation distribution index is fixed at 20; Step S3.3.5: To maintain the non-dominance and uniformity of the solution set, the population is updated in the following three steps, specifically including: Step S3.3.5.1: Merge the current parent population with the offspring population, perform fast non-dominated sorting on the merged population, and divide the non-dominated fronts into different levels according to the dominance relationship of individuals. The first level front is the current set of optimal candidate solutions. Step S3.3.5.2: In the three-dimensional target space of "water shortage - salinity impact - cost difference", preset uniformly distributed reference points, the number of which is 1.2 to 1.5 times the target population size, and calculate the correlation between the non-dominated solution and each reference point; Step S3.3.5.3: Select individuals from high to low correlation, and finally retain 100 to 150 individuals to form a new parent population, ensuring that the new population contains non-dominated solutions and can evenly cover different trade-off regions of the target space. Step S3.3.6: By dynamically monitoring the convergence and solution set distribution of the algorithm, determine the timing for iterative termination and output the optimal solution: Step S3.3.6.1: Dynamically calculate the average reverse generation distance (MIGD) and average hypervolume (MHV) of the solution set; Step S3.3.6.2: Terminate evolution when any of the following conditions are met: 1) The number of generations of evolution reaches the preset maximum value; 2) The average reverse generation distance (MIGD) for 50 consecutive generations is <0.001, and the hypervolume variation coefficient is <5%; Step S3.3.6.3: After terminating the evolution, output the final non-dominated solution set, containing 30-50 Pareto optimal solutions, each containing a complete set of solutions. and The allocation plan comprehensively covers the multi-objective trade-off space of "water shortage - salinity impact - cost difference", providing diversified alternatives for scheduling decisions.

5. The method for optimizing the scheduling of a regional water supply system considering the impact of saltwater intrusion according to claim 4, characterized in that, In step S3.3.2, the strategy includes: Demand-oriented strategy: Allocate water supply in reverse order based on water demand forecasts for different water-demand areas; Water supply capacity balancing strategy: allocate water according to the proportion of the maximum water supply capacity of each reservoir. ; Historical scheduling fitting strategy: Fit the scheduling data of the same period in the previous 3 years in the area affected by saltwater intrusion to generate an initial solution; Random focus strategy: Randomly select 1-2 water-demand areas to prioritize water supply; Regional priority strategy: Water allocation based on regional priority ; Reservoir balancing strategy: to keep the difference in water supply between reservoirs below 15%; Hybrid strategy: The above 6 strategies are weighted and combined. The weights are dynamically adjusted according to the scheduling deviation rate of the same period in the previous 3 years. If the scheduling deviation rate of a certain strategy is higher than 10%, its weight is reduced by 5% to 10%.

6. The method for optimizing the scheduling of a regional water supply system considering the impact of saltwater intrusion according to claim 5, characterized in that, In step S3.3.3.3, the engineering priority rules for the feasibility repair operator include: If the reservoir's water supply exceeds its limit, priority will be given to reducing the supply of saline water sources with the highest chloride concentration; for every 50 mg / L increase in concentration, the reduction ratio of the excess water supply from the corresponding water source will increase by 10%. If the water supply of a water plant is unbalanced, priority should be given to adjusting the water supply of the water plant that is geographically closest to the water demand area, and the adjustment amount in a single instance shall not exceed 15% of the water plant's designed water supply capacity. If the water shortage in the water-demand area exceeds the standard, priority will be given to the core area with the highest population density, and the water shortage in non-core areas will be distributed according to the population density of each area to balance the protection of people's livelihood and the fairness of water supply.

7. The method for optimizing the scheduling of a regional water supply system considering the impact of saltwater intrusion according to claim 6, characterized in that, In step S3.3.6.1, the average reverse generation distance (MIGD) for dynamically evaluating the convergence of the solution set and the average hypervolume (MHV) for evaluating the distribution of the solution set specifically include: Step S3.3.6.1.1: Calculate the mean reverse generation distance (MIGD) and convergence criteria: First, let the current optimal solution set be... S ={ x 1, x 2,..., x n };in, x i For the solution set of the first i Each Pareto candidate solution corresponds to a set of solutions including the water supply from the reservoir to the water treatment plant group. Water supply from water plant clusters to water demand areas A complete water supply scheduling plan n This represents the total number of solutions in the current solution set. Secondly, let the theoretically optimal solution set be... S *={ x 1*, x 2*,..., x k *};in, x j * is the first j One theoretically optimal solution; Third, calculate the first i candidate solutions x i Reverse generation distance IGD ( x i The mean reverse generation distance (MIGD) is as follows: （14） （15） In the above formula, IGD( x i )for x i To the theoretical optimal solution set S The three-dimensional Euclidean distance to the nearest solution in the *. f 1( x i ), f 2( x i ), f 3( x i ) are candidate solutions x i The target values ​​for water shortage, water supply affected by saltwater intrusion, and water supply cost differences; f 1( x j *) f 2( x j *) f 3( x j *) represent the theoretical optimal solutions. x j * The corresponding target value; Fourth, convergence judgment rule: when the calculated MIGD < 0.001, it is determined that the current solution set has converged to the theoretical optimal region and meets the algorithm convergence requirement; Step S3.3.6.1.2: Calculate the average hypervolume MHV and determine distribution equilibrium: First, with the "water shortage target" f 1. Saltwater intrusion affects water supply targets f 2. Target for difference in water supply costs f Using "3" as the coordinate axis, a three-dimensional Euclidean space is formed. O - f 1 f 2 f 3. Spatial origin O (0,0,0) represents the ideal optimal state of "zero water shortage, zero salinity impact, and zero cost difference"; Secondly, the effective hypervolume elements in the solution set are selected to form the effective hypervolume element set. V ={ V 1, V 2,..., V m };in, V j For the first j The volume of an effective hypervolume element, which is defined as an element that is not dominated by other candidate solutions and can independently contribute to the target space coverage. m The total number of effective hypervolume elements, m≤n; MHV is the average hypervolume, which is the sum of the volumes of all effective hypervolume elements; Third, calculate the first j Volume of an effective hypervolume unit V j Coefficient of variation of mean supervolume MHV CV Specifically: （16） （17） （18） In the above formula, x j For the candidate solution corresponding to the effective hypervolume element, its objective value is f 1( x j ), f 2( x j ), f 3( x j The values ​​are the water shortage, the impact of saltwater on water supply, and the difference in water supply costs for this solution, respectively. μ The mean of MHV. The standard deviation of the effective supervolume unit volume; Fourth, the distribution balance determination rule: when the calculated... CV When the percentage is less than 5%, it is determined that the current solution set is evenly distributed in the three-dimensional target space, with no obvious clustering or discontinuity, thus meeting the algorithm's distribution requirements.

Images