Scheduling method and system for long-distance water pumping station system considering multiple optimization

Abstract

The invention discloses a long-distance water diversion pump station system scheduling method and system considering multiple optimization. The method comprises the steps that pump station data are collected, an optimized operation model of a single pump station is constructed, a machine learning model is constructed, and the water diversion cost of each pump station is obtained through calculation; calculating the water demand of the area along the line and the water receiving area, inputting the water demand and the water guiding and lifting cost of each pump station into a pre-constructed multi-objective optimization model, solving the model to obtain a non-inferior solution set, and making a decision on the non-inferior solution set to obtain an optimal scheme, namely a water guiding and transferring scheme; optimizing the operation of the pump station system based on the lowest energy consumption of the total water diversion project system and the water diversion scheme to obtain a pump station operation scheme; and formulating a long-distance water diversion pump station system scheduling rule based on the water diversion scheme and the pump station operation scheme sample. According to the invention, through a multi-target scheduling strategy and advanced data processing capability, the operation efficiency of the pump station system is obviously improved, the energy consumption is reduced, and reasonable configuration and sustainable utilization of water resources are realized.

Description

Technical Field

[0001] The invention relates to a method for dispatching a long-distance water diversion pump station system, in particular to a method for dispatching a long-distance water diversion pump station system taking multiple optimizations into consideration. Background Art

[0002] With the acceleration of urbanization and the improvement of people's living standards, the demand for water resources continues to increase. Long-distance water diversion projects have become an important way to solve water shortages, especially in arid and semi-arid areas. The long-distance water diversion pump station system is responsible for transporting water from the water source to users or water treatment facilities through pipelines. Due to the long transportation distance and complex routes, the scheduling and energy consumption management of the pump station system face huge challenges. An effective scheduling strategy can significantly improve the operating efficiency of the pump station, reduce energy consumption, and reduce water waste.

[0003] The present invention proposes a long-distance water diversion pump station system scheduling method and system considering multiple optimizations to solve the above-mentioned problems currently existing. It uses machine learning algorithms to improve the accuracy of demand forecasting and optimize the operation scheduling of pump stations through historical data analysis and real-time monitoring. Summary of the invention

[0004] Purpose of the invention: To provide a method for dispatching a long-distance water diversion pump station system taking multiple optimizations into consideration, so as to solve the above-mentioned problems existing in the prior art. On the other hand, to provide a dispatching system for a long-distance water diversion pump station system taking multiple optimizations into consideration.

[0005] Technical solution: A long-distance water diversion pump station system dispatching method considering multiple optimizations includes the following steps: Step S1, collect pump station data, build an optimized operation model for a single pump station, calculate the mapping relationship between head, flow and energy consumption, build a machine learning model, and calculate the cost of water diversion and extraction at each pump station; Step S2, collecting water diversion and regulation project data, calculating the water demand of the areas along the line and the receiving areas, and inputting the water demand together with the water diversion and lifting costs of each pumping station into a pre-built multi-objective optimization model, solving the model to obtain a non-inferior solution set, making decisions on the non-inferior solution set to obtain the optimal solution, i.e., the water diversion and regulation solution; Step S3, optimizing the operation of the pump station system based on the minimum energy consumption of the total water diversion and regulation engineering system and the water diversion and regulation plan to obtain a pump station operation plan; Step S4, respectively construct a regression model, a long short-term memory network model, a reinforcement learning agent and a system dynamics model, based on the water diversion scheme and the pump station operation scheme samples, extract the regularity information of the optimal scheduling strategy, and formulate the scheduling rules for the long-distance water diversion pump station system.

[0006] According to one aspect of the present application, step S1 further comprises: Step S11, collect pump station data, and randomly sample the water level, flow rate, head and flow rate of each pump station based on flood control safety and slope stability to obtain a massive scenario (generally, more than 10,000 is considered massive); Step S12, constructing an optimized operation model of a single pump station; Step S13, inputting the massive scenarios into the optimization operation model of a single pump station in sequence, and obtaining the optimization operation plan and energy consumption of each pump station under the massive scenarios, that is, the head, flow rate and energy consumption mapping sample; Step S14: Based on the mapping samples of the head, flow rate and energy consumption of each pumping station in massive scenarios, a machine learning model is constructed to obtain the mapping relationship between the head, flow rate and energy consumption, and the cost of water extraction for each pumping station is calculated.

[0007] According to one aspect of the present application, the step S11 is further: Step S11a, collecting pump station data, fitting the two-dimensional joint distribution of head and flow, and randomly sampling to obtain a combination scenario of head and flow; Step S11b: based on the combined scenarios of water level, flow, head and flow allowed for flood control safety and slope stability, a weighted stratified sampling method is used to perform sampling to obtain a large number of scenarios.

[0008] According to one aspect of the present application, step S2 further comprises: Step S21, collecting water diversion project data, and calculating the water demand of the areas along the line and the receiving areas; Step S22, constructing a multi-objective optimization model, the objective function is: the highest water demand satisfaction, the lowest pump station system cost, and the minimum water loss along the way; Step S23, inputting the water demand of the area along the line, the water demand of the receiving area and the cost of water diversion and extraction of each pumping station into the multi-objective optimization model, and solving the multi-objective optimization model by using a multi-objective simulated annealing algorithm improved based on adaptive annealing temperature to obtain a non-inferior solution set; Step S24, using the analytic hierarchy process based on the Delphi method and cross-validation optimization combined with the TOPSIS method to perform multi-attribute decision-making on the non-inferior solution set, and obtaining the optimal solution is the water diversion plan.

[0009] According to one aspect of the present application, step S21 is further: Step S21a, collecting water diversion project data, including: regional lake water inflow data, water demand data of receiving areas, precipitation data, evaporation data, groundwater level data, land use data, population size, economic development index, and industry index; Step S21b, construct a water demand prediction model, the water demand of the area along the line is determined by precipitation, evaporation, groundwater level, and land use, and the water demand of the receiving area is determined by population, per capita water consumption, economic development level, and industrial index; Step S21c, solving the water demand prediction model to obtain the water demand of the areas along the line and the receiving areas.

[0010] According to one aspect of the present application, step S23 is further: Step S23a, setting the initial solution, initial temperature and parameters, calculating the objective function value of the current solution, and confirming whether it is a non-inferior solution; Step S23b, determining the domain of the current solution based on the annealing temperature, generating a new solution, and determining whether the new solution is better than the current solution in all objectives; Step S23c, adjusting the temperature based on the quality of the new solution generated by the current iteration; Step S23d, repeat the iteration until complete convergence, and obtain a non-inferior solution set.

[0011] According to one aspect of the present application, step S24 is further: Step S24a, constructing a decision hierarchy model, including: a target layer, an attribute layer, and a solution layer; Step S24b, using a cross-validation method to divide the experts into several groups to evaluate and construct a relative importance comparison matrix between the attributes, and using the Delphi method to evaluate and adjust the relative importance comparison matrix, and using the eigenvalue method to calculate the weight of each attribute; Step S24c, for each solution in the non-inferior solution set, calculate the performance value of the solution on each attribute in turn, standardize the values ​​in the performance matrix, and calculate the comprehensive score of each non-inferior solution based on the weight and the standardized performance value; Step S24d, set the optimal value of each attribute to the ideal solution, and the worst value of each attribute to the negative ideal solution. Use Euclidean distance to calculate the distance from each non-inferior set to the ideal solution and the negative ideal solution, calculate the degree of closeness to the ideal solution, and select the non-inferior solution with the greatest degree of closeness to the ideal solution as the optimal solution, that is, the water diversion plan.

[0012] According to one aspect of the present application, step S3 further comprises: Step S31, constructing an optimization operation model, the objective function is: the total energy consumption of the water diversion project system is the lowest; Step S32, inputting the mapping relationship between the head, flow rate and energy consumption of each pump station in the massive scenario into the optimization operation model, solving the model to obtain the initial operation plan of the pump station; Step S33: Optimize the initial operation plan of the pump station based on the water diversion plan to obtain the pump station operation plan.

[0013] According to one aspect of the present application, step S4 is further: Step S41, constructing a regression model, a long short-term memory network model, a reinforcement learning agent and a system dynamics model respectively; Step S42: using a regression model to predict water demand and pump station operation status, and using a long short-term memory network model to process time series to predict water demand and energy consumption; Step S43, the water demand predicted by the coupling regression model, the pump station operation status and the water demand and energy consumption predicted by the long short-term memory network model are used to construct the state equation of the pump station system, define the relationship between the water level, flow, energy consumption and the start, shutdown and operation speed of the pump station, define the feedback mechanism, simulate the dynamic changes of the system, and obtain the system simulation results; Step S44: Based on the system simulation results, an optimization algorithm is used to adjust the dispatching rules of the pump station operation, and a reinforcement learning agent is used to learn the best dispatching strategy to extract the dispatching rules of the long-distance water diversion pump station system.

[0014] According to another aspect of the present application, a long-distance water diversion pump station system scheduling system considering multiple optimizations is provided, comprising: at least one processor; and a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the long-distance water diversion pump station system scheduling method considering multiple optimizations as described in any of the above technical solutions.

[0015] Beneficial effects: The long-distance water diversion pump station system scheduling method considering multiple optimizations can improve the efficiency of water resource utilization, reduce energy consumption and operating costs, improve operational reliability, enhance adaptability and flexibility, contribute to the sustainable use of water resources, and promote the protection and restoration of the ecological environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of the present invention.

[0017] Figure 2 It is a flow chart of step S1 of the present invention.

[0018] Figure 3 It is a flow chart of step S2 of the present invention.

[0019] Figure 4 It is a flow chart of step S3 of the present invention.

[0020] Figure 5 It is a flow chart of step S4 of the present invention. DETAILED DESCRIPTION

[0021] like Figure 1 According to one aspect of the present application, a method for dispatching a long-distance water pump station system taking multiple optimizations into consideration is provided, comprising the following steps: Step S1, collect pump station data, build an optimized operation model for a single pump station, calculate the mapping relationship between head, flow and energy consumption, build a machine learning model, and calculate the cost of water diversion and extraction at each pump station; Step S2, collecting water diversion and regulation project data, calculating the water demand of the areas along the line and the receiving areas, and inputting the water demand together with the water diversion and lifting costs of each pumping station into a pre-built multi-objective optimization model, solving the model to obtain a non-inferior solution set, making decisions on the non-inferior solution set to obtain the optimal solution, i.e., the water diversion and regulation solution; Step S3, optimizing the operation of the pump station system based on the minimum energy consumption of the total water diversion and regulation engineering system and the water diversion and regulation plan to obtain a pump station operation plan; Step S4, respectively construct a regression model, a long short-term memory network model, a reinforcement learning agent and a system dynamics model, based on the water diversion scheme and the pump station operation scheme samples, extract the regularity information of the optimal scheduling strategy, and formulate the scheduling rules for the long-distance water diversion pump station system.

[0022] According to one aspect of the present application, step S1 further comprises: Step S11, collect pump station data, and randomly sample the water level, flow rate, and head and flow rate of each pump station based on flood control safety and slope stability to obtain a large number of scenarios; Step S12, constructing an optimized operation model of a single pump station; Step S13, inputting the massive scenarios into the optimization operation model of a single pump station in sequence, and obtaining the optimization operation plan and energy consumption of each pump station under the massive scenarios, that is, the head, flow rate and energy consumption mapping sample; Step S14: Based on the mapping samples of the head, flow rate and energy consumption of each pumping station in massive scenarios, a machine learning model is constructed to obtain the mapping relationship between the head, flow rate and energy consumption, and the cost of water extraction for each pumping station is calculated.

[0023] The pump station system follows the order of energy transfer. Electric energy enters the main transformer of the pump station from the substation outside the pump station through a dedicated high-voltage transmission line, and then supplies power to the water pump system and station auxiliary equipment along the power supply cable. Therefore, the calculation of the pump station cost includes: the energy consumption of the main pump system, the energy consumption of the transformer, the energy consumption of the transmission line system, and the energy consumption of the station auxiliary equipment; In this embodiment, a machine learning model is trained based on a large number of samples, and then a mapping relationship is obtained. The relationship between head, flow rate and energy consumption is obtained through scenarios, and then the complex physical process simulation calculation is refined through machine learning, which can save the step of finding an optimization solution in the middle, and can be more efficient in future use; The present invention calculates the water extraction cost of each pumping station in advance, which can be used as a boundary condition in subsequent optimization scheduling, thereby reducing the calculation amount of optimization scheduling.

[0024] According to one aspect of the present application, the step S11 is further: Step S11a, collecting pump station data, fitting the two-dimensional joint distribution of head and flow, and randomly sampling to obtain a combination scenario of head and flow; Step S11b: based on the combined scenarios of water level, flow, head and flow allowed for flood control safety and slope stability, a weighted stratified sampling method is used to perform sampling to obtain a large number of scenarios.

[0025] The weighted stratified sampling method combines stratified sampling and weighted sampling. It first divides the population into several strata to ensure the representativeness of the sample in each sub-group, and then uses a weighted method to extract samples within each stratum to better reflect the characteristics of the population.

[0026] In this embodiment, the use of weighted stratified sampling helps to reduce the bias caused by a single sampling method and improve the accuracy and reliability of data results. Through stratified sampling, it ensures that each subgroup has a reasonable proportion in the sample, making the sample more representative; through weighted sampling, it can more accurately reflect the importance of each stratum in the population.

[0027] In a certain embodiment, specifically: Collect operating parameter data of each pumping station and stratify them based on pumping station type, operating status, and pipeline stage; Based on historical operation data, the weight of each layer is determined, where: large pumping stations have a weight of 0.4, medium pumping stations have a weight of 0.3, small pumping stations have a weight of 0.2, and standby pumping stations have a weight of 0.1; Based on the weight of each layer, the sample size of each layer is determined. The current total sample size is 1000, then the large pumping station is 400; the medium pumping station is 300; the small pumping station is 200; the standby pumping station is 100; In each stratum, random sampling is performed.

[0028] According to one aspect of the present application, step S2 further comprises: Step S21, collecting water diversion project data, and calculating the water demand of the areas along the line and the receiving areas; Step S22, constructing a multi-objective optimization model, the objective function is: the highest water demand satisfaction, the lowest pump station system cost, and the minimum water loss along the way; In this embodiment, the degree of water demand satisfaction means that different regions have a priority order, and different characteristic indicators and weights are assigned to regions based on the degree of water shortage, population density, and economic structure. The weight of the region = the score of the region / the score of all regions and calculation; Step S23, inputting the water demand of the area along the line, the water demand of the receiving area and the cost of water diversion and extraction of each pumping station into the multi-objective optimization model, and solving the multi-objective optimization model by using a multi-objective simulated annealing algorithm improved based on adaptive annealing temperature to obtain a non-inferior solution set; Step S24, using the analytic hierarchy process based on the Delphi method and cross-validation optimization combined with the TOPSIS method to perform multi-attribute decision-making on the non-inferior solution set, and obtaining the optimal solution is the water diversion plan.

[0029] According to one aspect of the present application, step S21 is further: Step S21a, collecting water diversion project data, including: regional lake water inflow data, water demand data of receiving areas, precipitation data, evaporation data, groundwater level data, land use data, population size, economic development index, and industry index; Step S21b, construct a water demand prediction model, the water demand of the area along the line is determined by precipitation, evaporation, groundwater level, and land use, and the water demand of the receiving area is determined by population, per capita water consumption, economic development level, and industrial index; Step S21c, solving the water demand prediction model to obtain the water demand of the areas along the line and the receiving areas.

[0030] According to one aspect of the present application, step S23 is further: Step S23a, setting the initial solution, initial temperature and parameters, calculating the objective function value of the current solution, and confirming whether it is a non-inferior solution; Step S23b, determining the domain of the current solution based on the annealing temperature, generating a new solution, and determining whether the new solution is better than the current solution in all objectives; Step S23c, adjusting the temperature based on the quality of the new solution generated by the current iteration; Step S23d, repeat the iteration until complete convergence, and obtain a non-inferior solution set.

[0031] The simulated annealing algorithm is a randomized search algorithm. The traditional simulated annealing algorithm performs well when dealing with single-objective optimization problems. However, in this embodiment, due to the multi-objective optimization problem, its performance will be limited. Therefore, in this embodiment, an improved mechanism of adaptive annealing temperature is introduced, which significantly improves the algorithm in global search capability and convergence speed.

[0032] In this embodiment, since the water diversion problem usually involves multiple objectives, the multi-objective simulated annealing algorithm improved by adaptive annealing temperature can effectively balance multiple objectives. The algorithm can automatically adjust parameters according to real-time changes in water resources, adapt to different climatic conditions and changes in water demand, and ensure the dynamic and real-time optimization. In this embodiment, the multi-objective simulated annealing algorithm based on adaptive annealing temperature improvement can quickly provide high-quality scheduling solutions with short response time, which is suitable for emergency decision-making needs in practical applications.

[0033] In a certain embodiment, specifically: In order to ensure the water demand and reduce the operating energy consumption, the optimization goal is set as: maximizing water supply and minimizing energy consumption; Select pump stations A, B, and C, with a water supply demand of 1,000 cubic meters per day, pump A with a power of 10KW, pump B with a power of 15KW, and pump C with a power of 20KW; Each pump is assumed to run for 8 hours per day, with a water supply weight of 0.6 and an energy consumption weight of 0.4; Set the initial temperature to 1000 and the end temperature to 1; Randomly select the operation combination of pumps, for example: Pump A and Pump B start at the same time, and Pump C is turned off; Calculate the function value of water supply and energy consumption; Dynamically adjust temperature based on adaptive strategies; Perform perturbations on solutions and determine whether to accept new solutions; When the number of iterations reaches the set number or the temperature drops to the termination temperature, the iteration stops and a non-inferior solution set is obtained.

[0034] According to one aspect of the present application, step S24 is further: Step S24a, constructing a decision hierarchy model, including: a target layer, an attribute layer, and a solution layer; Step S24b, using a cross-validation method to divide the experts into several groups to evaluate and construct a relative importance comparison matrix between the attributes, and using the Delphi method to evaluate and adjust the relative importance comparison matrix, and using the eigenvalue method to calculate the weight of each attribute; Step S24c, for each solution in the non-inferior solution set, calculate the performance value of the solution on each attribute in turn, standardize the values ​​in the performance matrix, and calculate the comprehensive score of each non-inferior solution based on the weight and the standardized performance value; Step S24d, set the optimal value of each attribute to the ideal solution, and the worst value of each attribute to the negative ideal solution. Use Euclidean distance to calculate the distance from each non-inferior set to the ideal solution and the negative ideal solution, calculate the degree of closeness to the ideal solution, and select the non-inferior solution with the greatest degree of closeness to the ideal solution as the optimal solution, that is, the water diversion plan.

[0035] The analytic hierarchy process is a tool for quantitatively and qualitatively analyzing decision problems by constructing a hierarchical model. The TOPSIS method is a distance-based multi-attribute decision-making method that ranks decision plans by calculating how close they are to the ideal solution and the negative ideal solution. In this embodiment, the analytic hierarchy process and TOPSIS are used in combination to achieve more scientific weight allocation and plan evaluation in multi-attribute decision-making.

[0036] In this embodiment, since water diversion management involves multiple evaluation dimensions, the hierarchical analysis method based on the Delphi method and cross-validation optimization combined with the TOPSIS method can comprehensively consider the impact of various attributes and provide managers with multi-angle evaluation results. The supply and demand of water resources are changing. The hierarchical analysis method as a preliminary decision-making tool can quickly adapt to changes in new data, and the TOPSIS method can quickly judge the advantages and disadvantages of new decision-making methods to ensure the flexibility of decision-making.

[0037] In a certain embodiment, specifically: A certain region is facing a serious water shortage problem, which needs to be solved through different water diversion schemes Z, X, and Y. The decision attributes are set as: water supply, operating energy consumption, environmental impact, and investment cost; Evaluate the specific data of water diversion schemes Z, X, and Y in terms of water supply, operating energy consumption, environmental impact, and investment cost and make a table; Based on expert opinions, a hierarchy is set up to assess the importance of each attribute; The weights calculated based on the scores are 0.45 for water supply, 0.25 for operating energy consumption, 0.2 for environmental impact, and 0.1 for investment cost; The data of each plan is constructed into a decision matrix and standardized; Calculate a weighted decision matrix using weights; Calculate the ideal solution and negative ideal solution separately; Calculate the Euclidean distance of each solution to the ideal solution and the negative ideal solution, and obtain the priority of the solution.

[0038] According to one aspect of the present application, step S3 further comprises: Step S31, constructing an optimization operation model, the objective function is: the total energy consumption of the water diversion project system is the lowest; Step S32, inputting the mapping relationship between the head, flow rate and energy consumption of each pumping station in the massive scenario into the optimization operation model, solving the model to obtain the initial operation plan of the pumping station; Step S33: Optimize the initial operation plan of the pump station based on the water diversion plan to obtain the pump station operation plan.

[0039] According to one aspect of the present application, step S4 is further: Step S41, constructing a regression model, a long short-term memory network model, a reinforcement learning agent and a system dynamics model respectively; Multiple AI models include machine learning models and deep learning models, which can analyze and deeply understand problems from different angles. In this embodiment, they help overcome the limitations of a single model in handling complex scheduling tasks; The system dynamics model can simulate the interactions and delay effects within the system. In this embodiment, an in-depth analysis of the dynamic characteristics of the system can help better understand the long-term impact of scheduling decisions.

[0040] Step S42: using a regression model to predict water demand and pump station operation status, and using a long short-term memory network model to process time series to predict water demand and energy consumption; In this embodiment, optimization is performed for different objectives based on different AI models, and the overall scheduling efficiency is improved by synthesizing and considering multiple objectives, which can avoid the reduction in overall efficiency caused by unilateral optimization.

[0041] Step S43, the water demand predicted by the coupling regression model, the pump station operation status and the water demand and energy consumption predicted by the long short-term memory network model are used to construct the state equation of the pump station system, define the relationship between the water level, flow, energy consumption and the start, shutdown and operation speed of the pump station, define the feedback mechanism, simulate the dynamic changes of the system, and obtain the system simulation results; The graphical representation provided by the system dynamics model can help to more intuitively understand the scheduling rules and system behavior. By combining the decision-making process and results of different models and providing detailed explanations and basis, it can enhance the decision makers' trust and acceptance of the scheduling rules.

[0042] Step S44: Based on the system simulation results, an optimization algorithm is used to adjust the dispatching rules of the pump station operation, and a reinforcement learning agent is used to learn the best dispatching strategy to extract the dispatching rules of the long-distance water diversion pump station system.

[0043] By combining multiple AI models with system dynamics models, we can give full play to the advantages of each model to form a more powerful and flexible scheduling system. This approach not only improves the scientificity and accuracy of scheduling decisions, but also adapts to complex and changing operating environments and improves the efficiency and flexibility of the overall system.

[0044] In a certain embodiment, specifically: Water resources in a certain region are scarce, involving agricultural irrigation and industrial water use. It is necessary to optimize the distribution of water resources through a water diversion system. The region needs to maximize the efficiency of water resource utilization while meeting agricultural and industrial needs. First, using meteorological and water pollution data, a regression model was constructed to predict water flow and water quality changes; Used to dynamically optimize water allocation strategies and adjust water supply priorities; Integrate managers’ experience to develop short-term water scheduling strategies based on real-time water quality and demand; Construct a water resource management model to define the feedback relationship between reservoir water demand, irrigation demand and water quality changes, simulate water cycles and influencing factors, and evaluate the impact of different diversion strategies; Analyzing historical data through machine learning models, we found that water demand increases dramatically in spring, and water quality decreases in some cases; Dynamically adjust water diversion strategies based on real-time water quality, weather forecasts, and water demand through reinforcement learning models; Construct a feedback model to clearly depict the relationship between changes in reservoir water demand, water demand and water quality, simulate the impact of different water diversion volumes on reservoir water level and water quality, and identify the optimal water diversion strategy.

[0045] According to another aspect of the present application, a long-distance water diversion pump station system scheduling system considering multiple optimizations is provided, comprising: at least one processor; and a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement any of the above-mentioned methods for scheduling a long-distance water diversion pump station system considering multiple optimizations.

[0046] The preferred embodiments of the present invention are described in detail above; however, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A method for scheduling a long-distance water diversion pump station system considering multiple optimizations, characterized in that: The steps include: Step S1: Collect pump station data, build an optimized operation model for a single pump station, calculate the mapping relationship between head, flow rate, and energy consumption, build a machine learning model, and calculate the cost of water diversion and extraction at each pump station; Step S2: Collect water diversion and regulation project data, calculate the water demand of the areas along the route and the receiving areas, and input the data into a pre-built multi-objective optimization model together with the water diversion and extraction costs of each pumping station. Solve the model to obtain a set of non-inferior solutions, and make decisions on the non-inferior solutions to obtain the optimal solution, i.e., the water diversion and regulation solution. Step S3: Optimize the operation of the pumping station system based on the minimum energy consumption of the total water diversion and water diversion project system and the water diversion and water diversion plan to obtain a pumping station operation plan; Step S4: Construct a regression model, a long-short-term memory network model, a reinforcement learning agent, and a system dynamics model respectively. Based on the water diversion scheme and the pump station operation scheme samples, extract the regularity information of the optimal scheduling strategy and formulate the scheduling rules for the long-distance water diversion pump station system.

2. The method for scheduling a long-distance water diversion pump station system considering multiple optimizations according to claim 1, characterized in that: The step S1 is further as follows: Step S11: Collect pump station data and randomly sample the water level, flow rate, head, and flow rate of each pump station based on flood control safety and slope stability to obtain a large number of scenarios; Step S12: constructing an optimized operation model of a single pump station; Step S13: input the massive scenarios into the optimized operation model of a single pump station in sequence to obtain the optimized operation plan and energy consumption of each pump station under the massive scenarios, that is, the head, flow rate and energy consumption mapping sample; Step S14: Based on the mapping samples of the head, flow rate, and energy consumption of each pumping station in massive scenarios, a machine learning model is constructed to obtain the mapping relationship between the head, flow rate, and energy consumption, and calculate the cost of water diversion for each pumping station.

3. The method for scheduling a long-distance water diversion pump station system considering multiple optimizations according to claim 2, wherein: The step S11 is further as follows: Step S11a: collect pump station data, fit the two-dimensional joint distribution of head and flow, and randomly sample to obtain combined scenarios of head and flow; Step S11b: Based on the combined scenarios of water level, flow, head and flow allowed for flood control safety and slope stability, a weighted stratified sampling method is used to perform sampling to obtain a large number of scenarios.

4. The method for scheduling a long-distance water diversion pump station system considering multiple optimizations according to claim 1, wherein: The step S2 is further as follows: Step S21: Collect water diversion project data and calculate water demand in areas along the diversion project and receiving areas; Step S22: construct a multi-objective optimization model, with the objective function being: highest water demand satisfaction, lowest pumping station system cost, and minimum water loss along the way; Step S23: inputting the water demand of the areas along the line, the water demand of the receiving areas, and the cost of water diversion and extraction at each pumping station into a multi-objective optimization model, and solving the multi-objective optimization model using a multi-objective simulated annealing algorithm improved based on adaptive annealing temperature to obtain a non-inferior solution set; Step S24: Use the hierarchical analysis method based on the Delphi method and cross-validation optimization combined with the TOPSIS method to perform multi-attribute decision-making on the non-inferior solution set, and obtain the optimal solution, which is the water diversion scheme.

5. The method for scheduling a long-distance water diversion pump station system considering multiple optimizations according to claim 4, characterized in that: The step S21 is further as follows: Step S21a: Collect water diversion project data, including: regional lake water inflow data, water demand data of the receiving area, precipitation data, evaporation data, groundwater level data, land use data, population size, economic development index, and industry index; Step S21b: Construct a water demand prediction model. The water demand of the area along the line is determined by precipitation, evaporation, groundwater level, and land use. The water demand of the receiving area is determined by population, per capita water consumption, economic development level, and industrial index. Step S21c: Solve the water demand prediction model to obtain the water demand of the areas along the line and the water receiving areas.

6. The method for scheduling a long-distance water diversion pump station system considering multiple optimizations according to claim 4, characterized in that: The step S23 is further as follows: Step S23a: Set the initial solution, initial temperature and parameters, calculate the objective function value of the current solution, and confirm whether it is a non-inferior solution; Step S23b: determine the domain of the current solution based on the annealing temperature, generate a new solution, and determine whether the new solution is better than the current solution in all objectives; Step S23c: adjusting the temperature based on the quality of the new solution generated by the current iteration; Step S23d: Repeat the iteration until complete convergence, and obtain a non-inferior solution set.

7. The method for scheduling a long-distance water diversion pump station system considering multiple optimizations according to claim 4, characterized in that: The step S24 is further as follows: Step S24a: construct a decision hierarchy model, including: a target layer, an attribute layer, and a solution layer; Step S24b: using a cross-validation method to divide the experts into several groups to evaluate and construct a relative importance comparison matrix between the various attributes, and using the Delphi method to evaluate and adjust the relative importance comparison matrix, and using the eigenvalue method to calculate the weight of each attribute; Step S24c: For each solution in the non-inferior solution set, calculate the performance value of each solution on each attribute in turn, standardize the values ​​in the performance matrix, and calculate the comprehensive score of each non-inferior solution based on the weights and the standardized performance values; Step S24d: Set the optimal value of each attribute to the ideal solution, and the worst value of each attribute to the negative ideal solution. Use Euclidean distance to calculate the distance from each non-inferior set to the ideal solution and the negative ideal solution, calculate the degree of closeness to the ideal solution, and select the non-inferior solution with the greatest degree of closeness to the ideal solution as the optimal solution, that is, the water diversion plan.

8. The method for scheduling a long-distance water diversion pump station system considering multiple optimizations according to claim 1, wherein: The step S3 is further as follows: Step S31: construct an optimization operation model, with the objective function being: the energy consumption of the total water diversion and regulation engineering system is minimized; Step S32: Input the mapping relationship between the head, flow rate, and energy consumption of each pumping station under the massive scenario into the optimization operation model, and solve the model to obtain the initial operation plan of the pumping station; Step S33: Optimize the initial operation plan of the pumping station based on the water diversion plan to obtain the pumping station operation plan.

9. The method for dispatching a long-distance water diversion pump station system considering multiple optimizations according to claim 1, characterized in that: The step S4 is further as follows: Step S41: construct a regression model, a long short-term memory network model, a reinforcement learning agent, and a system dynamics model respectively; Step S42: using a regression model to predict water demand and pump station operating status, and using a long short-term memory network model to process time series to predict water demand and energy consumption; Step S43: The water demand predicted by the regression model, the pump station operating status, and the water demand and energy consumption predicted by the long short-term memory network model are coupled to construct a state equation for the pump station system, define the relationship between the water level, flow rate, energy consumption, and the start, shut-down, and operating speed of the pump station, define a feedback mechanism, simulate the dynamic changes of the system, and obtain the system simulation results; Step S44: Based on the system simulation results, an optimization algorithm is used to adjust the dispatching rules of the pump station operation, and a reinforcement learning agent is used to learn the optimal dispatching strategy to extract the dispatching rules of the long-distance water diversion pump station system.

10. A long-distance water diversion pump station system scheduling system considering multiple optimizations is characterized by: include: at least one processor; as well as a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the long-distance water diversion pump station system scheduling method considering multiple optimizations as described in any one of claims 1 to 9.

Images