Water energy scheduling method

Abstract

The present invention relates to the technical field of water energy scheduling, and specifically to a water energy scheduling method. Obtain the future water usage demands corresponding to each water diversion outlet in the target water conveyance pipeline within a preset future time period, and obtain the geographical location information, current flow data, and current pressure data corresponding to each water diversion outlet; obtain the current working parameter information corresponding to each water storage and power generation device in the target water conveyance pipeline; the water storage and power generation device is used for water storage or power generation, and output a target scheduling strategy according to the future water usage demands, geographical location information, current flow data, current pressure data corresponding to each water diversion outlet, and the current working parameter information corresponding to the water storage and power generation device; reduce manual intervention, improve the accuracy and timeliness of scheduling, reduce operating costs, improve the management level and operation efficiency of the entire water conveyance system, and safely and stably achieve the comprehensive utilization of water energy without affecting the water resource allocation function of the water diversion project itself.

Description

Technical Field

[0001] The present invention relates to the technical field of water energy scheduling, and particularly to a water energy scheduling method. Background Art

[0002] As a key measure to optimize the allocation of water resources and solve the problem of uneven distribution of regional water resources, water diversion and regulation projects have been widely constructed and applied in China. These projects usually have large-scale water flow drops and flows, containing rich water energy resources. However, in the past project operations, the functions of water diversion and regulation projects mainly focused on the allocation of water resources, with a low degree of utilization of the water energy therein, and a large amount of water energy resources were wasted. In the current situation where the concept of low-carbon and sustainable development has taken root in people's hearts, exploring the water energy potential in water diversion and regulation projects and achieving the full and efficient utilization of water energy have become important research directions in the field of water conservancy projects. This not only helps to improve the stability and sustainability of energy supply, but also can create additional energy benefits on the basis of water resource allocation, with significant economic and environmental benefits.

[0003] Currently, although some explorations have been carried out on the utilization of water energy in water diversion and regulation projects, there are still many technical bottlenecks. In terms of the construction of the water energy allocation system, there is a lack of a comprehensive, systematic and efficient design scheme, making it difficult to achieve the rational utilization of water energy.

[0004] Therefore, how to safely and stably achieve the comprehensive utilization of water energy without affecting the water resource allocation function of the water diversion and regulation project itself is also a key problem to be solved urgently. Summary of the Invention

[0005] In view of this, the present invention provides a water energy scheduling method to solve the problem of safely and stably achieving the comprehensive utilization of water energy without affecting the water resource allocation function of the water diversion and regulation project itself.

[0006] In a first aspect, the present invention provides a water energy scheduling method, which includes:

[0007] Obtain the future water usage demands corresponding to each water diversion port in the target water conveyance pipeline within a preset future time period, and obtain the geographical location information, current flow data, and current pressure data corresponding to each water diversion port;

[0008] Obtain the current working parameter information corresponding to each water storage and power generation device in the target water conveyance pipeline; the water storage and power generation device is used for water storage or power generation, and the current working parameters include current water storage working parameters and / or current power generation working parameters; at least one water diversion port is equipped with a water storage and power generation device;

[0009] Output a target scheduling strategy according to the future water demand, geographical location information, current flow data, current pressure data corresponding to each water diversion outlet, and the current working parameter information corresponding to the water storage and power generation equipment; the scheduling strategy includes a first scheduling strategy for controlling the valves corresponding to each water diversion outlet and / or a second scheduling strategy for controlling the water storage and power generation equipment corresponding to each water diversion outlet.

[0010] The water energy scheduling method provided by the embodiments of the present application obtains the future water demand corresponding to each water diversion outlet in the target water conveyance pipeline within a preset future time period. It can understand the water use changes in different regions in advance, so as to more accurately schedule water resources, avoid water supply shortages or surpluses, improve the utilization efficiency of water resources, and ensure the satisfaction of users' water demand. Then, obtain the geographical location information, current flow data, and current pressure data corresponding to each water diversion outlet. Obtaining the current working parameter information corresponding to each water storage and power generation equipment in the target water conveyance pipeline helps to flexibly adjust the working mode of the water storage and power generation equipment according to the actual water demand and the pipeline operation status. For example, during the low water use period, water storage can be increased to ensure sufficient water supply during the high water use period; at the same time, power generation can be reasonably arranged according to the water flow situation to improve the utilization efficiency of water energy, realize the comprehensive utilization of water resources, and increase energy output. Output a target scheduling strategy according to the future water demand, geographical location information, current flow data, current pressure data corresponding to each water diversion outlet, and the current working parameter information corresponding to the water storage and power generation equipment. Among them, the first scheduling strategy can accurately control the valve opening of each water diversion outlet to adjust the water flow size to meet the water demand of different regions; the second scheduling strategy can optimize the operation of the water storage and power generation equipment to achieve the best balance between water storage and power generation. This intelligent scheduling method reduces manual intervention, improves the accuracy and timeliness of scheduling, reduces operating costs, and improves the management level and operation efficiency of the entire water conveyance system. Without affecting the water resource allocation function of the water diversion project itself, it safely and stably realizes the comprehensive utilization of water energy. In addition, making scheduling decisions based on comprehensive information can prevent and respond to various possible situations in advance, such as local pipeline failures and sudden changes in water demand. By reasonably adjusting the operation of valves and water storage and power generation equipment, the impact of these sudden situations on the entire water conveyance system can be alleviated to a certain extent, enhancing the reliability and stability of the system, reducing the occurrence of water cut-off accidents, and ensuring water supply safety.

[0011] In an alternative embodiment, outputting a target scheduling strategy according to the future water demand, geographical location information, current flow data, current pressure data corresponding to each water diversion outlet, and the current working parameter information corresponding to the water storage and power generation equipment includes:

[0012] Construct a water diversion outlet state space corresponding to each water diversion outlet based on the geographical location information, future water demand, current flow data, and current pressure data corresponding to each water diversion outlet;

[0013] Construct the device state space based on the current working parameter information corresponding to each water storage and power generation device;

[0014] Based on the state space of each water diversion outlet, generate a first action space for controlling the valves of each water diversion outlet; the first action space includes multiple standby valve control actions;

[0015] Based on the state space of each device, generate a second action space for controlling each water storage and power generation device; the second action space includes multiple standby device control actions;

[0016] Construct an action reward calculation mechanism; the action reward calculation mechanism includes at least one of the water demand satisfaction reward, the device operation benefit reward, and the system stability reward;

[0017] Based on the action reward calculation mechanism, evaluate each action combination generated based on the first action space and the second action space;

[0018] According to the evaluation results, output the target scheduling strategy.

[0019] The water energy scheduling method provided by the embodiments of this application constructs a state space for each water diversion outlet based on the geographical location information, future water demand, current flow data, and current pressure data corresponding to each water diversion outlet; constructs an equipment state space based on the current working parameter information corresponding to each water storage and power generation device. By constructing the state space of the water diversion outlet and the equipment state space, various states of the water diversion outlet and the water storage and power generation device in the water conveyance system can be comprehensively and accurately described, enabling the water conveyance system to have a clear understanding of its own operating conditions and providing an accurate basis for subsequent decision-making. Based on the state space of each water diversion outlet, a first action space for controlling the valves of each water diversion outlet is generated, and based on the state space of each device, a second action space for controlling each water storage and power generation device is generated. This means that when facing different states, the water conveyance system has more options to achieve optimal scheduling, can be flexibly adjusted according to the actual situation, and improves the adaptability and flexibility of the system. A mechanism for calculating action rewards is constructed; the mechanism for calculating action rewards includes at least one of a water demand satisfaction reward, an equipment operation benefit reward, and a system stability reward. This multi-dimensional reward mechanism can prompt the system to not only focus on meeting water demand when formulating scheduling strategies, but also take into account the operation benefits of water storage and power generation equipment and the stability of the entire system, achieving multi-objective optimization of water resource utilization, energy production, and reliable system operation. Then, based on the mechanism for calculating action rewards, each action combination generated based on the first action space and the second action space is evaluated; according to the evaluation results, a target scheduling strategy is output. Thus, it can be ensured that the generated scheduling strategy is the optimal or relatively optimal choice after comprehensively considering various factors. By comparing and screening the reward values of different action combinations, the water conveyance system can find the scheduling plan that best meets the actual needs, improves the overall benefits, and ensures the stability of the system, thereby improving the operation efficiency and management level of the water conveyance system. Through the above series of measures, this method forms a complete closed-loop optimization system from the construction of the state space to the generation of the action space, and then to the evaluation based on the comprehensive reward mechanism and the output of the scheduling strategy. It can effectively improve the overall performance of the water conveyance system, including more accurately meeting water demand, improving the operation benefits of water storage and power generation equipment, enhancing the stability and reliability of the system, and ultimately achieving the efficient utilization of water resources and the sustainable operation of the system.

[0020] In an alternative embodiment, the formula for the water demand satisfaction reward is:

[0021] ;

[0022] where n is the number of water diversion outlets, Q 实际，i is the actual water supply of the i-th water diversion outlet; Q 需求，i is the future water demand of the i-th water diversion outlet; α 类型，i is the urgency coefficient of the water use type corresponding to the i-th water diversion outlet, and both a and b are coefficients.

[0023] The water energy scheduling method provided by the embodiments of this application, the reward formula for meeting water demand is Thus, the future water demands corresponding to each water diversion point can be met, and the urgency of the water use types corresponding to each water diversion point is considered.

[0024] In an alternative embodiment, the reward formula for equipment operation benefit is:

[0025] ;

[0026] ;

[0027] ;

[0028] ;

[0029] where γ is the electricity price fluctuation coefficient, E 预期 is the initial power generation revenue expectation, E 实际 is the actual power generation revenue, is the adjusted power generation revenue expectation after the power generation revenue expectation is dynamically adjusted according to the real-time electricity price, V 满足 is the water storage volume actually meeting the subsequent water allocation, V 需求 is the water storage demand for subsequent water allocation, is the seasonal water storage coefficient, is the maintenance cost coefficient, is the equipment life impact coefficient, and c, d, and f are all coefficients.

[0030] The water energy scheduling method provided by the embodiments of this application, the reward formula for equipment operation benefit takes into account the energy market price fluctuations, ensuring the accuracy of the calculated equipment power generation reward. And the seasonal water storage coefficient is introduced, ensuring the accuracy of the calculated equipment water storage reward. Based on the equipment power generation reward and the equipment water storage reward, the accuracy of the calculated equipment operation benefit reward is ensured.

[0031] In an alternative embodiment, the target water conveyance pipeline includes multiple flow and pressure monitoring nodes, and the system stability reward formula is:

[0032] ;

[0033] where m is the number of nodes monitoring flow and pressure, F 实际，j is the actual flow rate of the jth flow monitoring point, F 正常，j is the normal flow rate of the jth flow monitoring point, F 实际，k is the actual pressure of the kth pressure monitoring point, F 正常，k is the normal pressure of the kth flow monitoring point, is the system recovery ability index, and θ is the collaborative stability coefficient.

[0034] In the water energy scheduling method provided by the embodiments of the present application, the system stability reward formula introduces the system recovery ability index and the collaborative stability coefficient, ensuring the accuracy of the calculated system stability reward.

[0035] In an alternative embodiment, based on the state space of each water diversion outlet, a first action space for controlling the valves of each water diversion outlet is generated, including:

[0036] Input the state space of each water diversion outlet into the water diversion outlet agent corresponding to each water diversion outlet;

[0037] Each water diversion outlet agent generates a plurality of initial valve control actions based on the state space of the water diversion outlet;

[0038] Reward or punish the initial valve control actions according to the satisfaction of water use requirements, flow rate, and pressure stability;

[0039] According to the reward or punishment results, obtain a plurality of standby valve control actions and generate the first action space.

[0040] In the water energy scheduling method provided by the embodiments of the present application, the state space of each water diversion outlet is input into the water diversion outlet agent corresponding to each water diversion outlet; each water diversion outlet agent generates a plurality of initial valve control actions based on the state space of the water diversion outlet. Using the learning and decision-making capabilities of the agent to generate initial valve control actions can achieve intelligent decision-making for the water conveyance system. The agent can analyze and reason based on complex state information and generate control actions that conform to the current system state, reducing the subjectivity and limitations of manual decision-making and improving the accuracy and scientificity of decision-making. Reward or punish the initial valve control actions according to the satisfaction of water use requirements, flow rate, and pressure stability. An effective action optimization mechanism is established, which can guide the agent to continuously adjust and improve the control actions, making the generated standby valve control actions more in line with the optimization goals of the system. By rewarding actions that meet water use requirements and maintain stable flow rate and pressure and punishing actions that do not meet the requirements, the agent can gradually learn the optimal control strategy and improve the overall performance of the water conveyance system. According to the reward or punishment results, obtain a plurality of standby valve control actions and generate the first action space. Enable the water conveyance system to adapt to complex and changeable operating environments. In addition, the reward and punishment mechanism screens out a series of actions that can effectively control the valves in different situations, and the first action space composed of these actions provides rich choices for the subsequent generation of scheduling strategies. A high-quality action space helps the system find a better scheduling scheme, improve the control accuracy and flexibility of the water conveyance system, and better cope with various complex working conditions and emergencies.

[0041] In an alternative embodiment, based on the action reward calculation mechanism, each action combination generated based on the first action space and the second action space is evaluated, including:

[0042] Quantize each standby valve control action included in each first action space to generate a standby valve control action quantum state;

[0043] Quantize each standby device control action included in each second action space to generate a standby device control action quantum state;

[0044] Select one standby valve control action quantum state from each first action space and one standby device control action quantum state from each second action space to generate multiple standby combined action quantum states;

[0045] Simulate the execution process of each standby combined action quantum state in the target water conveyance pipeline, and calculate the reward value of each standby combined action quantum state based on the action reward calculation mechanism.

[0046] The water energy scheduling method provided by the embodiments of the present application quantizes each spare valve control action included in each first action space to generate a spare valve control action quantum state. Each spare device control action included in each second action space is quantized to generate a spare device control action quantum state. Thus, various control actions in the water conveyance system can be represented in a more precise manner. Quantum states can utilize the properties of quantum mechanics, such as superposition and entanglement, to more meticulously describe different characteristics and possibilities of actions. Compared with traditional representation methods, they can more accurately capture the essence and interrelationships of actions, providing richer information for subsequent decision-making. Select one spare valve control action quantum state from each first action space and one spare device control action quantum state from each second action space to generate multiple spare combined action quantum states. This greatly increases the diversity of action combinations. The quantized action space has more possibilities and flexibility, capable of covering a wider range of control strategies. This enables the water conveyance system to have more options to find the optimal or near-optimal scheduling scheme when facing complex water conveyance system operating conditions, improving the adaptability and optimization ability of the water conveyance system. In addition, simulate the execution process of each spare combined action quantum state in the target water conveyance pipeline, and calculate the reward value of each spare combined action quantum state based on the action reward calculation mechanism. This provides an effective means of evaluation and optimization. By simulating the execution, the possible effects of each spare combined action quantum state can be understood in advance, and the calculation of the reward value can quantify these effects, helping the water conveyance system quickly screen out action combinations with higher benefits. This way of simulating first and then evaluating can avoid a large number of trials and errors in the actual system, reduce the decision-making risk, and at the same time improve the optimization efficiency of the scheduling strategy, contributing to quickly finding the best scheduling scheme that meets multiple optimization goals. In addition, in combination with the action reward calculation mechanism, factors such as the satisfaction of water use demand, the operation benefit of water storage and power generation equipment, and system stability are incorporated into the calculation of the reward value, enabling the decision-making process to comprehensively consider multiple important indicators. This decision-making method that integrates multiple factors can ensure that the generated scheduling strategy not only focuses on the optimization of a single goal but also achieves balance and coordination among multiple goals. For example, while meeting the water use demand, taking into account the operation benefit of the water storage and power generation equipment and the stability of the system, thereby improving the overall performance of the water conveyance system and achieving a more ideal operation effect.

[0047] In an alternative embodiment, according to the evaluation result, output the target scheduling strategy, including:

[0048] Calculate the mean value and standard deviation of the reward values according to the reward values of each spare combined action quantum state;

[0049] Dynamically adjust the temperature drop rate according to the standard deviation of the reward value;

[0050] Adjust the quantum states of each standby combined action according to the adjusted temperature drop rate to generate updated combined action quantum states;

[0051] Calculate the reward values corresponding to each updated combined action quantum state;

[0052] Compare the reward values corresponding to each updated combined action quantum state with the reward values of the corresponding standby combined action quantum states;

[0053] If there are more than a preset proportion of updated combined action quantum states whose corresponding reward values are greater than those of the corresponding standby combined action quantum states, update each updated combined action quantum state based on the current quantum gate operation strategy until the target combined action quantum state with the maximum reward value is found;

[0054] Determine the target valve control action and target device control action corresponding to the target combined action quantum state;

[0055] Output the target scheduling strategy based on the target valve control action and target device control action.

[0056] The hydropower scheduling method provided in the embodiment of the present application can quantitatively evaluate and analyze the reward value of the standby combination action quantum state by calculating the mean and standard deviation of the reward value. The mean can reflect the overall reward level, and the standard deviation can reflect the discrete degree of the reward value, help understand the stability and reliability of different action combinations, and provide comprehensive data support for subsequent decision-making. The temperature drop rate is dynamically adjusted according to the standard deviation of the reward value, so that the water delivery system can flexibly adjust the optimization process according to the performance of the standby combination action quantum state. If the standard deviation is large, it means that the reward values of different action combinations are quite different, and there may be a large optimization space. At this time, the temperature drop rate is appropriately accelerated, and a better action combination can be explored more quickly; on the contrary, if the standard deviation is small, it means that the performance of the action combination is relatively stable. At this time, slowing down the temperature drop rate helps to search for the local optimal solution more finely, avoid missing potential better strategies, and thus improve the optimization efficiency and effect. Then, the standby combination action quantum state is adjusted to generate an updated combination action quantum state, and the reward values before and after the update are continuously compared, and the combination with a larger reward value is selected for further update. This gradual optimization mechanism can guide the system to continuously evolve in the direction of the optimal solution. Through repeated iterations, the system can gradually eliminate poor action combinations, retain and improve better combinations, and finally find the target combination action quantum state with the largest reward value, thereby obtaining the best scheduling strategy that meets multi-objective optimization. Finally, the target valve control action and target device control action corresponding to the target combination action quantum state are determined, and the target scheduling strategy is output, realizing the precise conversion from the quantized action space to the actual executable scheduling strategy. This precise decision-making can accurately apply the optimization results of the system at the quantum level to the actual water pipeline system, and accurately control the valves and water storage and power generation equipment to achieve the comprehensive goals of meeting water demand, improving equipment operation efficiency and ensuring system stability, effectively improving the operation management level and overall performance of the water supply system.

[0057] In an optional implementation, updating each update combination action quantum state based on the current quantum gate operation strategy includes:

[0058] The current quantum gate operation strategy updates the quantum state of each update combination action based on the following formula:

[0059] ;

[0060] Among them, s is the current update combination action quantum state, a is the quantum gate operation strategy adopted for the update combination action quantum state, Q(s,a) is the value function of taking action a under state s, which reflects the performance of the strategy in past iterations; R(s,a) is the reward value corresponding to the current update combination action quantum state, α is the learning rate, γ is the discount factor, and max a′Q(s′, a′) represents the one with the highest value among all possible actions a′ in the new state s′ entered after performing action a, which reflects the prediction of the optimal future strategy.

[0061] For the water energy scheduling method provided by the embodiments of the present application, the current quantum gate operation strategy updates each updated combined action quantum state based on the following formula, ensuring the accuracy of updating each updated combined action quantum state.

[0062] In an alternative implementation, obtaining the future water usage demands corresponding to each water diversion point in the target water conveyance pipeline within a preset future time period includes:

[0063] Obtaining the historical water usage data of each water diversion point within a preset time period before the current time, the precipitation data, weather data, temperature data within the preset future time period, and the time feature data corresponding to the preset future time period;

[0064] Inputting the historical water usage data, precipitation data, weather data, temperature data, and time feature data into a preset water usage demand prediction model;

[0065] The preset water usage demand prediction model outputs the future water usage demands corresponding to each water diversion point within the preset future time period.

[0066] For the water energy scheduling method provided by the embodiments of the present application, obtaining the historical water usage data of each water diversion point within a preset time period before the current time, the precipitation data, weather data, temperature data within the preset future time period, and the time feature data corresponding to the preset future time period can comprehensively consider various factors affecting the water usage demands of the water diversion points. Inputting the historical water usage data, precipitation data, weather data, temperature data, and time feature data into a preset water usage demand prediction model; the preset water usage demand prediction model outputs the future water usage demands corresponding to each water diversion point within the preset future time period, ensuring the accuracy of the output future water usage demands corresponding to each water diversion point within the preset future time period. It provides a scientific basis for the reasonable allocation and management of water resources. Managers can formulate scheduling plans in advance according to the prediction results, reasonably arrange the allocation of water resources, and ensure the efficient utilization of water resources and the stable operation of the system while meeting the water usage demands of each water diversion point. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0068] Figure 1 It is a schematic flowchart of a water energy scheduling method according to an embodiment of the present invention;

[0069] Figure 2 It is a schematic diagram corresponding to the target water conveyance pipeline according to an embodiment of the present invention;

[0070] Figure 3 It is a schematic diagram of the valve corresponding to one of the water diversion ports according to an embodiment of the present invention;

[0071] Figure 4 It is a schematic flowchart of another water energy scheduling method according to an embodiment of the present invention. Detailed implementation manners

[0072] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0073] It should be noted that for the water energy scheduling method provided in the embodiments of the present application, the execution subject may be a water energy scheduling device, and this water energy scheduling device may be implemented as part or all of a computer device through software, hardware, or a combination of software and hardware. Among them, the computer device may be a server or a terminal. Among them, the server in the embodiments of the present application may be a single server or a server cluster composed of multiple servers. The terminal in the embodiments of the present application may be other intelligent hardware devices such as a smart phone, a personal computer, a tablet computer, a wearable device, and a smart robot. In the following method embodiments, the execution subject is taken as an electronic device for illustration.

[0074] According to an embodiment of the present invention, an embodiment of a water energy scheduling method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings may be executed in a computer system such as a set of computer executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than here.

[0075] In this embodiment, a water energy scheduling method is provided, which can be used for the above-mentioned electronic device, Figure 1 It is a flowchart of the water energy scheduling method according to an embodiment of the present invention, as Figure 1 shown, and this process includes the following steps:

[0076] Step S101: Obtain the future water demands corresponding to each water diversion port in the target water conveyance pipeline within a future preset time period, and obtain the geographical location information, current flow data, and current pressure data corresponding to each water diversion port.

[0077] Specifically, the electronic device can receive the future water demands corresponding to each water diversion port in the target water conveyance pipeline within a future preset time period, as well as the geographical location information, current flow data, and current pressure data corresponding to each water diversion port input by the user. The electronic device can also receive the future water demands corresponding to each water diversion port in the target water conveyance pipeline within a future preset time period, as well as the geographical location information, current flow data, and current pressure data corresponding to each water diversion port sent by other devices.

[0078] In addition, the electronic device can also obtain the historical water usage data of each water diversion port in the target water conveyance pipeline within a preset duration before the current time, the precipitation data, weather data, temperature data within the future preset time period, and the time feature data corresponding to the future preset time period. Then, based on the obtained data, predict the future water demands corresponding to each water diversion port within the future preset time period. In addition, the electronic device obtains the current flow data and current pressure data corresponding to each water diversion port based on the flow monitoring devices and pressure monitoring devices installed at each water diversion port.

[0079] The content regarding obtaining the future water demands corresponding to each water diversion port in the target water conveyance pipeline within a future preset time period will be introduced in detail below.

[0080] Exemplarily, as Figure 2 shown, it is a schematic diagram corresponding to the target water conveyance pipeline.

[0081] Step S102: Obtain the current working parameter information corresponding to each water storage and power generation device in the target water conveyance pipeline.

[0082] Among them, the water storage and power generation device is used for water storage or power generation, and the current working parameters include current water storage working parameters and / or current power generation working parameters; at least one water diversion port is equipped with a water storage and power generation device.

[0083] Specifically, a variety of types of sensors are deployed on the water storage and power generation device to form a complete sensor network. These sensors can collect the current working parameters of the water storage and power generation device, then convert physical quantities (such as water level, flow rate, temperature, etc.) into electrical signals or digital signals, and transmit the data to the electronic device through wired or wireless communication methods.

[0084] Among them, the current water storage working parameters may include the water level information of the water storage pool corresponding to the water storage power generation equipment, the inlet flow rate and outlet flow rate of the water storage power generation equipment, and the detected water quality information of the water storage power generation equipment, etc. Among them, the water quality information includes acidity and alkalinity (pH value), dissolved oxygen content, turbidity, etc., which are related to the corrosion and wear conditions of the internal components of the equipment and are obtained by using water quality monitoring instruments. The current power generation working parameters may include information such as the current power generation power, the water turbine speed, the equipment temperature, and the vibration parameters. Among them, the equipment temperature includes the generator winding temperature, the bearing temperature, the water turbine volute temperature, etc. Excessive temperature may cause equipment failures, and multi-point monitoring is carried out using temperature sensors.

[0085] For example, the water level sensor uses a pressure type or ultrasonic type sensor to convert the water level height into an electrical signal and transmit it to the data acquisition module through a cable; the wireless sensor uses wireless communication technologies such as Bluetooth, Wi-Fi, or LoRa to send the data to a nearby gateway, and then the gateway aggregates and uploads it to the electronic device.

[0086] Step S103, according to the future water usage demands, geographical location information, current flow data, current pressure data corresponding to each water diversion port, and the current working parameter information corresponding to the water storage power generation equipment, output a target scheduling strategy.

[0087] Among them, the scheduling strategy includes a first scheduling strategy for controlling the valves corresponding to each water diversion port and / or a second scheduling strategy for controlling the water storage power generation equipment corresponding to each water diversion port. The first scheduling strategy specifies the opening adjustment plan of each water diversion port valve in different time periods and the emergency operation measures in special situations (such as abnormal flow, pressure fluctuation, etc.). For example, during the peak water usage period, the opening of some water diversion port valves is increased to increase the water supply; when the pipeline pressure is too high, the valve opening is appropriately reduced to relieve the pressure. The second scheduling strategy covers the operation mode switching time of the water storage power generation equipment, the power generation power adjustment plan, the time nodes and flow control of water storage and water release, etc. For example, during the low electricity consumption period, some water storage power generation equipment is switched to the water storage mode to increase the water storage volume; during the peak electricity consumption period, the equipment power generation power is adjusted to meet the electricity demand.

[0088] Specifically, the electronic device can comprehensively integrate the future water usage demands, geographical location information, current flow data, current pressure data corresponding to each water diversion port, and the current working parameter information of the water storage power generation equipment. Then, using data mining techniques, valuable features and patterns are extracted from the integrated data.

[0089] Then, based on the above data and evidence, a comprehensive mathematical model is established to describe the operating state and scheduling strategy of the water conveyance system. The model includes multiple sub-equations such as the water demand equation, the water flow motion equation, and the equipment operation equation. The water demand equation is established according to the water use types and predicted data at different water diversion outlets, reflecting the variation of water demand over time. The water flow motion equation combines the physical characteristics of the pipeline and the flow rate and pressure data to describe the flow law of water in the pipeline. The equipment operation equation is established according to the working parameters and operation modes of the water storage and power generation equipment, reflecting the operating state and energy conversion relationship of the equipment. By solving this mathematical model, the optimal scheduling strategy solution space under different constraint conditions is obtained. An optimization algorithm is used to search for the optimal target scheduling strategy in the solution space.

[0090] Exemplarily, as Figure 3 shown, it is a schematic diagram of the valve corresponding to one of the water diversion outlets.

[0091] This step will be introduced in detail below.

[0092] The water energy scheduling method provided by the embodiment of the present application obtains the future water usage demands corresponding to each water diversion outlet in the target water conveyance pipeline within a preset future time period. It can understand the water usage changes in different regions in advance, thereby more accurately scheduling water resources, avoiding water supply shortages or surpluses, improving the utilization efficiency of water resources, and ensuring the satisfaction of users' water usage demands. Then, obtain the geographical location information, current flow data, and current pressure data corresponding to each water diversion outlet. Obtaining the current working parameter information of each water storage and power generation device in the target water conveyance pipeline helps to flexibly adjust the working mode of the water storage and power generation devices according to the actual water usage demands and the pipeline operation status. For example, during the low water usage period, water storage can be increased to ensure sufficient water supply during the high water usage period; at the same time, power generation can be reasonably arranged according to the water flow situation to improve the utilization efficiency of water energy, realize the comprehensive utilization of water resources, and increase energy output. According to the future water usage demands, geographical location information, current flow data, current pressure data corresponding to each water diversion outlet, and the current working parameter information of the water storage and power generation devices, output the target scheduling strategy. Among them, the first scheduling strategy can precisely control the valve opening of each water diversion outlet to adjust the water flow size and meet the water usage demands of different regions; the second scheduling strategy can optimize the operation of the water storage and power generation devices to achieve the best balance between water storage and power generation. This intelligent scheduling method reduces manual intervention, improves the accuracy and timeliness of scheduling, reduces operating costs, enhances the management level and operation efficiency of the entire water conveyance system, and safely and stably realizes the comprehensive utilization of water energy without affecting the water resource allocation function of the water diversion project itself. In addition, making scheduling decisions based on comprehensive information can prevent and respond to various possible situations in advance, such as local pipeline failures and sudden changes in water usage demands. By reasonably adjusting the operation of valves and water storage and power generation devices, the impact of these sudden situations on the entire water conveyance system can be alleviated to a certain extent, enhancing the reliability and stability of the system, reducing the occurrence of water cut-off accidents, and ensuring water supply safety.

[0093] In this embodiment, a water energy scheduling method is provided, which can be used in the above-mentioned electronic device. Figure 4 It is a flowchart of the water energy scheduling method according to the embodiment of the present invention, as Figure 4 shown. The process includes the following steps:

[0094] Step S201, obtain the future water usage demands corresponding to each water diversion outlet in the target water conveyance pipeline within a preset future time period, and obtain the geographical location information, current flow data, and current pressure data corresponding to each water diversion outlet.

[0095] Specifically, the "obtain the future water usage demands corresponding to each water diversion outlet in the target water conveyance pipeline within a preset future time period" in the above step S201 may include the following steps;

[0096] Step S2011: Obtain the historical water consumption data of each water diversion outlet within a preset time period before the current time, the precipitation data, weather data, temperature data within a future preset time period, and the time feature data corresponding to the future preset time period.

[0097] Specifically, the electronic device can receive the historical water consumption data of each water diversion outlet within a preset time period before the current time, the precipitation data, weather data, temperature data within a future preset time period, and the time feature data corresponding to the future preset time period input by the user. It can also receive the historical water consumption data of each water diversion outlet within a preset time period before the current time, the precipitation data, weather data, temperature data within a future preset time period, and the time feature data corresponding to the future preset time period sent by other devices.

[0098] Among them, the time feature data can include time point information, time period information, special time identifiers, etc. Among them, the time point information such as specific year, month, date, hour, etc., these information can help the model identify the water consumption demand rules at different time scales. For example, the water consumption demands in summer and winter may have obvious differences, and the seasonal changes can be captured through the year and month information. The time period information includes the day of the week in a week, the time period in a day, etc. For example, the water consumption patterns on weekdays and weekends may be different, and the water consumption demands during the day and at night also vary. By encoding the day of the week and the time period, the model can learn these periodic rules. The special time identifiers are the identifiers of special days such as holidays and anniversaries. These special periods often lead to abnormal changes in water consumption demand, and the model can make more accurate predictions of the corresponding water consumption demands by identifying these special time identifiers.

[0099] Step S2012: Input the historical water consumption data, precipitation data, weather data, temperature data, and time feature data into a preset water consumption demand prediction model.

[0100] Specifically, the electronic device inputs the historical water consumption data, precipitation data, weather data, temperature data, and time feature data into a preset water consumption demand prediction model.

[0101] Step S2013: The preset water consumption demand prediction model outputs the future water consumption demands corresponding to each water diversion outlet within the future preset time period.

[0102] Specifically, the preset water consumption demand prediction model can perform feature extraction on the historical water consumption data, precipitation data, weather data, temperature data, and time feature data, and output the future water consumption demands corresponding to each water diversion outlet within the future preset time period.

[0103] Therefore, the above step S201 also includes:

[0104] Step S2014, obtain the geographical location information, current flow rate data, and current pressure data corresponding to each water diversion outlet.

[0105] Step S202, obtain the current working parameter information corresponding to each water storage and power generation device in the target water conveyance pipeline.

[0106] Among them, the water storage and power generation device is used for water storage or power generation, and the current working parameters include current water storage working parameters and / or current power generation working parameters; at least one water diversion outlet is equipped with a water storage and power generation device.

[0107] Step S203, based on the future water usage demands, geographical location information, current flow rate data, current pressure data corresponding to each water diversion outlet, and the current working parameter information corresponding to the water storage and power generation devices, output a target scheduling strategy.

[0108] Among them, the scheduling strategy includes a first scheduling strategy for controlling the valves corresponding to each water diversion outlet and / or a second scheduling strategy for controlling the water storage and power generation devices corresponding to each water diversion outlet.

[0109] Specifically, the above-mentioned Step S203 may include the following steps:

[0110] Step S2031, construct a water diversion outlet state space corresponding to each water diversion outlet based on the geographical location information, future water usage demands, current flow rate data, and current pressure data corresponding to each water diversion outlet.

[0111] Among them, the geographical location information corresponding to each water diversion outlet includes information such as the longitude and latitude of the water diversion outlet, altitude, slope, and terrain undulation.

[0112] Specifically, the electronic device can use geographic information system technology to perform three-dimensional modeling on the geographical location information of each water diversion outlet to generate a three-dimensional model of the water diversion outlet. It not only accurately presents the coordinate position of the water diversion outlet on the two-dimensional plane but also incorporates information such as the altitude, slope, and terrain undulation of the surrounding terrain. Through this three-dimensional modeling, the spatial relationship between the water diversion outlet and the surrounding geographical environment can be intuitively analyzed. For example, it can be determined whether the water diversion outlet is in a low-lying area prone to water accumulation or whether it is near a high mountain and may be affected by snowmelt runoff. The future water usage demand data is marked with virtual objects of different colors and shapes near the corresponding water diversion outlet positions in the three-dimensional model according to different water usage types (residential, industrial, agricultural, etc.), clearly showing the spatial distribution of different water usage demands. At the same time, combining the current flow rate data and pressure data, using dynamic visualization technology, the thickness of the water flow lines represents the flow rate size, and the depth of the color represents the pressure level in the three-dimensional model, realizing the intuitive fusion and display of multi-source data in the same three-dimensional space.

[0113] Then, the electronic device uses the convolutional neural network in deep learning to extract features from the current flow data and current pressure data. The current flow data and current pressure data are sliced and processed in time series to form a data matrix that is input into the convolutional neural network model. The convolution layer of the convolutional neural network model can automatically learn the local features in the current flow data and current pressure data, such as the mutation point of flow, the periodic fluctuation pattern of pressure, etc. Through multi-layer convolution and pooling operations, deep and abstract feature representations are extracted. For example, the convolutional neural network model can identify the regular change characteristics of flow due to increased agricultural irrigation water in a specific season or time period, as well as the abnormal pressure characteristics caused by aging or local blockage of water pipes. These extracted features are correlated with geographic location information and future water demand data to explore the potential connections between different types of data and provide rich feature information for building a more comprehensive water diversion state space.

[0114] Considering that future water demand, current flow data and current pressure data all have the characteristics of dynamic time changes, the electronic device builds a time series dynamic model for each water diversion. Using the time series model, the time series of future water demand, current flow data and current pressure data are modeled to generate time series features. The input of the time series dynamic model includes water demand, flow and pressure data in the past period of time. Through the loop structure and memory unit inside the network, the changing trend and dependency of the data over time are learned. For example, the LSTM model can capture the dynamic process of the gradual increase in residents' domestic water demand during the high temperature period in summer, and the corresponding changes in flow and pressure as the water demand increases. The model predicts the water demand, flow and pressure values at different time points in the future, and incorporates the prediction results into the state space of the water diversion, so that the state space can reflect the dynamic evolution of the water diversion state over time.

[0115] The electronic device constructs the water diversion outlet state space based on the water diversion outlet three-dimensional model, local features and time series features corresponding to each water diversion outlet.

[0116] Optionally, the electronic device can also associate the ecological environment data around the water diversion outlet with the future water demand, geographical location information, current flow data, and current pressure data corresponding to the water diversion outlet. For example, the electronic device collects ecological environment data such as the vegetation coverage around each water diversion outlet and river ecological health indicators (such as biodiversity, water quality eutrophication level, etc.). Among them, analyze the relationship between vegetation coverage and water demand. Areas with dense vegetation may have a certain demand for water due to transpiration, and at the same time, can play a role in conserving water sources and regulating runoff, affecting the flow and pressure of the water diversion outlet. Then, the electronic device incorporates the ecological environment data as a new dimension into the water diversion outlet state space. Through data visualization technology, the level of ecological environment indicators is represented by different colors or transparencies in the 3D GIS model, intuitively showing the mutual influence between the ecological environment and the water diversion outlet state. In this way, when considering the water diversion outlet scheduling strategy, the relationship between water resource utilization and ecological environment protection can be comprehensively weighed.

[0117] Step S2032, construct a device state space based on the current working parameter information corresponding to each water storage and power generation device.

[0118] Specifically, the electronic device can also obtain the current working parameter information corresponding to each water storage and power generation device. In addition, the electronic device can also, based on high-precision sensors installed on key components of the water storage and power generation device, such as turbine blades, generator windings, bearings, etc., real-time monitor microscopic working parameter data such as the stress, strain, vibration spectrum, and temperature field distribution of the components. For example, by pasting strain gauges on the surface of the turbine blade, obtain the stress distribution of the blade under different water flow impacts; use an infrared thermal imager to monitor the temperature field of the generator winding and timely detect potential hidden dangers of local overheating. Combine these microscopic working parameter data with traditional working parameters to form a multi-dimensional device state vector. In the device state space, the newly added microscopic structure health dimension can early warn of potential device failures. For example, by analyzing abnormal changes in the vibration spectrum, problems such as bearing wear or blade cracks can be predicted before the device shows obvious performance degradation, providing more comprehensive information for device maintenance and scheduling decisions.

[0119] In addition, the electronic device can also obtain the environmental monitoring data corresponding to the water storage power generation device. Among them, the environmental monitoring data includes humidity, air pressure, wind speed, water pollution indicators, etc. Changes in humidity and air pressure will affect the insulation performance and heat dissipation effect of the generator, and thus affect the power generation efficiency; wind speed may affect the stability of outdoor equipment; water pollution indicators (such as pH value, suspended solid content, etc.) will exacerbate the corrosion of internal components of the equipment. By establishing an association model between environmental factors and equipment operating parameters, for example, using multiple linear regression to analyze the influence law of humidity and air pressure on the temperature of the generator winding, the environmental factors are incorporated into the equipment state space as a new state dimension. When conducting equipment state assessment and scheduling decisions, comprehensively consider the influence of environmental factors on equipment performance, and take preventive or adjustment measures in advance to ensure the safe and stable operation of the equipment under different environmental conditions.

[0120] Then, the electronic device can map the high-dimensional device state vector and environmental monitoring data to a low-dimensional feature space based on a preset encoder, and extract the key features of the data. The preset decoder then reconstructs the low-dimensional feature space into the original data, thereby constructing the equipment state space.

[0121] Among them, during the training process, the electronic device can minimize the reconstruction error to enable the preset encoder to learn the most representative features in the data. For example, for multi-dimensional data including power generation power, water level, blade stress, winding temperature, environmental humidity, etc., the preset encoder can automatically extract the feature combination that has the greatest impact on the equipment state and remove redundant information. Using these low-dimensional features processed by the preset encoder as a new representation of the equipment state space not only reduces the data dimension, decreases the computational amount, but also improves the representation accuracy of the state space for the true state of the equipment, which is more conducive to subsequent state analysis and fault diagnosis.

[0122] Step S2033, based on the state space of each water diversion port, generate a first action space for controlling the valves of each water diversion port.

[0123] Among them, the first action space includes multiple standby valve control actions.

[0124] Specifically, the above step S2033 may include the following steps:

[0125] Step a1, input the state space of each water diversion port into the water diversion port intelligent body corresponding to each water diversion port.

[0126] Specifically, the electronic device inputs the state space of each water diversion port into the water diversion port intelligent body corresponding to each water diversion port.

[0127] Step a2, each water diversion port intelligent body generates multiple initial valve control actions based on the state space of the water diversion port.

[0128] Specifically, each water diversion port agent can generate multiple initial valve control actions based on the water diversion port state space.

[0129] Step a3, reward or punish the initial valve control actions according to the satisfaction of water usage demand, flow rate, and pressure stability.

[0130] Specifically, each water diversion port agent can dynamically adjust the weights of the reward and punishment factors by real-time monitoring the operating state of the target water conveyance pipeline and using the fuzzy logic algorithm, and multiply the satisfaction of water usage demand, flow rate stability, and pressure stability corresponding to each initial valve control action by the corresponding reward and punishment factors. Among them, the fuzzy logic algorithm infers the weight of each reward and punishment factor through fuzzy rule reasoning based on multiple input variables (such as time, season, pipeline maintenance status, etc.). For example, during the peak period of residential water use, the weight of the satisfaction of water usage demand is relatively high; during the maintenance period of the water conveyance pipeline, the weights of flow rate and pressure stability are more important. For example, when the time is between 7 - 9 pm in summer (the peak period of residential water use) and the system monitors that the pipeline is operating normally, the fuzzy logic algorithm increases the weight of the satisfaction of water usage demand to 0.6, sets the weight of flow rate stability to 0.2, and sets the weight of pressure stability to 0.2, making the reward and punishment more in line with the actual needs.

[0131] Step a4, obtain multiple alternative valve control actions according to the reward or punishment results, and generate the first action space.

[0132] Specifically, each water diversion port agent can encode the alternative valve control actions. For example, using the binary encoding method, each encoding bit represents a control parameter of the valve (such as the increase or decrease of the opening degree, the time point of opening or closing, etc.). According to the encoding rules, the electronic device generates an initial action population. Each individual in the population corresponds to an alternative valve control action. For example, for a valve control action with three control parameters (the adjustment range of the valve opening degree, the adjustment time interval, the adjustment priority), it can be encoded as a binary string with a length of 10 bits. The first 3 bits represent the size level of the adjustment range of the opening degree, the middle 3 bits represent the length level of the adjustment time interval, and the last 4 bits represent the order of the adjustment priority. By randomly generating a certain number of such binary strings, an initial action population is formed.

[0133] First, each water diversion port agent can perform a selection operation. According to the fitness value (i.e., the reward value minus the penalty value) obtained by each standby valve control action individual under the reward and penalty mechanism, methods such as roulette wheel selection are used to select the standby valve control action individuals with high fitness to enter the next generation. The standby valve control action individuals with high fitness have a greater probability of being selected, thus retaining excellent valve control action patterns. Then, each water diversion port agent performs a crossover operation, randomly selects two standby valve control action individuals, and exchanges some of their coding bits according to a certain crossover probability (such as 0.8) to generate new updated valve control action individuals, promoting the integration of different excellent action patterns. Finally, each water diversion port agent performs a mutation operation, and some coding bits of the updated valve control action individuals are inverted with a low mutation probability (such as 0.01) to introduce new action patterns and prevent the algorithm from falling into a local optimum. Through continuous iterative genetic operations, the action population gradually evolves to generate a better first action space. For example, after multiple generations of genetic operations, a series of standby valve control actions that can efficiently meet water use requirements, maintain stable flow and pressure, and are environmentally friendly appear in the action space, providing richer and better choices for subsequent scheduling decisions.

[0134] Step S2034, based on each device state space, generate a second action space for controlling each water storage and power generation device.

[0135] Among them, the second action space includes multiple standby device control actions.

[0136] Specifically, the electronic device can input each device state space into the device agent corresponding to each water storage and power generation device. Each device agent generates multiple initial device control actions based on the device state space. The initial valve control actions are rewarded or punished according to the increase in power generation revenue, the improvement of device stability, and the satisfaction of water use requirements. According to the reward or punishment results, multiple standby valve control actions are obtained to generate the first action space.

[0137] For the specific implementation process, reference can be made to the above specific introduction of step S2033, and details will not be elaborated here.

[0138] Step S2035, construct an action reward calculation mechanism.

[0139] Among them, the action reward calculation mechanism includes at least one of a water use requirement satisfaction reward, a device operation benefit reward, and a system stability reward.

[0140] In an alternative implementation manner, the water use requirement satisfaction reward formula is:

[0141] ;

[0142] where n is the number of water diversion outlets, Q 实际，i is the actual water supply of the i-th water diversion outlet; Q 需求，i is the future water demand of the i-th water diversion outlet; α 类型，i is the urgency coefficient of the water use type corresponding to the i-th water diversion outlet, and both a and b are coefficients.

[0143] Exemplarily, for different types of water use, α 类型，i has different urgency coefficients. For example, in case of emergencies (such as fires, emergency water use after a water supply notice, etc.) for domestic water use, the urgency coefficient α 类型，i can be set to 1.5; in the case of water shortage in key production links of industrial water use that may lead to major production accidents, the urgency coefficient α 类型，i is set to 1.2; in the critical growth period of crops for agricultural irrigation water use, the urgency coefficient α 类型，i is set to 1.1.

[0144] In an alternative embodiment, the equipment operation benefit reward formula is:

[0145] ;

[0146] ;

[0147] ;

[0148] ;

[0149] where γ is the electricity price fluctuation coefficient, E 预期 is the initial expected power generation income, E 实际 is the actual power generation income, is the adjusted expected power generation income after the expected power generation income is dynamically adjusted according to the real-time electricity price, V 满足 is the actual water storage volume that meets the subsequent water allocation, V 需求 is the water storage demand for the subsequent water allocation, is the seasonal water storage coefficient, is the maintenance cost coefficient, is the equipment life impact coefficient, and c, d, and f are all coefficients.

[0150] Exemplarily, when the electricity price rises, γ is greater than 1; when the electricity price falls, γ is less than 1. For the water storage equipment part, considering the different importance of water storage in different seasons, the seasonal water storage coefficient is introduced. In the season before the peak water use period, set to 1.3; in the off-peak water use season, set to 0.8. If the equipment operates in a highly efficient and stable state with low maintenance costs, Set it to 0.8; if the device frequently malfunctions and the maintenance cost is high, set it to 1.2. The equipment life impact coefficient is , obtained by comparing the equipment operation parameters with the equipment design life curve. If the equipment operation state is conducive to extending the life, set it to 1.1; if it accelerates the equipment aging, set it to 0.9.

[0151] In an alternative embodiment, the target water conveyance pipeline includes multiple flow and pressure monitoring nodes, and the system stability reward formula is:

[0152] ;

[0153] where m is the number of nodes monitoring flow and pressure, F 实际，j is the actual flow rate of the j-th flow monitoring point, F 正常，j is the normal flow rate of the j-th flow monitoring point, F 实际，k is the actual pressure of the k-th pressure monitoring point, F 正常，k is the normal pressure of the k-th flow monitoring point, is the system recovery ability index, and θ is the collaborative stability coefficient.

[0154] Exemplarily, when the system is subjected to external disturbances (such as a sudden surge in water demand, equipment failure, etc.), it can quickly recover to the normal stable state, set it to 1.2; if the recovery is slow, set it to 0.8. If the water conveyance system has a collaborative working relationship with the surrounding sewage treatment system, other water conveyance networks, etc., the collaborative stability coefficient θ is considered. When the collaboration with the surrounding systems is good and there are no abnormal flow rates and pressures caused by collaboration problems, θ is set to 1.1; if there are collaboration conflicts, θ is set to 0.9.

[0155] In an alternative embodiment of the present application, the electronic device can construct a hierarchical structure model, with the water demand satisfaction reward, equipment operation benefit reward, and system stability reward as the criterion layer, and the specific influencing factors under different operation scenarios as the index layer. Then, by means of expert scoring or data analysis, etc., the relative importance between the factors is determined, and a judgment matrix is constructed. The eigenvector method and other methods are used to calculate the eigenvector of the judgment matrix, so as to obtain the weight values of each reward under different operation scenarios. For example, in the peak water use scenario, invite water conservancy experts, electrical engineers, and system operation and maintenance personnel to score the relative importance of water demand satisfaction, equipment operation benefit, and system stability, construct a judgment matrix and calculate the corresponding weights. Then, based on the weight information corresponding to the water demand satisfaction reward, equipment operation benefit reward, and system stability reward respectively, an action reward calculation mechanism is generated.

[0156] Step S2036, based on the action reward calculation mechanism, evaluate each action combination generated based on the first action space and the second action space.

[0157] Specifically, the above step S2036 may include the following steps:

[0158] Step b1, quantize each standby valve control action included in each first action space to generate a standby valve control action quantum state.

[0159] Specifically, the electronic device can perform feature extraction on each standby valve control action in the first action space. In addition to extracting basic features such as valve opening degree, opening and closing times, it also determines the dynamic influence features on flow rate, pressure, water demand satisfaction degree, etc. after the execution of each standby valve control action. For example, analyze the amplitude of flow rate fluctuations, the trend of pressure changes, and the real-time satisfaction ratio of water demand caused by the action in different time periods. In addition, the electronic device can also introduce environment-related features, such as the influence of temperature, humidity, light, etc. on the valve control effect. Finally, fuse the extracted basic features, dynamic influence features, and environment-related features to generate a high-dimensional action feature vector.

[0160] Then, the electronic device performs qubit encoding on the generated action feature vector. Specifically, the electronic device can adopt a hierarchical encoding strategy, perform hierarchical processing on the action feature vector according to the importance and relevance of the features to obtain key features and secondary features. For key features, such as the precise control of valve opening degree, use more qubits for encoding to improve the encoding accuracy; for secondary features, such as the weak influence of environmental light on valve control, use fewer qubits for encoding. In this way, both the accurate representation of important information and the reasonable utilization of qubit resources are ensured. For example, divide the valve opening degree range into multiple intervals, and each interval corresponds to a specific qubit combination to achieve precise quantum encoding of the valve opening degree and generate a standby valve control action quantum state.

[0161] Step b2, quantize each standby device control action included in each second action space to generate a standby device control action quantum state.

[0162] Specifically, the electronic device extracts the control actions of each standby device and extracts basic features such as the power generation power adjustment amount, the set value of the water storage level, and the start / stop time of the device. In addition, the electronic device also introduces device operation state features (such as the vibration frequency mode and temperature change trend of the device), environmental features (such as local meteorological conditions and power grid load fluctuation conditions), and economic features (such as power generation cost and market electricity price). Through the multi-sensor network and data analysis technology, these feature information of different modalities are collected and integrated to construct a high-dimensional and rich action feature vector.

[0163] The electronic device can use a hybrid quantum-classical coding scheme to identify the action feature vector and determine the continuous features and discrete features in the action feature vector. For continuous features, such as the power generation power adjustment amount, the electronic device uses quantum amplitude coding technology to map the feature value to the amplitude of the quantum bit, so as to make full use of the superposition property of the quantum state to represent continuous values. For discrete features, such as the operation mode of the device (normal operation, maintenance mode, emergency mode), the electronic device uses the ground state combination of quantum bits for coding. At the same time, the electronic device introduces quantum entanglement states to code feature pairs with strong correlations, enhances the correlation representation between features, improves the efficiency and accuracy of coding, and thus generates the quantum state of the standby device control action.

[0164] Step b3: Select a quantum state of the standby valve control action from each first action space and select a quantum state of the standby device control action from each second action space to generate multiple quantum states of the standby combined actions.

[0165] Specifically, the electronic device can select a quantum state of the standby valve control action from each first action space and select a quantum state of the standby device control action from each second action space to generate multiple quantum states of the standby combined actions.

[0166] Step b4: Simulate the execution process of each quantum state of the standby combined action in the target water conveyance pipeline, and calculate the reward value of each quantum state of the standby combined action based on the action reward calculation mechanism.

[0167] Specifically, to simulate the execution of the standby combined action quantum state in the target water conveyance pipeline, the electronic device can first construct a quantum-classical hybrid model. Convert the control actions represented by the standby combined action quantum state (such as valve opening adjustment, equipment operation mode switching, etc.) into classical physical quantities so as to integrate with the classical physical model of the target water conveyance pipeline. For example, convert the valve opening quantum encoding in the standby valve control action quantum state into the actual valve opening value, and convert the power generation quantum representation in the standby equipment control action quantum state into the corresponding power generation value. Then, combined with the hydraulic model of the target water conveyance pipeline, including the water flow continuity equation, energy equation, and pipeline resistance characteristics, establish a comprehensive model that can reflect the impact of the quantum state control action on the water conveyance system.

[0168] Then, calculate the water demand satisfaction reward based on the above calculation formula for the water demand satisfaction reward. Calculate the equipment operation benefit reward based on the equipment operation benefit reward formula, calculate the system stability reward based on the system stability reward formula, and then calculate the reward value of each standby combined action quantum state according to the water demand satisfaction reward, equipment operation benefit reward, and system stability reward.

[0169] Step S2037, output the target scheduling strategy according to the evaluation result.

[0170] Specifically, the above step S2037 may include the following steps:

[0171] Step c1, calculate the mean value and standard deviation of the reward values according to the reward values of each standby combined action quantum state.

[0172] Specifically, the electronic device can calculate the mean value and standard deviation of the reward values according to the reward values of each standby combined action quantum state.

[0173] Step c2, dynamically adjust the temperature decrease rate according to the standard deviation of the reward values.

[0174] Specifically, the electronic device can dynamically adjust the temperature decrease rate according to the corresponding relationship between the standard deviation of the reward values and the temperature decrease rate, according to the standard deviation of the reward values.

[0175] Among them, the corresponding relationship between the standard deviation of the reward values and the temperature decrease rate can be as follows:

[0176] ;

[0177] Among them, T new is the adjusted temperature, T old is the current temperature, and k is a preset adjustment coefficient.

[0178] σ is the standard deviation of the reward value. When the standard deviation of the reward value σ is large, the value of 1 - k×σ is small, and the temperature drops rapidly; when the standard deviation of the reward value σ is small, the value of 1 - k×σ approaches 1, and the temperature drops slowly.

[0179] Step c3: Adjust each standby combined action quantum state according to the adjusted temperature drop rate to generate an updated combined action quantum state.

[0180] Specifically, the electronic device can establish a mapping relationship between the temperature drop rate and the quantum state adjustment intensity. When the temperature drops rapidly (corresponding to a large standard deviation of the reward value and high diversity of the action space), a strong adjustment strategy is adopted, that is, a large change is made to the standby combined action quantum state. This may include flipping multiple qubits simultaneously or applying multiple complex quantum gate operations to quickly explore different action combinations and generate an updated combined action quantum state. For example, when simulating the water conveyance pipeline scheduling, the qubit states corresponding to multiple valve control actions are adjusted simultaneously, and different valve opening combinations are tried to find new strategies that can significantly improve the satisfaction of water demand or power generation efficiency. When the temperature drops slowly (the standard deviation of the reward value is small and close to the local optimal solution), a weak adjustment strategy is adopted, and only a few key qubits are finely adjusted, or some quantum gates with less impact on the quantum state are applied to find a better solution within the current local area, thereby generating an updated combined action quantum state. For example, only a small phase adjustment is made to the key qubits that affect the system stability to optimize the system stability while not destroying the existing good balance between water demand satisfaction and power generation efficiency.

[0181] Step c4: Calculate the reward value corresponding to each updated combined action quantum state.

[0182] Specifically, the electronic device can calculate the reward value corresponding to each updated combined action quantum state based on the above action reward calculation mechanism.

[0183] Step c5: Compare the reward value corresponding to each updated combined action quantum state with the reward value of the corresponding standby combined action quantum state.

[0184] Specifically, the electronic device can establish corresponding reward value records for each standby combined action quantum state and the updated combined action quantum state generated after adjustment. Then, the electronic device compares the reward value corresponding to each updated combined action quantum state with the reward value of the corresponding standby combined action quantum state.

[0185] Step c6, if there are more than a preset proportion of update combined action quantum states whose corresponding reward values are greater than those of the corresponding standby combined action quantum states, update each update combined action quantum state based on the current quantum gate operation strategy until the target combined action quantum state with the maximum reward value is found.

[0186] Specifically, the electronic device statistically analyzes the comparison results. If there are more than a preset proportion of update combined action quantum states whose corresponding reward values are greater than those of the corresponding standby combined action quantum states, the electronic device updates each update combined action quantum state based on the current quantum gate operation strategy until the target combined action quantum state with the maximum reward value is found.

[0187] Among them, the preset proportion can be 50%, or 60%, or other values. The embodiments of the present application do not specifically limit the preset proportion.

[0188] In an alternative embodiment of the present application, the current quantum gate operation strategy updates each update combined action quantum state based on the following formula:

[0189] ;

[0190] Among them, s is the current update combined action quantum state, a is the quantum gate operation strategy adopted for the update combined action quantum state, and Q(s,a) is the value function of taking action a in state s, which reflects the performance of this strategy in past iterations; R(s,a) is the reward value corresponding to the current update combined action quantum state, α is the learning rate, γ is the discount factor, and max a′ Q(s′,a′) represents the one with the highest value among all actions a′ that can be taken in the new state s′ entered after executing action a, which reflects the estimation of the future optimal strategy.

[0191] Repeat the above quantum gate operation strategy optimization steps for quantum state adjustment under adaptive temperature regulation, continuously update the states of qubits and the quantum gate operation strategy, and continuously explore different action combination quantum states. In each iteration, the temperature is dynamically adjusted according to the current reward value distribution, and at the same time, the quantum gate operation strategy is optimized based on the reinforcement learning feedback.

[0192] Set termination conditions, such as reaching a preset maximum number of iterations N, or in M consecutive iterations, the increase in the reward value is less than a minimum value δ (that is, it is considered that a relatively optimal solution has been converged). When the termination condition is met, stop the iteration and output the target combined action quantum state with the highest current reward value. The action combination corresponding to this target combined action quantum state is the strategy that is most likely to maximize the reward value of the water conveyance system under the current conditions.

[0193] Step c7, determine the target valve control action and target device control action corresponding to the target combined action quantum state.

[0194] Specifically, the electronic device reads the qubit states related to valve control and device control in the target combined action quantum state. In the entire quantum state, clearly divide which qubits correspond to valve control and which correspond to device control. For example, assume that the first 5 qubits correspond to valve control actions and the last 3 qubits correspond to device control actions. Through technical means such as quantum measurement, obtain the actual states of these qubits. For example, the measured qubit state of the valve control part is |011〉, and the qubit state of the device control part is |101〉.

[0195] Then, the electronic device decodes the read valve control qubit state and device control qubit state into actual target valve control actions and target device control actions according to the pre-set mapping rules.

[0196] Step c8, output the target scheduling strategy based on the target valve control action and target device control action.

[0197] Finally, the electronic device outputs the target scheduling strategy based on the target valve control action and target device control action.

[0198] The water energy scheduling method provided by the embodiments of the present application obtains the historical water usage data of each water diversion outlet within a preset duration before the current time, the precipitation data, weather data, temperature data within a future preset time period, and the time feature data corresponding to the future preset time period, so as to comprehensively consider various factors affecting the water usage demand of the water diversion outlet. Input the historical water usage data, precipitation data, weather data, temperature data, and time feature data into a preset water usage demand prediction model; the preset water usage demand prediction model outputs the future water usage demands corresponding to each water diversion outlet within the future preset time period, ensuring the accuracy of the future water usage demands corresponding to each water diversion outlet output within the future preset time period. It provides a scientific basis for the reasonable allocation and management of water resources. Managers can formulate scheduling plans in advance according to the prediction results, reasonably arrange the allocation of water resources, and ensure the efficient utilization of water resources and the stable operation of the system while meeting the water usage demands of each water diversion outlet.

[0199] Then, based on the geographical location information, future water demand, current flow data, and current pressure data corresponding to each water diversion outlet, construct the state space of the water diversion outlet corresponding to each water diversion outlet; based on the current working parameter information corresponding to each water storage and power generation device, construct the device state space. By constructing the state space of the water diversion outlet and the device state space, various states of the water diversion outlet and the water storage and power generation device in the water conveyance system can be comprehensively and accurately described, enabling the water conveyance system to have a clear understanding of its own operating conditions and providing an accurate basis for subsequent decision-making. Input the state space of each water diversion outlet into the water diversion outlet intelligent agent corresponding to each water diversion outlet; each water diversion outlet intelligent agent generates multiple initial valve control actions based on the state space of the water diversion outlet. Using the learning and decision-making capabilities of the intelligent agent to generate initial valve control actions can achieve intelligent decision-making for the water conveyance system. The intelligent agent can analyze and reason based on complex state information, generate control actions that conform to the current system state, reduce the subjectivity and limitations of manual decision-making, and improve the accuracy and scientific nature of decision-making. Reward or punish the initial valve control actions according to the satisfaction of water demand, flow rate, and pressure stability. An effective action optimization mechanism is established, which can guide the intelligent agent to continuously adjust and improve the control actions, making the generated alternative valve control actions more in line with the optimization goals of the system. By rewarding actions that meet water demand, maintain flow rate and pressure stability, and punishing actions that do not meet the requirements, the intelligent agent can gradually learn the optimal control strategy and improve the overall performance of the water conveyance system. According to the reward or punishment results, obtain multiple alternative valve control actions and generate the first action space. Enable the water conveyance system to adapt to complex and changeable operating environments. In addition, the reward and punishment mechanism screens out a series of actions that can effectively control the valves in different situations, and the first action space composed of these actions provides rich choices for the subsequent generation of scheduling strategies. A high-quality action space helps the system find a better scheduling plan, improve the control accuracy and flexibility of the water conveyance system, and better cope with various complex working conditions and emergencies. Then, based on each device state space, generate the second action space for controlling each water storage and power generation device. This means that when the water conveyance system faces different states, it has more choices to achieve optimal scheduling, can be flexibly adjusted according to the actual situation, and improves the adaptability and flexibility of the system. Construct an action reward calculation mechanism; the action reward calculation mechanism includes at least one of water demand satisfaction reward, device operation benefit reward, and system stability reward. This multi-dimensional reward mechanism can prompt the system to not only focus on meeting water demand when formulating scheduling strategies, but also take into account the operation benefits of water storage and power generation devices and the stability of the entire system, achieving multi-objective optimization of water resource utilization, energy production, and reliable system operation. Then, quantize each alternative valve control action included in each first action space to generate the quantum state of the alternative valve control action. Quantize each alternative device control action included in each second action space to generate the quantum state of the alternative device control action.Thus, various control actions in the water conveyance system can be represented in a more precise manner. Select a spare valve control action quantum state from each of the first action spaces, and select a spare device control action quantum state from each of the second action spaces to generate multiple spare combined action quantum states. This greatly increases the diversity of action combinations. The quantized action space has more possibilities and flexibility, and can cover a wider range of control strategies. This enables the water conveyance system to have more options to find the optimal or near-optimal scheduling scheme when facing complex water conveyance system operation conditions, improving the adaptability and optimization ability of the water conveyance system. In addition, simulate the execution process of each spare combined action quantum state in the target water conveyance pipeline, and calculate the reward values of each spare combined action quantum state based on the action reward calculation mechanism. This provides an effective means of evaluation and optimization. Through simulation execution, the possible effects of each spare combined action quantum state can be understood in advance, and the calculation of reward values can quantify these effects, helping the water conveyance system quickly screen out action combinations with higher benefits. This way of simulating first and then evaluating can avoid a large number of trials and errors in the actual system, reduce the decision-making risk, and at the same time improve the optimization efficiency of the scheduling strategy, helping to quickly find the best scheduling scheme that meets multiple optimization goals. In addition, combined with the action reward calculation mechanism, factors such as the satisfaction of water demand, the operation benefits of water storage and power generation equipment, and system stability are incorporated into the calculation of reward values, enabling the decision-making process to comprehensively consider multiple important indicators. This decision-making method that integrates multiple factors can ensure that the generated scheduling strategy not only focuses on the optimization of a single target, but achieves balance and coordination among multiple targets. For example, while meeting the water demand, taking into account the operation benefits of water storage and power generation equipment and the stability of the system, thereby improving the overall performance of the water conveyance system and achieving a more ideal operation effect.

[0200] By calculating the mean and standard deviation of the reward values, the reward values of the standby combined action quantum states can be quantitatively evaluated and analyzed. Dynamically adjusting the temperature drop rate according to the standard deviation of the reward values enables the water conveyance system to flexibly adjust the optimization process based on the performance of the standby combined action quantum states. If the standard deviation is large, it indicates that the reward values of different action combinations vary greatly, and there may be a large optimization space. At this time, appropriately accelerating the temperature drop rate can more quickly explore better action combinations. On the contrary, if the standard deviation is small, it means that the performance of the action combinations is relatively stable. At this time, slowing down the temperature drop rate helps to more finely search for local optimal solutions, avoid missing potential better strategies, and thus improve the optimization efficiency and effect. Then, the standby combined action quantum states are adjusted to generate updated combined action quantum states, and the reward values before and after the update are continuously compared. The combination with a larger reward value is selected for further update. This gradually optimized mechanism can guide the system to continuously evolve towards the optimal solution. Through repeated iterations, the system can gradually eliminate inferior action combinations, retain and improve better combinations, and finally find the target combined action quantum state with the largest reward value, thereby obtaining the best scheduling strategy that meets multi-objective optimization. Finally, the target valve control actions and target device control actions corresponding to the target combined action quantum state are determined, and the target scheduling strategy is output, realizing the accurate conversion from the quantized action space to the actually executable scheduling strategy. This accurate decision-making can accurately apply the optimization results of the system at the quantum level to the actual water conveyance pipeline system, precisely control the valves and water storage and power generation equipment, so as to achieve comprehensive goals such as meeting water demand, improving the operation efficiency of equipment, and ensuring system stability, effectively enhancing the operation management level and overall performance of the water conveyance system.

[0201] By comparing and screening the reward values of different action combinations, the water conveyance system can find the scheduling plan that best meets the actual needs, improves the overall efficiency, and ensures the stability of the system, thereby improving the operation efficiency and management level of the water conveyance system. Through the above series of measures, this method forms a complete closed-loop optimization system from the construction of the state space to the generation of the action space, and then to the evaluation based on the comprehensive reward mechanism and the output of the scheduling strategy. It can effectively improve the overall performance of the water conveyance system, including more accurately meeting water demand, improving the operation efficiency of water storage and power generation equipment, enhancing the stability and reliability of the system, and ultimately realizing the efficient utilization of water resources and the sustainable operation of the system.

[0202] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A water energy scheduling method, characterized in that, The method includes: Obtaining the future water usage demands corresponding to each water diversion outlet in the target water conveyance pipeline within a preset future time period, and obtaining the geographical location information, current flow data, and current pressure data corresponding to each water diversion outlet; Obtaining the current working parameter information corresponding to each water storage and power generation device in the target water conveyance pipeline; the water storage and power generation device is used for water storage or power generation, and the current working parameters include current water storage working parameters and / or current power generation working parameters; at least one of the water diversion outlets is equipped with the water storage and power generation device; Constructing a water diversion outlet state space corresponding to each water diversion outlet based on the geographical location information, the future water usage demand, the current flow data, and the current pressure data corresponding to each water diversion outlet; Constructing a device state space based on the current working parameter information corresponding to each water storage and power generation device; Generating a first action space for controlling the valves of each water diversion outlet based on each water diversion outlet state space; the first action space includes multiple standby valve control actions; Generating a second action space for controlling each water storage and power generation device based on each device state space; the second action space includes multiple standby device control actions; Constructing an action reward calculation mechanism; the action reward calculation mechanism includes at least one of a water usage demand satisfaction reward, a device operation benefit reward, and a system stability reward; Evaluating each action combination generated based on the first action space and the second action space based on the action reward calculation mechanism; Outputting a target scheduling strategy according to the evaluation result; the target scheduling strategy includes a first scheduling strategy for controlling the valves corresponding to each water diversion outlet and / or a second scheduling strategy for controlling the water storage and power generation devices corresponding to each water diversion outlet; The formula for the water usage demand satisfaction reward is: ； where n is the number of water diversion outlets, and Q 实际，i is the actual water supply of the i-th water diversion outlet; Q 需求，i is the future water demand of the i-th water diversion outlet; α 类型，i is the urgency coefficient of the water use type corresponding to the i-th water diversion outlet, and both a and b are coefficients.

2. The method according to claim 1, wherein The formula for the device operation benefit reward is: ； ； ； ； Among them, γ is the electricity price fluctuation coefficient, and E 预期 is the initial expected power generation revenue, and E 实际 is the actual power generation revenue. is the adjusted expected power generation revenue after the expected power generation revenue is dynamically adjusted according to the real-time electricity price. Let V 满足 be the water storage volume actually meeting the subsequent water allocation, and V 需求 be the water storage demand for the subsequent water allocation. is the seasonal water storage coefficient. is the maintenance cost coefficient. is the equipment life impact coefficient. c, d, and f are all coefficients.

3. The method according to claim 1, wherein The target water conveyance pipeline includes multiple flow and pressure monitoring nodes, and the formula for the system stability reward is: ； Among them, m is the number of nodes for monitoring flow and pressure, F 实际，j is the actual flow rate of the j-th flow monitoring point, F 正常，j is the normal flow rate of the j-th flow monitoring point, F 实际，k is the actual pressure of the k-th pressure monitoring point, F 正常，k is the normal pressure of the k-th flow monitoring point, is the system recovery ability index, and θ is the collaborative stability coefficient.

4. The method according to claim 1, characterized in that The generating of the first action space for controlling the valves of each water diversion outlet based on each water diversion outlet state space includes: Inputting each water diversion outlet state space into the water diversion outlet intelligent body corresponding to each water diversion outlet; Each water diversion outlet intelligent body generates multiple initial valve control actions based on the water diversion outlet state space; Rewarding or punishing the initial valve control actions according to the water usage demand satisfaction situation, flow, and pressure stability; Obtaining multiple standby valve control actions according to the reward or punishment result, and generating the first action space.

5. The method according to claim 1, characterized in that The evaluating of each action combination generated based on the first action space and the second action space based on the action reward calculation mechanism includes: Quantizing each standby valve control action included in each first action space to generate a standby valve control action quantum state; Quantizing each standby device control action included in each second action space to generate a standby device control action quantum state; Select one standby valve control action quantum state from each of the first action spaces, select one standby device control action quantum state from each of the second action spaces, and generate multiple standby combined action quantum states; Simulate the execution process of each standby combined action quantum state in the target water conveyance pipeline, and calculate the reward value of each standby combined action quantum state based on the action reward calculation mechanism.

6. The method according to claim 5, wherein The output of the target scheduling strategy according to the evaluation result includes: Calculate the mean value and standard deviation of the reward values according to the reward values of each standby combined action quantum state; Dynamically adjust the temperature drop rate according to the standard deviation of the reward values; Adjust each standby combined action quantum state according to the adjusted temperature drop rate to generate updated combined action quantum states; Calculate the reward values corresponding to each updated combined action quantum state; Compare the reward values corresponding to each updated combined action quantum state with the reward values of the corresponding standby combined action quantum states; If there are more than a preset proportion of the reward values corresponding to the updated combined action quantum states greater than the reward values of the corresponding standby combined action quantum states, update each updated combined action quantum state based on the current quantum gate operation strategy until the target combined action quantum state with the largest reward value is found; Determine the target valve control action and target device control action corresponding to the target combined action quantum state; Output the target scheduling strategy based on the target valve control action and the target device control action.

7. The method according to claim 6, characterized in that, The update of each updated combined action quantum state based on the current quantum gate operation strategy includes: The current quantum gate operation strategy updates each updated combined action quantum state based on the following formula: ； Among them, s is the current updated combined action quantum state, a is the quantum gate operation strategy adopted for the updated combined action quantum state, and Q(s, a) is the value function of taking action a in state s, which reflects the performance of this strategy in past iterations; R(s, a) is the reward value corresponding to the current updated combined action quantum state, α is the learning rate, γ is the discount factor, and max a′ Q(s′, a′) represents the one with the highest value among all actions a′ that can be taken in the new state s′ entered after executing action a, which reflects the prediction of the optimal future strategy.

8. The method according to claim 1, characterized in that The acquisition of the future water usage demands corresponding to each water diversion port in the target water conveyance pipeline in a future preset time period includes: Obtain the historical water usage data of each water diversion port within a preset time period before the current time, the precipitation data, weather data, temperature data, and time feature data corresponding to the future preset time period; Input the historical water usage data, the precipitation data, the weather data, the temperature data, and the time feature data into a preset water usage demand prediction model; The preset water usage demand prediction model outputs the future water usage demands corresponding to each water diversion port in the future preset time period.

Images