A method and system for calculating the adjustable capacity of hydropower for virtual power plants

Abstract

The present invention proposes a method and system for calculating the adjustable capacity of hydropower for a virtual power plant, the method comprising: collecting the operating parameters of the hydropower station in the centralized control platform in real time through a sensor network; constructing a relationship model based on the water level and storage capacity of the reservoir in the centralized control platform, and constructing a comprehensive characteristic model based on the operation of the turbine of each unit of the reservoir power station in the centralized control platform; obtaining water level constraints; calculating the adjustable capacity of the target power station; calculating the total real-time adjustable capacity of the target power station in coordinated operation with the power grid; predicting changes in water volume and power demand in future time periods; according to the predicted scheduling needs of the virtual power plant, incorporating the adjustable capacity of the hydropower station into the overall energy resource pool, determining the optimal power generation plan of the hydropower station, and maximizing system efficiency and economic benefits. The present invention uses modern technical means such as big data, artificial intelligence diagnosis, and the Internet of Things to achieve data sharing such as safe operation, production management, and industry association among various hydropower stations.

Description

Technical Field

[0001] The present invention belongs to the technical field of virtual power plants, and in particular, relates to a method and system for calculating the adjustable capacity of hydropower used in virtual power plants. Background Art

[0002] With the access of large-scale random and intermittent distributed power sources, and the increase of flexible loads that can actively participate in the operation and control of the power grid, it is urgent to comprehensively coordinate distributed power sources and flexible loads during the operation of the distribution system. As a new operation mode at the distribution system level, virtual power plant (VPP) has broad application prospects in effectively integrating a large number of distributed power sources and flexible loads.

[0003] In the existing virtual power plant (VPP) system, the adjustable capacity calculation method of hydropower stations usually faces the following challenges:

[0004] The calculation of the adjustable capacity of a hydropower station is highly dependent on the accurate prediction of future water volume (including inflow and outflow) and the accurate prediction of current and future electricity demand. However, the existing prediction models may be limited by data quality and the limitations of the models themselves, making it difficult to provide high-precision prediction results in all cases, thus affecting the accuracy of the adjustable capacity calculation.

[0005] The accurate calculation of the adjustable capacity of a hydropower station often requires complex hydrological models and power system models, and the establishment and solution of these models require a lot of computing resources. In practical applications, the limitation of computing resources may lead to the simplification of the model, which in turn affects the accuracy and practicality of the calculation results.

[0006] The operation of hydropower stations is affected by both natural conditions (such as rainfall and snowmelt) and human scheduling (such as flood control and ecological flow release). These factors change frequently, requiring the scheduling strategy of hydropower stations to be highly flexible and responsive. Traditional computing methods may find it difficult to respond to these dynamic changes in real time and efficiently, especially in the face of extreme weather events.

[0007] In view of the above shortcomings, developing more advanced prediction models, adopting more flexible optimization algorithms, strengthening multi-source data fusion and analysis capabilities, realizing multi-objective comprehensive optimization, improving the personalized adaptability of the model, and strengthening cross-system communication and collaboration are important directions for improving the calculation method of adjustable capacity of hydropower stations in the future. Summary of the invention

[0008] The purpose of this invention is to propose a method and system for calculating the adjustable capacity of hydropower for a virtual power plant, which combines the centralized control center with modern technical means such as big data, artificial intelligence diagnosis, and the Internet of Things to realize data sharing such as safe operation, production management, and industry association among hydropower stations, so as to achieve the goal of unmanned operation.

[0009] In order to achieve the above object, a first aspect of the present invention provides a method for calculating the adjustable capacity of hydropower for a virtual power plant, the method comprising:

[0010] S1. Deploy a sensor network in the centralized control platform to collect the operating parameters of the hydropower station in the centralized control platform in real time through the sensor network;

[0011] S2. Build a relationship model based on the reservoir water level and storage capacity in the centralized control platform, and build a comprehensive characteristic model based on the operation of the turbine of each unit of the reservoir power station in the centralized control platform;

[0012] S3. Collect the parameters of the current water levels of all reservoirs in the centralized control platform and the water levels downstream of the power station to obtain water level constraints.

[0013] S4. Calculate the adjustable capacity of the target power station;

[0014] S5. Calculate the total real-time adjustable capacity of the target power station and the power grid in coordinated operation according to the water level constraint and the adjustable capacity of the target power station;

[0015] S6. Based on historical operating parameters and real-time operating parameters, a prediction and dispatching model of the centralized control platform based on reinforcement learning is constructed to predict the changes in water and electricity demand in the future period;

[0016] S7. According to the predicted dispatching demand of the virtual power plant, the adjustable capacity of the hydropower station is incorporated into the overall energy resource pool, and the optimal power generation plan of the hydropower station is determined through the improved simulated annealing algorithm to maximize system efficiency and economic benefits;

[0017] Wherein, in said S5, the total real-time adjustable capacity of the target power station and the power grid in coordinated operation is calculated according to the water level constraint condition and the adjustable capacity of the target power station, including:

[0018] The positive and negative peak-shaving capacity models of the centralized control platform are designed. The objective functions of the positive and negative peak-shaving capacity models of the centralized control platform are:

[0019]

[0020] Among them, P 正向(t) represents the forward adjustable load capacity, k represents the number of all reservoir power stations connected to the centralized control platform, n represents the number of units that can currently participate in the regulation, j represents the jth reservoir power station, i represents the i-th unit that can participate in the regulation, and P(j,H,t) represents the output value of the jth reservoir power station at time t when the current reservoir water level is H.

[0021] Furthermore, the method further comprises:

[0022] Constructing a virtual power plant risk assessment model:

[0023] Construct the set of risk decision variables of virtual power plant E = {e1, e2, e3, .....e i ,e n}, where e i represents the probability of the occurrence of the i-th risk event, let C max is the maximum adjustable capacity of the hydropower station under ideal conditions without any risk impact, then the objective function of the virtual power plant risk assessment model is designed as for:

[0024]

[0025] Where T represents the time set, t represents the time, represents the adjustable capacity when the i-th risk event occurs; γ represents the weight of the influencing factor, C i It represents the adjustable capacity after the impact of the first type of risk, and Δt represents the interval time;

[0026] The weighted average adjustable capacity is calculated using the total probability formula. Assuming that the risk events are independent of each other, the total adjusted adjustable capacity C j It is expressed as:

[0027]

[0028] Among them, 1-e i +e i ×β represents the actual adjustable capacity contribution when the i-th risk occurs. If the risk does not occur, it directly contributes 1, and if it occurs, it contributes 1-e i ; β represents the impact ratio on power generation capacity when the type of risk event occurs, reflecting the degree of production reduction or shutdown.

[0029] Furthermore, the operating parameters include water level, water flow velocity, current power generation and future short-term weather forecast data; in said S1, it also includes time window preprocessing of the operating data.

[0030] Furthermore, the relationship model includes a relationship proportional model between the reservoir water level and the reservoir volume and a relationship proportional model between the reservoir water level and the reservoir volume. The relationship proportional model between the reservoir water level and the reservoir volume is expressed as follows:

[0031]

[0032] in, represents the ratio of the kth reservoir water level to the reservoir volume, Z k represents the water level of the kth reservoir, V k represents the volume of the kth reservoir; the proportional model of the relationship between the reservoir water level and the reservoir volume is expressed as follows:

[0033]

[0034] in, represents the ratio of the kth reservoir water level to the reservoir area, F k represents the area of ​​the kth reservoir;

[0035] The turbine operation constructs a comprehensive characteristic model, which is expressed as follows:

[0036] Assume H represents the water head and P represents the turbine output. The water head and turbine output are expressed as follows:

[0037] P=K·H m Q n

[0038] Where K represents a constant, m and n represent indexes related to the turbine type, and Q represents the flow rate;

[0039] The turbine efficiency is obtained by fitting the experimental data and is expressed as follows:

[0040] η=f(H,P)

[0041] Where, η represents the turbine efficiency, and f represents the polynomial function;

[0042] Set H max and H min In order to ensure the maximum and minimum safe output under water head, the suction height of the turbine is set to +1.5m, +2.0m and -2.5m.

[0043] Furthermore, the S4 specifically includes:

[0044] Obtain the number of units that can currently participate in the regulation of the power station and the current real-time output of these units, and obtain the maximum output of the unit by establishing a comprehensive characteristic curve model of the turbine operation of each unit based on the current real-time water head;

[0045] According to the water level constraints:

[0046] Z min (m,t)≤Z(m,t)≤Z max (m,t)

[0047] Among them, Z min (m,t),Z max (m, t) are the upper and lower limits of the water level of the mth power station in the t period, which also include the normal water level and dead water level of the reservoir in the non-flood season and the flood limit water level and dead water level of the reservoir in the flood season;

[0048] When the upstream water level is Z min (m,t)<Z(m,t)<Z min (m, t), the positive adjustable load capacity is expressed as follows:

[0049]

[0050] Wherein, I represents the Ith unit;

[0051] Negative adjustable load capacity, expressed as follows:

[0052]

[0053] Where n represents the number of units that can currently participate in regulation in the mth power station; P max (m,I,t) is the current maximum output of the Ith unit in the mth power station during the tth period, obtained by establishing the comprehensive characteristic curve model of the turbine operation of each unit under the current real-time water head; P(m,I,t) is the current output value of the Ith unit in the mth power station during the tth period;

[0054] When the upstream water level Z(m,t) ≥ Z max (m, t), the positive adjustable load capacity is expressed as follows:

[0055]

[0056] Negative adjustable load capacity, P 反向 (m,t)=0;

[0057] When the upstream water level Z(m,t) ≥ Z min (m, t), the forward adjustable load capacity is calculated as follows:

[0058] P 正向 (m,t)=0

[0059] Negative adjustable load capacity, expressed as follows:

[0060]

[0061] Furthermore, the S5 specifically includes:

[0062] The overall adjustable capacity of the platform is calculated as follows:

[0063] Overall positive adjustable load capacity of the platform:

[0064] Overall negative adjustable load capacity of the platform:

[0065] Among them, l is the number of all reservoir power stations connected to the centralized control information platform, and j represents the jth reservoir power station.

[0066] Furthermore, the S6 specifically includes:

[0067] Constructing a prediction and scheduling model for the centralized control platform:

[0068] Define states t ∈S, represents the operating parameters at the current time t;

[0069] Action a t ∈A, represents the operation taken by the centralized control platform;

[0070] Design the reward function, expressed as follows:

[0071]

[0072] Among them, α and ω represent the conversion coefficients of the overall positive adjustable load capacity of the platform and the overall negative adjustable load capacity of the platform respectively;

[0073] Furthermore, the prediction and scheduling model of the centralized control platform is trained based on the Actor-Critic framework:

[0074] Sampling data from the experience pool R t ,s t ,s t+1 , r t}, design the value network weight of the Actor-Critic framework as θ Q , the policy network weight is θ μ ;

[0075] Introduce the white noise mechanism and set the action of the agent at time t to:

[0076] a t =μ t (s t )+Y t

[0077] Among them, μ t represents the output strategy of the policy network at time t, Y t represents white noise;

[0078] Execute action a t , will reward r t and the state s at the next moment t+1 Feedback to the agent, the current sample {a t ,s t ,s t+1 , r t}Deposit into experience pool R;

[0079] The loss function L of the value network is designed to be minimized according to the stochastic gradient descent method, which is expressed as follows:

[0080]

[0081] Among them, s i represents the i-th intelligent state, a i represents the action of the ith agent, y i represents the true value;

[0082] The policy network updates the policy with the help of the value network, where the updated gradient of the policy is expressed as follows:

[0083]

[0084] Among them, a represents action and s represents state;

[0085] Update the parameters of the target network, as follows:

[0086]

[0087] in, and represents the weight of the updated target network, and τ represents the hyperparameter.

[0088] Furthermore, the simulated annealing algorithm includes:

[0089] Initial state: Randomly initialize the power generation of each energy resource in each time period;

[0090] Temperature scheduling: Define the initial temperature δ and cooling rate δ. After each iteration, the temperature is updated as follows:

[0091] T new =δ·T old

[0092] Among them, T new represents the updated temperature, T old Indicates the old temperature;

[0093] State transfer: Try new power distribution based on the current state and certain probability rules;

[0094] Acceptance criteria: decide whether to accept the new state based on the cost difference between the new and old states and the current temperature;

[0095] Termination condition: reaching the preset number of iterations or the temperature is lower than a certain threshold.

[0096] In a second aspect of the present invention, a hydropower adjustable capacity calculation system for a virtual power plant is provided, the system comprising:

[0097] The data collection module is used to deploy a sensor network in the centralized control platform and collect the operating parameters of the hydropower station in the centralized control platform in real time through the sensor network;

[0098] The capacity calculation module is used to build a relationship model between the reservoir water level and storage capacity based on the centralized control platform, and to build a comprehensive characteristic model based on the operation of the turbines of each unit of the reservoir power station in the centralized control platform; to collect the parameters of the current water level of all reservoirs in the centralized control platform and the water level downstream of the power station, and to obtain the water level constraints; to calculate the adjustable capacity of the target power station; and to calculate the total real-time adjustable capacity of the target power station in coordinated operation with the power grid based on the water level constraints and the adjustable capacity of the target power station;

[0099] The capacity prediction module builds a prediction and dispatching model of the centralized control platform based on reinforcement learning based on historical and real-time operating parameters to predict changes in water and electricity demand in future periods. According to the predicted dispatching needs of the virtual power plant, the adjustable capacity of the hydropower station is incorporated into the overall energy resource pool, and the optimal power generation plan of the hydropower station is determined through an improved simulated annealing algorithm to maximize system efficiency and economic benefits.

[0100] The capacity calculation module calculates the total real-time adjustable capacity of the target power station and the power grid in coordinated operation according to the water level constraint and the adjustable capacity of the target power station, including:

[0101] The positive and negative peak-shaving capacity models of the centralized control platform are designed. The objective functions of the positive and negative peak-shaving capacity models of the centralized control platform are:

[0102]

[0103] Among them, P 正向 (t) represents the forward adjustable load capacity, k represents the number of all reservoir power stations connected to the centralized control platform, n represents the number of units that can currently participate in the regulation, j represents the jth reservoir power station, i represents the i-th unit that can participate in the regulation, and P(j,H,t) represents the output value of the jth reservoir power station at time t when the current reservoir water level is H.

[0104] The beneficial technical effects of the present invention are at least as follows:

[0105] According to the storage capacity of the reservoir power station, the present invention establishes a real-time adjustable storage capacity calculation model for the entire centralized control center to participate in the coordinated operation of the power grid, and obtains the real-time load capacity that the centralized control center can participate in the regulation of the power grid. The present invention can change the existing mode of the centralized control center passively accepting dispatching instructions. Through the real-time calculation model, the current adjustable power range is provided to the dispatching center with high accuracy, which has great practical significance for the regulation of the dispatching center's power grid load and provides good technical support for the coordinated operation of virtual power plants.

[0106] The centralized control center of the present invention utilizes modern technical means such as big data, artificial intelligence diagnosis, and the Internet of Things to achieve data sharing such as safe operation, production management, and industry association among various hydropower stations, so as to achieve the goal of unmanned duty. The construction of the centralized control center platform combines the power stations that were originally distributed in remote areas and were not connected to each other through a network mode. For the dispatching center, all the power stations in the centralized control center can be virtualized as one hydropower station, which is uniformly subject to the dispatching management of the centralized control center.

[0107] The present invention uses machine learning to help virtual power plants better manage and dispatch distributed energy resources, such as adjusting the power generation of hydropower stations according to predicted load demand, or coordinating the charging and discharging timing of energy storage equipment, thereby maximizing the economy and efficiency of energy assets. By intelligently allocating and adjusting power demand, energy waste can be reduced and energy conversion and utilization efficiency can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0108] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0109] Figure 1 The present invention is a flow chart of a method for calculating the adjustable capacity of hydropower used in a virtual power plant.

[0110] Figure 2 It is a schematic diagram of the water level storage capacity and water level area curve of the reservoir of the present invention.

[0111] Figure 3 It is a schematic diagram of the comprehensive characteristic curve of the operation of the turbine of the present invention.

[0112] Figure 4 This is a framework diagram of a hydropower adjustable capacity calculation system for a virtual power plant according to the present invention. DETAILED DESCRIPTION

[0113] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0114] like Figure 1 As shown, an embodiment of the present invention provides a method for calculating the adjustable capacity of hydropower for a virtual power plant, the method comprising:

[0115] S1. Deploy a sensor network in the centralized control platform to collect the operating parameters of the hydropower station in the centralized control platform in real time through the sensor network;

[0116] S2, such as Figure 2-Figure 3 As shown in the figure, a relationship model is built based on the reservoir water level and storage capacity in the centralized control platform, and a comprehensive characteristic model is built based on the turbine operation of each unit of the reservoir power station in the centralized control platform.

[0117] S3. Collect the parameters of the current water levels of all reservoirs in the centralized control platform and the water levels downstream of the power station to obtain water level constraints.

[0118] S4. Calculate the adjustable capacity of the target power station;

[0119] S5. Calculate the total real-time adjustable capacity of the target power station and the power grid in coordinated operation according to the water level constraint and the adjustable capacity of the target power station;

[0120] S6. Based on historical operating parameters and real-time operating parameters, a prediction and dispatching model of the centralized control platform based on reinforcement learning is constructed to predict the changes in water and electricity demand in the future period;

[0121] S7. According to the predicted dispatching demand of the virtual power plant, the adjustable capacity of the hydropower station is incorporated into the overall energy resource pool, and the optimal power generation plan of the hydropower station is determined through the improved simulated annealing algorithm to maximize system efficiency and economic benefits;

[0122] Wherein, in said S5, the total real-time adjustable capacity of the target power station and the power grid in coordinated operation is calculated according to the water level constraint condition and the adjustable capacity of the target power station, including:

[0123] The positive and negative peak-shaving capacity models of the centralized control platform are designed. The objective functions of the positive and negative peak-shaving capacity models of the centralized control platform are:

[0124]

[0125] Among them, P 正向(t) represents the forward adjustable load capacity, k represents the number of all reservoir power stations connected to the centralized control platform, n represents the number of units that can currently participate in the regulation, j represents the jth reservoir power station, i represents the i-th unit that can participate in the regulation, and P(j,H,t) represents the output value of the jth reservoir power station at time t when the current reservoir water level is H.

[0126] Furthermore, the method further comprises:

[0127] Constructing a virtual power plant risk assessment model:

[0128] Construct the set of risk decision variables of virtual power plant E = {e1, e2, e3, .....e i ,e n}, where e i represents the probability of the occurrence of the i-th risk event, let C max is the maximum adjustable capacity of the hydropower station under ideal conditions without any risk impact, then the objective function of the virtual power plant risk assessment model is designed as for:

[0129]

[0130] Where T represents the time set, t represents the time, represents the adjustable capacity when the i-th risk event occurs; γ represents the weight of the influencing factor, C i It represents the adjustable capacity after the impact of the first type of risk, and Δt represents the interval time;

[0131] The weighted average adjustable capacity is calculated using the total probability formula. Assuming that the risk events are independent of each other, the total adjusted adjustable capacity C j It is expressed as:

[0132]

[0133] Among them, 1-e i +e i ×β represents the actual adjustable capacity contribution when the i-th risk occurs. If the risk does not occur, it directly contributes 1, and if it occurs, it contributes 1-e i ; β represents the impact ratio on power generation capacity when the type of risk event occurs, reflecting the degree of production reduction or shutdown.

[0134] Furthermore, the operating parameters include water level, water flow velocity, current power generation and future short-term weather forecast data; in said S1, it also includes time window preprocessing of the operating data.

[0135] Specifically, assuming that the original operating parameters include the water level H t (Unit: meter), water velocity V t(Unit: m / s), current power generation P t (Unit: MW) and future short-term weather forecast data (for example, the precipitation in the next 24 hours R t , unit: mm). Normalization can be done by using the maximum and minimum value normalization or Z-score normalization. Take the maximum and minimum value normalization as an example:

[0136] Water level normalization:

[0137]

[0138] Normalized water velocity:

[0139]

[0140] Normalized power generation:

[0141]

[0142] Normalized precipitation:

[0143]

[0144] Among them, H max ,H min ,V max ,V min ,P max ,P min ,R max ,R min are the historical minimum and maximum values ​​of their respective parameters. t ′、V t ′、P t ′、R t ′R min are the normalized values ​​of their respective parameters.

[0145] Splitting continuous time series data into multiple windows of fixed length is usually to capture local patterns in the time series. Assuming that the window length is w time units (such as hours), and the present invention focuses on the data of w units of time from the current time t, the data of the kth time window can be expressed as:

[0146] W k ={(H t-k+w ′,V t-k+w ′,P t-k+w ′,R t-k+w ′),...,(H t ′,V t ′,P t ′,P t ′)}

[0147] Here, k starts from 0 and continues until k+w≤T, where T is the total length of the dataset.

[0148] In each time window W k In the sliding window, features can be further constructed, such as statistical features such as the average, maximum, minimum, and standard deviation within the time window, or the data sequence within the sliding window can be directly used as the input of models such as RNN (recurrent neural network).

[0149] Finally, each time window can be converted into a set of feature vectors for subsequent machine learning model input. These feature vectors not only capture the trend information of a single parameter, but also take into account the interaction between parameters and the dynamic characteristics of the time series, which is conducive to improving the accuracy and generalization ability of the prediction model.

[0150] Furthermore, the relationship model includes a relationship proportional model between the reservoir water level and the reservoir volume and a relationship proportional model between the reservoir water level and the reservoir volume. The relationship proportional model between the reservoir water level and the reservoir volume is expressed as follows:

[0151]

[0152] in, represents the ratio of the kth reservoir water level to the reservoir volume, Z k represents the water level of the kth reservoir, V k represents the volume of the kth reservoir; the proportional model of the relationship between the reservoir water level and the reservoir volume is expressed as follows:

[0153]

[0154] in, represents the ratio of the kth reservoir water level to the reservoir area, F k represents the area of ​​the kth reservoir;

[0155] The turbine operation constructs a comprehensive characteristic model, which is expressed as follows:

[0156] Assume H represents the water head and P represents the turbine output. The water head and turbine output are expressed as follows:

[0157] P=K·H m Q n

[0158] Where K represents a constant, m and n represent indexes related to the turbine type, and Q represents the flow rate;

[0159] The turbine efficiency is obtained by fitting the experimental data and is expressed as follows:

[0160] η=f(H,P)

[0161] Where, η represents the turbine efficiency, and f represents the polynomial function;

[0162] Set H max and H min In order to ensure the maximum and minimum safe output under water head, the suction height of the turbine is set to +1.5m, +2.0m and -2.5m.

[0163] Furthermore, the S4 specifically includes:

[0164] Obtain the number of units that can currently participate in the regulation of the power station and the current real-time output of these units, and obtain the maximum output of the unit by establishing a comprehensive characteristic curve model of the turbine operation of each unit based on the current real-time water head;

[0165] According to the water level constraints:

[0166] Z min (m,t)≤Z(m,t)≤Z max (m,t)

[0167] Among them, Z min (m,t),Z max (m, t) are the upper and lower limits of the water level of the mth power station in the t period, which also include the normal water level and dead water level of the reservoir in the non-flood season and the flood limit water level and dead water level of the reservoir in the flood season;

[0168] When the upstream water level is Z min (m,t)<Z(m,t)<Z min (m, t), the positive adjustable load capacity is expressed as follows:

[0169]

[0170] Wherein, I represents the Ith unit;

[0171] Negative adjustable load capacity, expressed as follows:

[0172]

[0173] Where n represents the number of units that can currently participate in regulation in the mth power station; P max (m,I,t) is the current maximum output of the Ith unit in the mth power station during the tth period, obtained by establishing the comprehensive characteristic curve model of the turbine operation of each unit under the current real-time water head; P(m,I,t) is the current output value of the Ith unit in the mth power station during the tth period;

[0174] When the upstream water level Z(m,t) ≥ Z max (m, t), the positive adjustable load capacity is expressed as follows:

[0175]

[0176] Negative adjustable load capacity, P 反向 (m,t)=0;

[0177] When the upstream water level Z(m,t) ≥ Z min (m, t), the forward adjustable load capacity is calculated as follows:

[0178] P 正向 (m,t)=0

[0179] Negative adjustable load capacity, expressed as follows:

[0180]

[0181] Furthermore, the S5 specifically includes:

[0182] The overall adjustable capacity of the platform is calculated as follows:

[0183] Overall positive adjustable load capacity of the platform:

[0184] Overall negative adjustable load capacity of the platform:

[0185] Among them, l is the number of all reservoir power stations connected to the centralized control information platform, and j represents the jth reservoir power station.

[0186] Furthermore, the S6 specifically includes:

[0187] Constructing a prediction and scheduling model for the centralized control platform:

[0188] Define states t ∈S, represents the operating parameters at the current time t;

[0189] Action a t ∈A, represents the operation taken by the centralized control platform;

[0190] Design the reward function, expressed as follows:

[0191]

[0192] Among them, α and ω represent the conversion coefficients of the overall positive adjustable load capacity of the platform and the overall negative adjustable load capacity of the platform respectively;

[0193] Furthermore, the prediction and scheduling model of the centralized control platform is trained based on the Actor-Critic framework:

[0194] Sampling data from the experience pool R t ,s t ,st+1 , r t}, design the value network weight of the Actor-Critic framework as θ Q , the policy network weight is θ μ ;

[0195] Introduce the white noise mechanism and set the action of the agent at time t to:

[0196] a t =μ t (s t )+Y t

[0197] Among them, μ t represents the output strategy of the policy network at time t, Y t represents white noise;

[0198] Execute action a t , will reward r t and the state s at the next moment t+1 Feedback to the agent, the current sample {a t ,s t ,s t+1 , r t}Deposit into experience pool R;

[0199] The loss function L of the value network is designed to be minimized according to the stochastic gradient descent method, which is expressed as follows:

[0200]

[0201] Among them, s i represents the i-th intelligent state, a i represents the action of the ith agent, y i represents the true value;

[0202] The policy network updates the policy with the help of the value network, where the updated gradient of the policy is expressed as follows:

[0203]

[0204] Among them, a represents action and s represents state;

[0205] Update the parameters of the target network, as follows:

[0206]

[0207] in, and represents the weight of the updated target network, and τ represents the hyperparameter.

[0208] Furthermore, the simulated annealing algorithm includes:

[0209] Initial state: Randomly initialize the power generation of each energy resource in each time period;

[0210] Temperature scheduling: Define the initial temperature δ and cooling rate δ. After each iteration, the temperature is updated as follows:

[0211] T new =δ·T old

[0212] Among them, T new represents the updated temperature, T old Indicates the old temperature;

[0213] State transfer: Try new power distribution based on the current state and certain probability rules;

[0214] Acceptance criteria: decide whether to accept the new state based on the cost difference between the new and old states and the current temperature;

[0215] Termination condition: reaching the preset number of iterations or the temperature is lower than a certain threshold.

[0216] Further, in S7, P 正向 The objective function of maximizing system efficiency and economic benefits by maximizing (t) is:

[0217]

[0218] Among them, Q i is the power of the first type of resources (including hydropower stations) at power p i The marginal cost function under i is the power generated by the resource type in a given time interval, is the number of resource types, and is the total time interval of the plan (in hours).

[0219] Another embodiment of the present invention is a hydropower adjustable capacity calculation system for a virtual power plant, the system comprising:

[0220] The data collection module 101 is used to deploy a sensor network in the centralized control platform and collect the operating parameters of the hydropower station in the centralized control platform in real time through the sensor network;

[0221] The capacity calculation module 102 is used to construct a relationship model between the reservoir water level and storage capacity based on the centralized control platform, and to construct a comprehensive characteristic model based on the operation of the turbine of each unit of the reservoir power station in the centralized control platform; to collect the parameters of the current water level of all reservoirs in the centralized control platform and the water level downstream of the power station, and to obtain the water level constraint conditions; to calculate the adjustable capacity of the target power station; and to calculate the total real-time adjustable capacity of the target power station in coordinated operation with the power grid based on the water level constraint conditions and the adjustable capacity of the target power station;

[0222] The capacity prediction module 103 builds a prediction and dispatching model of the centralized control platform based on reinforcement learning based on historical operating parameters and real-time operating parameters to predict the changes in water volume and power demand in the future period; according to the predicted dispatching needs of the virtual power plant, the adjustable capacity of the hydropower station is incorporated into the overall energy resource pool, and the optimal power generation plan of the hydropower station is determined through the improved simulated annealing algorithm to maximize system efficiency and economic benefits;

[0223] The capacity calculation module 102 calculates the total real-time adjustable capacity of the target power station in coordinated operation with the power grid according to the water level constraint and the adjustable capacity of the target power station, including:

[0224] The positive and negative peak-shaving capacity models of the centralized control platform are designed. The objective functions of the positive and negative peak-shaving capacity models of the centralized control platform are:

[0225]

[0226] Among them, P 正向 (t) represents the forward adjustable load capacity, k represents the number of all reservoir power stations connected to the centralized control platform, n represents the number of units that can currently participate in the regulation, j represents the jth reservoir power station, i represents the i-th unit that can participate in the regulation, and P(j,H,t) represents the output value of the jth reservoir power station at time t when the current reservoir water level is H.

[0227] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0228] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a division of logical functions. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0229] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0230] Although embodiments of the present invention have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for calculating the adjustable capacity of hydropower for a virtual power plant, characterized in that: The method comprises: S1. Deploy a sensor network in the centralized control platform to collect the operating parameters of the hydropower station in the centralized control platform in real time through the sensor network; S2. Build a relationship model based on the reservoir water level and storage capacity in the centralized control platform, and build a comprehensive characteristic model based on the operation of the turbine of each unit of the hydropower station in the centralized control platform; S3. Collect the parameters of the current water levels of all reservoirs in the centralized control platform and the water levels downstream of the hydropower station to obtain water level constraints; S4. Calculate the adjustable capacity of the target hydropower station; S5. Calculate the total real-time adjustable capacity of the target hydropower station in coordinated operation with the power grid according to the water level constraint and the adjustable capacity of the target hydropower station; S6. Based on historical operating parameters and real-time operating parameters, a prediction and dispatching model of the centralized control platform based on reinforcement learning is constructed to predict the changes in water and electricity demand in the future period; S7. According to the predicted dispatching demand of the virtual power plant, the adjustable capacity of the hydropower station is incorporated into the overall energy resource pool, and the optimal power generation plan of the hydropower station is determined through the improved simulated annealing algorithm to maximize system efficiency and economic benefits; Wherein, in said S5, the total real-time adjustable capacity of the target hydropower station in coordinated operation with the power grid is calculated according to the water level constraint condition and the adjustable capacity of the target hydropower station, including: The positive and negative peak-shaving capacity models of the centralized control platform are designed. The objective functions of the positive and negative peak-shaving capacity models of the centralized control platform are: Among them, P 正向 (t) represents the forward adjustable load capacity, k represents the number of all hydropower stations connected to the centralized control platform, j represents the jth hydropower station, and P(j,H,t) represents the output value of the jth hydropower station at time t when the current reservoir water level is H; The relationship model includes a proportional model between the reservoir water level and the reservoir volume and a proportional model between the reservoir water level and the reservoir area. The proportional model between the reservoir water level and the reservoir volume is expressed as follows: in, Indicates The relationship between the water level of a reservoir and its volume is proportional to the Indicates The water level of the reservoir, Indicates The volume of a reservoir; the proportional model of the relationship between the reservoir water level and the reservoir area is expressed as follows: in, Indicates The relationship between the water level of a reservoir and the area of ​​the reservoir, Indicates The area of ​​the reservoir; The turbine operation constructs a comprehensive characteristic model, which is expressed as follows: Assume Γ represents the water head and P represents the turbine output. The water head and turbine output are expressed as follows: Among them, K is a constant, m and They represent indices related to turbine types, Q represents the flow rate; The turbine efficiency is obtained by fitting the experimental data and is expressed as follows: η=f(Γ,P) Where, η represents the turbine efficiency, and f represents the polynomial function; Set H max and H min For the maximum and minimum safe output under water head Γ, the suction height of the turbine is set to +1.5m, +2.0m and -2.5m; Wherein, the S4 specifically includes: Obtain the number of units that can currently participate in regulation of the hydropower station and the current real-time output of these units, and obtain the maximum output of the unit by establishing a comprehensive characteristic curve model of the turbine operation of each unit based on the current real-time water head; According to the water level constraints: Z min (j,t)≤Z(j,t)≤Z max (j,t) Among them, Z max (j,t),Z min (j, t) are the upper and lower limits of the water level of the j-th hydropower station in period t, which also include the normal water storage level and dead water level of the reservoir in the non-flood season and the flood limit water level and dead water level of the reservoir in the flood season; When the upstream water level is Z min (j,t)<Z(j,t)<Z max (j, t), the positive adjustable load capacity is expressed as follows: Among them, n represents the number of units that can currently participate in regulation of the j-th hydropower station, and i represents the i-th unit that can participate in regulation; Negative adjustable load capacity, expressed as follows: Where n represents the number of units that can currently participate in regulation at the jth hydropower station; P max (j,i,t) is the current maximum output of the i-th unit of the j-th hydropower station in the t-th period, obtained by establishing the comprehensive characteristic curve model of the turbine operation of each unit under the current real-time water head; P(j,i,t) is the current output value of the i-th unit of the j-th hydropower station in the t-th period; When the upstream water level Z(j,t) ≥ Z max (j, t), the positive adjustable load capacity is expressed as follows: Negative adjustable load capacity, P 反向 (j,t)=0; When the upstream water level Z(j,t)≤Z min (j, t), the forward adjustable load capacity is calculated as follows: P 正向 (j,t)=0 Negative adjustable load capacity, expressed as follows:

2. A method for calculating the adjustable capacity of hydropower for a virtual power plant according to claim 1, characterized in that: The method further comprises: Constructing a virtual power plant risk assessment model: Construct the set of risk decision variables of virtual power plant E = {e1, e2, e3, ...., e w ,e n' }, where e w represents the probability of the w-th risk event occurring, let C max is the maximum adjustable capacity of the hydropower station under ideal conditions without any risk impact, then the objective function of the virtual power plant risk assessment model is designed as for: Where T represents the time set, t represents the time, represents the adjustable capacity when the w-th risk event occurs; γ represents the weight of the influencing factor, C w represents the adjustable capacity after the wth risk impact, and Δt represents the interval time; The weighted average adjustable capacity is calculated using the total probability formula. Assuming that the risk events are independent of each other, the total adjusted adjustable capacity C j It is expressed as: Among them, 1-e w +e w ×β represents the actual adjustable capacity contribution when the w-th risk occurs. If the risk does not occur, it directly contributes 1, and if it does occur, it contributes 1-e w +e w ×β; β represents the impact ratio on power generation capacity when the w-th risk event occurs, reflecting the degree of production reduction or shutdown.

3. A method for calculating the adjustable capacity of hydropower for a virtual power plant according to claim 1, characterized in that: The operating parameters include water level, water flow velocity, current power generation and future short-term weather forecast data; in said S1, it also includes time window preprocessing of the operating data.

4. A method for calculating the adjustable capacity of hydropower for a virtual power plant according to claim 1, characterized in that: The S5 specifically includes: The overall adjustable capacity of the platform is calculated as follows: Overall positive adjustable load capacity of the platform: Overall negative adjustable load capacity of the platform: Among them, k is the number of all hydropower stations connected to the centralized control information platform, and j represents the jth hydropower station.

5. The method for calculating the adjustable capacity of hydropower for a virtual power plant according to claim 1, characterized in that: The S6 specifically includes: Constructing a prediction and scheduling model for the centralized control platform: Define states t ∈S, represents the operating parameters at the current time t; Action a t ∈A, represents the operation taken by the centralized control platform; Design the reward function, expressed as follows: Among them, α and ω represent the conversion coefficients of the overall positive adjustable load capacity of the platform and the overall negative adjustable load capacity of the platform, respectively.

6. A method for calculating the adjustable capacity of hydropower for a virtual power plant according to claim 5, characterized in that: Based on the Actor-Critic framework, the prediction and scheduling model of the centralized control platform is trained: Sampling data from the experience pool R t ,s t ,s t+1 , r t }, design the value network weight of the Actor-Critic framework as θ Q , the policy network weight is θ μ ; Introduce the white noise mechanism and set the action of the agent at time t to: a t =μ t (s t )+Y t Among them, μ t represents the output strategy of the policy network at time t, Y t represents white noise; Execute action a t , will reward r t and the state s at the next moment t+1 Feedback to the agent, the current sample {a t ,s t ,s t+1 , r t }Deposit into experience pool R; The design minimizes the loss function L of the value network according to the stochastic gradient descent method, which is expressed as follows: Among them, s l represents the state of the lth agent, a l represents the action of the lth agent, y l represents the true value; The policy network updates the policy with the help of the value network, where the updated gradient of the policy is expressed as follows: Among them, a represents action and s represents state; Update the parameters of the target network, as follows: in, and represents the weight of the updated target network, and τ represents the hyperparameter.

7. A method for calculating the adjustable capacity of hydropower for a virtual power plant according to claim 1, characterized in that: The improved simulated annealing algorithm comprises: Initial state: Randomly initialize the power generation of each energy resource in each time period; Temperature scheduling: define the initial temperature T old And cooling rate δ, the temperature is updated after each iteration as: T new =δ·T old Among them, T new Indicates the updated temperature; State transfer: Try new power distribution based on the current state and certain probability rules; Acceptance criteria: decide whether to accept the new state based on the cost difference between the new and old states and the current temperature; Termination condition: reaching the preset number of iterations or the temperature is lower than a certain threshold.

8. A hydropower adjustable capacity calculation system for a virtual power plant, characterized in that: The system comprises: The data collection module is used to deploy a sensor network in the centralized control platform and collect the operating parameters of the hydropower station in the centralized control platform in real time through the sensor network; The capacity calculation module is used to build a relationship model between the reservoir water level and storage capacity based on the centralized control platform, and to build a comprehensive characteristic model based on the operation of the turbines of each unit of the hydropower station in the centralized control platform; collect the parameters of the current water level of all reservoirs in the centralized control platform and the downstream water level of the hydropower station to obtain the water level constraints; calculate the adjustable capacity of the target hydropower station; calculate the total real-time adjustable capacity of the target hydropower station in coordinated operation with the power grid based on the water level constraints and the adjustable capacity of the target hydropower station; The capacity prediction module builds a prediction and dispatching model of the centralized control platform based on reinforcement learning based on historical and real-time operating parameters to predict changes in water and electricity demand in future periods. According to the predicted dispatching needs of the virtual power plant, the adjustable capacity of the hydropower station is incorporated into the overall energy resource pool, and the optimal power generation plan of the hydropower station is determined through an improved simulated annealing algorithm to maximize system efficiency and economic benefits. Among them, in the capacity calculation module, the total real-time adjustable capacity of the target hydropower station and the power grid in coordinated operation is calculated according to the water level constraint condition and the adjustable capacity of the target hydropower station, including: The positive and negative peak-shaving capacity models of the centralized control platform are designed. The objective functions of the positive and negative peak-shaving capacity models of the centralized control platform are: Among them, P 正向 (t) represents the forward adjustable load capacity, k represents the number of all hydropower stations connected to the centralized control platform, j represents the jth hydropower station, and P(j,H,t) represents the output value of the jth hydropower station at time t when the current reservoir water level is H; The relationship model includes a proportional model between the reservoir water level and the reservoir volume and a proportional model between the reservoir water level and the reservoir area. The proportional model between the reservoir water level and the reservoir volume is expressed as follows: in, Indicates The relationship between the water level of a reservoir and its volume is proportional to the Indicates The water level of the reservoir, Indicates The volume of a reservoir; the proportional model of the relationship between the reservoir water level and the reservoir area is expressed as follows: in, Indicates The relationship between the water level of a reservoir and the area of ​​the reservoir, Indicates The area of ​​the reservoir; The turbine operation constructs a comprehensive characteristic model, which is expressed as follows: Assume Γ to represent the water head and P to represent the turbine output. The water head and turbine output are expressed as follows: Among them, K is a constant, m and They represent indices related to turbine types, Q represents the flow rate; The turbine efficiency is obtained by fitting the experimental data and is expressed as follows: η=f(Γ,P) Where, η represents the turbine efficiency, and f represents the polynomial function; Set H max and H min For the maximum and minimum safe output under water head Γ, the suction height of the turbine is set to +1.5m, +2.0m and -2.5m; The calculation of the adjustable capacity of the target hydropower station specifically includes: Obtain the number of units that can currently participate in regulation of the hydropower station and the current real-time output of these units, and obtain the maximum output of the unit by establishing a comprehensive characteristic curve model of the turbine operation of each unit based on the current real-time water head; According to the water level constraints: Z min (j,t)≤Z(j,t)≤Z max (j,t) Among them, Z max (j,t),Z min (j, t) are the upper and lower limits of the water level of the j-th hydropower station in period t, which also include the normal water storage level and dead water level of the reservoir in the non-flood season and the flood limit water level and dead water level of the reservoir in the flood season; When the upstream water level is Z min (j,t)<Z(j,t)<Z max (j, t), the positive adjustable load capacity is expressed as follows: Among them, n represents the number of units that can currently participate in regulation of the j-th hydropower station, and i represents the i-th unit that can participate in regulation; Negative adjustable load capacity, expressed as follows: Where n represents the number of units that can currently participate in regulation at the jth hydropower station; P max (j,i,t) is the current maximum output of the i-th unit of the j-th hydropower station in the t-th period, obtained by establishing the comprehensive characteristic curve model of the turbine operation of each unit under the current real-time water head; P(j,i,t) is the current output value of the i-th unit of the j-th hydropower station in the t-th period; When the upstream water level Z(j,t) ≥ Z max (j, t), the positive adjustable load capacity is expressed as follows: Negative adjustable load capacity, P 反向 (j,t)=0; When the upstream water level Z(j,t)≤Z min (j, t), the forward adjustable load capacity is calculated as follows: P 正向 (j,t)=0 Negative adjustable load capacity, expressed as follows:

Images