Scheduling method and system for electric power system containing wind, light and water storage based on hierarchical optimization

Abstract

The method comprises the following steps: carrying out mathematical modeling according to operation characteristics of wind power, photovoltaic power, hydroelectric power and energy storage, and establishing a dispatching model of the power system containing wind, light and water storage and considering constraint conditions of thermal power; establishing an upper-layer optimization objective function and a lower-layer optimization objective function according to the dispatching model of the power system containing wind, light and water storage; a KAN algorithm is used to predict the total load, wind power and photovoltaic output of the power system at each time point within a future preset duration; the MPC dynamic model combines with an optimization problem which takes the minimum total load fluctuation, wind power output fluctuation and photovoltaic output fluctuation as an upper layer optimization objective function, and solves the optimal output of water, electricity and energy storage at the current time point; and taking the predicted wind power output prediction and photovoltaic output prediction at the hourly moment and the hydropower optimal output and energy storage optimal output at the hourly moment as input of a lower-layer optimization objective function, and solving the lower-layer optimization objective function by adopting an optimization algorithm to obtain an optimal thermal power output scheduling strategy of the power system containing wind, light and water storage. According to the invention, through water-storage-fire cooperative adjustment, the scheduling flexibility and the new energy consumption level of the power system are effectively improved.

Description

Technical Field

[0001] This invention relates to a power system dispatching method and system based on hierarchical optimization, encompassing wind, solar, hydro, and storage, and belongs to the field of power system operation optimization. Background Technology

[0002] The volatility and intermittency of wind and solar power output are the core bottlenecks restricting the absorption of new energy sources. Their random fluctuations can easily lead to a mismatch between the total power output on the power supply side and the demand on the load side, significantly increasing the difficulty of real-time power system dispatch. Under the principle of prioritizing new energy dispatch, if the actual output of wind and solar power still cannot meet the total load demand of the system, the existing dispatch strategy usually relies solely on the peak-shaving capacity of hydropower and the charging and discharging flexibility of energy storage systems to supplement and adjust the wind and solar power output gap, thereby maintaining the real-time power balance of the power system.

[0003] However, limited by hydropower installed capacity, runoff conditions, and the capacity and charging / discharging time of energy storage systems, this mode relying solely on water-storage joint regulation has a clear upper limit on its ability to absorb wind and solar power output, and its application effect is significantly limited during periods of high wind and solar power generation or peak load. Considering that the current power system still relies mainly on thermal power, and that the regulation capacity of a single thermal power plant is far greater than that of conventional hydropower stations and energy storage stations, possessing stronger power support capabilities, it is urgent to incorporate thermal power units into the integrated wind-solar-hydro-storage dispatch system. Summary of the Invention

[0004] This invention provides a hierarchical optimization-based power system dispatching method for wind, solar, hydro, and energy storage systems. In such power systems, the optimal output schemes for hydropower and energy storage are first solved, and then the output plans of thermal power units are optimized. Through the coordinated regulation of water, energy storage, and thermal power, the dispatching flexibility and renewable energy absorption level of the power system are effectively improved.

[0005] The technical solution of this invention is:

[0006] According to a first aspect of the present invention, a hierarchical optimization-based power system dispatching method including wind, solar, hydro, and storage is provided, comprising:

[0007] Step 1: Based on the operational characteristics of wind power, photovoltaic power, hydropower, and energy storage, perform mathematical modeling to establish a power system dispatch model incorporating wind, solar, hydro, and energy storage, and consider the constraints of thermal power; based on the power system dispatch model incorporating wind, solar, hydro, and energy storage, establish upper-level and lower-level optimization objective functions;

[0008] Step 2: Using the KAN algorithm, predict the total load of the power system, wind power, and photovoltaic power output at each time point within a preset time period;

[0009] Step 3: The MPC dynamic model combines the optimization problem with minimizing the fluctuations of total load, wind power output, and photovoltaic output as the upper-level optimization objective function, and solves the optimal output of hydropower and energy storage at the current moment.

[0010] Step 4: Using the wind power output prediction and photovoltaic power output prediction at the hour in Step 2, and the optimal hydropower output and optimal energy storage output at the hour in Step 3 as inputs to the lower-level optimization objective function, the optimization algorithm is used to solve the lower-level optimization objective function to obtain the optimal thermal power output scheduling strategy for the power system containing wind, solar, hydro, and energy storage.

[0011] Furthermore, the upper-level optimization objective function is as follows:

[0012] ;

[0013] In the formula, This represents the upper-level optimization objective, namely the total load fluctuation of the power system including wind, solar, hydro, and energy storage. Wind power output fluctuation And fluctuations in photovoltaic power output Evaluation indicators; Corresponding time point; , This indicates the weighting coefficient.

[0014] Furthermore, the lower-level optimization objective function is as follows:

[0015] ;

[0016] In the formula: F represents the economic benefits, total power generation, and power generation reliability assessment indicators of a power system including wind, solar, hydro, and energy storage. For economic gain; Revenue from wind power, solar power, hydropower, and energy storage; For thermal power grid connection price, for At any given moment, the output of N thermal power units is given; A is the cost of coal consumption for thermal power; D is the total power generation of the power system including wind, solar, hydro, and storage; and E is the power generation reliability assessment index of the power system including wind, solar, hydro, and storage. , Corresponding to the hour.

[0017] Furthermore, the expression for the total power generation of the power system including wind, solar, hydro, and storage is as follows:

[0018] ;

[0019] In the formula, The total scheduling cycle; for Wind power output at all times for Photovoltaic power output at all times; for The hydroelectric power station is generating power at all times; for The output of N thermal power units at any given time; for The charging and discharging power of the energy storage battery at all times; for The wind and rain are constantly turning electricity into power; The scheduling period is [number].

[0020] Furthermore, the expression for the power generation reliability assessment index of the power system including wind, solar, hydro, and storage is as follows:

[0021] ;

[0022] In the formula, for Total system output at any given moment; Provide overall power to the system; The number of hours during which the system's total output exceeds its guaranteed output during the scheduling period; The total scheduling cycle.

[0023] According to a second aspect of the present invention, a power system dispatching system based on hierarchical optimization, including wind, solar, hydro, and storage, is provided, comprising modules of any of the methods described above.

[0024] According to a third aspect of the present invention, a terminal device is provided, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, the processor being configured to perform the steps of the method described in any one of the preceding inventions.

[0025] The beneficial effects of this invention are as follows: This invention adopts a hierarchical optimization method for power system dispatching including wind, solar, hydro, and energy storage. The upper-level optimization incorporates the load, wind power output, and solar power output predictions obtained from KAN into the MPC for rolling optimization, thereby achieving optimal power output dispatching for hydropower and energy storage. The lower-level optimization, based on the upper-level optimization, uses an improved dung beetle algorithm to obtain the optimal power output dispatching for thermal power. Through the optimal dispatching strategies for hydropower, energy storage, and thermal power obtained above, the limitations of energy storage capacity on wind power and solar power absorption and the limitations of its application effect can be compensated to a certain extent, thereby ultimately ensuring that the economic benefits, total power generation, and power generation reliability assessment indicators of the system including wind, solar, hydro, and energy storage are maximized. Attached Figure Description

[0026] Figure 1 This is a flowchart of the joint optimization process between the upper and lower layers of this invention.

[0027] Figure 2 This is the upper-level optimization flowchart of the present invention.

[0028] Figure 3 This is the lower-level optimization flowchart of the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other.

[0030] Example 1: As Figure 1-Figure 3 As shown, according to a first aspect of the present invention, a power system dispatching method based on hierarchical optimization, including wind, solar, hydro, and storage systems, is provided, comprising the following steps:

[0031] Step 1: Based on the operational characteristics of wind power, photovoltaic power, hydropower, and energy storage, perform mathematical modeling to establish a power system dispatch model including wind, solar, hydro, and energy storage, and consider the constraints of thermal power; based on the power system dispatch model including wind, solar, hydro, and energy storage, establish the upper-level optimization objective function and the lower-level optimization objective function.

[0032] Step 2: Using the KAN algorithm, integrate historical operation records to predict the total power system load, wind power, and photovoltaic output at each time point within a preset time period. For example, the preset time period is 24 hours, and time points are constructed with a step size of 5 minutes to predict the total power system load, wind power, and photovoltaic output at each time point within the next 24 hours.

[0033] Step 3: The MPC dynamic model combines the optimization problem with minimizing the fluctuations of total load, wind power output, and photovoltaic output as the upper-level optimization objective function, and solves the optimal output of hydropower and energy storage at the current moment.

[0034] Step 4: Using the predicted wind power output and photovoltaic power output at the hourly intervals from Step 2, and the optimal hydropower output and optimal energy storage output at the hourly intervals from Step 3, as inputs to the lower-level optimization objective function, an optimization algorithm is used to solve the lower-level optimization objective function to obtain the optimal thermal power output scheduling strategy for the power system including wind, solar, hydro, and energy storage. Lower-level optimization is performed once per hour. With time The interval between them is 1 hour, time ,time These are referred to as the optimization period.

[0035] Furthermore, the power system dispatch model in step 1, which includes wind, solar, hydro, and storage systems, includes the following:

[0036] 1) Power generation model and energy storage model

[0037] 1.1) Wind power output model

[0038] The output of a wind turbine is not constant, but varies dynamically with wind speed. This relationship is often described by a piecewise mathematical function, as shown in the following formula:

[0039] ;

[0040] ;

[0041] In the formula, This indicates the rated output of the wind turbine. This represents the wind speed measured at the height of the wind turbine hub at time t; , and These represent the cut-in velocity, rated velocity, and cut-out velocity of the fan, respectively. The actual output of the wind turbine at time t; This represents the maximum possible output of the wind turbine at that moment.

[0042] 1.2) Photovoltaic power output model

[0043] Constrained by irradiance flux, thermodynamic boundary conditions, and semiconductor material properties, the power output of photovoltaic systems exhibits strong nonlinear characteristics, and its standard characterization model is shown below:

[0044] P P V , t = P S T C ⋅ G t G S T C ⋅ [ 1 + k 1 ( T t − T S T C ) ] ;

[0045] In the formula, This represents the actual output of the photovoltaic array at time t; The output under standard test conditions; and These represent actual and standard light intensities, respectively. The power temperature coefficient; The operating temperature of the photovoltaic module; This is the standard reference temperature.

[0046] 1.3) Hydraulic output model

[0047] The formula for the hydropower output model is:

[0048] ;

[0049] In the formula, η represents the actual output of the hydropower station at time t; η represents the overall efficiency of the turbine generator unit, which is generally between 0.8 and 0.95, and is taken as 0.9 in this invention; ρ is the density of water; g is the acceleration due to gravity. The flow rate used for power generation is referred to as H; H is the effective net head, which is obtained by subtracting the head loss during water conveyance from the difference in water level between upstream and downstream.

[0050] 1.4) Energy Storage Model

[0051] The time-varying dynamic characteristics of the state of charge (SOC) in a battery energy storage system are described by the following model.

[0052] ;

[0053] ;

[0054] In the formula, Let be the state of charge at time t; and These are the charging and discharging efficiencies, respectively. This refers to the actual charge and discharge power of the energy storage battery. Δt represents the battery capacity; Δt represents the time step.

[0055] 2) Constraints

[0056] Power balance constraints:

[0057] ;

[0058] In the formula: The actual output of the fan at time t; This represents the actual output of the photovoltaic array at time t; The actual output of the hydropower station at time t; Let t be the actual charge / discharge power of the energy storage battery at time t. Battery charging, Battery discharge; Let N be the actual output of N thermal power units at time t; Let t be the actual total load of the power system including wind, solar, hydro, and storage.

[0059] Wind power output constraints:

[0060] ;

[0061] In the formula: and These represent the minimum and maximum output of wind power, respectively.

[0062] Photovoltaic output constraints:

[0063] ;

[0064] In the formula: and These represent the minimum and maximum output of photovoltaic power generation, respectively.

[0065] Hydropower output constraints:

[0066] ;

[0067] In the formula: and These represent the minimum and maximum output of hydropower, respectively.

[0068] The power and state of charge (SOC) limits for energy storage batteries are as follows:

[0069] ;

[0070] ;

[0071] In the formula, , These represent the maximum and minimum charge / discharge power limits of the battery at time t, respectively. , These represent the maximum and minimum states of charge of the battery at time t, respectively. The actual stored charge of the battery at time t; Let t be the rated stored capacity of the battery at time t.

[0072] Thermal power output constraints:

[0073] ;

[0074] In the formula: and These are the minimum and maximum outputs of the i-th thermal power unit, respectively.

[0075] Thermal power plant ramping constraints:

[0076] ;

[0077] In the formula: and Let $\frac{i}{i}$ be the downward climbing rate and the upward climbing rate of the $i$-th thermal power unit, respectively.

[0078] The upper-level optimization objective function and the lower-level optimization objective function are as follows:

[0079] The specific expression for the upper-level optimization objective function is shown below:

[0080] ;

[0081] In the formula, This represents the upper-level optimization objective, namely the total load fluctuation of the power system including wind, solar, hydro, and energy storage. Wind power output fluctuation And fluctuations in photovoltaic power output Evaluation indicators; Corresponding time point, Increment by 5 minutes; , The weighting coefficient represents the relative importance of total load fluctuation, wind power fluctuation, and photovoltaic power fluctuation in the upper-level optimization objective function.

[0082] The lower-level optimization objective function is geared towards power systems including wind, solar, hydro, and energy storage. It coordinates three objectives: economic benefits, total power generation, and power generation reliability assessment indicators. It covers the on-grid electricity price revenue and coal consumption costs of thermal power, the power generation revenue of wind power / solar power / hydro power / energy storage, and integrates the power output and power generation reliability assessment indicators of power systems including wind, solar, hydro, and energy storage. The mathematical expression of the lower-level optimization objective function is as follows:

[0083] ;

[0084] In the formula: F represents the economic benefits, total power generation, and power generation reliability assessment indicators of a power system including wind, solar, hydro, and energy storage. For economic gain; Revenue from wind power, solar power, hydropower, and energy storage; For thermal power grid connection price, for At any given time, the output of N thermal power units is given; A is the cost of coal consumption for thermal power; D is the total power generation of the system; and E is the system's power generation reliability assessment index.

[0085] The expressions for the revenue generated by wind power, solar power, hydropower, and energy storage are as follows:

[0086]

[0087] In the formula: Revenue from wind power, solar power, hydropower, and energy storage; For wind power grid connection price, for Wind power output at all times , Corresponding to the exact hour (for example, h1 corresponds to 1 o'clock), (corresponding to 24:00) Increment in increments of 1 hour; For photovoltaic feed-in tariffs, for Solar power output at all times; For hydropower grid connection price, for Wind power output at any given moment; For energy storage grid connection tariff, for Wind power output at all times.

[0088] The expression for the cost of coal consumption in thermal power generation is:

[0089] ;

[0090] In the formula: A is the cost of coal consumption for thermal power generation; N is the number of all thermal power units; for The output of the i-th thermal power unit at time i; , and Let be the coal consumption characteristic coefficient of the i-th thermal power unit.

[0091] The expression for the total power generation of a power system including wind, solar, hydro, and storage:

[0092] ;

[0093] In the formula, D is the total power generation of the system; The total scheduling cycle is 24 hours. , They are respectively Real-time photovoltaic and wind power output (output power). for The hydroelectric power station is generating power at all times; for The output of N thermal power units at any given time; for The charging and discharging power of the energy storage battery at all times; for The wind and rain are constantly turning electricity into power; The scheduling cycle is 1 hour.

[0094] The expression for the power generation reliability assessment index of a power system including wind, solar, hydro, and storage is:

[0095] ;

[0096] In the formula, E is the system power generation reliability assessment index; for Total system output at any given moment; Provide overall power to the system; The number of hours during which the system's total output exceeds its guaranteed output during the scheduling period; The total scheduling cycle is 24 hours.

[0097] It should be noted that, in order to improve the power supply reliability of power systems that include new energy sources such as wind power and photovoltaics, a system power generation reliability assessment index is introduced as an objective in the lower-level optimization objective function of the power system, so as to improve the reliability of dispatching strategies.

[0098] Furthermore, KAN in step 2 is described in detail below:

[0099] KAN's innovative design is rooted in the classic Kolmogorov-Arnold representation theorem in mathematical analysis. This theorem reveals the essential structure of multivariate continuous functions, for n-variable continuous functions... It can be represented as:

[0100] ;

[0101] In the formula, It is an inner single-variable function and ϕ u , v : [ 0 , 1 ] → R ; It is an inner single-variable function and ; 2n+1 represents the number of neurons in the hidden layer.

[0102] Represented as control points, Represented as nodes, the m-th order B-spline function can be constructed using a recursive formula:

[0103] ;

[0104] ;

[0105] In the formula, ; .

[0106] The equation for the B-spline is:

[0107] ;

[0108] In the KAN algorithm, the activation function for:

[0109] ;

[0110] ;

[0111] In the formula, and Initialize using the Xavier method.

[0112] Assuming the depth of the KAN algorithm is L, it can be represented by an integer array. [ n 0 , n 1 , ⋯ , n L ] To represent its shape. Let This refers to the activation function that connects the l-th layer and the (l+1)-th layer. Specifically, The output of the l-th layer can be represented as:

[0113] x ( l + 1 ) = [ ϕ 1 , 1 , l ( ⋅ ) ϕ 1 , 2 , l ( ⋅ ) ⋯ ϕ 1 , n l , l ( ⋅ ) ϕ 2 , 1 , l ( ⋅ ) ϕ 2 , 2 , l ( ⋅ ) ⋯ ϕ 2 , n l , l ( ⋅ ) ⋮ ⋮ ⋮ ϕ n l + 1 , 1 , l ( ⋅ ) ϕ n l + 1 , 2 , l ( ⋅ ) ⋯ ϕ n l + 1 , n l , l ( ⋅ ) ] x ( l ) ;

[0114] In the formula, It is the input of the l-th layer.

[0115] Based on the above theory, the KAN algorithm is used to predict the total load of a power system including wind, solar, hydro, and energy storage for the next 24 hours (the prediction for wind power and solar power is similar), specifically including:

[0116] (1) Input layer. The historical total load sequence and its associated features (such as timestamps, temperature, humidity, holiday markers, etc.) are used as input, and the dimension value depends on the total number of selected features.

[0117] (2) Intermediate layer. It consists of multiple stacked KAN layers, each containing a learnable edge activation function. Each KAN layer is defined as a one-dimensional function matrix, which is a unary function parameterized as B-splines. At the same time, a residual connection structure is introduced to perform a layer-by-layer fine decomposition of the input signal, effectively alleviating the gradient decay problem in deep networks.

[0118] (3) Output layer. The total load forecast for the power system including wind, solar, hydro, and storage for the next 24 hours is achieved using the following KAN forecast formula:

[0119] ;

[0120] In the formula, Forecast the total load of a power system including wind, solar, hydro, and energy storage at time t within the next 24 hours. For historical data feature matrix, The output dimension of this layer is 1, which is used to directly generate point predictions of the total load of the power system at future times.

[0121] Similarly, wind power output prediction can be derived from KAN. Photovoltaic power output forecast .

[0122] Furthermore, the optimization of the power system containing wind, solar, hydro, and storage through MPC in step 3 is as follows:

[0123] MPC includes dynamic models and rolling optimization. The following example uses MPC at time k.

[0124] 1) The dynamic model of MPC at time k is characterized by the following formula.

[0125] The control input vector at time k, the state vectors at times k+1 and k+2, and the control input vector at time k+1 are predicted based on the actual value at time k.

[0126] 1.1) Predict the control input vector at time k and the state vector at time k+1.

[0127] ; x k + 1 | k = [ P W T , k + 1 | k P P V , k + 1 | k P L , k + 1 | k ] ; x k | k = [ P W T , k | k = P W T , k P P V , k | k = P P V , k P L , k | k = P L , k ] ; u k | k = [ P H , k | k P T H , k | k P S , k | k ] ;

[0128] The component form is [ P W T , k + 1 | k P P V , k + 1 | k P L , k + 1 | k ] = A 3 × 3 [ P W T , k | k P P V , k | k P L , k | k ] + B 3 × 3 [ P H , k | k P T H , k | k P S , k | k ] ;

[0129] In the formula, For example, the predicted value of the state vector at time k+1 from time k. middle This represents the predicted photovoltaic output at time k+1 based on the value at time k. This is the predicted value of the state vector at time k, for example... middle This represents the predicted photovoltaic output at time k. Since the actual value at time k is already available, no further prediction is needed. middle Equals the actual output of wind power at time k , This equals the actual power output of the photovoltaic system at time k. , Equals the actual total load of the power system including wind, solar, hydro, and storage at time k. . For example, the predicted value of the control input vector at time k is given by time k. middle This represents the predicted hydropower output at time k. Including hydropower output Thermal power output and energy storage output These three predictive outputs act as three decision variables in the upper-level optimization. A and B are system matrices; matrix A describes the self-evolutionary relationship of the states, and matrix B represents the influence of the control inputs on the states. As can be seen from the component form, Each element in is , and The function.

[0130] 1.2) Predict the control input vector at time k+1, and predict the state vector at time k+2:

[0131] ; x k + 2 | k = [ P W T , k + 2 | k P P V , k + 2 | k P L , k + 2 | k ] ; x k + 1 | k = [ P W T , k + 1 | k P P V , k + 1 | k P L , k + 1 | k ] ; u k + 1 | k = [ P H , k + 1 | k P T H , k + 1 | k P S , k + 1 | k ]

[0132] The component form is [ P W T , k + 2 | k P P V , k + 2 | k P L , k + 2 | k ] = A 3 × 3 [ P W T , k + 1 | k P P V , k + 1 | k P L , k + 1 | k ] + B 3 × 3 [ P H , k + 1 | k P T H , k + 1 | k P S , k + 1 | k ]

[0133] In the formula, For example, the predicted value of the state vector at time k+2 from time k. middle This represents the predicted photovoltaic output at time k+2 from time k. This is the predicted value of the state vector at time k+1 from time k, for example... middle This represents the predicted photovoltaic output at time k+1 based on the value at time k. For example, the predicted value of the control input vector at time k is given by time k. middle This represents the predicted hydropower output at time k+1 from time k. Including hydropower output Thermal power output and energy storage output These three predictive outputs act as three additional decision variables in the upper-level optimization. A and B are system matrices; matrix A describes the self-evolutionary relationship of the states, and matrix B represents the influence of the control inputs on the states. As can be seen from the component form, Each element in is , , and , , The function.

[0134] 2) Optimization at time k

[0135] 2.1) The more detailed expression of the upper-level optimization objective function is shown below:

[0136] ;

[0137] ;

[0138] Expanding the above equation with respect to t=k, t=k+1, t=k+2, we get:

[0139] ;

[0140] In the formula, This represents the upper-level optimization objective, namely the total load fluctuation, wind power and photovoltaic output fluctuation of the power system including wind, solar, hydro and storage at three time points starting from time k (the three time points are k, k+1 and k+2 respectively). This represents the fluctuation between the k-time predicted value obtained from the k-time MPC dynamic model and the k-time predicted value of the 24-hour prediction result obtained from KAN. This represents the fluctuation between the predicted value at time k+1 obtained from the MPC dynamic model at time k and the predicted value at time k+1 of the 24-hour prediction result obtained from KAN. This represents the fluctuation between the predicted value at time k+2 obtained from the MPC dynamic model at time k and the predicted value at time k+2 of the 24-hour prediction result obtained from KAN. The 24-hour total load forecast is obtained from KAN. , The 24-hour wind power output and solar power output forecasts are obtained from KAN.

[0141] The state vector of the dynamic model in 1) Substituting each element into the upper-level optimization objective function, the optimal control input vector is obtained by solving. u k | k ∗ = [ P H ,k | k ∗ P TH ,k | k ∗ P S ,k | k ∗ ] and u k+ 1 | k ∗ = [ P H ,k + 1 | k ∗ P TH ,k + 1 | k ∗ P S ,k + 1 | k ∗ ] Then just execute China Hydropower's optimal output and optimal energy storage output This means adjusting the hydropower output and energy storage output of the system to... and And this will continue until the next time step (i.e., time k+1) is executed, at which point the hydropower output will be optimal. and optimal energy storage output So far, other optimal solutions are ignored and do not need to be executed.

[0142] 3) Scrolling optimization.

[0143] Steps 3.1) and 2) are repeated every 5 minutes. The time when steps 3.1) and 2) are executed is called the optimization time. After time k... At time k+1, the system repeats steps 1) and 2) of step 3. It is 5 minutes. At time k, 5 minutes have passed. The system repeats the above process at time k+2. ... The optimization at multiple times constitutes rolling optimization, and k, k+1, k+2... are called optimization times.

[0144] For example, KAN and MPC constitute upper-level optimization, performing upper-level optimization on a power system containing wind, solar, hydro, and storage for a day, as follows:

[0145] A day is 24 hours, or 1440 minutes. =The time is divided into 288 time points with a step size of 5 minutes. At each time point, 00:00 AM is time point 1, 00:05 AM is time point 2, 00:10 AM is time point 3, ..., 11:55 PM is time point 288. These 288 time points are called optimization time points.

[0146] By combining historical data with KAN, the total load forecast for the 24 hours from 00:00 on the first day to 00:00 on the second day can be obtained. Wind power output forecast Photovoltaic power output forecast .For example, This is the KAN total load forecast for time 5 on the first day, which is 00:20 AM on the first day. The KAN forecast was obtained before the start of the first day.

[0147] For example, to optimize a single moment in MPC, take moment 3 of the first day, which is 00:10 AM on the first day, execute steps 1) and 2) of step 3.

[0148] S3.1: The dynamic model at time 3, which predicts the control input vector at time 3, the state vector and control input vector at time 4, and the state vector at time 5.

[0149] S3.1.1: Predicting the control input vector at time 3 for time 3, and predicting the state vector at time 4 for time 3:

[0150] ; x 4 | 3 = [ P W T , 4 | 3 P P V , 4 | 3 P L , 4 | 3 ] ; x 3 | 3 = [ P W T , 3 | 3 = P W T , 3 P P V , 3 | 3 = P P V , 3 P L , 3 | 3 = P L , 3 ] ; u 3 | 3 = [ P H , 3 | 3 P T H , 3 | 3 P S , 3 | 3 ] ;

[0151] The component form is [ P W T , 4 | 3 P P V , 4 | 3 P L , 4 | 3 ] = A 3 × 3 [ P W T , 3 | 3 P P V , 3 | 3 P L , 3 | 3 ] + B 3 × 3 [ P H , 3 | 3 P T H , 3 | 3 P S , 3 | 3 ] ;

[0152] S3.1.2: Predict the control input vector at time 3 for time 4, and predict the state vector at time 3 for time 5.

[0153] ; x 5 | 3 = [ P W T , 5 | 3 P P V , 5 | 3 P L , 5 | 3 ] ; x 4 | 3 = [ P W T , 4 | 3 P P V , 4 | 3 P L , 4 | 3 ] ; u 4 | 3 = [ P H , 4 | 3 P T H , 4 | 3 P S , 4 | 3 ] ;

[0154] The component form is [ P W T , 5 | 3 P P V , 5 | 3 P L , 5 | 3 ] = A 3 × 3 [ P W T , 4 | 3 P P V , 4 | 3 P L , 4 | 3 ] + B 3 × 3 [ P H , 4 | 3 P T H , 4 | 3 P S , 4 | 3 ] ;

[0155] S3.2: The state vector obtained in S3.1 x 3 | 3 = [ P W T , 3 | 3 = P W T , 3 P P V , 3 | 3 = P P V , 3 P L , 3 | 3 = P L , 3 ] , x 4 | 3 = [ P W T , 4 | 3 P P V , 4 | 3 P L , 4 | 3 ] , x 5 | 3 = [ P W T , 5 | 3 P P V , 5 | 3 P L , 5 | 3 ] Each element is substituted into the upper-level optimization objective function:

[0156] ;

[0157] The optimal control input vector is obtained by solving the problem. u 3 | 3 ∗ = [ P H , 3 | 3 ∗ P TH , 3 | 3 ∗ P S , 3 | 3 ∗ ] and u 4 | 3 ∗ = [ P H , 4 | 3 ∗ P TH , 4 | 3 ∗ P S , 4 | 3 ∗ ] Then just execute China Hydropower's optimal output and optimal energy storage output This means adjusting the output of hydropower and energy storage in the system to... and And this will continue until the next time step (i.e., time k+1) is executed, at which point the hydropower output will be optimal. and optimal energy storage output So far, other optimal solutions are ignored and do not need to be executed.

[0158] S3.3: Scrolling optimization.

[0159] Processes S3.1 and S3.2 are repeated every 5 minutes. The time when S3.1 and S3.2 are executed is called the optimization time. After time 3... The system repeats processes S3.1 and S3.2 at the next four time points. It is 5 minutes. At time 3... The system repeats the above process after 5 time points. The optimization at multiple time points together constitutes rolling optimization, and time points 3, 4, 5... are called optimization time points.

[0160] The optimization algorithm is the classic dung beetle algorithm and the improved dung beetle algorithm.

[0161] The classic dung beetle algorithm includes behaviors such as rolling a ball, dancing, breeding, foraging, and stealing. The algorithm's implementation process is as follows:

[0162] (1) Set hyperparameters such as population size, maximum number of iterations and upper and lower bounds of solution space, and randomly initialize dung beetle individuals in the search space;

[0163] (2) The fitness of each dung beetle individual is evaluated based on the preset objective function (i.e., the lower-level optimization objective), and the calculation method is given by the corresponding fitness formula;

[0164] (3) Simulate four types of biological behaviors of dung beetles: rolling balls, dancing, foraging, and stealing, and drive the individual's location update accordingly;

[0165] (4) Check whether the updated position exceeds the defined boundary. If it does, perform the corresponding processing.

[0166] (5) Compare the current individual fitness with the historical best value. If it is better than the current best solution, replace the global best solution; otherwise, leave it unchanged.

[0167] (6) Return to step (2) and repeat until the preset maximum number of iterations is reached. Finally, the global optimal solution and its corresponding fitness value are output.

[0168] The execution flow of the improved dung beetle algorithm (HDBO) is as follows:

[0169] (4.1) First, set hyperparameters such as population size, maximum number of iterations and upper and lower bounds of solution space, and randomly initialize dung beetle individuals in the search space;

[0170] (4.2) Use the lower-level optimization objective function to evaluate the fitness of all dung beetle individuals. By comparing their objective values ​​horizontally, identify and record the current global best individual and its corresponding objective function value.

[0171] (4.3) In the rolling ball behavior stage, the traditional fixed path mode is abandoned and a position update formula with integrated variable spiral search mechanism is adopted to dynamically adjust the individual's trajectory to enhance the exploration ability;

[0172] (4.4) After entering the foraging and reproductive behavior stage, nonlinear dynamic factors are introduced to reconstruct the boundary convergence coefficient. The behavior boundary is delineated by combining the restricted oviposition area formula and the optimal foraging domain definition formula. The egg ball position update rule and the larval foraging position evolution formula are called respectively to complete the individual position update.

[0173] (4.5) During the theft stage, the spatial distance between each theft individual and the global and local optimal solutions is first calculated. Then, the position perturbation term of standard Brownian motion and the Levy flight strategy with both long-range jump and local walk characteristics are combined to construct a hybrid position update mechanism to improve the ability to escape local extreme values.

[0174] (4.6) Before the end of each iteration, the embedded hybrid reverse learning strategy is used to reconstruct the positions of some individuals, re-evaluate their objective function values, and compare them with the existing optimal solutions. If the new solution is better, it is replaced.

[0175] (4.7) Finally, determine whether the termination condition is met (i.e., the preset maximum number of iterations or error threshold is reached). If it is met, output the optimal result; otherwise, return to step (4.2) to continue the next round of iteration.

[0176] (4.8) Output the globally optimal individual and its function value.

[0177] I. Formula for the behavior of a rolling ball:

[0178] ;

[0179] ;

[0180] In the formula, Indicates the current iteration round; The i-th dung beetle individual at time Based on the initial value of time, the first The position during the round of iteration; The sign coefficient is ±1; parameter K ∈ ( 0 , 0 . 2 ] A fixed constant to control the magnitude of directional deflection; parameters Another preset constant; Δx is used to simulate changes in light intensity, with a larger amplitude corresponding to stronger light intensity and a smaller amplitude corresponding to weaker light intensity; variables The value of reflects whether the dung beetle deviates from the original rolling ball path: when When, the direction of motion remains unchanged; when If the value is zero, it indicates that the direction has shifted. The selection of this coefficient is dynamically determined based on a probability mechanism. For the optimal individual, D refers to the distance between the optimal individual and the current individual, and l is a random number in the range [0,1]. Spiral amplitude used for dynamic adjustment This represents the maximum number of iterations. When n is 0... for There is no iteration of -1. use Approximate substitution.

[0181] II. Formula for Dancing Behavior:

[0182] ;

[0183] In the formula, Indicates the current iteration round; The i-th dung beetle individual at time... Based on the initial value of time, the first The position it is in during the round of iteration. The deflection angle characterizing the direction of movement of an individual dung beetle. i ∈ [ 0 , π ] As can be seen from this formula, an individual's position in the next iteration is highly dependent on its position in the previous moment; when The value is 0. or When this happens, its spatial coordinates will remain unchanged, that is, no displacement occurs.

[0184] III. Formula for foraging behavior:

[0185] The mathematical definition of a foraging area is as follows:

[0186] ;

[0187] The formula for updating the position of the dung beetle during foraging is shown below:

[0188] ;

[0189] In the formula, This indicates the optimal global foraging location for an individual dung beetle. and Let represent the lower and upper boundaries of the foraging area, respectively. In the formula, This indicates the maximum number of iterations. The i-th dung beetle refers to the time... Based on the initial value of time, the first The position it is in during the round of iteration. For random numbers that follow a normal distribution, express A random vector.

[0190] IV. Formula for Reproductive Behavior:

[0191] The feasible spawning areas are defined. The corresponding mathematical expression is as follows:

[0192] ;

[0193] In the formula, and Define the lower and upper bounds of the global optimization problem respectively; and This corresponds to the local lower and upper boundaries of the female dung beetle's egg-laying area; It represents the local optimal position recorded by the current individual; This indicates the preset maximum number of iterations.

[0194] The update of the oocyte position follows the formula:

[0195] ;

[0196] In the formula, The i-th dung beetle individual at time... Based on the initial value of time, the first The position it is in during the round of iteration. and Indicates two sizes of Independent random vectors, whose values ​​range from 1 to 2. D represents the dimension of the problem that needs to be solved.

[0197] V. Formula for theft:

[0198] In this stage, the corresponding position update formulas for Brownian motion and Lévy flight are introduced as follows:

[0199] ;

[0200] ;

[0201] In the formula, The i-th dung beetle individual at time... Based on the initial value of time, the first The position during the round of iteration. This represents the current global optimum. S is a random constant value, g is a random vector of size 1×D following a normal distribution, and D represents the dimension of the problem to be solved. This refers to the local optimum position of the current dung beetle individual. Mean μ = 0, variance... .

[0202] The more detailed expression of the lower-level optimization objective function is shown below:

[0203] ;

[0204] ;

[0205] Predicting wind power output at the top of the hour in step 2. Photovoltaic power output forecast And the optimal hydropower output at the top of the hour in step 3 and optimal energy storage output Replacing the lower-level objective function respectively From this, we obtain the following formula:

[0206] ;

[0207] 4) with Based on the initial value at time n, the output iteration value of the i-th thermal power unit in N thermal power units is... Let n=0 represent the initial value for iteration. Divide the N thermal power units into a 4:2:2:2 ratio into a ball-rolling group (ball-rolling behavior, dancing behavior), a breeding group, a foraging group, and a stealing group. Iterate through the formulas for ball-rolling behavior, dancing behavior, breeding behavior, foraging behavior, and stealing behavior in the improved dung beetle algorithm, respectively.

[0208] For all N thermal power units, the first iteration is performed, and the iteration values ​​of the first thermal power units are then fed into the lower-level optimization objective function to obtain the objective function value of the first iteration; then, the second iteration is performed, and the iteration values ​​of the second thermal power units are fed into the lower-level optimization objective function to obtain the objective function value of the second iteration; ... finally, the second iteration is performed, and ... Next, then the first The iteration value of thermal power is fed into the lower-level optimization objective function to obtain the value of the th iteration. The objective function value of the next iteration is compared. The iteration value that maximizes the objective function is used as the optimal output of thermal power.

[0209] For example, when N=10, time... Maximum number of iterations There are 4 thermal power units in the ball rolling group, 2 thermal power units in the breeding group, 2 thermal power units in the foraging group, and 2 thermal power units in the thieving group. That is, the 1st to 4th units are the ball rolling group, the 5th to 6th units are the breeding group, the 7th to 8th units are the foraging group, and the 9th to 10th units are the thieving group.

[0210] set up This is the initial output value of the first unit in the rolling ball group. This is the initial output value of the second unit in the rolling ball group. This is the initial output value for the third unit of the rolling ball unit. This is the initial output value for the third unit of the rolling ball assembly. This is the initial output value of the first unit in the breeding group. This is the initial output value for the second unit of the breeding group. This is the initial output value for the first unit of the foraging group. The initial output value for the second generator unit in the foraging group; This is the initial output value of the first generator unit in the theft group. The initial output value of the second generator set for the theft group; calculations are performed by substituting the formulas for the actions of different groups:

[0211] For example, the first unit in the rolling ball group. Let the initial output value of the first unit in the ball rolling group be substituted into the ball rolling behavior formula. ; The first iteration output of the first unit in the rolling ball group was obtained. Next, the output of units 2 through 10 in the first iteration is calculated and substituted into the objective function to obtain the first iteration value of the objective function. After iterating in this way for 7 times, the iteration values ​​of the objective function in the 7 iterations are compared, and the one with the largest value is the optimal output of thermal power in the lower-level objective function.

[0212] In the example above, the dung beetle algorithm is used to obtain... The feasible solution involves 10 generating units in 7 groups. Each group contains 10 outputs, and the 7 groups contain 70 outputs. The outputs of each group are directly substituted into the objective function for calculation. The group whose output maximizes the objective function is the optimal thermal power output. In other words, the above-mentioned improved dung beetle algorithm is used to solve the lower-level optimization objective function, yielding the optimal thermal power output scheduling strategy for a power system including wind, solar, hydro, and storage.

[0213] According to a second aspect of the present invention, a hierarchical optimization-based power system dispatching system incorporating wind, solar, hydro, and energy storage is provided, comprising modules of any of the methods described above. Each module in the hierarchical optimization-based power system dispatching system incorporating wind, solar, hydro, and energy storage can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0214] According to a third aspect of the present invention, a terminal device is provided, including a processor, a memory, and a computer program stored in the memory and executable on the processor, the processor being configured to perform the steps of the method described in any one of the above embodiments.

[0215] The specific embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A hierarchical optimization-based power system dispatching method including wind, solar, hydro, and storage systems, characterized in that, include: Step 1: Based on the operational characteristics of wind power, photovoltaic power, hydropower, and energy storage, perform mathematical modeling to establish a power system dispatch model incorporating wind, solar, hydro, and energy storage, and consider the constraints of thermal power; based on the power system dispatch model incorporating wind, solar, hydro, and energy storage, establish upper-level and lower-level optimization objective functions; Step 2: Using the KAN algorithm, predict the total load of the power system, wind power, and photovoltaic power output at each time point within a preset time period; Step 3: The MPC dynamic model combines the optimization problem with minimizing the fluctuations of total load, wind power output, and photovoltaic output as the upper-level optimization objective function, and solves the optimal output of hydropower and energy storage at the current moment. Step 4: Using the wind power output prediction and photovoltaic power output prediction at the hour in Step 2, and the optimal hydropower output and optimal energy storage output at the hour in Step 3 as inputs to the lower-level optimization objective function, the optimization algorithm is used to solve the lower-level optimization objective function to obtain the optimal thermal power output scheduling strategy for the power system containing wind, solar, hydro, and energy storage.

2. The power system dispatching method based on hierarchical optimization including wind, solar, hydro, and storage as described in claim 1, characterized in that, The objective function for the upper-level optimization is shown below: ； In the formula, This represents the upper-level optimization objective, namely the total load fluctuation of the power system including wind, solar, hydro, and energy storage. Wind power output fluctuation And fluctuations in photovoltaic power output Evaluation indicators; Corresponding time point; , This indicates the weighting coefficient.

3. The power system dispatching method based on hierarchical optimization including wind, solar, hydro, and storage as described in claim 1, characterized in that, The lower-level optimization objective function is as follows: ； In the formula, These are indicators for evaluating the economic benefits, total power generation, and power generation reliability of power systems that include wind, solar, hydro, and energy storage. For economic gain; Revenue from wind power, solar power, hydropower, and energy storage; For thermal power grid connection price, for The output of N thermal power units at any given time; D represents the cost of coal consumption for thermal power generation; E represents the total power generation of the power system including wind, solar, hydro, and storage; and E represents the reliability assessment index of the power generation of the power system including wind, solar, hydro, and storage. Corresponding to the hour, This represents the total scheduling cycle and the corresponding hourly time point.

4. The power system dispatching method based on hierarchical optimization including wind, solar, hydro, and storage as described in claim 3, characterized in that, The expression for the total power generation of the power system including wind, solar, hydro, and storage is as follows: ； In the formula, for Wind power output at all times for Photovoltaic power output at all times; for The hydroelectric power station is generating power at all times; for At any given moment, the output of N thermal power units; for The charging and discharging power of the energy storage battery at all times; for The wind and rain are constantly turning electricity into power; The scheduling period is [number].

5. The power system dispatching method based on hierarchical optimization including wind, solar, hydro, and storage according to claim 3, characterized in that, The expression for the power generation reliability assessment index of the power system including wind, solar, hydro, and storage is as follows: ； In the formula, for Total system output at any given moment; Provide overall power to the system; This refers to the number of hours during which the system's total output exceeds its guaranteed output during the scheduling period.

6. A power system dispatching system based on hierarchical optimization, incorporating wind, solar, hydro, and storage, characterized in that, The module includes the method described in any one of claims 1-5.

7. A terminal device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that, The processor is configured to perform the steps of the method according to any one of claims 1-5.

Images