Power distribution network power balance control method, system, equipment and medium
By combining the improved particle swarm optimization algorithm and support vector machine with the gray wolf optimization algorithm, an output prediction model is constructed, and the power optimization model for distributed resource access to the distribution network is optimized. This solves the problem of distribution network instability caused by distributed generation resources, and achieves the goals of minimizing network loss, minimizing node voltage deviation, and minimizing operating costs, thereby improving power supply quality and economy.
Patent Information
- Application Number
- CN202310583557.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-23
- Publication Date
- 2024-11-26
AI Technical Summary
Existing technologies cannot effectively solve the problems caused by distributed generation resources in the distribution network, such as voltage levels, short-circuit capacity, relay protection, power output and power supply stability, leading to unstable operation of the distribution network and a decline in power quality.
An improved particle swarm optimization algorithm and support vector machine combined with the gray wolf optimization algorithm are used to construct a power output prediction model. The power optimization model for distributed resource access to the distribution network is optimized by combining the analytic hierarchy process and chaotic mapping initialization, so as to achieve the goals of minimizing network loss, minimizing node voltage deviation and minimizing operating cost.
It improves the stability of power output and power quality of the distribution network, ensures the economic efficiency of system operation, and realizes power coordination optimization with a high proportion of distributed resources accessing the distribution network.
Smart Images

Figure CN119030031A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of power distribution network voltage control, and in particular relates to a power distribution network power balance control method, system, equipment and medium. Background Art
[0002] At present, with the continuous advancement of smart grid construction, the high proportion of distributed generation resources in the power grid is bound to become the future development trend. However, distributed generation resources are intermittent and volatile, which have a great impact on voltage level, short-circuit capacity, relay protection, power output, power supply stability and power quality. In addition, due to the influence of distribution automation and dispatching methods, the distribution network's absorption capacity for distributed generation is severely constrained.
[0003] In order to solve the above problems, scholars have conducted relevant research around the coordinated operation of distribution networks and accumulated some results. Some studies have proposed a dispatch optimization strategy for intelligent distribution networks based on a multi-interaction mode; based on a hierarchical dispatching system architecture, a cross-regional power grid source-grid-load coordinated dispatching mode has been developed; and some literature has proposed a two-stage optimization method of open points and energy storage systems based on interval optimization to improve the operational economy of active distribution systems. However, the results have not achieved the expected results, and the safety and stability of the distribution network cannot be guaranteed. It can be seen that the current power coordination optimization of distribution network operation still faces a series of problems. Summary of the invention
[0004] In order to overcome the above-mentioned deficiencies of the prior art, the present invention proposes a power balance control method for a distribution network, comprising:
[0005] Obtain the output power forecast value of the distributed resources connected to the distribution network;
[0006] Based on the output power prediction value, an improved particle swarm algorithm is used to solve the pre-built power optimization model to obtain a power coordination optimization solution for the distribution network;
[0007] The power optimization model is constructed with the goal of minimizing network loss, minimizing network node voltage deviation, and minimizing distribution network operation cost.
[0008] Preferably, the power optimization model includes the following construction process:
[0009] The objective function is constructed with the goal of minimizing network loss, minimizing voltage deviation at network nodes, and minimizing the operating cost of the distribution network;
[0010] According to the objective function, corresponding constraint conditions are formulated;
[0011] Based on the objective function and the constraint conditions, a power optimization model is obtained.
[0012] Preferably, the constraints include at least one or more of the following: flow-related constraints, output regulation constraints of distributed resources and equal reserve rate constraints; the flow-related constraints include at least one or more of the following: flow equation constraints, node voltage amplitude inequality constraints and branch thermal constraints.
[0013] Preferably, the expression corresponding to the objective function is as follows:
[0014] min F=μ1f1+μ2f2+μ3f3
[0015] Among them, F represents the objective function; f1 is the first objective corresponding to network loss; μ1 is the weight of the first objective f1; f2 is the second objective corresponding to network node voltage deviation; μ2 is the weight of the second objective f2; f3 is the third objective corresponding to the operating cost of the distribution network; μ3 is the weight of the third objective f3.
[0016] Preferably, the expression corresponding to the first target f1 is as follows:
[0017]
[0018] The expression corresponding to the second target f2 is as follows:
[0019] f2={max(1-v i )}
[0020] The expression corresponding to the third target f3 is as follows:
[0021] f3=max(E-Cost)
[0022] in, represents the active power flowing through branch i; represents the reactive power flowing through branch i; represents the node voltage of the terminal node of branch i; R i represents the resistance of branch i; I represents the number of branches; i=1,…,I; v i represents the node voltage of branch i in the distribution network; E represents the economic benefit of distributed resource power generation; Cost represents the maintenance cost of distributed resource power generation.
[0023] Preferably, the improved particle swarm algorithm is used to solve the pre-built power optimization model to obtain a power coordination optimization scheme for the distribution network, including:
[0024] Step S1: using the output power prediction value as an input item of the power optimization model;
[0025] Step S2: randomly generate a particle swarm, use chaotic mapping to initialize the position and velocity of the particle swarm, initialize the inertia weight and learning factor; use the objective function to represent the fitness function of the particle; and use the power coordination parameters of the distributed resources connected to the distribution network as the optimization variables of the particles; wherein the optimization variables include at least one or more of the following: distributed power output, generator terminal voltage, node voltage amplitude;
[0026] Step S3: Calculate the fitness function values of all particles, and select the particle with the largest fitness function value as the global optimal solution;
[0027] Step S4: Determine whether the global optimal solution converges. If so, determine that the optimization variable corresponding to the global optimal solution is the power coordination optimization scheme of the distribution network, and use the power coordination scheme of the distribution network as the output item of the power optimization model, and stop executing; if not, iteratively update the position and speed of the particle, and execute step S5;
[0028] Step S5: updating the inertia weight and the learning factor based on the preset inertia weight and the learning factor update expression;
[0029] Step S6: Based on the updated inertia weight, learning factor and the fitness function value of the updated particle, the particle velocity and position adaptive update strategy is adopted to update the global optimal solution, and step S4 is executed.
[0030] Preferably, the inertia weight and learning factor update expressions are as follows:
[0031]
[0032] Where w represents the inertia weight; w max Indicates the maximum value of the preset inertia weight; w min Indicates the minimum value of the preset inertia weight; k indicates the current number of iterations; represents the fitness function value corresponding to the global optimal solution at the kth iteration; A represents the associated parameter; MaxIter represents the maximum number of iterations; c1 represents the first learning factor; c2 represents the second learning factor.
[0033] Preferably, the output power prediction value of the distributed resource accessing the power distribution network includes the following acquisition process:
[0034] Obtaining power impact data of distributed resources connected to the distribution network;
[0035] Based on the power impact data, using a pre-built output prediction model, an output power prediction value corresponding to the power impact data is obtained;
[0036] The output prediction model is constructed based on the support vector machine and the grey wolf optimization algorithm, using the historical power impact data and the corresponding actual output power values as training data.
[0037] Preferably, the output prediction model includes the following construction process:
[0038] Selecting a preset group from historical power impact data as input of training data;
[0039] The actual value of the output power corresponding to the selected power impact data is used as the output of the training data;
[0040] Based on the input and output of the training data, a preset support vector machine is trained to obtain an output vector;
[0041] Determining a parameter to be optimized according to the output vector;
[0042] The parameters to be optimized are optimized using the Gray Wolf Optimization Algorithm, the optimal parameters are determined, and an output prediction model is obtained;
[0043] The parameters to be optimized include at least one or more of the following: regularization parameters and kernel function parameters.
[0044] Based on the same inventive concept, the present invention also provides a power distribution network power balance control system, comprising:
[0045] Prediction value acquisition module: used to obtain the output power prediction value of the distributed resources connected to the distribution network;
[0046] Power coordination optimization module: used to solve the pre-built power optimization model based on the output power prediction value by using an improved particle swarm algorithm to obtain a power coordination optimization solution for the distribution network;
[0047] The power optimization model in the power coordination optimization module is constructed with the goal of minimizing network loss, minimizing network node voltage deviation, and minimizing distribution network operation cost.
[0048] Preferably, the power optimization model in the power coordination optimization module includes the following construction process:
[0049] The objective function is constructed with the goal of minimizing network loss, minimizing voltage deviation at network nodes, and minimizing the operating cost of the distribution network;
[0050] According to the objective function, corresponding constraint conditions are formulated;
[0051] Based on the objective function and the constraint conditions, a power optimization model is obtained.
[0052] Preferably, the constraints in the power coordination optimization module include at least one or more of the following: flow-related constraints, output regulation constraints of distributed resources and equal reserve rate constraints; the flow-related constraints include at least one or more of the following: flow equation constraints, node voltage amplitude inequality constraints and branch thermal constraints.
[0053] Preferably, the expression corresponding to the objective function in the power coordination optimization module is as follows:
[0054] min F=μ1f1+μ2f2+μ3f3
[0055] Among them, F represents the objective function; f1 is the first objective corresponding to network loss; μ1 is the weight of the first objective f1; f2 is the second objective corresponding to network node voltage deviation; μ2 is the weight of the second objective f2; f3 is the third objective corresponding to the operating cost of the distribution network; μ3 is the weight of the third objective f3.
[0056] Preferably, the expression corresponding to the first objective f1 in the power coordination optimization module is as follows:
[0057]
[0058] The expression corresponding to the second target f2 is as follows:
[0059] f2={max(1-v i )}
[0060] The expression corresponding to the third target f3 is as follows:
[0061] f3=max(E-Cost)
[0062] in, represents the active power flowing through branch i; represents the reactive power flowing through branch i; represents the node voltage of the terminal node of branch i; R i represents the resistance of branch i; I represents the number of branches; i=1,…,I; v i represents the node voltage of branch i in the distribution network; E represents the economic benefit of distributed resource power generation; Cost represents the maintenance cost of distributed resource power generation.
[0063] Preferably, the power coordination optimization module adopts an improved particle swarm algorithm to solve the pre-built power optimization model to obtain a power coordination optimization solution for the distribution network, including:
[0064] Step S1: using the output power prediction value as an input item of the power optimization model;
[0065] Step S2: randomly generate a particle swarm, use chaotic mapping to initialize the position and velocity of the particle swarm, initialize the inertia weight and learning factor; use the objective function to represent the fitness function of the particle; and use the power coordination parameters of the distributed resources connected to the distribution network as the optimization variables of the particles; wherein the optimization variables include at least one or more of the following: distributed power output, generator terminal voltage, node voltage amplitude;
[0066] Step S3: Calculate the fitness function values of all particles, and select the particle with the largest fitness function value as the global optimal solution;
[0067] Step S4: Determine whether the global optimal solution converges. If so, determine that the optimization variable corresponding to the global optimal solution is the power coordination optimization scheme of the distribution network, and use the power coordination scheme of the distribution network as the output item of the power optimization model, and stop executing; if not, iteratively update the position and speed of the particle, and execute step S5;
[0068] Step S5: updating the inertia weight and the learning factor based on the preset inertia weight and the learning factor update expression;
[0069] Step S6: Based on the updated inertia weight, learning factor and the fitness function value of the updated particle, the particle velocity and position adaptive update strategy is adopted to update the global optimal solution, and step S4 is executed.
[0070] Preferably, the inertia weight and learning factor update expressions in the power coordination optimization module are as follows:
[0071]
[0072] Where w represents the inertia weight; w max Indicates the maximum value of the preset inertia weight; w min Indicates the minimum value of the preset inertia weight; k indicates the current number of iterations; represents the fitness function value corresponding to the global optimal solution at the kth iteration; A represents the associated parameter; MaxIter represents the maximum number of iterations; c1 represents the first learning factor; c2 represents the second learning factor.
[0073] Preferably, the output power prediction value of the distributed resources connected to the power distribution network in the prediction value acquisition module includes the following acquisition process:
[0074] Obtaining power impact data of distributed resources connected to the distribution network;
[0075] Based on the power impact data, using a pre-built output prediction model, an output power prediction value corresponding to the power impact data is obtained;
[0076] The output prediction model is constructed based on the support vector machine and the grey wolf optimization algorithm, using the historical power impact data and the corresponding actual output power values as training data.
[0077] Preferably, the output prediction model in the prediction value acquisition module includes the following construction process:
[0078] Selecting a preset group from historical power impact data as input of training data;
[0079] The actual value of the output power corresponding to the selected power impact data is used as the output of the training data;
[0080] Based on the input and output of the training data, a preset support vector machine is trained to obtain an output vector;
[0081] Determining a parameter to be optimized according to the output vector;
[0082] The parameters to be optimized are optimized using the Gray Wolf Optimization Algorithm, the optimal parameters are determined, and an output prediction model is obtained;
[0083] The parameters to be optimized include at least one or more of the following: regularization parameters and kernel function parameters.
[0084] Based on the same inventive concept, the present invention also provides a computer device, including: one or more processors;
[0085] A memory for storing one or more programs;
[0086] When the one or more programs are executed by the one or more processors, a power distribution network power balance control method as described above is implemented.
[0087] Based on the same inventive concept, the present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed, a power balance control method for a distribution network as described above is implemented.
[0088] Compared with the closest prior art, the present invention has the following beneficial effects:
[0089] The present invention provides a power balance control method, system, device and medium for a distribution network, comprising: obtaining an output power prediction value of distributed resources connected to a distribution network; based on the output power prediction value, using an improved particle swarm algorithm to solve a pre-constructed power optimization model to obtain a power coordination optimization scheme for the distribution network; wherein the power optimization model is constructed with the goal of minimizing network loss, minimizing network node voltage deviation and minimizing distribution network operation cost; the present invention pre-establishes a power optimization model with the goal of minimizing network loss, minimizing network node voltage deviation and minimizing distribution network operation cost, which is beneficial to improving the stability of power output of the distribution network, thereby ensuring the power supply quality and the economy of system operation; by using an improved particle swarm algorithm to solve the power optimization model, it is possible to solve multi-objective mixed problems, which is beneficial to improving the solution efficiency of the model, thereby realizing coordinated optimization of power when a high proportion of distributed resources are connected to the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS
[0090] Figure 1 A schematic flow chart of a power distribution network power balance control method provided by the present invention;
[0091] Figure 2 An algorithm flow chart of an output prediction model in a power distribution network power balance control method provided by the present invention;
[0092] Figure 3 A schematic diagram of the overall framework of a power distribution network power balance control method provided by the present invention;
[0093] Figure 4 A schematic diagram of the structural composition of a power distribution network power balance control system provided by the present invention. DETAILED DESCRIPTION
[0094] The specific implementation modes of the present invention are further described in detail below with reference to the accompanying drawings.
[0095] Embodiment 1:
[0096] The present invention provides a power distribution network power balance control method, the flow chart is as follows Figure 1 As shown, including:
[0097] Step 1: Obtain the output power prediction value of the distributed resources connected to the distribution network;
[0098] Step 2: Based on the output power prediction value, an improved particle swarm algorithm is used to solve the pre-built power optimization model to obtain a power coordination optimization solution for the distribution network;
[0099] The power optimization model is constructed with the goal of minimizing network loss, minimizing network node voltage deviation, and minimizing distribution network operation cost.
[0100] Specifically, the output power prediction value of the distributed resource connected to the distribution network in step 1 includes the following acquisition process:
[0101] Obtaining power impact data of distributed resources connected to the distribution network;
[0102] Based on the power impact data, using a pre-built output prediction model, an output power prediction value corresponding to the power impact data is obtained;
[0103] The output prediction model is constructed based on the support vector machine and the gray wolf optimization algorithm, using the historical power impact data and the corresponding actual output power value as training data;
[0104] The output prediction model includes the following construction process:
[0105] Selecting a preset group from historical power impact data as input of training data;
[0106] The actual value of the output power corresponding to the selected power impact data is used as the output of the training data;
[0107] Based on the input and output of the training data, a preset support vector machine is trained to obtain an output vector;
[0108] Determining a parameter to be optimized according to the output vector;
[0109] The parameters to be optimized are optimized using the Gray Wolf Optimization Algorithm, the optimal parameters are determined, and an output prediction model is obtained;
[0110] The parameters to be optimized include at least one or more of the following: regularization parameters and kernel function parameters.
[0111] Since distributed resources are intermittent energy sources with randomness and volatility, it is necessary to establish a prediction model for the output power of distributed resources connected to the distribution network. The present invention adopts the supervised learning method SVM support vector machine to achieve accurate power prediction. In order to improve the performance of the model, the radial basis function (RBF) kernel is adopted, and a meta-heuristic method is used, that is, based on the Grey Wolf Optimization (GWO) algorithm to find the optimal values of the parameters involved.
[0112] Support Vector Machine (SVM) is one of the most popular and widely used machine learning techniques for classification and regression. However, in practical applications, SVM is difficult to perform at its best due to the complex parameter selection problem. The present invention selects the optimal parameters of the SVM support vector machine by applying the GWO gray wolf optimization algorithm to obtain the best prediction accuracy. There are two types of parameters in the SVM support vector machine, including regularization parameters and kernel function parameters, and the values of these parameters can directly affect the performance of the SVM support vector machine. The comparison results of the GWO gray wolf optimization algorithm with some of the most famous evolutionary algorithms such as particle swarm optimization and genetic algorithms show that the GWO gray wolf optimization algorithm can provide very competitive results in terms of improved local optimal avoidance. Therefore, the present invention selects the GWO gray wolf optimization algorithm to solve the SVM support vector machine parameter optimization problem. The specific parameter optimization process is as follows:
[0113] 1. Establish a basic output prediction model and determine the parameters to be optimized.
[0114] The present invention uses SVM support vector machine to complete the prediction task, which is specifically defined as follows:
[0115]
[0116] Where f represents the prediction function; W is the weight vector perpendicular to the hyperplane; ξ represents the control classification error tolerance; ξ m represents the classification error tolerance of the mth training set; b represents the offset; l represents the number of training sets, m = 1,…,i; C represents the regularization parameter; X m represents the input vector of the mth training set; Y m Represents the output vector of the mth training set;
[0117] In the present invention, the RBF radial basis function is selected as the kernel function of the SVM support vector machine, and the corresponding expression is as follows:
[0118]
[0119] Where J represents the radial basis function; X m represents the input vector of the mth training set; Y m Represents the output vector of the mth training set; the RBF radial basis function involves the kernel function parameter σ, whose value varies between 1 and 100.
[0120] The data required for model training may be the power generation data and meteorological information of a specific photovoltaic power station over a period of time. The data samples should include cloud cover information corresponding to the time point, daily maximum temperature, daily minimum temperature, maximum humidity, minimum humidity and other environmental related characteristics, as well as irradiance, photovoltaic array area and photoelectric conversion efficiency, etc., and also include the corresponding power output variables at the corresponding time point; in addition, the present invention chooses to use the mean absolute percentage error and root mean square error to measure the model prediction performance.
[0121] 2. Use the GWO Grey Wolf Optimization algorithm to assist the deep learning training process and improve its prediction accuracy.
[0122] The present invention adopts the GWO gray wolf optimization algorithm to optimize the regularization parameter C and the kernel function parameter σ in the SVM support vector machine; the GWO gray wolf optimization algorithm is a new meta-heuristic technology; the method can be used to solve the optimization problem, and the specific process of the gray wolf optimization algorithm includes:
[0123] The algorithm mimics the leadership hierarchy and hunting mechanism of gray wolves. To simulate the leadership hierarchy, there are four types of gray wolves, namely γ, η, λ and τ; these four types can be used to simulate the leadership hierarchy, and hunting is optimization, guided by three wolves (γ, η and λ).
[0124] It is well known that grey wolves surround their prey during hunting; mathematically, this can be expressed as follows:
[0125]
[0126]
[0127] in, represents the distance vector between the gray wolf and the prey; t represents the current iteration number; Represents the position vector of the gray wolf at the tth iteration; represents the position vector of the prey at the tth iteration; and is the coefficient vector;
[0128] Among them, the coefficient vector and The corresponding calculation formula is as follows:
[0129]
[0130]
[0131] in, represents the convergence factor, the coefficient vector The value of Shadow, The component of decreases linearly from 2 to 0 during the iteration process; is a random vector in [0,1]; is the coefficient vector.
[0132] To simulate the hunting process of the gray wolf, it is assumed that γ (the best candidate), η, and λ have a better understanding of the possible location of the prey; therefore, the three best solutions obtained so far are saved and other search agents (including τ) are forced to update their positions based on the position of the best search agent. To update the position of the gray wolf, the following formula is used:
[0133]
[0134]
[0135]
[0136] in, represents the distance vector between the gray wolf population τ and the gray wolf population γ; and represents a random vector; represents the position vector of the gray wolf γ; Represents the current position vector of the gray wolf; represents the distance vector between the gray wolf population τ and the gray wolf population η; represents the position vector of the gray wolf η; represents the distance vector between the gray wolf population τ and the gray wolf population λ; represents the position vector of the gray wolf λ; It indicates the position vector that the gray wolf τ needs to adjust due to the influence of the gray wolf γ; and represents the coefficient vector; It represents the position vector that the gray wolf τ needs to adjust due to the influence of the gray wolf η; It indicates the position vector that the gray wolf τ needs to adjust due to the influence of the gray wolf λ; represents the position vector of the gray wolf at the t+1th iteration, and also represents the final position of the gray wolf τ;
[0137] like Figure 2The figure shows the implementation flow chart of the output prediction model used in the present invention, whose input is the feature vector of the training data set and the corresponding label, the GWO initialization parameter, and the output is the optimal support vector machine parameter and prediction result in the test data set; the present invention uses the regularization parameter C and the kernel function parameter σ of the SVM support vector machine as the position parameter of the individual wolf in the GWO gray wolf optimization algorithm, and uses the prediction result obtained by the initial SVM support vector machine to calculate the initial values of γ, η and λ in the GWO gray wolf optimization algorithm through the fitness function, and iteratively updates the position of τ according to the values of the first three until the termination condition is met, and uses the corresponding value of the position of τ as the final optimization result of the two parameters of the SVM support vector machine, thereby realizing the accurate prediction of the photovoltaic output power. The gray wolf optimization algorithm is used to optimize the parameters to be optimized and determine the optimal parameters, which specifically includes:
[0138] Step S11: randomly generate a gray wolf population, initialize the gray wolf population, the maximum number of iterations, the convergence factor and the coefficient vector; represent the position of the individuals in the gray wolf population as the parameter to be optimized; and set the fitness function according to the average error between the predicted output and the expected output of the output prediction model;
[0139] Step S12: Calculate the fitness function value of each individual gray wolf, take the top three gray wolves (γ, η and λ) with the best fitness function values as the optimal solution set, and take the remaining gray wolf population τ as the candidate solution set;
[0140] Step S13: updating the position of the individual gray wolf in the optimal solution set;
[0141] Step S14: updating the position of the gray wolf population in the candidate solution set according to the position of the gray wolf individual in the optimal solution set;
[0142] Step S15: updating the convergence factor and coefficient vector, and calculating the fitness function value of each individual gray wolf;
[0143] Step S16: Update the optimal solution set;
[0144] Step S17: Determine whether the maximum number of iterations has been reached. If so, take the position of the gray wolf individual in the optimal solution set as the optimal parameter and end; if not, execute step S14.
[0145] The power optimization model in step 2 includes the following construction process:
[0146] The objective function is constructed with the goal of minimizing network loss, minimizing voltage deviation at network nodes, and minimizing the operating cost of the distribution network;
[0147] According to the objective function, corresponding constraint conditions are formulated;
[0148] Based on the objective function and the constraint conditions, a power optimization model is obtained;
[0149] The constraint conditions include at least one or more of the following: power flow related constraints, output regulation constraints of distributed resources and equal reserve rate constraints; the power flow related constraints include at least one or more of the following: power flow equation constraints, node voltage amplitude inequality constraints and branch thermal constraints;
[0150] The expression corresponding to the objective function is as follows:
[0151] minF=μ1f1+μ2f2+μ3f3
[0152] Wherein, F represents the objective function; f1 is the first objective corresponding to network loss; μ1 is the weight of the first objective f1; f2 is the second objective corresponding to network node voltage deviation; μ2 is the weight of the second objective f2; f3 is the third objective corresponding to the operating cost of the distribution network; μ3 is the weight of the third objective f3;
[0153] The expression corresponding to the first target f1 is as follows:
[0154]
[0155] The expression corresponding to the second target f2 is as follows:
[0156] f2={max(1-v i )}
[0157] The expression corresponding to the third target f3 is as follows:
[0158] f3=max(E-Cost)
[0159] in, represents the active power flowing through branch i; represents the reactive power flowing through branch i; represents the node voltage of the terminal node of branch i; R i represents the resistance of branch i; I represents the number of branches; i=1,…,I; v i represents the node voltage of branch i in the distribution network; E represents the economic benefit of distributed resource power generation; Cost represents the maintenance cost of distributed resource power generation.
[0160] The present invention determines the objective function by considering network loss, voltage deviation and operating cost, adds equal reserve rate constraint to coordination constraint condition, and transforms multi-objective problem solving into single-objective problem solving with the help of hierarchical analysis. The specific steps include: 1. Determine the optimization objective function to ensure the economic benefits and power quality of system operation.
[0161] First, ensure that network losses are minimized.
[0162] Specifically, a function is established to describe how to minimize the network loss of the distribution network. The expression is as follows:
[0163]
[0164] Where f1 represents the first target; P loss Represents network loss; R i represents the resistance of branch i; represents the active power flowing through branch i; represents the reactive power flowing through branch i; represents the node voltage of the terminal node of branch i; I represents the number of branches; i=1,…,I.
[0165] Secondly, the voltage deviation of the distribution network is minimized, and the expression is as follows:
[0166] f2=minΔv i =min{max(1-v i )}
[0167] Where f2 represents the second target; v i represents the node voltage of branch i; Δv i Represents the node voltage deviation in branch i. Finally, an expression is established to describe the minimization of the distribution network operation cost. The corresponding expression is as follows:
[0168] f3=max(E-Cost)
[0169] In the formula,
[0170]
[0171]
[0172] Where f3 represents the third goal; E represents the economic benefits of distributed resource power generation; Cost represents the maintenance cost of distributed resource power generation; α i represents the distributed resource generation revenue coefficient in branch i; β i represents the cost coefficient of maintenance and operation of distributed resources in branch i; P1 is the output value of distributed resources; I represents the number of branches; i=1,…,I.
[0173] 2. Determine the relevant constraints of the optimization problem involved in the present invention, including power flow related constraints and output regulation constraints of distributed resources.
[0174] First, determine the constraints related to the power flow. Specifically, the expression corresponding to the power flow equation constraint is as follows:
[0175]
[0176] in, represents the active power of distributed resources in branch i; They represent the reactive power of distributed resources in branch i respectively; represents the active power of load in branch i; represents the reactive power of the load in branch i; U i Represents the voltage of branch i; U j Represents the voltage of branch j, where j is another branch in the total branch I except branch i, j = 1,…,I; G ij represents the real part of the node admittance matrix of branch ij; B ij represents the imaginary part of the node admittance matrix in branch ij; θ ij Represents the phase angle value of branch ij.
[0177] The expression corresponding to the node voltage amplitude inequality constraint is as follows:
[0178] U i,min ≤U i ≤U i,max
[0179] Among them, U i The voltage of branch i; U i,min Indicates the minimum voltage in branch i; U i,max represents the maximum voltage in branch i;
[0180] The expression corresponding to the branch thermal constraint is as follows:
[0181]
[0182]
[0183] in, represents the apparent power flowing through branch h; represents the active power flowing through branch h; represents the reactive power flowing through branch h; Indicates the upper limit of the apparent power flowing through branch h; U i Represents the voltage of branch i; U j represents the voltage of branch j, where j is another branch in the total branch I except branch i; G ij represents the real part of the node admittance matrix of branch ij; B ij represents the imaginary part of the node admittance matrix in branch ij; θ ij Represents the phase angle value of branch ij.
[0184] Secondly, consider the output regulation constraints of distributed resources. In order to make full use of the active power generation of distributed resources, both distributed and centralized distributed resources operate normally at rated power; the regulation of the active power output of distributed resources is maintained or reduced, and the reactive power output is zero, because the default assumption is that the distributed resource system operates near the unity power factor. The expression corresponding to the output regulation constraint is as follows:
[0185]
[0186]
[0187] in, represents the active power output by the distributed resources in branch i; represents the minimum active power output by the distributed resources in branch i; represents the maximum active power output by the distributed resources in branch i; Represents the reactive power output by the distributed resources in branch i.
[0188] Inverters can be controlled to provide or absorb reactive power. In addition, for reactive power control of distributed resources, a specific power factor can be provided for reactive power through power converters according to the adjustment scheme of the control / dispatch center.
[0189] In the high-ratio distributed resource integration scenario, in order to prevent the compression ratio of a single distributed resource from exceeding its own regulation capability, an equal reserve ratio constraint is proposed. The reserve ratio of a distributed resource refers to the ratio of the actual active power reduction value to the maximum active power value that can be generated; the equal reserve ratio can be used to ensure that the reserve ratio of each distributed resource maintains a similar ratio, and its mathematical definition is:
[0190]
[0191] in, is the active output reduction value of the distributed resources of branch i after optimization; ε represents the maximum reserve rate value of the distributed resources; r QCR Indicates the set value of equal reserve ratio; Represents the maximum active power output by the distributed resources in branch i.
[0192] 3. Normalize the multi-objective function and take the harmony of the three objectives as the final objective function.
[0193] The present invention takes the reconciliation of the three objectives of minimizing the network loss of the distribution network, minimizing the network voltage deviation rate, and minimizing the network operation cost of the distribution network as the final objective function. In order to achieve the optimal regulation of distributed resources, it is necessary to determine the weights of the three objectives separately, and finally reconcile and unify them to achieve a successful transformation from a multi-objective optimization problem to a single-objective optimization problem.
[0194] The present invention uses the hierarchical analysis method to judge the importance of the three goals, using Z uv Denotes the objective function f u For the objective function f v The degree of effect, where u and v represent different numbers of the objective function, the corresponding expression is as follows:
[0195]
[0196] Then, according to the obtained element values, construct the corresponding judgment matrix A:
[0197]
[0198] Among them, n in matrix A represents the number of objective functions to be optimized. Through this matrix, the value Z of the diagonal in matrix A can be calculated uu Calculate the proportion of each target, u = 1...n, Z uu The corresponding geometric mean π u The calculation formula is as follows:
[0199]
[0200] The final weight of each goal is calculated as follows:
[0201]
[0202] Where v represents the number of the objective function; n represents the number of objective functions; Z uu represents the values on the diagonal of matrix A; μ u Represents Z uu The corresponding weight; π u Represents Z uu The corresponding geometric mean; π v represents the geometric mean of the vth objective function, v = 1…n.
[0203] Finally, the single objective function expression is obtained, which is as follows:
[0204] min F=μ1f1+μ2f2+μ3f3
[0205] Wherein, F represents the objective function; f1 is the first objective corresponding to network loss; μ1 is the weight of the first objective f1; f2 is the second objective corresponding to network node voltage deviation; μ2 is the weight of the second objective f2; f3 is the third objective corresponding to the operating cost of the distribution network; μ3 is the weight of the third objective f3;
[0206] The improved particle swarm algorithm is used to solve the pre-built power optimization model to obtain the power coordination optimization scheme of the distribution network, including:
[0207] Step S1: using the output power prediction value as an input item of the power optimization model;
[0208] Step S2: randomly generate a particle swarm, use chaotic mapping to initialize the position and velocity of the particle swarm, initialize the inertia weight and learning factor; use the objective function to represent the fitness function of the particle; and use the power coordination parameters of the distributed resources connected to the distribution network as the optimization variables of the particles; wherein the optimization variables include at least one or more of the following: distributed power output, generator terminal voltage, node voltage amplitude;
[0209] Step S3: Calculate the fitness function values of all particles, and select the particle with the largest fitness function value as the global optimal solution;
[0210] Step S4: Determine whether the global optimal solution converges. If so, determine that the optimization variable corresponding to the global optimal solution is the power coordination optimization scheme of the distribution network, and use the power coordination scheme of the distribution network as the output item of the power optimization model, and stop executing; if not, iteratively update the position and speed of the particle, and execute step S5;
[0211] Step S5: updating the inertia weight and the learning factor based on the preset inertia weight and the learning factor update expression;
[0212] Step S6: Based on the updated inertia weight, learning factor and the fitness function value of the updated particle, the particle velocity and position adaptive update strategy is adopted to update the global optimal solution, and step S4 is executed.
[0213] The inertia weight and learning factor update expressions are as follows:
[0214]
[0215] Where w represents the inertia weight; w max Indicates the maximum value of the preset inertia weight; w min Indicates the minimum value of the preset inertia weight; k indicates the current number of iterations; represents the fitness function value corresponding to the global optimal solution at the kth iteration; A represents the associated parameter; MaxIter represents the maximum number of iterations; c1 represents the first learning factor; c2 represents the second learning factor.
[0216] Traditional particle swarm optimization (PSO) has limitations in solving continuous space optimization problems and is prone to fall into local optimality. These problems are particularly evident in solving distribution network optimization problems. In order to improve the performance and efficiency of the algorithm, the present invention has taken four improvement measures, as follows.
[0217] 1. Chaos theory is introduced to initialize the particle swarm algorithm to solve the problem of slow convergence and easy falling into local optimality in the later stage of optimization.
[0218] The particle swarm algorithm can randomly assign particles, and the optimal solution is highly related to particle initialization. The more uniform the initial particle distribution is, the richer the diversity of the swarm is, and the faster the optimal solution is obtained.
[0219] The present invention utilizes chaotic mapping to extract the initial population of the particle swarm algorithm, so that particles are evenly distributed in the solution space.
[0220] Chaos refers to nonlinear motion that can pass through all situations within a specified range. Chaotic sequences can represent all states in a specified space, which is usually generated by mapping. Many researchers have found that chaotic mapping has unpredictable characteristics when studying chaotic mapping relationships, but certain laws can still be used to predict the mapping process. The most commonly used chaotic mapping method is logical mapping, and the corresponding expression is as follows:
[0221] Z ψ+1 =μZ ψ (1-Z ψ )ψ=0,1,2,...,N;μ∈(0,4]
[0222] Among them, Z ψ represents the value obtained by the ψ-times logistic mapping of the initial value Z0, 0≤Z0≤1; μ represents the control variable between (0,4], when o=4, it is in a completely chaotic state; the range of the chaotic space is [0,1].
[0223] The steps to initialize a particle swarm using a chaotic map are summarized as follows:
[0224] First, randomly generate an n-dimensional vector Z1=(z1,z2,...,z n ), and use the logical mapping equation to map other vectors to generate the chaotic sequence Z1, Z2, ..., Z N Then the chaotic sequence z ψ Inverse mapping from the chaotic space [0,1] to the space where the optimal solution is located [a,b]; the particle position is x ψε =a+(ba)z ψε,ε=1,2,...,M;ψ=1,2,...,N;where ψ represents the ψth particle; N is the size of the particle group; ε represents the εth dimension; M represents the spatial dimension; z ψε represents the chaotic sequence of the ψth particle in the εth dimension;
[0225] Finally, the particles with higher fitness function values are determined as the initial particles of the population.
[0226] When the particle swarm algorithm falls into a local optimum, the algorithm will select the historical optimal value of the particle during the iteration process and convert it into a chaotic sequence through inverse mapping to obtain the optimal position of the particle, and then randomly replace a particle position in the current search space to make the algorithm jump out of the local optimum. Whether a particle falls into premature maturity depends on the variance of the population fitness, which is calculated as follows:
[0227]
[0228] Where N represents the size of the population; f ψ represents the fitness function value of the ψth particle; f avg Represents the average fitness function value of the current group of particles; population variance κ 2 reflects the premature state of the particle; when κ 2 When it is less than a certain threshold, the particle algorithm is calculated to fall into premature maturity. Then chaos is applied to process the optimal particles to increase the diversity of the population. The detailed steps are described as follows.
[0229] (1) Select the best position in the iterative process and map it to the chaotic space [0,1] using a logistic function;
[0230] (2) Generate new sequences using logical variables and reverse map the sequences to the population;
[0231] (3) Calculate the best fitness function value of the particle, determine whether the particle has jumped out of the local optimum, record the best fitness function value, and set the corresponding particle to α, β, and δ;
[0232] (4) Use the current optimal chaotic particle to manage the particles in the particle swarm so that the particles leave the local optimum.
[0233] After the particle swarm performs the chaotic initialization operation, the particles are more evenly distributed in the search space, and the chaotic sequence can be used to reduce prematureness, increase the diversity of particles, and improve the convergence speed of the algorithm.
[0234] 2. Adaptive inertia weight is used to balance the local search and global search capabilities of the particle swarm algorithm.
[0235] As is known to all, the solving performance of the particle swarm algorithm is closely related to the inertia weight, wherein the smaller the inertia weight, the higher the local search capability of the algorithm, and the larger the inertia weight, the stronger the global search capability of the algorithm. In order to enable the algorithm to maintain a high search capability throughout the entire operation process, many methods have been proposed to adjust the inertia weight. However, the existing inertia weight improvement method is closely related to the number of iterations and cannot adapt well to nonlinear changes. To this end, the present invention processes the inertia weight by using the ratio of the fitness function values of adjacent particles; the updating and calculation methods of the inertia weight are as follows:
[0236]
[0237]
[0238] Among them, k represents the current number of iterations; MaxIter represents the maximum number of iterations; Represents the fitness function value corresponding to the global optimal solution of the kth iteration; represents the fitness function value corresponding to the global optimal solution of the k-1th iteration; w represents the inertia weight; w max Indicates the maximum value of the preset inertia weight, preferably, w max =0.9; w min Indicates the minimum value of the preset inertia weight, preferably, w min =0.4; q represents and The ratio of is used to represent the search ability of the algorithm. It can be seen from the above formula that q is larger at the beginning of the iteration, so the algorithm has a stronger global search ability, and q gradually decreases in the later stage of the iteration, so the algorithm has a stronger local search ability.
[0239] In summary, the inertia weight combined with the ratio of the fitness function values of adjacent particles can adaptively adjust the size of the inertia weight w according to the number of iterations, so that the algorithm has a higher search ability.
[0240] 3. By improving the learning factor, the algorithm can obtain a more accurate optimal solution.
[0241] The learning factors c1 and c2 of the particle swarm algorithm respectively determine the impact of individual particle experience and group experience on the particle trajectory, and can reflect the information exchange between particle swarms. These two values are usually fixed rather than set according to different stages.
[0242] The present invention adopts a learning factor based on sine and cosine and combined with inertia weight as the learning factor of the particle swarm algorithm, so that there is a relationship of increase and decrease between c1 and c2. The specific mathematical expression is:
[0243]
[0244] Among them, k represents the current number of iterations; MaxIter represents the maximum number of iterations; c1 and c2 both represent learning factors; by adjusting the learning factors, particles are searched in a large range in the initial stage to obtain a batch of high-quality particles, so that in the later stage, global optimization can be continued to be learned, and the interference of local extreme values can be eliminated as much as possible, thereby improving the accuracy of the solution.
[0245] 4. Adopt particle velocity and position adaptive update strategy to further improve the algorithm convergence speed.
[0246] The present invention adopts a new particle velocity and position adaptive update strategy, the expression is as follows:
[0247] First update expression:
[0248]
[0249] Second update expression:
[0250]
[0251] Among them, v ψε (k+1) represents the velocity of the ψth particle in the εth dimension in the k+1th iteration; x ψε (k+1) represents the position of the ψth particle in the εth dimension in the k+1th iteration; w represents the inertia weight; c1 and c2 are learning factors, preferably, both of which are 2; r1 and r2 are random numbers uniformly distributed in [0,1]; Represents the fitness function value corresponding to the global optimal solution of the ψth particle in the ε-dimensional space at the k-th iteration.
[0252] The calculation method of the decision condition for adaptive update is shown in the following formula:
[0253]
[0254] Where f(·) is the fitness function value of the particle; ρ is the adaptive judgment expression; N represents the population size, ψ=1…N; x ψ(t) represents the position of the ψth particle at time t; when the ρ value is large, the fitness function value of the current particle is much larger than the average fitness function value of all particles in the population, indicating that the algorithm is in the early stage of the search or the current particle distribution is relatively dispersed. At this time, the first update expression should be used to update the particle speed and position to make the search space wider. When the ρ value is small, the fitness function value of the current particle is not much different from the average fitness function value of all particles in the population, indicating that the algorithm is in the middle and late stages of the search or the current particle distribution is relatively concentrated. At this time, the second update expression should be used to update the particle speed and position to prevent the algorithm from falling into the local optimum when solving complex multi-peak functions, and introduce the particle average dimension information into the particle speed update formula to improve the convergence speed of the algorithm.
[0255] like Figure 3 The figure shows a schematic diagram of the overall framework of a power balance control method for a distribution network provided by the present invention. First, the present invention establishes an output prediction model, which is based on an SVM support vector machine, and uses a GWO grey wolf optimization algorithm to optimize the parameters in the SVM support vector machine, and finally obtains an output prediction model to achieve accurate estimation of the output power of distributed resources connected to the distribution network; secondly, considering network loss, voltage deviation and operating cost, the objective function is determined, the equal reserve rate constraint is added to the coordination constraint condition, and the multi-objective problem is transformed into a single-objective problem with the help of hierarchical analysis, which can ensure the power supply quality while ensuring the economy of system operation; finally, the improved particle swarm algorithm is used to solve the multi-objective and multi-constrained distribution network power coordination hybrid problem with step size calculation involved in the present invention, which is beneficial to improve the solution efficiency of the power optimization model, thereby realizing coordinated optimization of power when a high proportion of distributed resources are connected to the distribution network.
[0256] Embodiment 2:
[0257] Based on the same inventive concept, the present invention also provides a power distribution network power balance control system, the system structure is as follows: Figure 4 As shown, including:
[0258] Prediction value acquisition module: used to obtain the output power prediction value of the distributed resources connected to the distribution network;
[0259] Power coordination optimization module: used to solve the pre-built power optimization model based on the output power prediction value by using an improved particle swarm algorithm to obtain a power coordination optimization solution for the distribution network;
[0260] The power optimization model in the power coordination optimization module is constructed with the goal of minimizing network loss, minimizing network node voltage deviation, and minimizing distribution network operation cost;
[0261] The output power prediction value of the distributed resources connected to the power distribution network in the prediction value acquisition module includes the following acquisition process:
[0262] Obtaining power impact data of distributed resources connected to the distribution network;
[0263] Based on the power impact data, using a pre-built output prediction model, an output power prediction value corresponding to the power impact data is obtained;
[0264] The output prediction model is constructed based on the support vector machine and the gray wolf optimization algorithm, using the historical power impact data and the corresponding actual output power value as training data;
[0265] The output prediction model in the prediction value acquisition module includes the following construction process:
[0266] Selecting a preset group from historical power impact data as input of training data;
[0267] The actual value of the output power corresponding to the selected power impact data is used as the output of the training data;
[0268] Based on the input and output of the training data, a preset support vector machine is trained to obtain an output vector;
[0269] Determining a parameter to be optimized according to the output vector;
[0270] The parameters to be optimized are optimized using the Gray Wolf Optimization Algorithm, the optimal parameters are determined, and an output prediction model is obtained;
[0271] The parameters to be optimized include at least one or more of the following: regularization parameters and kernel function parameters.
[0272] The power optimization model in the power coordination optimization module includes the following construction process:
[0273] The objective function is constructed with the goal of minimizing network loss, minimizing voltage deviation at network nodes, and minimizing the operating cost of the distribution network;
[0274] According to the objective function, corresponding constraint conditions are formulated;
[0275] Based on the objective function and the constraint conditions, a power optimization model is obtained.
[0276] The constraints in the power coordination optimization module include at least one or more of the following: flow-related constraints, output regulation constraints of distributed resources and equal reserve rate constraints; the flow-related constraints include at least one or more of the following: flow equation constraints, node voltage amplitude inequality constraints and branch thermal constraints.
[0277] The expression corresponding to the objective function in the power coordination optimization module is as follows:
[0278] min F=μ1f1+μ2f2+μ3f3
[0279] Among them, F represents the objective function; f1 is the first objective corresponding to network loss; μ1 is the weight of the first objective f1; f2 is the second objective corresponding to network node voltage deviation; μ2 is the weight of the second objective f2; f3 is the third objective corresponding to the operating cost of the distribution network; μ3 is the weight of the third objective f3.
[0280] The expression corresponding to the first objective f1 in the power coordination optimization module is as follows:
[0281]
[0282] The expression corresponding to the second target f2 is as follows:
[0283] f2={max(1-v i )}
[0284] The expression corresponding to the third target f3 is as follows:
[0285] f3=max(E-Cost)
[0286] in, represents the active power flowing through branch i; represents the reactive power flowing through branch i; represents the node voltage of the terminal node of branch i; R i represents the resistance of branch i; I represents the number of branches; i=1,…,I; v i represents the node voltage of branch i in the distribution network; E represents the economic benefit of distributed resource power generation; Cost represents the maintenance cost of distributed resource power generation.
[0287] The power coordination optimization module adopts an improved particle swarm algorithm to solve the pre-built power optimization model to obtain a power coordination optimization solution for the distribution network, including:
[0288] Step S1: using the output power prediction value as an input item of the power optimization model;
[0289] Step S2: randomly generate a particle swarm, use chaotic mapping to initialize the position and velocity of the particle swarm, initialize the inertia weight and learning factor; use the objective function to represent the fitness function of the particle; and use the power coordination parameters of the distributed resources connected to the distribution network as the optimization variables of the particles; wherein the optimization variables include at least one or more of the following: distributed power output, generator terminal voltage, node voltage amplitude;
[0290] Step S3: Calculate the fitness function values of all particles, and select the particle with the largest fitness function value as the global optimal solution;
[0291] Step S4: Determine whether the global optimal solution converges. If so, determine that the optimization variable corresponding to the global optimal solution is the power coordination optimization scheme of the distribution network, and use the power coordination scheme of the distribution network as the output item of the power optimization model, and stop executing; if not, iteratively update the position and speed of the particle, and execute step S5;
[0292] Step S5: updating the inertia weight and the learning factor based on the preset inertia weight and the learning factor update expression;
[0293] Step S6: Based on the updated inertia weight, learning factor and the fitness function value of the updated particle, the particle velocity and position adaptive update strategy is adopted to update the global optimal solution, and step S4 is executed.
[0294] The inertia weight and learning factor update expressions in the power coordination optimization module are as follows:
[0295]
[0296] Where w represents the inertia weight; w max Indicates the maximum value of the preset inertia weight; w min Indicates the minimum value of the preset inertia weight; k indicates the current number of iterations; represents the fitness function value corresponding to the global optimal solution at the kth iteration; A represents the associated parameter; MaxIter represents the maximum number of iterations; c1 represents the first learning factor; c2 represents the second learning factor.
[0297] Embodiment 3:
[0298] Based on the same inventive concept, the present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function, so as to implement the steps of a power balance control method for a distribution network in the above embodiment.
[0299] Embodiment 4:
[0300] Based on the same inventive concept, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It can be understood that the computer-readable storage medium here can include both built-in storage media in a computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of a power balance control method for a distribution network in the above embodiment.
[0301] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0302] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0303] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0304] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0305] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit its protection scope. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that after reading the present invention, those skilled in the art can still make various changes, modifications or equivalent substitutions to the specific implementation methods of the application, but these changes, modifications or equivalent substitutions are all within the protection scope of the claims to be approved.
Claims
1. A power balance control method for a distribution network, characterized in that: include: Obtain the output power forecast value of the distributed resources connected to the distribution network; Based on the output power prediction value, an improved particle swarm algorithm is used to solve the pre-built power optimization model to obtain a power coordination optimization solution for the distribution network; The power optimization model is constructed with the goal of minimizing network loss, minimizing network node voltage deviation, and minimizing distribution network operation cost.
2. The method according to claim 1, characterized in that The power optimization model includes the following construction process: The objective function is constructed with the goal of minimizing network loss, minimizing voltage deviation at network nodes, and minimizing the operating cost of the distribution network; According to the objective function, corresponding constraint conditions are formulated; Based on the objective function and the constraint conditions, a power optimization model is obtained.
3. The method according to claim 2, characterized in that The constraints include at least one or more of the following: flow-related constraints, output regulation constraints of distributed resources and equal reserve rate constraints; the flow-related constraints include at least one or more of the following: flow equation constraints, node voltage amplitude inequality constraints and branch thermal constraints.
4. The method according to claim 2, characterized in that The expression corresponding to the objective function is as follows: minF=μ1f1+μ2f2+μ3f3 Among them, F represents the objective function; f1 is the first objective corresponding to network loss; μ1 is the weight of the first objective f1; f2 is the second objective corresponding to network node voltage deviation; μ2 is the weight of the second objective f2; f3 is the third objective corresponding to the operating cost of the distribution network; μ3 is the weight of the third objective f3.
5. The method according to claim 4, characterized in that The expression corresponding to the first target f1 is as follows: The expression corresponding to the second target f2 is as follows: f2={max(1-v i )} The expression corresponding to the third target f3 is as follows: f3=max(EC ost) in, represents the active power flowing through branch i; represents the reactive power flowing through branch i; represents the node voltage of the terminal node of branch i; R i represents the resistance of branch i; I represents the number of branches; i=1,…,I; v i represents the node voltage of branch i in the distribution network; E represents the economic benefit of distributed resource power generation; Cost represents the maintenance cost of distributed resource power generation.
6. The method according to claim 2, characterized in that The improved particle swarm algorithm is used to solve the pre-built power optimization model to obtain the power coordination optimization scheme of the distribution network, including: Step S1: using the output power prediction value as an input item of the power optimization model; Step S2: randomly generate a particle swarm, use chaotic mapping to initialize the position and velocity of the particle swarm, initialize the inertia weight and learning factor; use the objective function to represent the fitness function of the particle; and use the power coordination parameters of the distributed resources connected to the distribution network as the optimization variables of the particles; wherein the optimization variables include at least one or more of the following: distributed power output, generator terminal voltage, node voltage amplitude; Step S3: Calculate the fitness function values of all particles, and select the particle with the largest fitness function value as the global optimal solution; Step S4: Determine whether the global optimal solution converges. If so, determine that the optimization variable corresponding to the global optimal solution is the power coordination optimization scheme of the distribution network, and use the power coordination scheme of the distribution network as the output item of the power optimization model, and stop executing; if not, iteratively update the position and speed of the particle, and execute step S5; Step S5: updating the inertia weight and the learning factor based on the preset inertia weight and the learning factor update expression; Step S6: Based on the updated inertia weight, learning factor and the fitness function value of the updated particle, the particle velocity and position adaptive update strategy is adopted to update the global optimal solution, and step S4 is executed.
7. The method according to claim 6, characterized in that The inertia weight and learning factor update expressions are as follows: Where w represents the inertia weight; w max Indicates the maximum value of the preset inertia weight; w min Indicates the minimum value of the preset inertia weight; k indicates the current number of iterations; represents the fitness function value corresponding to the global optimal solution at the kth iteration; A represents the associated parameter; MaxIter represents the maximum number of iterations; c1 represents the first learning factor; c2 represents the second learning factor.
8. The method according to claim 1, characterized in that The output power prediction value of the distributed resource accessing the power distribution network includes the following acquisition process: Obtaining power impact data of distributed resources connected to the distribution network; Based on the power impact data, using a pre-built output prediction model, an output power prediction value corresponding to the power impact data is obtained; The output prediction model is constructed based on the support vector machine and the grey wolf optimization algorithm, using the historical power impact data and the corresponding actual output power values as training data.
9. The method according to claim 8, characterized in that The output prediction model includes the following construction process: Selecting a preset group from historical power impact data as input of training data; The actual value of the output power corresponding to the selected power impact data is used as the output of the training data; Based on the input and output of the training data, a preset support vector machine is trained to obtain an output vector; Determining a parameter to be optimized according to the output vector; The parameters to be optimized are optimized using the Gray Wolf Optimization Algorithm, the optimal parameters are determined, and an output prediction model is obtained; The parameters to be optimized include at least one or more of the following: regularization parameters and kernel function parameters.
10. A power balance control system for a distribution network, characterized in that: include: Prediction value acquisition module: used to obtain the output power prediction value of the distributed resources connected to the distribution network; Power coordination optimization module: used to solve the pre-built power optimization model based on the output power prediction value by using an improved particle swarm algorithm to obtain a power coordination optimization solution for the distribution network; The power optimization model in the power coordination optimization module is constructed with the goal of minimizing network loss, minimizing network node voltage deviation, and minimizing distribution network operation cost.
11. The system according to claim 10, characterized in that The power optimization model in the power coordination optimization module includes the following construction process: The objective function is constructed with the goal of minimizing network loss, minimizing voltage deviation at network nodes, and minimizing the operating cost of the distribution network; According to the objective function, corresponding constraint conditions are formulated; Based on the objective function and the constraint conditions, a power optimization model is obtained.
12. A computer device, characterized in that: include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, a power distribution network power balance control method as described in any one of claims 1 to 9 is implemented.
13. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, a power balance control method for a distribution network as described in any one of claims 1 to 9 is implemented.
Citation Information
Cited By
New energy distributed power control method and system
CN119651788A
Multi-objective optimization method and device for operation of power distribution system, and electronic equipment
CN121906526A
