Data-driven multi-region power system economic environmental dispatching method and system

Abstract

The application discloses a data-driven-based multi-region power system economic environmental scheduling method and system, electronic equipment and a computer readable storage medium, and belongs to the technical field of power systems. The method comprises the following steps: constructing a multi-objective optimization model of multi-region power system economic environmental scheduling considering multiple constraint conditions with the aim of minimizing the emission of pollution gases and the cost of fossil fuels; constructing a data-driven proxy model to transform the multi-objective optimization model of multi-region power system economic environmental scheduling according to the target and multiple constraint conditions; solving the transformed multi-objective optimization model of multi-region power system economic environmental scheduling by using a multi-objective ant lion algorithm; and obtaining a global optimal solution according to the solving result. Meanwhile, the convergence speed and accuracy of the optimization algorithm are improved, the operation time is reduced as much as possible, the problem of the increase of the dimension of multi-region power system economic emission scheduling is solved, and the problems of the occupation of a large amount of computing resources and the consumption of a large amount of computing time are solved.

Description

Technical Field

[0001] This application relates to the field of power system technology, and in particular to a data-driven method and system for economic and environmental dispatching of multi-regional power systems. Background Technology

[0002] The statements in this section merely refer to the background art relevant to this application and do not necessarily constitute prior art.

[0003] With the expansion of the power system and the dispersion of load centers, the establishment of a multi-regional power system interconnected with multiple load centers is crucial for the safe and stable operation of the power system. Many countries have granted the power industry more autonomy in its operations, which has effectively contributed to economic operation and emission reduction. Based on these two factors, economic dispatch of multi-regional power systems can provide the power system with higher stability and economy. As the power industry emits the most polluting gases, how to balance environmental and economic benefits while ensuring that the power supply quality is not affected is a major technical problem worthy of research and a key aspect of promoting energy conservation and emission reduction in power enterprises. This problem has been preliminarily studied, and its main research ideas are: (1) simultaneously considering the power generation cost and polluting gas emission of thermal power units, (2) considering different constraints in the multi-regional power generation and dispatch process, and (3) selecting appropriate algorithms for analysis and solution based on meeting the dispatch time requirements.

[0004] However, due to the expansion of the power grid and the decentralization of load centers, the dimensionality of the economic emission dispatch problem of multi-regional power systems will inevitably increase, thus occupying a large amount of computing resources and consuming a large amount of computing time.

[0005] Therefore, how to make the optimized scheduling strategy more suitable for the economic emission scheduling problem of multi-regional power systems, that is, to simultaneously improve the convergence speed and accuracy of the optimization algorithm and minimize the operation time, is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this application provides a data-driven multi-regional power system economic and environmental dispatch method, system, electronic equipment, and computer-readable storage medium. For any multi-regional power grid topology, after determining the structural parameters, the method aims to simultaneously minimize power generation costs and environmental pollution, taking into account the actual power flow of the power grid, while improving the convergence speed and accuracy of the optimization algorithm and minimizing the operation time.

[0007] Firstly, this application provides a data-driven method for economic and environmental dispatching of multi-regional power systems;

[0008] Data-driven multi-regional power system economic environment dispatching methods include:

[0009] To minimize pollutant emissions and fossil fuel costs, a multi-objective optimization model for economic and environmental dispatching of multi-regional power systems is constructed, taking into account multiple constraints.

[0010] Based on the objectives and multiple constraints, a data-driven proxy model is constructed to transform the multi-objective optimization model of economic environment dispatching of multi-regional power systems. The transformed multi-objective optimization model of economic environment dispatching of multi-regional power systems is then solved using the multi-objective antlion algorithm.

[0011] Based on the solution results, the global optimal solution is obtained as the decision-making basis for the optimal scheduling of each region of the power system in the current period.

[0012] Furthermore, the multi-objective optimization model for economic environment dispatching of multi-regional power systems is expressed as follows:

[0013] Minimize [F(P),E(P)],

[0014] Subject to:g i (P) = 0, i = 1, ..., M1

[0015] h j (P)≤0,j=1,…,M2,

[0016] Where F(P) is the objective function for optimizing fossil fuel costs, E(P) is the objective function for optimizing pollutant emissions, and g i (P) represents the inequality constraints involved, h j (P) represents the equality constraints involved.

[0017] Furthermore, the multiple constraints include generator active power upper and lower limit constraints, power balance constraints, node voltage amplitude constraints, line power flow constraints, transmission line safety constraints, regional hot reserve transfer constraints, and line loss constraints.

[0018] Preferably, the upper and lower limits of generator active power are expressed as follows:

[0019]

[0020] The power balance constraint is expressed as:

[0021]

[0022] The node voltage magnitude constraint is expressed as:

[0023]

[0024] Line power flow constraints are expressed as:

[0025]

[0026] Transmission line security constraints are expressed as follows:

[0027]

[0028] The regional hot standby transfer constraint is expressed as:

[0029]

[0030] Line loss constraints are expressed as follows:

[0031]

[0032] Furthermore, the specific steps for constructing the data-driven agent model to transform the multi-objective optimization model for economic environment dispatching of multi-regional power systems include:

[0033] Based on multiple constraints, several feasible solutions that satisfy the constraints are obtained; based on the feasible solutions and the multi-regional power system economic environment dispatch multi-objective optimization model, the optimization objective value corresponding to the feasible solution is calculated.

[0034] Based on feasible solutions and corresponding optimization objective values, a data-driven proxy model is constructed to replace the multi-objective optimization model for economic environment dispatching of multi-regional power systems.

[0035] Furthermore, the specific steps for solving the transformed multi-regional power system economic environment dispatch multi-objective optimization model using the multi-objective antlion algorithm include:

[0036] Initialize the ant colony and antlion colony, and randomly initialize the power generation of each generator in each area;

[0037] Assign the ant colony and antlion colony the corresponding randomly initialized power generation of each generator in each region, and solve the non-dominated solution for each optimization objective;

[0038] Traverse all individuals in the population and calculate non-dominated solutions based on individual optimization results; while traversing all individuals in the population, perform a single-dimensional retention strategy to update the ant population; sort the non-dominated solutions according to the initial values.

[0039] Furthermore, the specific steps for implementing the single-dimensional retention strategy to update the ant colony include:

[0040] Based on the fitness value of the optimization objective, the ant population is divided into four parts;

[0041] Find the minimum fitness value of the optimization objective in the population, perform position movement operations on the four parts of the ant population respectively, and obtain the intermediate position variable;

[0042] Based on the intermediate position variable, obtain the new population position;

[0043] The single-dimensional optimal retention strategy is implemented, and for each dimension of each particle, it is determined whether to update the dimension of that ion.

[0044] Furthermore, the specific steps for obtaining the global optimal solution based on the solution results include:

[0045] For the optimization objective function of pollutant gas emissions and fossil fuel costs, the membership function value corresponding to its non-dominated solution is calculated using the membership function.

[0046] The membership function values ​​are regularized and solved to obtain the global optimal solution.

[0047] Secondly, this application provides a data-driven multi-regional power system economic environment dispatching system;

[0048] A data-driven multi-regional power system economic environment dispatching system includes:

[0049] The model building module is configured to construct a multi-objective optimization model for economic and environmental dispatching of a multi-regional power system, taking into account multiple constraints, with the goal of minimizing pollutant emissions and fossil fuel costs.

[0050] The solution module is configured to: construct a data-driven proxy model based on the objectives and constraints to transform the multi-regional power system economic environment scheduling multi-objective optimization model, and solve the transformed multi-regional power system economic environment scheduling multi-objective optimization model through the multi-objective antlion algorithm;

[0051] The decision-making module is configured to obtain the global optimal solution based on the solution results, and use it as the basis for decision-making on the optimal scheduling of each region of the power system in the current period.

[0052] Thirdly, this application provides an electronic device;

[0053] An electronic device includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the processor executes the computer instructions, the computer instructions perform the steps of the above-described data-driven multi-regional power system economic environment dispatch method.

[0054] Fourthly, this application provides a computer-readable storage medium;

[0055] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps of the above-described data-driven multi-regional power system economic environment dispatching method.

[0056] Compared with the prior art, the beneficial effects of this application are:

[0057] 1. This application provides a data-driven, multi-regional power system economic and environmental dispatching method based on data-driven, transfer learning and antlion algorithm. It can minimize the power generation cost of generator sets, minimize the emission of pollutants, minimize the use of fossil fuels, and significantly reduce the computation time while meeting the power load. It can provide power companies with timely decision-making basis.

[0058] 2. This application introduces line power flow constraints and maximum transmission line constraints into the constraints, which are combined with the actual operation of the power grid, making the optimization scheduling strategy more practically instructive.

[0059] 3. This application is based on a deep belief network model and uses a data-driven agent-assisted method to reconstruct the objective function, replacing the original computationally expensive objective function with a computationally fast data-driven deep belief network agent regression model; it is based on a pre-training-fine-tuning transfer learning method, by pre-training and fine-tuning the network part of the deep belief network, the agent model of other regions is quickly established using the network information of the agent model of one region; it is based on the multi-objective antlion algorithm, overcoming the drawbacks of traditional weights and methods, and introducing a single-dimensional optimal retention strategy to adapt to multi-objective optimization methods, which improves the convergence, uniformity and extensibility of the Pareto front solution during the iteration period, and calculates the membership values ​​of all non-dominated solutions, with the non-dominated solution with the largest membership value being the decision basis for the optimal scheduling in this period. Attached Figure Description

[0060] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute an undue limitation of this application.

[0061] Figure 1 A flowchart illustrating a data-driven multi-regional power system economic environment dispatching system provided in this application embodiment;

[0062] Figure 2 A flowchart illustrating the multi-target antlion algorithm provided in this application embodiment;

[0063] Figure 3 A schematic diagram of the simulation object four-area forty generator test system provided in the embodiments of this application;

[0064] Figure 4 The diagram shows the Pareto optimal frontier generated under all constraints for a test system of forty generators in four regions (load of 10500MW) provided in the embodiments of this application. Detailed Implementation

[0065] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0066] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations according to this application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or apparatus.

[0067] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0068] Terminology Explanation:

[0069] Fitness value: The calculated value of the objective function;

[0070] Non-dominated solution: Unit output decision, i.e. the active power output of each generator in each region.

[0071] Example 1

[0072] In the prior art, due to the expansion of the power grid and the decentralization of load centers, the dimensionality of the economic emission dispatch problem of multi-regional power systems inevitably increases, occupying a large amount of computing resources and consuming a large amount of computing time; therefore, this application provides a data-driven economic environmental dispatch method for multi-regional power systems.

[0073] The data-driven multi-regional power system economic and environmental dispatching method provided in this application is based on the following idea: For any multi-regional power grid topology, after determining the structural parameters, with the goal of simultaneously minimizing power generation cost and minimizing environmental pollution, and considering the actual power grid operation flow, firstly, the original multi-regional power generation cost model and multi-regional environmental pollution model are replaced with a data-driven deep belief network regression model. Secondly, a transfer learning method based on pre-training and fine-tuning is used to quickly establish surrogate models between different regions. Finally, an improved multi-objective antlion algorithm is applied for iterative calculation to obtain a set of non-dominated solutions for each region, forming the Pareto optimal frontier of the two optimization objectives of "power generation cost - environmental pollution" for different regions. Then, the membership function method is used to obtain a set of optimal dispatching decision criteria for a single region in the current time period.

[0074] Next, combined Figure 1-4 This embodiment provides a detailed description of the data-driven multi-regional power system economic environment dispatch method disclosed herein. This data-driven multi-regional power system economic environment dispatch method includes the following steps:

[0075] S1. Determine the optimization objectives and constraints, and construct a multi-objective optimization model for economic and environmental dispatch of a multi-regional power system. Specifically, with the objective of simultaneously minimizing pollutant emissions and fossil fuel costs, select generator active power upper and lower limits constraints, power balance constraints, node voltage amplitude constraints, line power flow constraints, transmission line safety constraints, regional hot reserve transfer constraints, and line loss constraints as constraints to construct a multi-objective optimization model for economic and environmental dispatch of a multi-regional power system.

[0076] For example, optimization objective 1 is to minimize fossil fuel costs, and the optimization objective function for fossil fuel costs is:

[0077]

[0078] Among them, a ij b ij c ij d ij e ij Let N and N' be the fuel cost coefficients for the j-th generator in the i-th region. g P represents the number of areas and generators participating in the scheduling. ij Let be the active power output of the j-th generator in the i-th region.

[0079] Optimization objective 2 is to minimize pollutant emissions. The objective function for optimizing pollutant emissions is:

[0080]

[0081] Where, α ij βij γ ij ε ij , λ ij Let P be the pollutant emission coefficient of the j-th generator in the i-th region. ij Let N be the active power output of the j-th generator in the i-th region, and N be the active power output of the generator. g This indicates the number of areas and generators participating in the scheduling.

[0082] The active power upper and lower limits of the generator are represented as follows:

[0083]

[0084] Among them, S ti Represents the power on the i-th line. N represents the maximum allowed transmission power on the i-th line. line This represents the number of lines.

[0085] The power balance constraint is expressed as:

[0086]

[0087] Among them, P di P represents the load of the i-th region. lossi T represents the network loss in the i-th region. ip This represents the transfer power of the link between region i and region p.

[0088] The node voltage magnitude constraint is expressed as:

[0089]

[0090] in, and V represents the minimum allowable node voltage and the maximum allowable node voltage, respectively. i It is the voltage amplitude at node i.

[0091] Line power flow constraints are expressed as:

[0092]

[0093] Among them, S ti Represents the power on the i-th line. Let N represent the maximum allowed transmission power on the i-th line, and Nline represent the number of lines. Transmission line safety constraints are expressed as:

[0094]

[0095] in, This represents the maximum power that can be safely transmitted between the i-th region and the p-th region.

[0096] The regional hot standby transfer constraint is expressed as:

[0097]

[0098] Among them, S pj S is the hot reserve power of the j-th generator in the p-th region; p,req It is the required thermal reserve power for the p-th region, RC ip It represents the heat reserve power transferred between the i-th and q-th regions.

[0099] The above constraints are all inequality constraints. In this embodiment, equality constraints are also considered, namely power balance constraints:

[0100] Line losses are obtained by solving the following power flow equations:

[0101]

[0102]

[0103] Consequently, line loss

[0104]

[0105] Among them, G ij θ is the conductance of the branch connecting nodes i and j. ij It is the voltage phase difference between nodes i and j, V j P is the voltage amplitude at node j. loss For line loss, P i Let Q be the active power output of the i-th generator. i Let be the reactive power output of the i-th generator.

[0106] In summary, the multi-objective optimization model for economic environment dispatching of multi-regional power systems in this embodiment can be described as follows:

[0107] Minimize [F(P),E(P)],

[0108] Subject to:g i (P) = 0, i = 1, ..., M1

[0109] h j (P)≤0,j=1,…,M2,

[0110] Where F(P) and E(P) represent optimization objective 1 and optimization objective 2, respectively; g(P) and h(P) are the equality and inequality constraints involved, respectively; and M1 and M2 are the number of equality and inequality constraints, respectively.

[0111] S2. Based on the objectives and multiple constraints, construct a data-driven proxy model to replace the multi-objective optimization model for economic environment dispatching of multi-regional power systems. Specifically, using data-driven proxy assistance techniques and known feasible solutions of optimization functions, construct a data-driven proxy model based on deep belief networks to replace the original two objective functions. Specific steps include:

[0112] S201. Based on the constraints, find several possible solutions that satisfy the constraints.

[0113] S202. Using the possible solutions and the calculated optimization objective value, a data-driven agent model based on deep belief network is constructed to replace the optimization objective model. The model is divided into a deep belief network part, a logistic regression part, and a neural network part.

[0114] S203. Calculate the fitness value of the data used to train the multi-objective optimization model of economic environment dispatch for multi-regional power systems and the fitness value calculated using the data-driven surrogate model.

[0115] S204. Calculate the cosine similarity between the two fitness values. If the cosine similarity is greater than 0.9, the data-driven proxy model for the first region is established. Otherwise, return to step S202 to re-establish the data-driven proxy model. Specifically, the data-driven proxy model is a regression proxy model that combines a deep belief network and a backpropagation neural network.

[0116] S205. After building the data-driven agent model for the first region, apply the pre-training-fine-tuning transfer learning technique to quickly build agent models for other regions.

[0117] Furthermore, in some embodiments, a pre-trained-fine-tuned transfer learning technique is used to quickly construct a data-driven agent model for each sub-region. Specific steps include:

[0118] S2051. Decompose the multi-regional power system into several dispatch sub-regions, and divide the data-driven agent model of the pre-constructed dispatch sub-regions into three parts: a deep belief network part, a logistic regression layer part, and an error backpropagation neural network.

[0119] S2052. Copy the deep belief network part of the data-driven agent model of the first scheduling sub-region obtained in step S204, fine-tune the deep belief network part according to the actual data of the corresponding scheduling sub-region, and reconstruct the logistic regression part and the neural network part.

[0120] S2053. Perform the operations in step S2052 for each scheduling sub-region until the data-driven agent model for all regions is obtained.

[0121] S3. Solve the above data-driven surrogate model using an improved multi-objective antlion algorithm; the improvement of the multi-objective antlion algorithm lies in using a single-dimensional preservation optimization strategy to update the ant population. This step specifically includes:

[0122] S301. Initialize the relevant parameters of the multi-objective antlion algorithm. The relevant parameters of the multi-objective antlion algorithm include the population sample and the algorithm parameters. When initializing the population sample, the upper and lower limits of generator active power and the power balance constraint should be satisfied at the same time. At the same time, the power generation of each generator in each region is randomly initialized to represent the possible solutions of the optimization objective.

[0123] S302. Based on the initial values, calculate the fitness values ​​of all individuals. That is, for different initial values, according to the optimization objective function in the multi-regional power system economic environment dispatch multi-objective optimization model, calculate the fossil fuel cost and pollutant emissions of each generator in each region.

[0124] S303. Assign the ant colony and antlion colony the power generation of each generator in each region according to the corresponding random initialization. Calculate the non-dominated solutions based on the fitness values ​​and store the obtained non-dominated solutions in the external archive. If the number of non-dominated solutions exceeds the preset value of the external archive, execute the crowding sorting rule and delete the redundant non-dominated solutions.

[0125] S304. Using the data-driven proxy model constructed by the application, recalculate the virtual fitness value corresponding to the initial value, obtain non-dominated solutions based on the virtual fitness value, and store the obtained non-dominated solutions in the internal archive; if the number of non-dominated solutions exceeds the preset value of the internal archive, execute the crowding sorting rule to delete the redundant non-dominated solutions.

[0126] S305, Iteration begins:

[0127] S3051. Traverse all individuals in the population. Specifically, after performing the single-dimensional preservation strategy operation, obtain the position of the next generation ant, update the current antlion position, handle the generator and region transfer constraints, update the position of the next generation ant, calculate the virtual fitness value corresponding to all individuals, update the internal archive using the internal archive and the virtual fitness value of the next generation ant, and the traversal ends.

[0128] Furthermore, in some embodiments, the single-dimensional retention strategy specifically includes:

[0129] (1) The population is divided into four parts: the population in which the fitness values ​​of both target 1 and target 2 are less than the average fitness values ​​of the corresponding target 1 and target 2 in the current population is named "win-win population"; the population in which only the fitness value of target 1 is less than the average fitness value of the corresponding target 1 in the current population is named "win-win population"; the population in which only the fitness value of target 2 is less than the average fitness value of the corresponding target 2 in the current population is named "win-win population"; the population in which both the fitness values ​​of target 1 and target 2 are greater than the average fitness values ​​of the corresponding target 1 and target 2 in the current population is named "failure population";

[0130] (2) Find the minimum fitness values ​​of target one and target two in the population and name them "minimum one" and "minimum two". The corresponding individuals are named "minimum decision one" and "minimum decision two".

[0131] (3) Perform different operations on the four populations: the "win-win population" moves towards the direction of the vector sum of "first minimum decision" and "second minimum decision"; the "first winning population" moves towards "first minimum decision"; the "second winning population" moves towards "second minimum decision"; and the "failure population" moves towards the direction of the vector sum of "first minimum decision" and "second minimum decision". The overall population position is named P1.

[0132] (4) Find the minimum fitness values ​​of target one and target two in the population and name them "maximum one" and "maximum two". The corresponding individuals are named "maximum one decision" and "maximum two decision".

[0133] (5) Perform the "profit-seeking and loss-avoidance" operation on the four population groups as follows: the "win-win population" moves towards the direction of the vector sum of "first minimum decision" and "second minimum decision" and moves away from the direction of the vector sum of "first maximum decision" and "second maximum decision"; the "first winning population" moves towards "first minimum decision" and moves away from the direction of "first maximum decision"; the "second winning population" moves towards "second minimum decision" and moves away from "second maximum decision"; the "failure population" moves towards the direction of the vector sum of "first minimum decision" and "second minimum decision" and moves away from the direction of the vector sum of "first maximum decision" and "second maximum decision". The overall population position is named P2.

[0134] (6) Find the particle with the smallest habitat radius in the population, and move all particles toward the particle with the smallest habitat radius. Name the overall population position P3.

[0135] (7) The new population positions are the vector sum of P1, P2 and P3, where the sum of the coefficients of P1, P2 and P3 is 1.

[0136] (8) Implement the single-dimensional optimal retention strategy. For a particle, judge each dimension. First, generate a random number that is uniformly distributed between 0 and 1. If the random number is greater than 0.2, the particle’s dimension will not change. If the random number is less than or equal to 0.2, the particle’s dimension will be changed to the corresponding dimension of the new population position.

[0137] S3052. Calculate non-dominated solutions based on individual optimization results, and store the obtained non-dominated solutions in the internal archive. If the number of non-dominated solutions exceeds the preset value of the internal archive, execute the crowding sorting rule and delete the redundant non-dominated solutions.

[0138] S3053. Calculate the corresponding fitness value based on the updated individual values ​​in the internal archive; find the individual with the minimum value corresponding to each of the two objectives, and store the fitness value of the individual with the minimum value in the external archive; if the number of non-dominated solutions exceeds the preset value of the external archive, execute the crowding sorting rule and delete the redundant non-dominated solutions.

[0139] S3054. The iteration ends, and the Pareto optimal frontier for each region is generated.

[0140] During the implementation of the numerical example, the initialization parameters used in the improved multi-objective antlion algorithm could initially only be assigned values ​​based on experience, but these initial values ​​were clearly not optimal. During algorithm execution, one parameter could be adjusted appropriately while keeping others constant, and the simulation results observed. This method allowed for continuous adjustment of each parameter to achieve the best simulation effect.

[0141] S4. Based on the solution results, obtain the globally optimal solution as the decision-making basis for the optimal scheduling of each region of the power system in the current time period. This embodiment presents a discrimination method based on the membership function, in which the non-dominated solution with the largest membership value is selected as the scheduling decision-making basis for the current time period. Specific steps include:

[0142] S401. For each optimization objective function, calculate the membership function value corresponding to its non-dominated solution:

[0143]

[0144] Among them, F i,k For the k-th solution of the i-th optimization objective, and These are the minimum and maximum values ​​of the i-th optimization objective, respectively.

[0145] S402. For each individual nondominated solution, μ i,k Regularization yields μ j :

[0146]

[0147] Where N1 = 2 is the number of optimization objectives, and M is the number of non-dominated solutions.

[0148] S403, Solve for μ j Obtain the non-dominated solution corresponding to the maximum value, and use it as the basis for scheduling decisions in the current time period.

[0149] In this embodiment, the obtained power grid operation parameters include: ① node parameters, mainly including the distribution of PQ, PV and reference nodes; ② active load during the dispatch period; ③ node voltage amplitude, phase angle, and the maximum and minimum voltage that the node can withstand; ④ active power output of generator nodes and the maximum and minimum active power that the node can withstand; ⑤ maximum active power output allowed by each generator; ⑥ branch parameters: branch resistance, reactance, per-unit susceptance, the allowable capacity of long (short) distance transmission branches, and the maximum and minimum phase angle allowed by the branch; ⑦ maximum allowable transmission capacity of transmission lines; ⑧ regional hot standby requirements.

[0150] The power grid parameters used in the calculation data of this scheduling method can be actual power grid operating parameters or any simulation object, such as a four-region, forty-generator test system. This simulation example uses a four-region, forty-generator test system, the structure of which is shown in the diagram below. Figure 3 As shown. When using Figure 3 As a case study, the load during the current scheduling period is 10500MW, and the extreme value solutions of the two optimization objectives involved are shown in Table 1 (fuel cost is in RMB / h, and pollutant emission is in ton / h).

[0151] Table 1. Solution to extreme values ​​of fuel cost

[0152]

[0153]

[0154] exist Figure 4 In China, targeting Figure 3 In the simulation example, the Pareto optimal frontier of each region obtained by this invention consists of 10 non-dominated solutions. This is a set of compromise solutions formed for the two optimization objectives of environment and economy. All non-dominated solutions are the optimal solutions for the current scheduling period. Figure 4 The extreme value solution is (827228.68, 177386.3). The size and breadth of the extreme value solution directly determine the advancement of the optimization method.

[0155] In Table 1, the minimum fuel cost obtained using the method of this invention is ¥827,228.68 / h, which is ¥6,411.46 / h, ¥6,977.18 / h, and ¥435.31 / h lower than the other three methods, respectively. The other extreme value, i.e., the pollutant emission of 177,386.3 tons / h, is also the lowest among all methods. Therefore, this invention is more advanced than other existing methods.

[0156] The experimental data above demonstrates that, compared to previous economic and environmental dispatching methods, this invention achieves superior performance while meeting basic power supply requirements. It also effectively reduces power generation costs and pollutant emissions. Taking the extreme point in Table 1 as an example, if this solution is used as the dispatching basis, assuming the load remains stable at 10500MW, and comparing it with the improved multi-objective antlion algorithm and the NSOS algorithm, this invention can save approximately 167,452 yuan in fossil fuel costs per day, and approximately 61,120,097 yuan per year.

[0157] Example 2

[0158] This embodiment discloses a multi-regional power system economic environment dispatching system, including:

[0159] The model building module is configured to construct a multi-objective optimization model for economic and environmental dispatching of a multi-regional power system, taking into account multiple constraints, with the goal of minimizing pollutant emissions and fossil fuel costs.

[0160] The solution module is configured to: construct a data-driven proxy model based on the objectives and constraints to transform the multi-regional power system economic environment scheduling multi-objective optimization model, and solve the transformed multi-regional power system economic environment scheduling multi-objective optimization model through the multi-objective antlion algorithm;

[0161] The decision-making module is configured to obtain the global optimal solution based on the solution results, and use it as the basis for decision-making on the optimal scheduling of each region of the power system in the current period.

[0162] It should be noted that the model building module, solution module, and decision module described above correspond to the steps in Embodiment 1. The examples and application scenarios implemented by these modules and their corresponding steps are the same, but they are not limited to the content disclosed in Embodiment 1. It should also be noted that these modules, as part of the system, can be executed in a computer system, such as a set of computer-executable instructions.

[0163] Example 3

[0164] Embodiment 3 of the present invention provides an electronic device, including a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the computer instructions are executed by the processor, they complete the steps of the above-described data-driven multi-regional power system economic environment dispatching method.

[0165] Example 4

[0166] Embodiment 4 of the present invention provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps of the above-described data-driven multi-regional power system economic environment dispatching method.

[0167] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0168] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0169] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment, whereby a series of operational steps are performed to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0170] The descriptions of each embodiment in the above embodiments have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0171] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A data-driven multi-regional power system economic environment dispatching method, characterized by: include: To minimize pollutant emissions and fossil fuel costs, a multi-objective optimization model for economic and environmental dispatching of multi-regional power systems is constructed, taking into account multiple constraints. Based on the objectives and multiple constraints, a data-driven proxy model is constructed to transform the multi-objective optimization model for economic environment dispatching of multi-regional power systems. The specific steps include: solving for several feasible solutions that satisfy the constraints; calculating the optimization objective value corresponding to each feasible solution based on the feasible solutions and the multi-objective optimization model for economic environment dispatching of multi-regional power systems; and constructing a data-driven proxy model based on deep belief networks to replace the multi-objective optimization model for economic environment dispatching of multi-regional power systems based on the feasible solutions and their corresponding optimization objective values. The multiple constraints include generator active power upper and lower limits, power balance constraints, node voltage amplitude constraints, line power flow constraints, transmission line safety constraints, regional hot reserve transfer constraints, and line loss constraints. The data-driven proxy model is a regression proxy model that combines a deep belief network and a backpropagation neural network. After constructing the data-driven proxy model for the first region, a pre-training-fine-tuning transfer learning technique is applied to quickly build proxy models for other regions. The multi-objective antlion algorithm is used to solve the transformed multi-regional power system economic environment dispatch multi-objective optimization model. The specific steps include: initializing the ant and antlion populations, and randomly initializing the power generation of each generator in each region; assigning the ant and antlion populations to the corresponding randomly initialized power generation of each generator in each region, and solving for the non-dominated solution for each optimization objective; traversing all individuals in the population, and calculating the non-dominated solution based on the individual optimization results; wherein, when traversing all individuals in the population, a single-dimensional retention strategy is performed to update the ant population; sorting the non-dominated solutions according to the initial values; and obtaining the globally optimal solution based on the solution results as the decision-making basis for the optimal dispatch of each region of the power system in the current time period. The single-dimensional retention strategy specifically includes: The population is divided into four parts: the population where the fitness values ​​of both Target 1 and Target 2 are less than the average fitness values ​​of the corresponding Target 1 and Target 2 in the current population is named the win-win population; the population where only the fitness value of Target 1 is less than the average fitness value of the corresponding Target 1 in the current population is named the first winning population; the population where only the fitness value of Target 2 is less than the average fitness value of the corresponding Target 2 in the current population is named the second winning population; and the population where the fitness values ​​of both Target 1 and Target 2 are greater than the average fitness values ​​of the corresponding Target 1 and Target 2 in the current population is named the loser population. Find the minimum fitness values ​​of objective one and objective two in the population, and name them as minimum one and minimum two. The corresponding individuals are named minimum decision one and minimum decision two. Different operations are performed on the four population groups: the win-win population moves towards the direction of the vector sum of the first and second minimum decisions; the first-win population moves towards the first minimum decision; the second-win population moves towards the second minimum decision; and the losing population moves towards the direction of the vector sum of the first and second minimum decisions. The overall population position is named... ; Find the maximum fitness values ​​of objective one and objective two in the population, and name them as maximum one and maximum two. The corresponding individuals are named maximum one decision and maximum two decision. For the four population groups, the following actions are taken to determine which groups are more advantageous and which are less advantageous: The winning population moves towards the direction of the sum of the vectors of the first and second minimum decisions and away from the direction of the sum of the vectors of the first and second maximum decisions; the first-victorious population moves towards the direction of the first minimum decision and away from the direction of the first maximum decision; the second-victorious population moves towards the direction of the second minimum decision and away from the direction of the second maximum decision; and the losing population moves towards the direction of the sum of the vectors of the first and second minimum decisions and away from the direction of the sum of the vectors of the first and second maximum decisions. The overall population position is named as follows: ; Find the particle with the smallest niche radius in the population, and move all particles towards the particle with the smallest niche radius. Name the overall population position as... ; The new population location is , and The vector sum, where , and The corresponding coefficients sum to 1; The single-dimensional optimal retention strategy is implemented. For a particle, each dimension is judged. First, a random number that is uniformly distributed between 0 and 1 is generated. If the random number is greater than 0.2, the dimension of the particle is not changed. If the random number is less than or equal to 0.2, the dimension of the particle is changed to the corresponding dimension of the new population position.

2. The data-driven multi-regional power system economic environment dispatching method as described in claim 1, characterized in that, The multi-objective optimization model for economic environment dispatching of multi-regional power systems is expressed as follows: in, Let E(P) be the objective function for optimizing fossil fuel costs, and let E(P) be the objective function for optimizing pollutant emissions. For the inequality constraints involved, For the equality constraints involved.

3. The data-driven multi-regional power system economic environment dispatching method as described in claim 1, characterized in that, The specific steps for obtaining the global optimal solution based on the solution results include: For the optimization objective function of pollutant gas emissions and fossil fuel costs, the membership function value corresponding to its non-dominated solution is calculated using the membership function. The membership function values ​​are regularized and solved to obtain the global optimal solution.

4. A multi-regional power system economic environment dispatching system that implements the data-driven multi-regional power system economic environment dispatching method as described in any one of claims 1-3, characterized in that, include: The model building module is configured to construct a multi-objective optimization model for economic and environmental dispatching of a multi-regional power system, taking into account multiple constraints, with the goal of minimizing pollutant emissions and fossil fuel costs. The solution module is configured to: construct a data-driven proxy model based on the objectives and constraints to transform the multi-regional power system economic environment scheduling multi-objective optimization model, and solve the transformed multi-regional power system economic environment scheduling multi-objective optimization model through the multi-objective antlion algorithm; The decision-making module is configured to obtain the global optimal solution based on the solution results, and use it as the basis for decision-making on the optimal scheduling of each region of the power system in the current period.

5. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the steps of the method according to any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, perform the steps of the method according to any one of claims 1-3.

Images