Power distribution network optimization scheduling method based on overload scene prediction

Abstract

According to the power distribution network optimization scheduling method based on overload scene prediction, through cooperation of multiple modules in a power distribution network electrical model, the accuracy of power distribution network operation can be effectively improved, the power distribution network electrical model is constructed firstly, then a precise load prediction result is output by using historical load data, and the scene type is judged in combination with the model; and when the three typical scenes of heavy load, overload and heavy overload are identified, minimizing the number of the heavy load, overload and heavy overload branches is taken as a target function, the optimal operation mode and distribution load of the line under the three typical scenes of heavy load, overload and heavy overload are optimized and calculated, and finally an optimized scheduling scheme is generated, so that the intelligent management level of the power distribution network is improved, and the power distribution network is optimized. Risks of heavy load and overload are reduced and stable and reliable operation of the power distribution network is guaranteed.

Description

Technical Field

[0001] This application relates to the field of power system distribution network optimization control technology, and in particular to a distribution network optimization scheduling method based on overload scenario prediction. Background Technology

[0002] With the advancement of the "dual-carbon" strategy, distributed power sources (such as rooftop photovoltaics and small wind power) and diversified loads (electric vehicle charging stations, industrial, commercial, and residential loads) are being connected to the 10kV distribution network on a large scale. This has led to frequent problems such as equipment and line overload, making the traditional "source follows load" dispatching model ill-suited to the dynamically changing multi-source characteristics. The core issues are concentrated in the following three aspects: Weak scenario identification capabilities lead to a disconnect between dispatch strategies and actual needs. The operating status of the distribution network is affected by multiple factors such as load distribution, renewable energy output, and user electricity consumption habits, exhibiting significant temporal and spatial heterogeneity. Existing applications have weak capabilities in classifying different load scenarios, and the differentiation of different load scenarios mostly relies on manual methods. There is a lack of intelligent means for hierarchical control of different load scenarios, and the control methods lack predictability and have insufficient theoretical basis. Some control schemes analyzed manually have poor implementation effects and are difficult to meet real-time response requirements.

[0003] Multi-objective optimization scheduling needs to simultaneously satisfy constraints such as load factor, voltage qualification rate, and number of switching actions. Traditional genetic algorithms (GA) have strong global search capabilities but slow convergence speed, while particle swarm optimization (PSO) has high efficiency in local optimization but is prone to getting trapped in local optima. In contrast, mixed-integer linear programming (MILP) has advantages such as global optimum guarantee, flexible constraint handling, and simple parameter tuning, making it more suitable for real-time topology calculations in distribution networks, but it has not yet been widely used in load scheduling driven by typical scenarios.

[0004] There is insufficient coordination between load forecasting and scheduling optimization. In existing technologies, load forecasting is disconnected from subsequent operational performance optimization, making it difficult to directly and efficiently apply the forecast results to scheduling strategy formulation, thus limiting the overall improvement in operational performance.

[0005] Therefore, it is necessary to address the common heavy overload problem by selectively improving the predictability, scientific rigor, and intelligence of scheduling methods to meet the operational requirements of accurate prediction and rapid response coordination. Summary of the Invention

[0006] To address the problems of weak scenario identification capability, poor scheduling efficiency, and insufficient coordination between load forecasting and scheduling optimization in existing distribution networks, this invention proposes a distribution network optimization scheduling method based on overload scenario forecasting.

[0007] To achieve the above-mentioned technical effects, the technical solution of the present invention is as follows: A distribution network optimization scheduling method based on overload scenario prediction includes the following steps: S1: Construct the electrical model of the power distribution network; S2: Historical load data is used to construct a hybrid forecasting model to output load forecasting results; S3: Import the load forecast results into the distribution network electrical model to identify three typical scenarios: heavy load, overload, and severe overload. S4: With the objective function of minimizing the number of branches under heavy load, overload, and severe overload, optimize the optimal operation mode and load distribution of the line under three typical scenarios: heavy load, overload, and severe overload.

[0008] Compared with the prior art, the beneficial effects of the technical solution of the present invention are: This invention proposes a distribution network optimization scheduling method based on overload scenario prediction. By coordinating multiple modules in the distribution network electrical model, the accuracy of distribution network operation can be effectively improved. First, the distribution network electrical model is constructed, and then historical load data is used to output accurate load prediction results. The model is combined with the scenario type to determine the scenario type. When three typical scenarios of heavy load, overload, and severe overload are identified, the optimal operating mode and load allocation of the lines under the three typical scenarios of heavy load, overload, and severe overload are optimized with the objective function of minimizing the number of branches under these scenarios. Finally, an optimized scheduling scheme is generated, which improves the intelligent management level of the distribution network, reduces the risks of heavy load and overload, and ensures the stable and reliable operation of the distribution network. Attached Figure Description

[0009] Figure 1 This is a flowchart of a distribution network optimization scheduling method based on overload scenario prediction in one embodiment; Figure 2 Here is a diagram of the hybrid prediction model structure in one embodiment; Figure 3 This is a flowchart of a mixed-integer linear programming algorithm in one embodiment. Detailed Implementation

[0010] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0011] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0012] It should be understood that although the terms first, second, third, etc., may be used in this invention to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of this invention, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0013] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0014] Example 1 This embodiment proposes a distribution network optimization scheduling method based on overload scenario prediction, the flowchart of which is as follows: Figure 1 As shown, it includes the following steps: S1: Construct the electrical model of the power distribution network; S2: Historical load data is used to construct a hybrid forecasting model to output load forecasting results; Furthermore, the hybrid prediction model is an LSTM-attention mechanism hybrid prediction model, and its model structure diagram is as follows. Figure 2 As shown.

[0015] S3: Import the load forecast results into the distribution network electrical model to identify three typical scenarios: heavy load, overload, and severe overload. S4: With the objective function of minimizing the number of branches under heavy load, overload, and severe overload, optimize the optimal operation mode and load distribution of the line under three typical scenarios: heavy load, overload, and severe overload.

[0016] In this embodiment, the accuracy of distribution network operation can be effectively improved through the collaboration of multiple modules in the distribution network electrical model. First, the distribution network electrical model is constructed, and then historical load data is used to output accurate load prediction results. The model is combined to determine the scenario type. When three typical scenarios of heavy load, overload, and severe overload are identified, the objective function is to minimize the number of branches under the three typical scenarios of heavy load, overload, and severe overload. The optimal operation mode of the line and the load allocation under the three typical scenarios of heavy load, overload, and severe overload are optimized and calculated. Finally, an optimized scheduling scheme is generated, which improves the intelligent management level of the distribution network, reduces the risks of heavy load and overload, and ensures the stable and reliable operation of the distribution network.

[0017] Example 2 This embodiment further explains the present invention based on Embodiment 1.

[0018] In one optional embodiment, the power distribution network electrical model consists of a power supply module, a line module, a load module, a switch module, and a measurement module.

[0019] In one optional embodiment, the power module collects the operation and maintenance log data of the substation to which the target distribution network belongs and the parameters of the 110kV / 10kV main transformer in the substation to which the target distribution network belongs, and builds a power unit based on the three-phase voltage source model. The 110kV output voltage of the three-phase voltage source model is converted into 10kV distribution network voltage through the π-type equivalent transformer model, and the first-end bus voltage is output to the line module. For example, a project file is created on the Matlab Simulink platform, the simulation step size is set to milliseconds, and a variable step size solver suitable for rigid networks is selected; the power supply module is a substation, the load module includes industrial and commercial loads and residential loads, the line module contains impedance parameters, the switch module performs opening and closing control, and the measurement module performs current and voltage monitoring. For example, the substation operation and maintenance log is collected in the power supply module. Main transformer capacity Short-circuit impedance A three-phase voltage source is connected via a transformer, according to the formula. Calculate the voltage at the first bus; Furthermore, the power module also acquires Main transformer rated capacity Short-circuit impedance With parameters such as these, the three-phase voltage source model is as follows: π-type equivalent transformer model turns ratio The target distribution network's historical load data is collected from the 5-minute data stored in the distribution network dispatch automation system; The line module collects line topology data, obtains positive sequence resistance and reactance from the operation and maintenance ledger data, calculates line impedance using an equivalent model, calculates node voltage based on line impedance and head bus voltage, and outputs it to the load module. The load module obtains the rated power and power factor of industrial, commercial and residential loads from the operation and maintenance ledger data, calculates the load current using an equivalent model and outputs it to the switch module; The switching module collects the rated current, controls the opening and closing of the circuit breaker through logic signals, outputs the switch status according to the logic of controlling the opening and closing of the circuit breaker, and corrects the topology of the line module. The measurement module calculates the voltage and current measured in the power distribution network electrical model based on Fourier transform. RMS values ​​of voltage and current at each stage; The power distribution network electrical model also includes a verification module, which performs power flow calculations on historical load data and iterates using the Newton-Raphson method to ensure that the node voltage error is less than or equal to a preset value, for example, the node voltage error is less than or equal to 1%.

[0020] In one optional embodiment, the power module collects operation and maintenance log data of the substations belonging to the target distribution network and parameters of the 110kV / 10kV main transformers within the substations. Based on a three-phase voltage source model, it builds a power unit and converts the 110kV output voltage of the three-phase voltage source model into a 10kV distribution network voltage using an equivalent transformer model, outputting the head-end bus voltage; its expression is:

[0021] in, This indicates the voltage at the first bus. This represents the 110kV output voltage of a three-phase voltage source model. Indicates the transformer turns ratio. Indicates the voltage drop across the transformer; The collected line topology data is used to obtain the positive sequence resistance and reactance from the maintenance log data, and the line impedance is calculated using an equivalent model; its expression is:

[0022] in, Indicates line impedance. Indicates positive sequence resistance. represents an imaginary number, Indicates reactance, Indicates the length of the line; Furthermore, the line topology data includes length L and conductor type, using... The equivalent model is used to input the positive and zero sequence impedance parameters. The line length is input segment by segment according to the line topology, and the impedance is calculated according to the above formula.

[0023] The process involves obtaining the rated power and power factor of industrial, commercial, and residential loads from the maintenance ledger data, and then calculating the load current using an equivalent model; the expression is as follows:

[0024] in, Indicates the load current. Indicates active power. Indicates reactive power. Indicates node voltage; Active power is calculated based on the rated power of industrial, commercial, and residential loads, power factor, and ZIP equivalent model coefficients; its expression is:

[0025] in, Indicates the load baseline active power. Represents the coefficients of the ZIP equivalent model. This indicates the voltage of the power distribution network.

[0026] Furthermore, Indicates the rated power of industrial, commercial, and residential loads. Indicates the power factor; For example, industrial ( ) are established using ZIP equivalent models respectively. , , ),Business( , , ),resident( , , The three subsystems have rated power and power factor taken from the line operation and maintenance log.

[0027] The rated current is collected, and the opening and closing of the circuit breaker is controlled by logic signals. The logic expression for controlling the opening and closing of the circuit breaker is as follows:

[0028] in, Indicates the open / closed status. Indicates the load current. Indicates the rated current.

[0029] Furthermore, the switch module is also configured with a programmable circuit breaker, exemplarily model ZW20-12, rated current 630A, when... The time indicates that the circuit is closed. The time indicates the circuit breaker to open; the opening / closing time is set to milliseconds. Remote control is achieved through Simulink logic signals. ( ), ( ) Perform the switch action.

[0030] In one alternative embodiment, the step involves calculating based on the voltage and current measured in the distribution network electrical model and using Fourier transform. The effective values ​​of the voltage and current at each stage; their expressions are:

[0031] in, Indicates the effective value of the voltage. Indicates the power frequency period, Indicates the instantaneous voltage value. Indicates the effective value of the current. This represents the instantaneous value of the current.

[0032] Furthermore, measuring elements are arranged at the beginning and end of the main branch line. For example, three-phase voltage and current measuring elements are arranged, and the effective value at the 5-minute level and the power frequency cycle are calculated according to the above formula. .

[0033] In one optional embodiment, identifying the three typical scenarios of heavy load, overload, and severe overload includes selecting feature quantities as the basis for scenario division, and includes the following steps: The load forecast results are input into the distribution network electrical model, and the power flow calculation is run to obtain the actual current of the line. The characteristic quantities are calculated based on the actual current of the line and the collected rated current of the line. The characteristic quantity is the ratio of the actual current of the line to the sampled rated current of the line; its expression is:

[0034] in, Representing characteristic quantities, Indicates the actual current of the line. This indicates the rated current being collected; Based on the characteristic threshold, three typical scenarios—heavy load, overload, and severe overload—are classified.

[0035] Further, exemplarily, the distribution network scenario types are classified according to feature threshold values, and the distribution network scenario types include: Normal scenario: Feature quantity K ≤ 80%; Heavy-load scenarios: 80% < feature quantity K ≤ 100%; Overload scenario: 100% < feature K ≤ 120%; Heavy overload scenario: Feature quantity K > 120%.

[0036] In one optional embodiment, S3 specifically includes the following steps: S31: Import the load forecast results into the distribution network electrical model and replace the original historical load data in the distribution network electrical model to obtain a new distribution network electrical model; S32: Based on the line module in the new distribution network electrical model The equivalent model and the ZIP model of the load module are combined with power flow calculation to calculate the actual line current. Based on the actual line current and the collected rated current, the characteristic quantities corresponding to the load prediction results are calculated. S33: Based on the feature threshold, the feature quantity corresponding to the load prediction result is automatically output as a scene number through Matlab logic judgment to complete scene recognition.

[0037] Further, for example, the load forecast results are imported into the distribution network electrical model to calculate the current topology of the main line and branch lines under the forecast results. Based on the classification threshold of the distribution network scenario, Matlab is used to determine the scenario type of different forecast results. Based on the forecast results and the distribution network electrical model, scenario identification is achieved through data integration calculation, as follows: Data import: Import the 24-hour 5-minute load forecast values ​​(288 numerical points, restored by inverse normalization) of the output load forecast results into the distribution network electrical model, replacing the historical data of the load module; Main and branch line current calculation: Perform power flow calculations in the distribution network electrical model, based on line modules. Type equivalent model ( Using the load module ZIP model, the actual current of each main branch is calculated iteratively. I real (Current source: Based on the power distribution network electrical model circuit module) Iterative calculations of the ZIP model and load module ), that is, the output of the measurement module I rms , combined I rated Calculate the characteristic quantity K; Scene identification: Use a Matlab logic judgment program to identify the scene type; After determining the scenario type, the optimal scheduling scheme for the line is calculated through simulation using the mixed integer linear programming (MILP) algorithm.

[0038] In this embodiment, electrical simulation calculations of the main line and branch line currents under different load scenarios are used to establish four typical scenario classification standards: normal, heavy load, overload, and severe overload. Based on the constructed load prediction model, future load data of the line is obtained. Through electrical line topology simulation calculations, comprehensive prediction of the main line and branch line currents is achieved. Logical judgments are made based on the prediction results, and the optimal line scheduling scheme is calculated through MILP algorithm simulation based on the judgment results.

[0039] Furthermore, by combining dynamic identification technology with the analysis of typical scenarios in distribution networks, the optimal operating mode is calculated based on the key characteristics of distribution networks under typical scenarios such as normal, overload, heavy load, and severe overload. The portability and implementation efficiency of the project are significantly improved by using scenario classification thresholds. The collaborative research on predictive scheduling optimization throughout the entire process and the modular processing research model provide theoretical guidance and practical basis for intelligent scheduling strategies of distribution networks, thereby improving the level of intelligent management of distribution networks.

[0040] Example 3 This embodiment details the use of the Mixed Integer Linear Programming (MILP) algorithm based on Embodiments 1 and 2, and its flowchart is shown below. Figure 3 As shown, the calculation and identification results indicate the optimal operating mode of the line under three typical scenarios: heavy load, overload, and severe overload, and the load is allocated to control the load rate of each line and equipment.

[0041] In one optional embodiment, S4 specifically includes the following steps: S41: Establish Boolean variables for the automated switches used for power transfer. ,in Indicates that the circuit is closed. This indicates that the circuit breaker is open; the other switches remain in their original states. S42: The objective function is to minimize the number of heavily loaded, overloaded, and heavily overloaded branches; its expression is:

[0042] in, Describe the objective function. Indicates the first A side road, Represents a binary variable, where, When the first a side road The value is 1 if the condition is met, otherwise it is 0. S43: Set constraints for node voltage, characteristic quantities, and switching actions. The node voltage constraint is as follows: The characteristic constraints are: The switching action constraint is: the number of switching actions within a single scheduling cycle. .

[0043] Furthermore, node voltage constraints Feature constraints: Switch action constraints: .

[0044] In an optional embodiment, S4 further includes the following step: S44: Use a mixed-integer linear programming algorithm to solve for the optimal switching action vector, and obtain the optimal operation mode and load allocation of the line based on the optimal switching action vector; S44 specifically includes the following steps: S441: Using Matlab intlinprog to globally solve for the optimal switching action vector For the optimal switching action vector Perform a 2-opt neighborhood search, exchange adjacent switch states while maintaining constraints, obtain the optimal line operation mode and load allocation based on the optimal switch action vector, and generate a distribution network optimization scheduling scheme based on the optimal line operation mode and load allocation. S442: When computing power is limited, the generated distribution network optimization scheduling scheme degenerates into a heuristic reverse supply strategy combined with a local switching strategy.

[0045] In an optional embodiment, S4 further includes the following step: S45: Substitute the optimal operating mode of the line and the distributed load into the distribution network electrical model for verification; S45 specifically includes the following steps: The generated distribution network optimization scheduling scheme is substituted into the distribution network electrical model for verification. If the characteristic quantity is no more than 80% and the node voltage meets the node voltage constraint, the verification is passed and the generated distribution network optimization scheduling scheme is output; otherwise, the verification fails. If the verification fails, the suboptimal running mode that minimizes the objective function under the current constraints will be output.

[0046] Furthermore, the states of adjacent switches are exchanged to reduce network losses. And maintain constraints.

[0047] Furthermore, if the verification fails, the suboptimal scheduling scheme that minimizes the objective function under the constraints of the current node voltage, load rate (i.e., characteristic quantity), and switching action is output.

[0048] Furthermore, the simulation verification standard is as follows: the calculation results (optimal scheduling scheme) are substituted into the distribution network electrical model for verification, and the following conditions must be met: the characteristic quantity K of each branch is ≤80%, and the node voltage meets the constraints; If the above requirements are met, confirm the validity of the operation mode switch; if not, output the suboptimal operation mode that minimizes the objective function F under the current constraints as the optimal scheduling scheme. The entire process runs on general-purpose computing hardware in ≤20s.

[0049] In this embodiment, mixed integer linear programming (MILP) is used to solve the prediction results globally to achieve the optimal solution. This can simultaneously handle multiple constraints such as load rate, voltage, network loss, and number of switching actions, ensuring that an executable heavy overload elimination solution is provided within 20 seconds.

[0050] In this embodiment, the line topology is adjusted experimentally based on the calculation results to verify the load rate control effect and ensure the effectiveness of the adjustment method. In actual situations, when the load forecast results are identified as having three types of conditions: heavy load, overload, and severe overload, the system automatically substitutes the forecast data into the line calculation to obtain the optimal scheduling scheme. After the pre-testing of the adjustment scheme proves feasible, the system automatically adjusts the line operation mode and executes the load-splitting distribution network performance improvement method.

[0051] Example 4 This embodiment provides a detailed description of the process of building, training, and using a hybrid prediction model.

[0052] Step 1: Historical Load Data Acquisition and Preprocessing: Data is collected from the distribution network dispatch automation system. Historical load data (time span) (Year, including workdays / restdays, seasons, and extreme weather loads); missing values ​​were filled using Lagrange interpolation. Guidelines ( Remove outliers, Min-Max normalization ( Processing data; Step 2: Input the preprocessed historical load data into the LSTM layer of the hybrid prediction model to obtain the hidden state at the current time; including the following steps: First, calculate the output of the forget gate ( ): Through the Sigmoid activation function ( Handling the hidden state from the previous moment ( ) and the current input ( After splicing and The product, plus The result is that Used to determine the cell state at the previous moment ( The retention ratio of the input gate is then calculated. Next, the input gate output (...) is calculated. and candidate cell states: (Determines the update ratio of the input information at the current moment). Then update the cell state ( ): ( This is an element-wise multiplication, with the cell state updated by combining the results of the forget gate and the input gate. Finally, the hidden state at the current time step is calculated. ): Output As an intermediate result of the LSTM layer, it is passed to the subsequent additive attention layer.

[0053] in, This represents the forget gate weight matrix. This represents the input gate weight matrix. This represents the output gate weight matrix. This represents the candidate cell state bias term. Indicates the output gate bias term. This represents the input gate bias term.

[0054] For example, the input layer has 7 dimensions (historical load data, industrial and commercial load ratio, residential load ratio, temperature, season, holidays, and day of the week), and the hidden layer has 64 neurons.

[0055] Furthermore, each neuron corresponds to the aforementioned temporal computation process, and the outputs of all neurons together constitute the complete hidden state sequence of the LSTM layer. (T is the time step), used for attention layer calculation.

[0056] Step 3: Input the hidden state sequence into the additive attention layer to obtain the load prediction result.

[0057] Furthermore, the hidden state sequence output by the LSTM layer is obtained. , For the current time step, This is used as the input to the additive attention layer, and the attention score is calculated according to the additive attention formula. ): ,in The current hidden state is respectively Historical moments hidden state The weight matrix, As a bias term, the tanh function is used to map the attention score to the interval [-1, 1]. Attention weight normalization: The attention score is normalized using the Softmax function to obtain the attention weights. ): This ensures that the sum of the attention weights of all historical moments to the current moment is 1, highlighting the contribution of key temporal information. Attention-weighted output: Calculate the attention-weighted hidden state (…). ): , The key information of the historical hidden state is incorporated and passed as the output of the additive attention layer to the first fully connected layer to obtain the feature vector. The feature vector is then input into the second fully connected layer to obtain the load prediction result.

[0058] Furthermore, the dataset used by the hybrid prediction model is divided into training / validation / test sets in a 7:2:1 ratio, and the Adam optimizer (learning rate) is applied. Minimize RMSE loss ( After training, input future feature data and output... The load forecast results are used for subsequent scenario identification and optimized scheduling.

[0059] The various embodiments in this invention are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The device embodiments described above are merely exemplary. The modules described as separate components may or may not be physically separate. When implementing the present invention, the functions of each module can be implemented in one or more software and / or hardware. Alternatively, some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0060] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A distribution network optimization scheduling method based on overload scenario prediction, characterized in that, Includes the following steps: S1: Construct the electrical model of the power distribution network; S2: Historical load data is used to construct a hybrid forecasting model to output load forecasting results; S3: Import the load forecast results into the distribution network electrical model to identify three typical scenarios: heavy load, overload, and severe overload. S4: With the objective function of minimizing the number of branches under heavy load, overload, and severe overload, optimize the optimal operation mode and load distribution of the line under three typical scenarios: heavy load, overload, and severe overload.

2. The distribution network optimization scheduling method based on overload scenario prediction according to claim 1, characterized in that, The power distribution network electrical model consists of a power supply module, a line module, a load module, a switch module, and a measurement module.

3. The distribution network optimization scheduling method based on overload scenario prediction according to claim 2, characterized in that, The power module collects operation and maintenance log data of the substations belonging to the target distribution network and parameters of the 110kV / 10kV main transformers in the substations belonging to the target distribution network. Based on the three-phase voltage source model, it builds a power unit and converts the 110kV output voltage of the three-phase voltage source model into the 10kV distribution network voltage through the equivalent transformer model, and outputs the first-end bus voltage to the line module. The line module collects line topology data, obtains positive sequence resistance and reactance from the operation and maintenance ledger data, calculates line impedance using an equivalent model, calculates node voltage based on line impedance and head bus voltage, and outputs it to the load module. The load module obtains the rated power and power factor of industrial, commercial and residential loads from the operation and maintenance ledger data, calculates the load current using an equivalent model and outputs it to the switch module; The switching module collects the rated current, controls the opening and closing of the circuit breaker through logic signals, outputs the switch status according to the logic of controlling the opening and closing of the circuit breaker, and corrects the topology of the line module. The measurement module calculates the voltage and current measured in the power distribution network electrical model based on Fourier transform. RMS values ​​of voltage and current at each stage; The power distribution network electrical model also includes a verification module, which performs power flow calculations on historical load data and uses the Newton-Raphson method for iteration to ensure that the node voltage error is less than or equal to a preset value.

4. The distribution network optimization scheduling method based on overload scenario prediction according to claim 3, characterized in that, The power module collects operation and maintenance log data from the substations belonging to the target distribution network and parameters of the 110kV / 10kV main transformers within those substations. Based on a three-phase voltage source model, it builds a power unit and converts the 110kV output voltage of the three-phase voltage source model into a 10kV distribution network voltage using an equivalent transformer model, outputting the head-end bus voltage. Its expression is: in, This indicates the voltage at the first bus. This represents the 110kV output voltage of a three-phase voltage source model. Indicates the transformer turns ratio. Indicates the voltage drop across the transformer; The collected line topology data is used to obtain the positive sequence resistance and reactance from the maintenance log data, and the line impedance is calculated using an equivalent model; its expression is: in, Indicates the line impedance. Indicates positive sequence resistance. represents an imaginary number, Indicates reactance, Indicates the length of the line; The process involves obtaining the rated power and power factor of industrial, commercial, and residential loads from the maintenance ledger data, and then calculating the load current using an equivalent model; the expression is as follows: in, Indicates the load current. Indicates active power. Indicates reactive power. Indicates node voltage; Active power is calculated based on the rated power of industrial, commercial, and residential loads, power factor, and ZIP equivalent model coefficients; its expression is: in, Indicates the load baseline active power. Represents the coefficients of the ZIP equivalent model. This indicates the voltage of the power distribution network.

5. The distribution network optimization scheduling method based on overload scenario prediction according to claim 3, characterized in that, The voltage and current measured in the power distribution network electrical model are used to calculate based on Fourier transform. RMS values ​​of voltage and current at each stage; Its expression is: in, Indicates the effective value of the voltage. Indicates the power frequency period, Indicates the instantaneous voltage value. Indicates the effective value of the current. This represents the instantaneous value of the current.

6. The distribution network optimization scheduling method based on overload scenario prediction according to claim 1, characterized in that, Identifying three typical scenarios—heavy load, overload, and severe overload—involves selecting feature quantities as the basis for scenario classification, and includes the following steps: The load forecast results are input into the distribution network electrical model, and the power flow calculation is run to obtain the actual current of the line. The characteristic quantities are calculated based on the actual current of the line and the collected rated current of the line. The characteristic quantity is the ratio of the actual current of the line to the sampled rated current of the line; its expression is: in, Representing characteristic quantities, Indicates the actual current of the line. This indicates the rated current being collected; Based on the characteristic threshold, three typical scenarios—heavy load, overload, and severe overload—are classified.

7. The distribution network optimization scheduling method based on overload scenario prediction according to claim 6, characterized in that, S3 specifically includes the following steps: S31: Import the load forecast results into the distribution network electrical model and replace the original historical load data in the distribution network electrical model to obtain a new distribution network electrical model; S32: Based on the line module in the new distribution network electrical model The equivalent model and the ZIP model of the load module are combined with power flow calculation to calculate the actual line current. Based on the actual line current and the collected rated current, the characteristic quantities corresponding to the load prediction results are calculated. S33: Based on the feature threshold, the feature quantity corresponding to the load prediction result is automatically output as a scene number through Matlab logic judgment to complete scene recognition.

8. The distribution network optimization scheduling method based on overload scenario prediction according to claim 1, characterized in that, S4 specifically includes the following steps: S41: Establish Boolean variables for the automated switches used for power transfer; S42: The objective function is to minimize the number of heavily loaded, overloaded, and heavily overloaded branches; its expression is: in, Describe the objective function. Indicates the first A side road, Represents a binary variable; S43: Set constraints for node voltage, characteristic quantities, and switching actions. The node voltage constraint is as follows: The characteristic constraints are: The switching action constraint is: the number of switching actions within a single scheduling cycle. .

9. A distribution network optimization scheduling method based on overload scenario prediction according to claim 8, characterized in that, S4 also includes the following steps: S44: Use a mixed-integer linear programming algorithm to solve for the optimal switching action vector, and obtain the optimal operation mode and load allocation of the line based on the optimal switching action vector; S44 specifically includes the following steps: S441: Using Matlab intlinprog to globally solve for the optimal switching action vector For the optimal switching action vector Perform a 2-opt neighborhood search, exchange adjacent switch states while maintaining constraints, obtain the optimal line operation mode and load allocation based on the optimal switch action vector, and generate a distribution network optimization scheduling scheme based on the optimal line operation mode and load allocation. S442: When computing power is limited, the generated distribution network optimization scheduling scheme degenerates into a heuristic reverse supply strategy combined with a local switching strategy.

10. A distribution network optimization scheduling method based on overload scenario prediction according to claim 9, characterized in that, S4 also includes the following steps: S45: Substitute the optimal operating mode of the line and the distributed load into the distribution network electrical model for verification; S45 specifically includes the following steps: The generated distribution network optimization scheduling scheme is substituted into the distribution network electrical model for verification. If the characteristic quantity is no more than 80% and the node voltage meets the node voltage constraint, the verification is passed and the generated distribution network optimization scheduling scheme is output; otherwise, the verification fails. If the verification fails, the suboptimal running mode that minimizes the objective function under the current constraints will be output.

Images