Power distribution network double-layer optimization method for comprehensive bearing capacity improvement

Abstract

The invention provides a power distribution network double-layer optimization method for comprehensive bearing capacity improvement, and belongs to the technical field of power distribution networks. A power distribution network comprehensive bearing capacity three-layer evaluation system is constructed, and a double-layer optimization framework is adopted to take site selection and sizing of an energy storage system and a reactive power compensation device as upper-layer optimization; the minimum active power loss and voltage offset are taken as targets, equipment operation scheduling is taken as lower-layer optimization, the maximum comprehensive bearing capacity score is taken as a target, an improved grey wolf optimization algorithm is utilized to solve an upper-layer and lower-layer optimization problem, and collaborative optimization of equipment configuration and an operation strategy is realized through an iterative feedback mechanism. Meanwhile, a rolling optimization time window and model prediction control are introduced to improve the optimization precision and the adaptive capacity to the fluctuation of the distributed power supply, and the technical problem that equipment configuration optimization and operation scheduling optimization are difficult to realize at the same time in a large-scale distributed power supply access scene of the power distribution network is solved.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network technology, and more specifically, relates to a two-layer optimization method for power distribution networks aimed at improving comprehensive carrying capacity. Background Technology

[0002] With the rapid development of renewable energy power generation technologies, the penetration rate of distributed power sources in distribution networks is constantly increasing. Distribution networks need to improve their carrying capacity through optimized configurations of equipment such as energy storage systems and reactive power compensation devices. Traditional distribution network optimization methods typically separate equipment location and capacity determination from operation and scheduling. They first determine the installation location and capacity parameters of the equipment, and then optimize operation based on a fixed equipment configuration. This approach ignores the coupling relationship between equipment configuration schemes and operational strategies. Since equipment configuration decisions directly affect the subsequent operation and scheduling effects, and the quality of operation and scheduling, in turn, affects the rationality of the equipment configuration scheme, separate optimization cannot guarantee global optimality. In existing technologies, distribution network planners often configure equipment based on experience or a single objective, lacking a comprehensive evaluation of multi-dimensional performance such as safety, flexibility, economy, and low carbon emissions. This leads to a disconnect between equipment configuration schemes and actual operational needs, failing to fully leverage the coordinated control capabilities of the equipment. In other words, existing technologies present a technical problem: distribution networks struggle to simultaneously achieve equipment configuration optimization and operation and scheduling optimization in scenarios with large-scale distributed power source integration. Summary of the Invention

[0003] In view of this, the present invention provides a two-layer optimization method for distribution networks aimed at improving comprehensive carrying capacity, which can solve the technical problem in the prior art that it is difficult to simultaneously achieve equipment configuration optimization and operation scheduling optimization in distribution networks under large-scale distributed power source access scenarios.

[0004] This invention is implemented as follows: This invention provides a two-layer optimization method for distribution networks aimed at improving comprehensive carrying capacity, comprising the following steps: Constructing a three-layer evaluation system for the comprehensive carrying capacity of the distribution network; collecting node voltage values, distributed power output values, and node load values; calculating the original index values ​​of the index layer and performing normalization processing, positiveizing negative indicators, and calculating index weight values ​​using the analytic hierarchy process (AHP); establishing a comprehensive carrying capacity scoring calculation model; constructing an upper-layer optimization model; using an improved gray wolf optimization algorithm to optimize the location and capacity of energy storage systems and reactive power compensation devices; when active power loss exceeds a preset loss threshold, dividing the 24-hour optimization period into four rolling optimization time windows and re-executing the upper-layer optimization; constructing a lower-layer optimization model. The model, with the objective function of maximizing the comprehensive carrying capacity score, employs an improved gray wolf optimization algorithm to coordinate and optimize the energy storage system's charging and discharging power sequence, capacitor switching group sequence, static var compensator reactive power regulation capacity sequence, and distributed generation output sequence. When the distributed generation output deviation percentage exceeds 15%, a model predictive control algorithm is used to predict the node voltage prediction sequence and adjust the reactive power regulation capacity sequence and switching group sequence. The lower-level optimization results are substituted into the comprehensive carrying capacity score calculation model to calculate the optimized comprehensive carrying capacity score and fed back to the upper level for re-iteration. The iteration terminates when the change in the optimized comprehensive carrying capacity score obtained after three consecutive iterations is less than 0.5% or the number of iterations reaches 50.

[0005] In the comprehensive carrying capacity three-layer evaluation system, the target layer is the distributed power source access adaptability, the performance layer includes safety, flexibility, economy and low carbon emissions, and the indicator layer includes voltage deviation rate, voltage qualification rate, net load fluctuation rate, renewable energy utilization rate, operating cost, electricity purchase cost and total carbon emissions.

[0006] The voltage offset rate refers to the voltage offset rate of a single node obtained by dividing the absolute value of the difference between the node voltage value and the rated voltage value at each time step by the rated voltage value. The voltage offset rate is obtained by taking the arithmetic mean of the voltage offset rates of all nodes and the single node voltage offset rates at 24 time steps.

[0007] The voltage qualification rate refers to determining whether the voltage value of each node at each time point is within the range of 0.95 times the rated voltage value to 1.05 times the rated voltage value. When the voltage value is within the qualified voltage range, the node voltage qualification flag is set to 1; when the voltage value is outside the qualified voltage range, the node voltage qualification flag is set to 0. The voltage qualification rate is obtained by calculating the arithmetic mean of the node voltage qualification flags of all nodes and the node voltage qualification flags at 24 time points.

[0008] The net load volatility refers to the net load value calculated at each moment within 24 hours, the standard deviation of the net load values ​​at 24 moments, the arithmetic mean of the net load values ​​at 24 moments, and the net load average value. The net load volatility is obtained by dividing the net load standard deviation by the net load average value.

[0009] The analytic hierarchy process (AHP) constructs an indicator comparison matrix to compare indicators pairwise. Each element in the indicator comparison matrix represents the relative importance of the indicator, which is expressed on a scale of 1 to 9. The indicator comparison matrix needs to undergo a consistency test. When the consistency ratio is less than 0.1, the indicator comparison matrix is ​​considered to have passed the consistency test.

[0010] The calculation of the indicator weight value adopts the geometric mean method. The product of all matrix elements in each row of the indicator comparison matrix is ​​obtained to obtain the row product. The nth root of the row product is used to obtain the initial value of the indicator weight. The sum of the initial values ​​of the indicator weights of all indicators is used to obtain the sum of the initial values ​​of the weights. The initial value of the indicator weight of each indicator is divided by the sum of the initial values ​​of the weights to obtain the indicator weight value.

[0011] The comprehensive carrying capacity scoring calculation model refers to multiplying the positive index value of each indicator by the index weight value to obtain the index weighted score, and then summing the index weighted scores of all indicators and multiplying by 100 to obtain the comprehensive carrying capacity score.

[0012] The upper-level optimization model takes the minimum weighted sum of active power loss and voltage offset as the upper-level objective function. The active power loss is multiplied by the active power loss weight coefficient to obtain the active power loss weight term, the voltage offset is multiplied by the voltage offset weight coefficient to obtain the voltage offset weight term, and the active power loss weight term plus the voltage offset weight term is the upper-level objective function value.

[0013] The improved gray wolf optimization algorithm is an improved algorithm based on the traditional gray wolf optimization algorithm by introducing a nonlinear convergence factor and a dimension learning strategy. The nonlinear convergence factor adopts the form of a cosine function and decreases with the number of iterations. The dimension learning strategy randomly selects some dimensions for learning and updating in each iteration while keeping the other dimensions unchanged.

[0014] In the rolling optimization time window, each rolling optimization time window is 6 hours long and contains 6 1-hour time periods. The optimization results of the first 4 time periods in each rolling optimization time window are used as the final running strategy, and the optimization results of the last 2 time periods are used as the initial state of the next rolling optimization time window.

[0015] The energy storage system's charge and discharge power sequence must satisfy the following constraints: mutual exclusion of charge and discharge, power of charge and discharge, state of charge (POC) constraint, and consistency of the first and last POCs. Mutual exclusion of charge and discharge means that the sum of the charging state flag and the discharging state flag is less than or equal to 1. The POC constraint includes the dynamic equation of the POC and the range constraint of the POC.

[0016] The sequence of capacitor switching groups must satisfy capacitor operation constraints. The reactive power injected into the capacitor at a given time is equal to the number of capacitor switching groups at that time multiplied by the reactive power compensation of a single capacitor group. The number of capacitor switching groups at a given time is greater than or equal to 0 and less than or equal to the maximum number of capacitor switching groups.

[0017] The output deviation percentage is calculated by subtracting the predicted output value from the actual output value of the distributed power source and taking the absolute value to obtain the absolute value of the output deviation. The absolute value of the output deviation is then divided by the rated capacity of the distributed power source and multiplied by 100% to obtain the output deviation percentage.

[0018] The model predictive control algorithm solves for the optimal control sequence in the future prediction time domain based on the current state and the predicted value of future disturbances at each control time. It only executes the control action at the current time and re-predicts and optimizes based on the new state information at the next control time.

[0019] The change range is calculated by dividing the absolute value of the difference between the optimized comprehensive bearing capacity score of the k-th iteration and the optimized comprehensive bearing capacity score of the k-1-th iteration by the optimized comprehensive bearing capacity score of the k-1-th iteration, and then multiplying by 100% to obtain the change range of the k-th iteration. The iteration termination condition is met when the change ranges of the k-th iteration, the k-1-th iteration, and the k-2-th iteration are all less than 0.5%.

[0020] This invention constructs a two-layer optimization framework, using the location and capacity determination of energy storage systems and reactive power compensation devices as the upper-layer optimization, and the operation and scheduling of equipment as the lower-layer optimization. A comprehensive carrying capacity scoring model is established as a feedback mechanism between the two layers, achieving coordinated optimization of equipment configuration and operation strategies. The upper-layer optimization uses an improved gray wolf optimization algorithm to solve for the installation location and capacity parameters of the equipment, aiming to minimize active power loss and voltage deviation, and determines the physical configuration scheme of the equipment. The lower-layer optimization, based on the equipment configuration determined in the upper layer, uses the same algorithm to coordinate and optimize energy storage charging and discharging, capacitor switching, static var compensator adjustment, and distributed power output, aiming to maximize the comprehensive carrying capacity score and generate the optimal operation strategy. The comprehensive carrying capacity score obtained from the lower-layer optimization is fed back to the upper layer to re-optimize the equipment configuration. Through multiple iterations, the equipment configuration scheme and operation scheduling strategy adapt to each other and tend towards global optimum. This invention establishes a dynamic feedback mechanism between the equipment configuration layer and the operation scheduling layer through a two-layer optimization framework. This ensures that equipment configuration fully considers the actual operating effect, and operation scheduling makes full use of equipment configuration resources. This solves the technical problem mentioned in the background art that it is difficult to simultaneously achieve equipment configuration optimization and operation scheduling optimization in the scenario of large-scale distributed power source access in the distribution network. Attached Figure Description

[0021] Figure 1 This is a flowchart of a two-layer optimization strategy for distribution networks aimed at improving overall carrying capacity, according to an embodiment of the present invention.

[0022] Figure 2 This is a diagram of a two-layer planning model in an embodiment of the present invention.

[0023] Figure 3 This is a system diagram of the IEEE-33 node standard computational example in an embodiment of the present invention.

[0024] Figure 4 This is a radar chart showing the positive comprehensive bearing capacity index in an embodiment of the present invention.

[0025] Figure 5 This is the optimized voltage diagram of Scheme 3 in the embodiment of the present invention.

[0026] Figure 6 This is a diagram showing the charging and discharging power of energy storage 1 in an embodiment of the present invention.

[0027] Figure 7 This is a diagram showing the charging and discharging power of energy storage 2 in an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0029] This invention provides a two-layer optimization method for distribution networks aimed at improving overall carrying capacity, comprising the following steps: S01. Construct a three-layer evaluation system for the comprehensive carrying capacity of the distribution network. The target layer is the adaptability of distributed power source access, the performance layer includes safety, flexibility, economy and low carbon emissions, and the indicator layer includes voltage deviation rate, voltage qualification rate, net load fluctuation rate, renewable energy utilization rate, operating cost, electricity purchase cost and total carbon emissions. S02. Collect the node voltage value, distributed power output value and node load value of N nodes in the distribution network at each moment within 24 hours, where N is the total number of nodes in the distribution network. Calculate the original index values ​​of the voltage deviation rate, the voltage qualification rate, the net load fluctuation rate, the renewable energy utilization rate, the operating cost, the electricity purchase cost and the total carbon emissions. S03. Normalize the original index values ​​to obtain normalized index values, and perform negative index-positiveing ​​processing on the normalized index values ​​to obtain positive index values. Use the analytic hierarchy process to construct an index comparison matrix and calculate the index weight values ​​of each index to establish a comprehensive carrying capacity scoring calculation model. S04. Construct an upper-level optimization model. With the weighted sum of active power loss and voltage deviation as the upper-level objective function, use the improved Grey Wolf optimization algorithm to optimize the location and capacity of energy storage system installation nodes, energy storage system charging and discharging power capacity, energy storage system energy capacity, reactive power compensation device installation node location, and reactive power compensation device reactive power compensation capacity. Output the energy storage system installation node location, energy storage system charging and discharging power capacity, energy storage system energy capacity, reactive power compensation device installation node location, and reactive power compensation device reactive power compensation capacity. S05. Calculate the relationship between the active power loss and the preset network loss threshold. When the active power loss exceeds the preset network loss threshold, divide the 24-hour optimization period into 4 rolling optimization time windows. The length of each rolling optimization time window is 6 hours. Within each rolling optimization time window, re-execute step S04 to obtain the installation node location of the energy storage system, the charging and discharging power capacity of the energy storage system, the energy capacity of the energy storage system, the installation node location of the reactive power compensation device, and the reactive power compensation capacity of the reactive power compensation device. S06. Based on the energy storage system installation node location, energy storage system charging and discharging power capacity, energy storage system energy capacity, reactive power compensation device installation node location, and reactive power compensation device reactive power compensation capacity output in step S04 or step S05, a lower-level optimization model is constructed. The objective function of the lower-level model is to maximize the comprehensive carrying capacity score calculated by the comprehensive carrying capacity score calculation model. The improved gray wolf optimization algorithm is used to coordinate and optimize the energy storage system charging and discharging power sequence, capacitor switching group sequence, static var compensator reactive power regulation capacity sequence, and distributed power output sequence. S07. Collect the actual output value of the distributed power source, calculate the output deviation percentage between the actual output value and the predicted output value of the distributed power source, and when the output deviation percentage exceeds 15%, use the model predictive control algorithm to predict the node voltage prediction value sequence for the next 30 minutes, and adjust the static var compensator reactive power regulation capacity sequence and the capacitor switching group number sequence according to the node voltage prediction value sequence. S08. Substitute the energy storage system charging and discharging power sequence, the capacitor switching group sequence, the static var compensator reactive power regulation capacity sequence, and the distributed power source output sequence obtained in step S06 into the comprehensive carrying capacity scoring calculation model to calculate the optimized comprehensive carrying capacity score value. Feed the optimized comprehensive carrying capacity score value back to step S04 and repeat steps S04 to S08. When the change of the optimized comprehensive carrying capacity score value obtained in three consecutive iterations is less than 0.5% or the number of iterations reaches 50, terminate the iteration and output the final energy storage system charging and discharging power sequence, the capacitor switching group sequence, the static var compensator reactive power regulation capacity sequence, and the distributed power source output sequence.

[0030] Wherein, the voltage deviation rate represents the average deviation of the node voltage value of all nodes in the distribution network from the rated voltage value at all times. The calculation formula for the voltage deviation rate is as follows: For each node i at each time t, calculate the node voltage value. With the rated voltage value The absolute value of the difference divided by the rated voltage value The voltage offset rate of a single node is obtained, and the arithmetic mean of the voltage offset rates of all N nodes and 24 time points is calculated to obtain the voltage offset rate.

[0031] Wherein, the voltage qualification rate represents the proportion of the number of nodes in the distribution network whose voltage values ​​are within the qualified voltage range to the total number of nodes in the distribution network. The calculation formula for the voltage qualification rate is as follows: For each node i at each time t, determine the node voltage value. Is it at 0.95? Up to 1.05 Within the specified range, the node voltage is within the acceptable voltage range. A value of 1 indicates that the node voltage is within the acceptable voltage range. A value of 0 represents the pass / fail flag for the node voltage across all N nodes and at 24 time points. The arithmetic mean is used to obtain the voltage qualification rate.

[0032] The net load volatility represents the degree of fluctuation in the net load value after subtracting the output value of the distributed power source from the total load of the distribution network. The formula for calculating the net load volatility is as follows: calculate the net load value at each moment within 24 hours, calculate the standard deviation of the net load value at each of the 24 moments to obtain the net load standard deviation, calculate the arithmetic mean of the net load value at each of the 24 moments to obtain the net load average, and divide the net load standard deviation by the net load average to obtain the net load volatility.

[0033] The renewable energy utilization rate represents the ratio of the actual renewable energy consumption in the distribution network to the predicted renewable energy power generation. The calculation formula for the renewable energy utilization rate is as follows: sum the renewable energy consumption at all times within 24 hours to obtain the total renewable energy utilization; sum the predicted renewable energy power generation at all times within 24 hours to obtain the total predicted renewable energy; and divide the total renewable energy utilization by the total predicted renewable energy to obtain the renewable energy utilization rate.

[0034] The operating costs include the operating costs of the energy storage system, capacitors, static var compensators, and distributed generation. The calculation formula for the operating costs is as follows: For each time t, calculate the charging and discharging power of the energy storage system. The absolute value multiplied by the unit operating cost of the energy storage system Obtain the real-time operating cost of the energy storage system and calculate the capacitor switching power. Multiply by the unit operating cost of the capacitor Obtain the constant operating cost of the capacitor and calculate the reactive power regulation of the static var compensator. The absolute value multiplied by the unit operating cost of the static var compensator Obtain the operating cost of the static var compensator at any given time, and calculate the output value of the distributed power source. Multiply by the unit operating cost of distributed power sources The operating cost of the distributed power source at any given time is obtained by summing the operating costs of the energy storage system, the capacitor, the static var compensator, and the distributed power source at any given time over 24 time periods.

[0035] Wherein, the electricity purchase cost represents the economic expenditure of the distribution network in purchasing electricity from the superior power grid, and the calculation formula for the electricity purchase cost is as follows: For each time t, calculate the active power output of the superior power grid. Multiply by the electricity price at any time The electricity purchase cost at each time point is obtained, and the electricity purchase cost at each time point is summed over 24 time points to obtain the total electricity purchase cost.

[0036] The total carbon emissions refer to the carbon dioxide emissions generated by various energy consumptions during the operation of the power distribution network. The formula for calculating the total carbon emissions is as follows: For each energy source l at each time t, calculate the active power output of the energy source. Multiply by the energy carbon emission factor The carbon emissions at each energy moment are obtained, and the total carbon emissions are obtained by summing the carbon emissions at each energy moment over all L types of energy and 24 time moments.

[0037] The normalization process refers to a data standardization method to eliminate the influence of different units of measurement. The calculation formula for the normalization process is as follows: For each index i, calculate the original index value. Subtract the minimum value of the indicator Obtain the index difference and calculate the index maximum value. Subtract the minimum value of the indicator The normalized index value is obtained by dividing the index difference by the index range. .

[0038] The "positive transformation of negative indicators" refers to a data transformation method that converts negative indicators (where smaller values ​​are better) into positive indicators (where larger values ​​are better). The calculation formula for this positive transformation is as follows: For negative indicators, calculate the maximum value of the normalized indicator. Subtract the normalized index value The positive index value is obtained. For positive indicators, the positive indicator value Equal to the normalized index value .

[0039] The analytic hierarchy process (AHP) is a multi-indicator decision-making method that combines qualitative and quantitative analysis. It constructs an indicator comparison matrix to compare indicators pairwise. Each element in the indicator comparison matrix... This indicates the relative importance of indicator i relative to indicator j. The relative importance is represented by a scale from 1 to 9, where 1 indicates equal importance, 3 indicates slightly important, 5 indicates significantly important, 7 indicates strongly important, 9 indicates extremely important, and 2, 4, 6, and 8 are the median values ​​for adjacent judgments.

[0040] The indicator comparison matrix needs to undergo a consistency test, which includes calculating the consistency index and the consistency ratio. The formula for calculating the consistency index CI is as follows: Calculate the largest eigenvalue of the indicator comparison matrix. Subtracting the matrix order n yields the eigenvalue difference. Dividing the eigenvalue difference by the matrix order n minus 1 yields the consistency index CI. The formula for calculating the consistency ratio CR is as follows: Dividing the consistency index CI by the random consistency index RI yields the consistency ratio CR. When the consistency ratio CR is less than 0.1, the index comparison matrix is ​​considered to have passed the consistency test.

[0041] The calculation of the indicator weight value adopts the geometric mean method, and the calculation formula of the indicator weight value is expressed as follows: Take the product of all matrix elements in each row of the indicator comparison matrix to obtain the row product, and take the nth root of the row product to obtain the initial value of the indicator weight. Where n is the total number of indicators, and the initial weights of the indicators for all indicators are... The summation yields the initial weight values, and the initial weight value for each indicator is calculated separately. The index weight value is obtained by dividing by the sum of the initial weight values. .

[0042] The comprehensive carrying capacity scoring calculation model is used to quantitatively evaluate the comprehensive carrying capacity of the distribution network. The expression of the comprehensive carrying capacity scoring calculation model is as follows: The positive index value of each indicator... Multiply by the weight value of the indicator The weighted scores of the indicators are obtained, and the sum of the weighted scores of all n indicators is multiplied by 100 to obtain the comprehensive bearing capacity score S.

[0043] The active power loss refers to the active power loss caused by line resistance during power transmission in the distribution network. The calculation formula for the active power loss is as follows: For each line connecting node i and node j in the distribution network, at each time t, calculate the voltage of node i. The square of the sum of the voltages at node j The square of the sum of the squares of the voltages is obtained by calculating the voltage at node i. Multiply by node j voltage Multiply by the cosine of the nodal phase angle difference The voltage product term is obtained by multiplying the sum of squared voltages by the voltage product term, and then multiplying the sum of squared voltages by the line conductance. The active power loss at each time point of the line is obtained, and the active power loss at each time point of the line is summed over all lines and 24 time points to obtain the active power loss.

[0044] Wherein, the total voltage offset represents the difference between the node voltage value of all nodes in the distribution network at all times and the rated voltage value. The sum of deviations, the total voltage offset, is calculated using the following formula: For each node i at each time t, calculate the node voltage value. With the rated voltage value The absolute value of the difference is used to obtain the single-node voltage offset value. The total voltage offset is obtained by summing the single-node voltage offset values ​​of all N nodes and 24 time points.

[0045] Wherein, the upper-level objective function represents the optimization objective of the upper-level optimization model, and the expression of the upper-level objective function is as follows: the active power loss multiplied by the active power loss weight coefficient. The active power loss weighting term is obtained by multiplying the total voltage offset by the voltage offset weighting coefficient. The voltage offset weighting term is obtained, and the active power loss weighting term is added to the voltage offset weighting term to obtain the upper-level objective function value f, wherein the active power loss weighting coefficient... With the voltage offset weighting coefficient The sum of them equals 1.

[0046] The improved gray wolf optimization algorithm is an improved algorithm based on the traditional gray wolf optimization algorithm by introducing a nonlinear convergence factor and a dimension learning strategy. The nonlinear convergence factor adopts the form of a cosine function and decreases with the number of iterations. The dimension learning strategy randomly selects some dimensions for learning and updating in each iteration while keeping the other dimensions unchanged.

[0047] The location of the energy storage system installation node indicates the node number in the distribution network where the energy storage system is installed, and is indicated by a node installation marker. Indicates whether node i has an energy storage system installed, the node installation flag. A value of 1 indicates that an energy storage system is installed at node i, and the node installation flag is... A value of 0 indicates that no energy storage system is installed at node i.

[0048] Wherein, the energy storage system charge / discharge power capacity represents the maximum charging power or maximum discharging power of the energy storage system per unit time, and the energy storage system charge / discharge power capacity is denoted as... The energy storage system's charge / discharge power capacity The constraints are ,in This represents the minimum charge / discharge power capacity of the energy storage system. This refers to the maximum charge / discharge power capacity of the energy storage system.

[0049] Wherein, the energy capacity of the energy storage system represents the maximum electrical energy that the energy storage system can store, and the energy capacity of the energy storage system is denoted as... The energy storage system has an energy capacity of The constraints are ,in This represents the minimum energy capacity of the energy storage system. This represents the maximum energy capacity of the energy storage system.

[0050] The reactive power compensation device includes two types: capacitors and static var compensators. The installation node location of the reactive power compensation device indicates the node number in the distribution network where the reactive power compensation device is installed.

[0051] The reactive power compensation capacity of the reactive power compensation device is expressed as the capacitor reactive power compensation capacity for a capacitor. The reactive power compensation capacity of the capacitor The constraints are ,in This is the minimum reactive power compensation capacity of the capacitor. This represents the maximum reactive power compensation capacity of the capacitor.

[0052] Wherein, the reactive power compensation capacity of the reactive power compensation device is expressed as the reactive power compensation capacity of the static var compensator (SVC). The reactive power compensation capacity of the static var compensator The constraints are ,in This represents the minimum reactive power compensation capacity of the static var compensator. This represents the maximum reactive power compensation capacity of the static var compensator.

[0053] The preset network loss threshold represents the maximum allowable active power loss limit of the distribution network. The preset network loss threshold is calculated based on the rated capacity of the distribution network and line parameters, and is set to 3% to 5% of the rated capacity of the distribution network.

[0054] The rolling optimization time window represents a time period division unit in multi-timescale coordinated optimization. Each rolling optimization time window contains 6 one-hour time periods. The optimization results of the first 4 time periods in each rolling optimization time window are used as the final running strategy, and the optimization results of the last 2 time periods are used as the initial state of the next rolling optimization time window.

[0055] Wherein, the lower-level objective function represents the optimization objective of the lower-level optimization model, and the lower-level objective function is to maximize the comprehensive carrying capacity score value S.

[0056] Wherein, the energy storage system charge / discharge power sequence represents the arrangement of the charging or discharging power of the energy storage system at each moment within 24 hours, and the energy storage system charge / discharge power sequence is denoted as... ,in Let be the charging power at time t. Let be the discharge power at time t.

[0057] The charging and discharging power sequence of the energy storage system must satisfy a charging and discharging mutual exclusion constraint. This constraint indicates that the energy storage system cannot charge and discharge simultaneously. The expression for the charging and discharging mutual exclusion constraint is as follows: Charging status flag With discharge status flag The sum is less than or equal to 1, wherein the charging status flags The value 1 indicates that charging begins at time t, and the discharge state flag is... A value of 1 indicates that the discharge occurs at time t.

[0058] The energy storage system's charge / discharge power sequence must satisfy a charge / discharge power constraint, which is expressed as follows: the charging power at time t. Greater than or equal to 0 and less than or equal to the charging status flag Multiplied by the charging and discharging power capacity of the energy storage system Discharge power at time t Greater than or equal to 0 and less than or equal to the discharge state flag Multiplied by the charging and discharging power capacity of the energy storage system .

[0059] The energy storage system's charge and discharge power sequence must satisfy state of charge (SOC) constraints, which include a dynamic equation for SOC and a range constraint for SOC. The dynamic equation for SOC is expressed as follows: the state of charge of the energy storage system at time t... The state of charge of the energy storage system at time t-1 is equal to the state of charge of the energy storage system. Add the charging power at time t Multiply by charging efficiency Subtract the discharge power at time t Divide by discharge efficiency The state of charge range constraint is described as follows: the state of charge of the energy storage system Greater than or equal to 0.1 times the energy capacity of the energy storage system And less than or equal to 0.9 times the energy capacity of the energy storage system. .

[0060] The energy storage system's charge and discharge power sequence must satisfy the consistency constraint between the first and last states of charge. This consistency constraint is described as follows: the energy storage system's state of charge at time 0... The state of charge of the energy storage system at time 24 .

[0061] Wherein, the capacitor switching group sequence represents the arrangement of the number of capacitor groups put into operation at each moment within 24 hours, and the capacitor switching group sequence is denoted as... ,in Let t be the number of capacitor banks switched at time t.

[0062] The sequence of capacitor switching groups must satisfy capacitor operation constraints, which are expressed as follows: Reactive power injected into the capacitor at time t. The number of capacitor banks switched at time t Multiply by the reactive power compensation of a single capacitor bank The number of capacitor banks switched at time t Greater than or equal to 0 and less than or equal to the maximum number of capacitor switching groups .

[0063] Wherein, the static var compensator reactive power regulation capacity sequence represents the reactive power arrangement provided by the static var compensator at each moment within 24 hours, and the static var compensator reactive power regulation capacity sequence is denoted as... ,in Let t be the reactive power regulation capacity of the static var compensator at time t.

[0064] The reactive power regulation capacity sequence of the static var compensator (SVC) must satisfy the SVC operating constraints, which are expressed as follows: The reactive power regulation capacity of the SVC at time t. Greater than or equal to the minimum reactive power regulation capacity of the static var compensator And less than or equal to the maximum reactive power regulation capacity of the static var compensator. .

[0065] Wherein, the distributed power generation output sequence represents the active power output arrangement of the distributed power source at each moment within a 24-hour period, and the distributed power generation output sequence is denoted as... ,in Let t be the output value of the distributed power source at time t.

[0066] The distributed power generation output sequence must satisfy the distributed power generation output constraint, which is expressed as follows: the distributed power generation output value at time t. Greater than or equal to the minimum output of the distributed power source And less than or equal to the maximum output of the distributed power source .

[0067] The actual output value of the distributed power source refers to the active power output value measured by the distributed power source during actual operation. The actual output value of the distributed power source is collected in real time by the distribution network monitoring system.

[0068] The predicted output value of the distributed power source represents the active power output value of the distributed power source predicted by the prediction model. The predicted output value of the distributed power source is calculated based on meteorological forecast data and historical output data.

[0069] The output deviation percentage represents the percentage of the difference between the actual output value and the predicted output value of the distributed power source relative to the rated capacity of the distributed power source. The formula for calculating the output deviation percentage is as follows: the absolute value of the output deviation is obtained by subtracting the predicted output value from the actual output value of the distributed power source. The absolute value of the output deviation is then divided by the rated capacity of the distributed power source and multiplied by 100% to obtain the output deviation percentage.

[0070] The model predictive control algorithm is a control algorithm that predicts the future state based on a dynamic model of the system and performs rolling optimization. At each control time, the model predictive control algorithm solves for the optimal control sequence in the future prediction time domain based on the current state and the predicted value of future disturbances. It only executes the control action at the current time and re-predicts and optimizes based on the new state information at the next control time.

[0071] The node voltage prediction sequence represents the predicted node voltage value of each node in the distribution network for each 5-minute period within the next 30 minutes. The node voltage prediction sequence is calculated based on the predicted output value of the distributed power source, the predicted load value, and the distribution network power flow calculation model.

[0072] The optimized comprehensive carrying capacity score represents the comprehensive carrying capacity score calculated by substituting the optimized energy storage system charging and discharging power sequence, the capacitor switching group sequence, the static var compensator reactive power regulation capacity sequence, and the distributed power output sequence into the comprehensive carrying capacity score calculation model.

[0073] Wherein, the change range represents the relative degree of change among the optimized comprehensive bearing capacity score values ​​obtained in three consecutive iterations, and the calculation formula for the change range is as follows: Calculate the optimized comprehensive bearing capacity score value in the k-th iteration. Compared with the optimized comprehensive bearing capacity score value of the (k-1)th iteration The absolute value of the difference divided by the optimized comprehensive bearing capacity score value of the (k-1)th iteration Multiply by 100% to obtain the change magnitude of the k-th iteration. The iteration termination condition is met when the change magnitudes of the k-th, (k-1)-th, and (k-2)-th iterations are all less than 0.5%.

[0074] The power flow equation constraints represent the active power balance and reactive power balance conditions of each node in the distribution network. The expression for the power flow equation constraints is as follows: the active power of node i at time t. equal to the voltage at node i Multiply by the voltage of all connected nodes j With the real part of the line admittance Multiply by the cosine of the nodal phase angle difference Add the imaginary part of the line admittance Multiply by the sine of the nodal phase angle difference The sum of the values ​​over all connected nodes is the node reactive power of node i at time t. equal to the voltage at node i Multiply by the voltage of all connected nodes j With the real part of the line admittance Multiply by the sine of the nodal phase angle difference Subtract the imaginary part of the line admittance Multiply by the cosine of the nodal phase angle difference The difference is summed over all connected nodes.

[0075] The node voltage constraint means that the voltage value of each node in the distribution network must be between the allowable minimum and maximum node voltage values. The expression for the node voltage constraint is as follows: the node voltage value of node i at time t. Greater than or equal to the minimum node voltage And less than or equal to the maximum node voltage. The minimum node voltage It is 0.95 The maximum value of the node voltage It is 1.05 .

[0076] Wherein, the charging efficiency The charging efficiency refers to the efficiency with which input electrical energy is converted into stored electrical energy during the charging process of an energy storage system. The value ranges from 0.9 to 0.95. The discharge efficiency... The discharge efficiency represents the efficiency with which stored electrical energy is converted into output electrical energy during the discharge process of an energy storage system. The value ranges from 0.9 to 0.95. The reactive power compensation of a single capacitor bank... This indicates the reactive power compensation capacity provided by each switching of a capacitor bank, and the reactive power compensation amount of a single capacitor bank. The line conductance is determined based on the capacitor equipment parameters. The real part of the line admittance connecting nodes i and j is given by the line conductance. The real part of the line admittance is calculated based on the line resistance and line reactance. With the line conductance These represent the same physical quantity. The imaginary part of the line admittance... The imaginary part of the line admittance connecting node i and node j is represented by the imaginary part of the line admittance. The value is calculated based on the line resistance and line reactance. The cosine value of the node phase angle difference is... This represents the voltage phase angle difference between node i and node j at time t. The cosine function value. The sine value of the node phase angle difference. This represents the voltage phase angle difference between node i and node j at time t. The sine function value. The distribution network monitoring system is a system for real-time monitoring of the operating status of the distribution network. The system collects node voltage values, actual output values ​​of distributed generation sources, and node load values ​​through voltage sensors, current sensors, and power sensors installed at each node of the distribution network. The meteorological forecast data includes wind speed forecast data, light intensity forecast data, and temperature forecast data, provided by the meteorological forecast system. The historical output data represents the active power output records of distributed generation sources over historical periods, stored in the distribution network data management system. The load forecast value represents the load value of each node in the distribution network in future periods, predicted by the load forecast model, calculated based on the historical output data and historical load data. The distribution network power flow calculation model is a mathematical model used to calculate the voltage of each node and the power of the lines in the distribution network, established based on the constraints of the power flow equations.

[0077] Optionally, the present invention also provides a method for implementing a two-layer optimization system for distribution networks aimed at improving comprehensive carrying capacity through a computer. The computer is equipped with a readable storage medium, which stores program instructions. When the program instructions are run in the computer, they can execute the above-described method.

[0078] It should be noted that this invention also solves the following technical problems: First, it solves the technical problem of maintaining voltage stability in distribution networks under the condition of distributed power generation output fluctuations. This invention monitors the deviation between the actual output and the predicted output of distributed power sources in real time. When the output deviation percentage exceeds 15%, it uses a model predictive control algorithm to predict the node voltage change trend in the next 30 minutes and adjusts the reactive power compensation capacity of static var compensators and capacitors in advance, thus achieving forward-looking control of voltage fluctuations and avoiding the voltage over-limit problem caused by the lag of traditional feedback control methods. Second, it solves the technical problem of poor optimization effect caused by excessive active power loss during distribution network optimization. This invention sets a preset network loss threshold. When the active power loss exceeds the threshold, it automatically divides the 24-hour optimization period into four 6-hour rolling optimization time windows. The upper-level optimization is re-executed in each time window, which shortens the optimization time span, improves the adaptability to load and distributed power generation output changes, makes the optimization results closer to the actual operating state, and effectively reduces the network loss level.

[0079] Specifically, the principle of this invention is as follows: The invention solves the aforementioned technical problems by establishing a bidirectional coupling optimization mechanism between equipment configuration and operation scheduling. Traditional methods separate equipment configuration from operation scheduling, leading to unpredictable actual operational effects during equipment configuration and limitations in achieving optimal performance during operation scheduling due to constraints imposed by existing equipment configurations. This invention, after determining the physical configuration of the equipment through upper-level optimization, immediately verifies the comprehensive carrying capacity performance of this configuration scheme in actual operation through lower-level optimization. If the comprehensive carrying capacity score is unsatisfactory, the score result is fed back to the upper level to adjust the equipment configuration scheme. This iterative feedback mechanism allows equipment configuration decisions to fully consider multi-dimensional performance requirements at the operational level. Simultaneously, the three-layer comprehensive carrying capacity evaluation system constructed by this invention covers four performance dimensions: safety, flexibility, economy, and low carbon emissions. By quantifying the weights of each indicator through the analytic hierarchy process (AHP), the multi-objective optimization problem is transformed into a single comprehensive score maximization problem, providing a unified evaluation standard for upper and lower level optimization. Furthermore, this invention introduces a rolling optimization time window to address situations with excessive active power loss. The 24-hour optimization period is subdivided into multiple 6-hour windows for separate optimization, which improves the optimization accuracy and adaptability to fluctuations. This aligns with the actual operating characteristics of the distribution network and ensures the feasibility and effectiveness of the optimization scheme.

[0080] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0081] The specific implementation of step S01 is to construct a three-layer evaluation system for the comprehensive carrying capacity of the distribution network. This evaluation system adopts a hierarchical structure. The target layer is the adaptability of distributed power source access. The performance layer includes four dimensions: safety, flexibility, economy and low carbon. The indicator layer includes seven specific evaluation indicators: voltage deviation rate, voltage qualification rate, net load fluctuation rate, renewable energy utilization rate, operating cost, electricity purchase cost and total carbon emissions. Through this three-layer evaluation system, a comprehensive quantitative evaluation of the comprehensive carrying capacity of the distribution network can be achieved.

[0082] The specific implementation of step S02 is to collect data from the power distribution network. The node voltage, distributed generation output, and node load values ​​of each node at each moment within a 24-hour period. Given the total number of distribution network nodes, calculate the original index values ​​based on the collected data. Voltage offset rate. The calculation formula is expressed as follows: ; In the formula, Voltage offset rate, dimensionless; For nodes At any moment The node voltage value, in units of ; This is the rated voltage value, in units of... ; This represents the total number of nodes in the distribution network. This is a time marker, with a value range of 1 to 24; The node number is determined by a value ranging from 1 to 1. Voltage qualification rate The calculation formula is expressed as follows: ; In the formula, Voltage pass rate, dimensionless; For nodes At any moment The node voltage qualification mark is dimensionless. The value is 1 if the condition is met, and 0 otherwise. Net load volatility The calculation formula is expressed as follows: ; In the formula, Net load volatility, dimensionless; For a moment Net load value, in units of The calculation method is as follows ; For a moment Total load value, in units It is obtained by summing the load power of each node in real time through the distribution network monitoring system; For a moment The output value of distributed power sources, in units of ; This is the average net load, in units of The calculation method is as follows Renewable energy utilization rate The calculation formula is expressed as follows: ; In the formula, The utilization rate of renewable energy is dimensionless. For a moment The amount of renewable energy used, in units of This represents the actual renewable energy power absorbed by the distribution network; For a moment Forecasted renewable energy generation, in units of This represents the renewable energy generation capacity predicted based on a meteorological forecasting model. Operating costs. The calculation formula is based on a 1-hour time step and is expressed as follows: ; In the formula, Operating costs, in yuan; For a moment The charging and discharging power of the energy storage system, in units of Positive values ​​indicate charging, and negative values ​​indicate discharging. The time step is 1 hour. The unit operating cost of the energy storage system is expressed in yuan per unit. ; For a moment Capacitor switching power, in units of ; The unit operating cost of the capacitor is expressed in yuan per unit. ; For a moment The reactive power regulation of the static var compensator, in units of ; The unit operating cost of the static var compensator is expressed in yuan per unit. ; For a moment The renewable energy output value, in units of ; The unit operating cost of distributed power sources is expressed in yuan per unit. Electricity purchase cost The calculation formula is expressed as follows: ; In the formula, The cost of electricity is expressed in yuan. For a moment The active power output of the upstream power grid, in units of ; For a moment The electricity price is in yuan per unit. Total carbon emissions The calculation formula is expressed as follows: ; In the formula, Total carbon emissions, in units of ; Total number of energy types; The energy type is numbered, with a value ranging from 1 to... ; For energy At any moment Contributions, in units of ; For energy The carbon emission factor, in units of Every .

[0083] The specific implementation of step S03 involves normalizing the original index values ​​to obtain normalized index values. The calculation formula for normalization is as follows: ; In the formula, The value is a normalized index, dimensionless. As an indicator The original index values; As an indicator The minimum value among all samples; As an indicator The maximum value among all samples; Assign an indicator number. The calculation formula for turning negative indicators into positive ones is expressed as follows: For negative indicators, For positive indicators, In the formula The positive index value is dimensionless. The maximum value of the normalized index is dimensionless and typically takes the value of 1. An index comparison matrix is ​​constructed using the analytic hierarchy process (AHP). Matrix elements Indicators relative to indicators The relative importance is dimensionless and expressed on a scale of 1 to 9, where 1 indicates equal importance, 3 indicates slightly important, 5 indicates significantly important, 7 indicates strongly important, and 9 indicates extremely important. 2, 4, 6, and 8 are the median values ​​for adjacent judgments. Consistency index. The calculation formula is expressed as follows: ; In the formula, It is a consistency indicator and is dimensionless. The largest eigenvalue of the index comparison matrix is ​​dimensionless. The matrix order is equal to the total number of indicators. Consistency ratio. The calculation formula is expressed as follows: ; In the formula, The consistency ratio is dimensionless. It is a random consistency index, dimensionless, and its value increases with the order of the matrix. hour ,when hour ,when hour ,when hour ,when hour ,when The indicator comparison matrix is ​​considered to have passed the consistency test. Indicator weight values. The calculation formula is expressed as follows: ; ; In the formula, As an indicator The initial values ​​of the weights are dimensionless; As an indicator The indicator weight values ​​are dimensionless. For the index comparison matrix, the first... Line 1 The elements of the column are dimensionless; The column number is 1 to 1. ; For summation numbering, the value range is 1 to 1. The expression for the comprehensive bearing capacity scoring calculation model is as follows: ; In the formula, This is a dimensionless comprehensive bearing capacity score. The total number of indicators; the constant 100 is used to scale the score values ​​to the range of 0 to 100.

[0084] The specific implementation of step S04 is to construct an upper-level optimization model, the upper-level objective function. The expression, in its normalized form, is as follows: ; In the formula, The value of the upper-level objective function is dimensionless. This is the active power loss weighting coefficient, dimensionless, with a default value of 0.6; This is the voltage offset weighting coefficient, dimensionless, with a default value of 0.4. ; For active network loss, the unit is ; For reference, the unit is active power loss. The value is usually taken as 2% of the rated capacity of the distribution network multiplied by 24 hours; Total voltage offset, in units of ; Total reference voltage offset, in units of The value is usually taken as Active network loss The calculation formula is expressed as follows: ; In the formula, A collection of distribution network lines; Indicates the connection node and nodes The route; For the line Conductivity, in units of ; For a moment node With nodes The phase angle difference between them, in units of Total voltage offset The calculation formula is expressed as follows: ; An improved gray wolf optimization algorithm is used to optimize the location and capacity of energy storage system installation nodes, charging and discharging power capacity, energy capacity, reactive power compensation device installation node location, and reactive power compensation capacity. The improved gray wolf optimization algorithm introduces a nonlinear convergence factor. Nonlinear convergence factor Using the cosine function form, the calculation formula is expressed as follows: ; In the formula, is a nonlinear convergence factor, dimensionless; This represents the current iteration number; This represents the maximum number of iterations, typically set to 100. The dimension learning strategy randomly selects a subset of dimensions for learning and updating in each iteration; the dimension selection probability... The calculation formula is expressed as follows: ; In the formula, The probability for each dimension is dimensionless. The formula for updating the wolf pack position is as follows: ; In the formula, For the first Sekiro in the 1990s The position vector of the next iteration; For the alpha wolf in the The position vector of the next iteration; For the first Sekiro in the 1990s The position vector of the next iteration; The wolf's number; The coefficient vector is calculated as follows: ; The coefficient vector is calculated as follows: ; and A dimensionless random number between 0 and 1. Energy storage system charge / discharge power capacity. The constraints are Energy storage system energy capacity The constraints are Capacitor reactive power compensation capacity The constraints are Static var compensator reactive power compensation capacity The constraints are In the formula For nodes The charging and discharging power capacity of the energy storage system is expressed in units of... , The minimum charge / discharge power capacity of the energy storage system, in units of , The maximum charge / discharge power capacity of the energy storage system, in units of , For nodes Energy capacity of the energy storage system, in units of , The minimum energy capacity of the energy storage system, in units of , The maximum energy capacity of the energy storage system, in units of , For nodes The reactive power compensation capacity of the capacitor is expressed in units of [unit missing]. , This is the minimum reactive power compensation capacity of the capacitor, in units of , This represents the maximum reactive power compensation capacity of the capacitor, in units of... , For nodes The reactive power compensation capacity of the static var compensator is expressed in units of 1. , This represents the minimum reactive power compensation capacity of the static var compensator, in units of... , This represents the maximum reactive power compensation capacity of the static var compensator, in units of... .

[0085] The specific implementation of step S05 is to calculate the active power loss. With preset network loss threshold The size relationship, the preset network loss threshold is set to the rated capacity of the distribution network. 3% to 5% multiplied by 24 hours, that is In the formula The preset network loss threshold is in units of , Rated capacity of the distribution network, in units of ,when The 24-hour optimization period is divided into 4 rolling optimization time windows, each with a length of 6 hours. Step S04 is re-executed within each rolling optimization time window to obtain the optimization results.

[0086] The specific implementation of step S06 involves constructing a lower-level optimization model based on the energy storage system installation node location, energy storage system charging and discharging power capacity, energy storage system energy capacity, reactive power compensation device installation node location, and reactive power compensation device reactive power compensation capacity output from step S04 or step S05. The objective function of the lower-level model is to maximize the comprehensive carrying capacity score. An improved gray wolf optimization algorithm is used to coordinate and optimize the charging and discharging power sequence, capacitor switching group sequence, static var compensator reactive power regulation capacity sequence, and distributed generation output sequence of the energy storage system. The charging and discharging power sequence of the energy storage system must satisfy charging and discharging mutual exclusion constraints. Charging and discharging power constraints and Dynamic equations of charged state State of charge range constraints Consistent first and last charge states constraint In the formula For a moment The charging status indicator is dimensionless and takes a value of 0 or 1. For a moment The discharge status indicator is dimensionless and takes a value of 0 or 1. For a moment Charging power, in units of , For a moment The discharge power, in units of , For a moment The state of charge of the energy storage system, in units of , For a moment The state of charge of the energy storage system, in units of , Charging efficiency, dimensionless, default value is 0.95. Discharge efficiency, dimensionless, default value is 0.95. The state of charge of the energy storage system at time 0, in units of , The state of charge of the energy storage system at time 24, in units of The sequence of capacitor switching groups must satisfy... and In the formula For a moment The reactive power injected into the capacitor, in units of , For a moment The number of capacitor banks switched on, dimensionless. This refers to the reactive power compensation of a single capacitor bank, in units of... , This represents the maximum number of capacitor banks that can be switched on and off, and is dimensionless. The reactive power regulation capacity sequence of the static var compensator must meet the following requirements: The output sequence of distributed power sources must meet the following requirements. In the formula For a moment The reactive power regulation capacity of the static var compensator, in units of , This represents the minimum reactive power regulation capacity of the static var compensator, in units of... , This represents the maximum reactive power regulation capacity of the static var compensator, in units of... , For a moment The minimum output of a distributed power source, in units of , For a moment The maximum output of the distributed power source, in units of .

[0087] The specific implementation of step S07 is to collect the actual output value of the distributed power source. Calculate the percentage of output force deviation. The calculation formula is expressed as follows: ; In the formula, This represents the percentage of output deviation, dimensionless. This represents the actual output value of the distributed power source, in units of [unit missing]. This is obtained through real-time measurement by the power distribution network monitoring system; The predicted output value of distributed generation is given in units of . It is predicted through meteorological forecasting models and historical data; Rated capacity of distributed power sources, in units of .when At that time, a model predictive control algorithm is used to predict the node voltage forecast sequence for the next 30 minutes. The prediction model adopts the distribution network power flow equation, and the node voltage forecast values ​​are... The calculation formula is expressed as follows: ; In the formula, For nodes At any moment The predicted node voltage, in units of ; For the time domain of prediction, the value ranges from 1 to 6, corresponding to the next 5 to 30 minutes, with a step size of 5 minutes; The model predictive control function is established based on the power flow equations; For nodes At any moment The current voltage value, in units of ; For a moment The predicted output value of distributed power sources, in units of ; For a moment The load forecast value, in units of ; For a moment The reactive power regulation capacity of the static var compensator, in units of ; For a moment The number of capacitor banks switched on / off is dimensionless. The static var compensator reactive power regulation capacity sequence and the capacitor bank switching number sequence are adjusted based on the node voltage prediction value sequence.

[0088] The specific implementation of step S08 involves substituting the energy storage system charging and discharging power sequence, capacitor switching group number sequence, static var compensator reactive power regulation capacity sequence, and distributed power output sequence obtained in step S06 into the comprehensive carrying capacity scoring calculation model to calculate the optimized comprehensive carrying capacity score value. Variation range The calculation formula is expressed as follows: ; In the formula, For the first The magnitude of change in each iteration is dimensionless. For the first The optimized comprehensive bearing capacity score after the next iteration is dimensionless. For the first The optimized comprehensive bearing capacity score after the next iteration is dimensionless. and and The iteration may terminate when the number of iterations reaches 50, outputting the final energy storage system charging and discharging power sequence, capacitor switching group number sequence, static var compensator reactive power regulation capacity sequence, and distributed power generation output sequence, where... For the first The magnitude of change in each iteration is dimensionless. For the first The magnitude of change in each iteration is dimensionless.

[0089] The expressions for the power flow equation constraints are as follows, and the nodal active power balance equations are: ; The reactive power balance equation at the node is: ; In the formula, For nodes At any moment The node active power, in units of ; For nodes At any moment The node reactive power, in units of ; For nodes A set of connected nodes; For nodes At any moment The node voltage value, in units of ; For nodes The numbers of the connected nodes; For the line susceptance, in units of The expression for the node voltage constraint is as follows: ; In the formula, For nodes The minimum voltage, in units of The value is ; For nodes The maximum voltage, in units of The value is .

[0090] To better understand and implement this invention, a specific application scenario, Example 2, is provided below: A technical team is responsible for a comprehensive capacity enhancement and renovation project for a typical urban power distribution network. This power distribution network adopts the IEEE-33 node standard system architecture, such as... Figure 3 As shown, the distribution network comprises 33 distribution nodes and 32 distribution lines, with a rated voltage of 12.66kV and a total load capacity of 3715kW. Multiple distributed generation sources have been integrated into the network, including a 900kW wind power system at node 13, and photovoltaic power systems at nodes 14 and 31 with capacities of 870kW and 690kW, respectively. With the increasing proportion of distributed generation sources, the distribution network faces a series of technical challenges, such as intensified voltage fluctuations, difficulties in renewable energy integration, and rising operating costs. The technical team decided to adopt the two-layer optimization method for distribution networks proposed in this invention, which aims to enhance overall carrying capacity. By optimizing the configuration of energy storage systems and reactive power compensation devices, and coordinating and optimizing the active and reactive power operation strategies of the system, the overall carrying capacity of the distribution network can be comprehensively improved.

[0091] The technical team followed Figure 1 The flowchart shown illustrates a two-layer optimization strategy for distribution networks aimed at improving overall carrying capacity. First, a three-layer evaluation system for the overall carrying capacity of the distribution network was constructed. The target layer is defined as the distribution network's adaptability to distributed generation access; the performance layer includes four dimensions: security, flexibility, economy, and low carbon emissions; and the indicator layer includes seven specific quantitative indicators. Security indicators include voltage deviation rate and voltage qualification rate; flexibility indicators include net load volatility and renewable energy utilization rate; economic indicators include operating costs and electricity purchase costs; and the low carbon emissions indicator is total carbon emissions. The technical team collected hourly node voltage values, distributed generation output values, and node load values ​​for 33 nodes in the distribution network over a continuous 24-hour period using the distribution network monitoring system. The collected data shows that the maximum daily load of the distribution network occurs at 18:00, reaching 3580kW, and the minimum daily load occurs at 3:00, at 1650kW. The daily power generation of the wind power system is 15840kWh, and the daily power generation of the photovoltaic system is 10260kWh.

[0092] The technical team calculated the raw values ​​of various evaluation indicators based on the collected operational data. The voltage deviation rate was calculated to be 0.0327, indicating that the average deviation of the distribution network node voltage from the rated voltage was 3.27%. The voltage compliance rate was calculated to be 0.823, indicating that only 82.3% of the node voltages met the standard range of 0.95 to 1.05. The net load volatility was calculated by dividing the standard deviation of the 24-hour net load (658kW) by the mean (2156kW) to obtain 0.305, reflecting relatively severe net load fluctuations. The renewable energy utilization rate was calculated to be 0.867, indicating that the actual renewable energy consumption was 86.7% of the predicted power generation, with 13.3% of wind and solar power being curtailed. Operating costs, including the operating costs of the energy storage system, capacitors, static var compensators, and distributed power sources, totaled 2830 yuan. The electricity purchase cost was calculated by multiplying the active power output of the upstream grid at each time point over 24 hours by the corresponding electricity price, resulting in 18560 yuan. The total carbon emissions are calculated by multiplying the consumption of various energy sources by the corresponding carbon emission factors and summing them up to 12.6t.

[0093] The technical team normalized the original indicator values ​​and positiveened negative indicators. Normalization eliminated the influence of different indicator dimensions, placing all indicators on a uniform scale of 0 to 1. Voltage deviation rate, net load fluctuation rate, operating cost, electricity purchase cost, and total carbon emissions, which are negative indicators, were transformed into positive indicators with higher values ​​being better. The technical team used the analytic hierarchy process (AHP) to construct an indicator comparison matrix, and expert evaluation determined the relative importance of each indicator. The largest eigenvalue of the comparison matrix was 7.086, the consistency index was 0.014, the random consistency index was 1.32, and the consistency ratio was 0.011, which is less than 0.1, thus passing the consistency test. The weight values ​​of each indicator were calculated using the geometric mean method, as shown in Table 1.

[0094] Table 1. Weight values ​​of evaluation indicators

[0095] The technical team established a comprehensive carrying capacity scoring model based on the positiveized values ​​and weights of each indicator. Before optimization, the comprehensive carrying capacity score of the distribution network was 68.34, with a safety score of 53.72, a flexibility score of 49.86, an economic score of 38.45, and a low-carbon score of 36.21. The overall carrying capacity was at a lower-middle level.

[0096] The technical team followed Figure 2The illustrated two-layer programming model constructs the upper-layer location and capacity optimization model, with the optimization objective being the minimum weighted sum of active power loss and voltage offset. The active power loss weight coefficient is set to 0.6, and the voltage offset weight coefficient is set to 0.4. The technical team uses an improved Grey Wolf optimization algorithm to optimize the location and capacity of the energy storage system and reactive power compensation device. The improved Grey Wolf optimization algorithm introduces a nonlinear convergence factor and a dimensionality learning strategy. The convergence factor decreases linearly from 2 to 0 using a cosine function, and the dimensionality learning strategy randomly selects 60% of the dimensions for learning and updating in each iteration. The algorithm sets the population size to 30 and the maximum number of iterations to 100. The energy storage system's charging and discharging power capacity range is set to 100 to 800 kW, and the energy capacity range is set to 1000 to 8000 kWh. The reactive power compensation of a single capacitor bank is 150 kvar, and the maximum number of switching banks is 10. The reactive power compensation capacity range of the static var compensator is set to -600 to +600 kvar. The upper-level optimization model must satisfy the constraints of power flow equations, node voltage constraints, energy storage system capacity constraints, capacitor capacity constraints, and static var compensator capacity constraints.

[0097] The optimized configuration at the upper level involves installing an energy storage system with a charging / discharging power capacity of 600kW and an energy capacity of 6000kWh at node 11, an energy storage system with a charging / discharging power capacity of 500kW and an energy capacity of 5000kWh at node 30, a capacitor bank with 10 switching groups at node 8, a capacitor bank with 7 switching groups at node 26, and a static var compensator (SVC) with a reactive power compensation capacity of ±400kvar at node 17. After configuration, the technical team calculated the active power loss to be 186.5kW, a 27.8% reduction compared to the original 258.3kW. The preset network loss threshold was set at 148.9kW based on the rated capacity of the distribution network. Since the optimized active power loss of 186.5kW exceeded the preset threshold of 148.9kW, the technical team initiated a multi-timescale rolling optimization strategy. The 24-hour optimization period is divided into four rolling optimization time windows, each lasting six hours: 0:00 to 6:00, 6:00 to 12:00, 12:00 to 18:00, and 18:00 to 24:00. Upper-level addressing and capacity optimization is re-executed within each rolling optimization time window.

[0098] The first time window optimization yielded the following results: Node 11 energy storage system has a charge / discharge power capacity of 550kW and an energy capacity of 5500kWh; Node 30 energy storage system has a charge / discharge power capacity of 450kW and an energy capacity of 4500kWh; Node 8 has 9 capacitor banks; Node 26 has 6 capacitor banks; and Node 17 static var compensator has a reactive power compensation capacity of ±350kvar. The second time window optimization yielded the following results: Node 11 energy storage system has a charge / discharge power capacity of 580kW and an energy capacity of 5800kWh; Node 30 energy storage system has a charge / discharge power capacity of 520kW and an energy capacity of 5200kWh; Node 8 has 10 capacitor banks; Node 26 has 7 capacitor banks; and Node 17 static var compensator has a reactive power compensation capacity of ±420kvar. The third time window optimization yielded the following results: Node 11 energy storage system charging / discharging power capacity of 620kW and energy capacity of 6200kWh; Node 30 energy storage system charging / discharging power capacity of 530kW and energy capacity of 5300kWh; Node 8 capacitor switching group number of 10 groups; Node 26 capacitor switching group number of 8 groups; and Node 17 static var compensator reactive power compensation capacity of ±450kvar. The fourth time window optimization yielded the following results: Node 11 energy storage system charging / discharging power capacity of 600kW and energy capacity of 6000kWh; Node 30 energy storage system charging / discharging power capacity of 500kW and energy capacity of 5000kWh; Node 8 capacitor switching group number of 9 groups; Node 26 capacitor switching group number of 7 groups; and Node 17 static var compensator reactive power compensation capacity of ±400kvar. After rolling optimization, the active power loss in each time window is 142.6kW, 145.8kW, 147.3kW and 146.1kW, respectively, all of which meet the preset network loss threshold requirements.

[0099] The technical team constructed a lower-level operational optimization model based on the configuration scheme determined by rolling optimization, with the goal of maximizing the comprehensive carrying capacity score. An improved Grey Wolf optimization algorithm was used to coordinate and optimize the energy storage system's charging and discharging power sequence, capacitor switching sequence, static var compensator (SVC) reactive power regulation capacity sequence, and distributed generation output sequence. The energy storage system's charging and discharging efficiency was set to 0.9, the state of charge (SOC) variation range was 0.1 to 0.9, and the initial SOC was set to 0.5. The energy storage system's charging and discharging must satisfy charging and discharging mutual exclusion constraints, charging and discharging power constraints, SOC constraints, and consistency constraints between the first and last SOCs. Capacitor operation must satisfy switching sequence constraints and reactive power constraints. SVC operation must satisfy reactive power regulation capacity constraints. Distributed generation output must satisfy upper and lower output limit constraints. The lower-level optimization model also needs to satisfy power flow equation constraints and node voltage constraints.

[0100] The lower-level optimization yields the charge and discharge power sequence of the node 11 energy storage system over 24 hours, as follows: Figure 6 As shown. From Figure 6 As shown in the charge / discharge power diagram of energy storage system 1, the system charges at a power of 200 to 350 kW during the off-peak nighttime period from 0:00 to 6:00, and discharges at a power of 400 to 550 kW during the peak load periods from 8:00 to 10:00 and from 17:00 to 20:00. The charge / discharge power sequence of the energy storage system at node 30 over 24 hours is as follows: Figure 7 As shown. From Figure 7 As shown in the energy storage 2 charge / discharge power diagram, the energy storage system charges at a power of 150 to 300 kW during the off-peak hours from 1:00 to 5:00 AM, and discharges at a power of 350 to 480 kW during the peak load periods from 11:00 to 1:00 PM and from 6:00 to 9:00 PM. The capacitor switching sequence at node 8 shows 8 to 10 groups activated between 6:00 to 9:00 AM and 5:00 to 10:00 PM, and 3 to 5 groups activated between 0:00 to 5:00 AM and 11:00 to 12:00 AM. The capacitor switching sequence at node 26 shows 6 to 8 groups activated between 7:00 to 10:00 AM and 6:00 to 9:00 PM, and 2 to 4 groups activated during other periods. The static var compensator (SVC) reactive power regulation capacity sequence at node 17 shows that it provides 300 to 420 kvar of inductive reactive power support during peak load periods and absorbs 150 to 280 kvar of capacitive reactive power during off-peak periods. The output sequence of distributed power sources is determined based on the wind and solar resource forecast curves and system optimization requirements. The output of the wind power system reaches a peak of 720kW at night, and the output of the photovoltaic system reaches a peak of 780kW at noon.

[0101] During the lower-level optimization process, the technical team continuously monitored the deviation between the actual and predicted output values ​​of distributed power sources. At one point, the actual output of the wind power system at node 13 was 485kW, while the predicted output was 620kW, resulting in a deviation of 15%, which did not exceed the 15% threshold. Subsequently, at another point, the actual output of the photovoltaic system at node 14 was 532kW, while the predicted output was 748kW, resulting in a deviation of 24.8%, exceeding the 15% threshold. The technical team immediately activated the model predictive control algorithm to predict the node voltage values ​​for the next 30 minutes. The model predictive control algorithm uses a 5-minute prediction step size and a 6-step prediction time domain. Based on the node voltage prediction sequence, the technical team adjusted the reactive power regulation capacity sequence of the static var compensator (SVC) and the sequence of capacitor switching groups. The reactive power regulation capacity of the SVC at node 17 was increased from 380kvar to 450kvar, and the number of capacitor switching groups at node 8 was increased from 8 to 10, effectively suppressing voltage fluctuations caused by photovoltaic output fluctuations.

[0102] The technical team substituted the optimized energy storage system charging and discharging power sequence, capacitor switching group sequence, static var compensator reactive power regulation capacity sequence, and distributed power output sequence into the comprehensive carrying capacity scoring model, calculating an optimized comprehensive carrying capacity score of 85.73. The team then fed this score back to the upper-level optimization model, initiating a two-layer optimization iteration process. After the first iteration, the comprehensive carrying capacity score was 85.73; after the second iteration, it was 86.18; after the third iteration, it was 86.42; after the fourth iteration, it was 86.56; and after the fifth iteration, it was 86.63. The changes in the third to fifth iterations were 0.28%, 0.33%, and 0.16%, respectively, with three consecutive changes less than 0.5%, meeting the termination condition for iteration. The team output the final optimization result: a comprehensive carrying capacity score of 86.63, representing a 26.7% improvement compared to the original score of 68.34.

[0103] The technical team conducted a detailed analysis of the optimization results. In terms of safety, the optimized voltage offset rate decreased to 0.0168, a 48.6% reduction compared to before optimization, and the voltage compliance rate increased to 0.967, a 17.5% improvement compared to before optimization. Figure 5 The voltage diagram after optimization of Scheme 3 shows that the voltage at each node remains within the acceptable range of 0.96 to 1.04, indicating a significant improvement in voltage quality. The voltage curves at each node are stable, and the fluctuation amplitude is significantly reduced. In terms of flexibility, the net load volatility decreased to 0.186 after optimization, a reduction of 39.0% compared to before optimization. The renewable energy utilization rate increased to 0.956, an increase of 10.3% compared to before optimization, and the wind and solar curtailment rate decreased from 13.3% to 4.4%. In terms of economics, the operating cost increased to 3245 yuan after optimization, while the electricity purchase cost decreased to 15680 yuan, and the total cost decreased from 21390 yuan to 18925 yuan, a reduction of 11.5%. In terms of low carbon emissions, the total carbon emissions decreased to 10.8 tons after optimization, a reduction of 14.3% compared to before optimization. The performance level scores are as follows: safety score 66.52, flexibility score 55.50, economic score 42.34, and low carbon emission score 41.37. Figure 4 The radar chart showing the positive comprehensive carrying capacity index clearly demonstrates the comprehensive improvement in performance across all dimensions. The envelope area of ​​Scheme 3 in the radar chart is significantly larger than that of Scheme 1 and Scheme 2, indicating that the distribution network has achieved significant improvements in four dimensions—safety, flexibility, economy, and low carbon emissions—after adopting the active and reactive power coordination optimization strategy.

Claims

1. A two-layer optimization method for distribution networks aimed at improving comprehensive carrying capacity, characterized in that, Includes the following steps: A three-tiered evaluation system for the comprehensive carrying capacity of the distribution network is constructed. This system collects data on distribution network node voltage, distributed generation output, and node load. The original index values ​​are calculated and normalized, and negative indices are normalized. The analytic hierarchy process (AHP) is used to calculate index weights, and a comprehensive carrying capacity scoring model is established. An upper-level optimization model is constructed, and an improved gray wolf optimization algorithm is used to optimize the location and capacity of energy storage systems and reactive power compensation devices. When active power loss exceeds a preset loss threshold, the 24-hour optimization period is divided into four rolling optimization time windows, and the upper-level optimization is re-executed. A lower-level optimization model is constructed with the objective function of maximizing the comprehensive carrying capacity score. An improved gray wolf optimization algorithm is used to coordinate and optimize the energy storage system's charging and discharging power sequence, capacitor switching group sequence, static var compensator reactive power regulation capacity sequence, and distributed generation output sequence. When the distributed generation output deviation percentage exceeds 15%, a model predictive control algorithm is used to predict the node voltage prediction value sequence and adjust the reactive power regulation capacity sequence and switching group sequence. The lower-level optimization results are substituted into the comprehensive carrying capacity score calculation model to calculate the optimized comprehensive carrying capacity score and fed back to the upper level for re-iteration. The iteration is terminated when the change in the optimized comprehensive carrying capacity score obtained after three consecutive iterations is less than 0.5% or the number of iterations reaches 50.

2. The method according to claim 1, characterized in that, In the three-tiered comprehensive carrying capacity assessment system, the target layer is the distributed power source access adaptability, the performance layer includes safety, flexibility, economy and low carbon emissions, and the indicator layer includes voltage deviation rate, voltage qualification rate, net load fluctuation rate, renewable energy utilization rate, operating cost, electricity purchase cost and total carbon emissions.

3. The method according to claim 2, characterized in that, The voltage offset rate refers to the voltage offset rate of a single node obtained by dividing the absolute value of the difference between the node voltage value and the rated voltage value at each node at each time step by the rated voltage value. The voltage offset rate is obtained by taking the arithmetic mean of the voltage offset rates of all nodes and the single node voltage offset rates at 24 time steps.

4. The method according to claim 3, characterized in that, The voltage qualification rate refers to determining whether the voltage value of each node at each time point is within the range of 0.95 times to 1.05 times the rated voltage value. When the voltage value is within the qualified voltage range, the node voltage qualification flag is set to 1; when the voltage value is outside the qualified voltage range, the node voltage qualification flag is set to 0. The voltage qualification rate is obtained by calculating the arithmetic mean of the node voltage qualification flags of all nodes and at 24 time points.

5. The method according to claim 4, characterized in that, The net load volatility refers to the net load value calculated at each moment within 24 hours. The standard deviation of the net load values ​​at 24 moments is calculated to obtain the net load standard deviation. The arithmetic mean of the net load values ​​at 24 moments is calculated to obtain the net load average. The net load standard deviation is divided by the net load average to obtain the net load volatility.

6. The method according to claim 5, characterized in that, The analytic hierarchy process (AHP) constructs an indicator comparison matrix to compare indicators pairwise. Each element in the indicator comparison matrix represents the relative importance of the indicator, which is expressed on a scale of 1 to 9. The indicator comparison matrix needs to undergo a consistency test. When the consistency ratio is less than 0.1, the indicator comparison matrix is ​​considered to have passed the consistency test.

7. The method according to claim 6, characterized in that, The calculation of the indicator weight values ​​adopts the geometric mean method. The product of all matrix elements in each row of the indicator comparison matrix is ​​obtained to obtain the row product. The nth root of the row product is used to obtain the initial value of the indicator weight. The sum of the initial values ​​of the indicator weights of all indicators is used to obtain the total initial value of the weights. The initial value of the indicator weight of each indicator is divided by the total initial value of the weights to obtain the indicator weight value.

8. The method according to claim 7, characterized in that, The comprehensive carrying capacity scoring calculation model refers to multiplying the positive index value of each indicator by the index weight value to obtain the index weighted score, and then summing the index weighted scores of all indicators and multiplying by 100 to obtain the comprehensive carrying capacity score.

9. The method according to claim 8, characterized in that, The upper-level optimization model takes the minimum weighted sum of active power loss and voltage offset as the upper-level objective function. Active power loss is multiplied by the active power loss weight coefficient to obtain the active power loss weight term, and the total voltage offset is multiplied by the voltage offset weight coefficient to obtain the voltage offset weight term. The active power loss weight term plus the voltage offset weight term is the upper-level objective function value.

10. The method according to claim 9, characterized in that, The improved gray wolf optimization algorithm is an improved algorithm based on the traditional gray wolf optimization algorithm by introducing a nonlinear convergence factor and a dimension learning strategy. The nonlinear convergence factor adopts the form of a cosine function and decreases with the number of iterations. The dimension learning strategy randomly selects some dimensions for learning and updating in each iteration while keeping the other dimensions unchanged.

Images