Multi-objective power distribution network source-network-load-storage collaborative planning method and system

Abstract

This invention provides a multi-objective distribution network source-grid-load-storage coordinated planning method and system, relating to the field of power supply systems. The method includes: acquiring input data, including at least configuration data, source-load time-series data, distribution network operating parameters, and economic and low-carbon parameters; modeling distributed wind power, distributed photovoltaics, energy storage systems, smart soft switches, and demand-side response resources; constructing economic evaluation indicators, flexibility evaluation indicators, and low-carbon evaluation indicators; constructing a two-layer optimization model for distribution network source-grid-load-storage coordinated planning; generating candidate planning schemes based on the input data using an improved harmony search algorithm in the planning layer of the two-layer optimization model; and generating operational evaluation results of the candidate planning schemes using a genetic algorithm based on Logistic chaotic mapping in the operational layer of the two-layer optimization model, thereby generating the optimal planning scheme. This method has the advantage of improving the effectiveness of distribution network source-grid-load-storage coordinated planning.

Description

Technical Field

[0001] This invention relates to the field of power supply systems, and in particular to a multi-objective method and system for coordinated planning of power generation, grid, load and storage in distribution networks. Background Technology

[0002] With the rapid growth of new energy installed capacity, the proportion of distributed power sources such as distributed wind power and distributed photovoltaic power in the distribution network is constantly increasing. The power flow of the distribution network has changed from a unidirectional radial pattern to a multidirectional dynamic interaction. The coupling relationship between the randomness of power output on the source side, the difference in energy consumption on the load side and the network operation constraints has become more complex. Problems such as local reverse power flow, node voltage exceeding the limit, branch heavy load and limited new energy consumption are becoming increasingly prominent.

[0003] Existing distribution network planning is mainly driven by load growth and relies primarily on grid expansion and equipment capacity increase. It rarely considers the impact of high-proportion renewable energy output fluctuations, energy storage timing regulation, demand-side response, and grid-side flexible interconnection equipment on the operating status. Equipment configuration is based solely on static capacity margins, which can easily lead to situations where the planning scheme is statically feasible but frequently triggers voltage, power flow, or regulation capacity constraints in actual operation. Therefore, coordinated planning of power generation, grid, load, and storage has become an important direction. Existing technologies propose a multi-dimensional coordinated planning method for distribution network sources, grids, and storage that considers flexibility. However, they do not incorporate low-carbon performance as an independent objective into a unified optimization framework. Furthermore, the flexibility indicators used in the comprehensive decision-making process are biased towards static capacity evaluation at the planning level, making it difficult to reflect the impact of energy storage state of charge, actual active power transmission capacity of SOPs, and operational constraints on actual regulation capabilities. Existing technologies propose a collaborative planning model for distribution network sources, grids, loads, and storage that considers both low carbon emissions and economics. This model aims to minimize carbon emissions and overall costs, employs a three-layer serial structure of planning, scheduling, and reconfiguration, and considers price-based demand response. However, it does not construct a systematic flexibility evaluation index, and the operational verification results in the three-layer serial structure are difficult to provide closed-loop feedback to the planning scheme. Moreover, it does not consider flexible interconnected equipment with continuous active power transfer capabilities, such as SOPs, as key planning objects.

[0004] In summary, existing technologies still have the following shortcomings: First, the three objectives of economy, flexibility, and low carbon emissions have not yet been balanced in a unified manner within the same distribution network planning framework; second, although some schemes have constructed flexibility indicators, the indicators used for comprehensive decision-making are biased towards static capacity evaluation, which is difficult to reflect the actual available adjustment capacity during operation; third, although some schemes have considered low carbon emissions and demand response, they lack closed-loop feedback between the planning and operation layers, and have not fully considered the active power transfer function of grid-side flexible interconnection equipment such as SOPs.

[0005] Therefore, there is a need to provide a multi-objective distribution network source-grid-load-storage coordinated planning method and system to improve the effectiveness of distribution network source-grid-load-storage coordinated planning. Summary of the Invention

[0006] This invention provides a multi-objective distribution network source-grid-load-storage coordinated planning method, comprising: acquiring input data, wherein the input data includes at least configuration data, source-load time-series data, distribution network operating parameters, and economic and low-carbon parameters; modeling distributed wind power, distributed photovoltaic, energy storage systems, smart soft switches, and demand-side response resources; constructing economic evaluation indicators, flexibility evaluation indicators, and low-carbon evaluation indicators; constructing a two-layer optimization model for distribution network source-grid-load-storage coordinated planning, wherein the two-layer optimization model includes a planning layer and an operation layer; generating candidate planning schemes based on the input data using an improved harmony search algorithm in the planning layer of the two-layer optimization model; generating operation evaluation results of the candidate planning schemes using a genetic algorithm based on Logistic chaotic mapping in the operation layer of the two-layer optimization model; and generating an optimal planning scheme based on the economic evaluation indicators, flexibility evaluation indicators, low-carbon evaluation indicators, and the operation evaluation results of the candidate planning schemes.

[0007] Furthermore, through the planning layer of the two-layer optimization model, an improved harmony search algorithm is adopted to generate candidate planning schemes based on the input data. This includes: generating an initial harmony memory library based on the input data, wherein each harmony in the initial harmony memory library corresponds to a candidate planning scheme, and the candidate planning schemes include the configuration results of distributed wind power, distributed photovoltaic, energy storage systems, and grid-side interconnection equipment; and generating candidate planning schemes based on the initial harmony memory library by adopting an adaptive parameter scheduling strategy, a harmony library structured reorganization strategy, and an optimal solution approximation out-of-bounds repair strategy.

[0008] Furthermore, the adaptive parameter scheduling strategy includes: adaptively updating the memory selection probability and the tone fine-tuning probability with the number of iterations.

[0009] Furthermore, the harmonic library structured reorganization strategy includes: when memory selection is triggered, extracting harmonic vectors from the harmonic memory library to form a harmonic sub-library, and using the diagonal elements of the harmonic sub-library to form a new harmony; the out-of-bounds repair strategy for approximating the optimal solution includes: when a variable in the new harmony goes out of bounds, using the current optimal harmony to repair the out-of-bounds components of the new harmony.

[0010] Furthermore, the decision variables of the operation layer of the two-layer optimization model include at least the active power output of distributed wind power and distributed photovoltaic power, energy storage charging and discharging power, energy storage state of charge, active power transmission of smart soft switching, load reduction ratio, and load transfer ratio of loads that can be shifted in each time period; the constraint set of the operation layer of the two-layer optimization model includes at least the power balance constraint, network operation constraint, distributed power output and curtailment constraint, energy storage operation constraint, smart soft switching operation constraint, and demand-side response constraint. The objective function of the running layer of the two-layer optimization model is: , in, For the fuzzy comprehensive target value, For the fuzzy membership degree of operating costs, Fuzzy membership degree for short-term flexibility For fuzzy membership of carbon emissions, , and, As weight.

[0011] Furthermore, the genetic algorithm based on Logistic chaotic mapping introduces chaotic mutation and jump search mechanisms during the genetic evolution process to enhance the local search of individuals.

[0012] Furthermore, the economic evaluation indicators include annualized comprehensive cost indicators and typical daily operating costs; the annualized comprehensive cost consists of annual investment cost, annual operation and maintenance cost, electricity purchase and sale cost, demand-side management cost, and wind and solar curtailment penalty cost; the typical daily operating cost consists of daily electricity purchase and sale cost, daily demand-side management cost, and daily wind and solar curtailment penalty cost.

[0013] Furthermore, the flexibility evaluation indicators include medium- and long-term configuration flexibility evaluation indicators and comprehensive operational flexibility evaluation indicators; the medium- and long-term configuration flexibility evaluation indicators consist of medium- and long-term power regulation flexibility and medium- and long-term grid structure regulation flexibility; the comprehensive operational flexibility evaluation indicators consist of daily power regulation flexibility and daily grid structure regulation flexibility.

[0014] Furthermore, the low-carbon performance evaluation indicators include annualized comprehensive equivalent emissions and daily equivalent emissions.

[0015] This invention provides a multi-objective distribution network source-grid-load-storage collaborative planning system, comprising: a data acquisition module for acquiring input data, wherein the input data includes at least configuration data, source-load time-series data, distribution network operating parameters, and economic and low-carbon parameters; a unified modeling module for modeling distributed wind power, distributed photovoltaics, energy storage systems, smart soft switches, and demand-side response resources; an index construction module for constructing economic evaluation indicators, flexibility evaluation indicators, and low-carbon evaluation indicators; a model construction module for constructing a two-layer optimization model for distribution network source-grid-load-storage collaborative planning, wherein the two-layer optimization model includes a planning layer and an operation layer; a two-layer optimization module for generating candidate planning schemes based on the input data using an improved harmony search algorithm in the planning layer of the two-layer optimization model; the two-layer optimization module is also used to generate operation evaluation results of the candidate planning schemes using a genetic algorithm based on Logistic chaotic mapping in the operation layer of the two-layer optimization model; and a planning generation module for generating an optimal planning scheme based on the economic evaluation indicators, flexibility evaluation indicators, low-carbon evaluation indicators, and the operation evaluation results of the candidate planning schemes.

[0016] Compared with existing technologies, the multi-objective distribution network source-grid-load-storage coordinated planning method and system provided by this invention has at least the following beneficial effects: 1. By incorporating three evaluation dimensions—annualized comprehensive cost, medium- and long-term configuration flexibility, and annualized comprehensive equivalent emissions—at the planning level, distribution network planning is no longer limited to a single standard but rather involves a comprehensive trade-off between economic efficiency, regulation capacity, and low-carbon operation. The candidate planning schemes are ranked in a three-dimensional manner, and the output optimal planning scheme is the result of a balance among the three, rather than a one-sided scheme where one indicator is optimal but other indicators are severely sacrificed. This approach helps to obtain an optimal planning scheme that better meets the actual engineering needs.

[0017] 2. After generating candidate schemes based on configuration data and source-load time-series data using an improved harmony search algorithm, the planning layer does not directly output the results but instead sends them to the operation layer for further verification. The operation layer employs a genetic algorithm based on Logistic chaotic mapping to perform time-series scheduling simulations under the predicted curves of wind power output, photovoltaic power output, active load, and reactive load on typical days, verifying the schemes daily. Only schemes that meet the operational constraints on all typical days will have their operational evaluation results fed back to the planning layer for comprehensive decision-making. This mechanism effectively avoids the problem in traditional planning where static capacity is feasible but actual operation exceeds limits. Attached Figure Description

[0018] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein: Figure 1 This is a flowchart illustrating a multi-objective distribution network source-grid-load-storage coordinated planning method according to some embodiments of this specification; Figure 2 This is a schematic diagram of the solution process for a two-layer optimization model of coordinated planning of power distribution network sources, grids, loads and storage, as shown in some embodiments of this specification. Figure 3 This is a flowchart illustrating an improved harmony search algorithm according to some embodiments of this specification; Figure 4 This is a flowchart illustrating a chaotic genetic algorithm according to some embodiments of this specification; Figure 5 This is a schematic diagram of the IEEE-33 node power distribution system according to some embodiments of this specification; Figure 6 This is a schematic diagram of wind power and photovoltaic output time series curves according to some embodiments of this specification; Figure 7This is a schematic diagram of load timing curves according to some embodiments of this specification; Figure 8 This is a schematic diagram of the convergence curve of the planning and execution layer algorithm according to some embodiments of this specification; Figure 9 This is a schematic diagram of a multi-objective distribution network source-grid-load-storage coordinated planning system according to some embodiments of this specification. Detailed Implementation

[0019] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0020] Figure 1 This is a flowchart illustrating a multi-objective distribution network source-grid-load-storage coordinated planning method according to some embodiments of this specification, such as... Figure 1 As shown, the multi-objective distribution network source-grid-load-storage coordinated planning method may include the following steps.

[0021] Step 110: Obtain input data.

[0022] The input data includes at least configuration data, source-load time sequence data, distribution network operation parameters, and economic and low-carbon parameters.

[0023] Specifically, the configuration data may include the access locations, rated capacities of individual units, and upper and lower limits of the number of distributed wind power, distributed photovoltaic, and energy storage systems at each candidate node in the distribution network; the connection node pairs, capacity boundaries of interconnection equipment, and selection range of interconnection equipment types (e.g., smart soft switches, ordinary interconnection switches, etc.) at candidate grid-side interconnection locations; and the load nodes participating in demand-side response and their adjustable upper limits. Here, the access location refers to the specific node number connected when the three types of source / storage equipment (distributed wind power, distributed photovoltaic, and energy storage systems) are actually connected to the distribution network, representing the result of the site selection and capacity determination decision at the planning level. The capacity boundaries of interconnection equipment are the upper and lower limits of the allowed rated capacity values ​​of grid-side interconnection equipment (smart soft switches, SOPs, or ordinary interconnection switches) during planning, serving as the constraint boundaries for the planning level when determining the capacity of grid-side equipment.

[0024] Source-load time-series data can include predicted curves for wind power output, solar power output, active load, and reactive load on typical days. Typical days are obtained by scene segmentation and clustering based on historical wind speed, solar irradiance, and load data. Source-load time-series data are the input conditions used to verify planning schemes; essentially, they are a set of typical day curves representing different operating conditions throughout the year. Specifically, the typical day wind power output curve describes the fluctuation process of wind turbine power caused by wind speed changes over 24 hours; the typical day solar power output curve describes the change process of solar power generation corresponding to changes in solar irradiance from morning to evening; the typical day active load curve describes the time-varying characteristics of user electricity demand throughout the day; and the typical day reactive load curve describes the variation of reactive component in the load over time. A typical day refers to a representative date selected from historical wind speed, solar irradiance, and load measurement data from the past year or even several years, after scene segmentation and clustering, that represents a certain type of operating characteristic. First, historical data is divided into several scenarios based on dimensions such as season, weather type, and weekday / holiday period (e.g., windy days in summer, sunny days in winter, cloudy days in spring and autumn, weekdays, weekends, etc.). Then, the daily data within each scenario are grouped using a clustering algorithm (e.g., K-means). From each group, the day closest to the cluster center is selected as the typical day for that scenario. Selecting a small number of typical days can represent the source load variation characteristics of the entire 365 days of the year, ensuring the comprehensiveness of the verification while avoiding the computational explosion caused by daily simulation.

[0025] Distribution network operating parameters may include distribution network topology, node number, branch impedance, allowable range of node voltage, and upper limit of branch transmission capacity.

[0026] Economic and low-carbon parameters may include peak, flat, and valley time-of-use electricity prices, demand-side management (DSM) compensation unit price, wind and solar curtailment penalty unit price, and equivalent emission conversion factor corresponding to electricity purchased from the upstream grid.

[0027] The application scenarios of the multi-objective distribution network source-grid-load-storage collaborative planning method are not limited to the IEEE 33-node distribution system, but can also be applied to distribution networks with other node sizes, feeder structures, or topologies. Typical daily scenarios are not limited to the four types of typical days in spring, summer, autumn, and winter. Monthly typical days, multiple operating scenarios, or representative scenario sets obtained by scenario reduction can be constructed based on the historical wind speed, light intensity, and load data of the actual region.

[0028] Step 120: Model distributed wind power, distributed photovoltaics, energy storage systems, smart soft switches, and demand-side response resources.

[0029] Specifically, distributed wind power is achieved by wind turbine generators (WG), whose output is mainly affected by wind speed. Distributed photovoltaic power is achieved by photovoltaic (PV) units, whose output is mainly affected by solar irradiance and module operating temperature.

[0030] Typical daily wind and solar power output curves are used to describe the time-series fluctuation characteristics of renewable energy sources, and the maximum available output for the corresponding time period is determined based on the installed capacity of each candidate node. During operation-level scheduling, the actual output of wind and solar power must not exceed the maximum available output for the corresponding time period. When renewable energy cannot be fully absorbed due to network constraints or regulation capacity limitations, wind and solar curtailment variables are allowed, and the amount of unabsorbed renewable energy is limited through the penalty cost of wind and solar curtailment.

[0031] The energy storage system employs a State of Charge (SOC) model to describe the temporal coupling relationship between charging / discharging power and energy state. The SOC of the energy storage system at each time period is jointly determined by the SOC of the previous time period, charging power, discharging power, charging / discharging efficiency, and rated capacity. To ensure the safe operation of the energy storage system, the model includes upper and lower limits for charging / discharging power, upper and lower limits for SOC, mutual exclusion constraints for charging / discharging, and SOC constraints at the beginning and end of the day. The energy storage system achieves energy time-shifting and power balance support through charging and discharging regulation, thereby realizing power regulation flexibility.

[0032] Soft Open Points (SOPs), as flexible interconnection devices on the grid side, primarily utilize their continuously controllable bidirectional active power transmission capabilities to achieve power balance and flow regulation between feeders. In grid-side modeling, only the active power transmission characteristics of the SOP are considered; reactive power regulation variables and reactive power capacity constraints are not introduced. At each candidate tie location, the SOP and ordinary tie switches (e.g., tie disconnectors) participate in the planning layer selection as optional devices.

[0033] Demand-side response includes two categories: shiftable loads and loads that can be reduced. Shiftable loads participate in peak shaving and valley filling by transferring a portion of electricity demand between different time periods; loads that can be reduced participate in system regulation by reducing a portion of the load during a specified time period. The access nodes and callable limits of demand-side response resources are pre-defined boundary parameters, not decision variables at the planning level, but rather participate in timing optimization as schedulable resources at the operation level. The amount of demand-side response calls must meet constraints such as response ratio, response time period, load transfer conservation, and maximum delay time.

[0034] By inputting data and unifying the modeling of distributed wind power, distributed photovoltaics, energy storage systems, smart soft switches, and demand-side response resources, it is possible to incorporate the renewable energy output on the source side, the energy time shift on the storage side, the active power transfer on the grid side, and the response capability on the load side into a unified source-grid-load-storage collaborative optimization framework, providing a foundation for subsequent calculations of economic efficiency, flexibility, and low-carbon indicators.

[0035] Source-side resources can be distributed wind power, distributed photovoltaic power, or other distributed generation units that meet grid-connected operation constraints; storage-side resources can be electrochemical energy storage or other equivalent energy storage devices that meet charging and discharging power, energy state, and efficiency constraints; load-side demand response can be one or a combination of load reduction and load shifting, and the response ratio, response period, and duration can be set according to actual user-side conditions.

[0036] The locations of candidate interconnections on the grid side, the number of smart soft switches installed, their rated capacity, and the configuration of traditional interconnection switches can be adjusted according to the actual distribution network structure and investment conditions. At some candidate interconnection locations, only smart soft switches can be configured, or smart soft switches and ordinary interconnection switches can be selected together as optional equipment at the planning level.

[0037] Step 130: Construct economic evaluation indicators, flexibility evaluation indicators, and low-carbon evaluation indicators.

[0038] Specifically, the economic evaluation indicators are used to reflect the investment and operating cost level of the planning scheme, the flexibility evaluation indicators are used to reflect the ability of the planning scheme to adjust and adapt to changes in source load fluctuations and operating constraints, and the low-carbon evaluation indicators are used to reflect the comprehensive equivalent emission level of the planning scheme.

[0039] Economic evaluation indicators include annualized comprehensive cost and typical daily operating cost. The planning level uses the annualized comprehensive cost as the economic target to evaluate the comprehensive cost level of different planning schemes within the planning period; the operation level uses the typical daily operating cost as the scheduling evaluation indicator to reflect the intraday operational economy under a given candidate planning scheme.

[0040] The annualized comprehensive cost consists of annual investment cost, annual operation and maintenance cost, electricity purchase and sale cost, demand-side management cost, and wind and solar curtailment penalty cost. Specifically, annual investment cost represents the annual equivalent expenditure after annual value conversion for equipment such as distributed wind power, distributed photovoltaics, energy storage systems, smart soft switches, and ordinary interconnection switches; annual operation and maintenance cost represents the annual operation and maintenance expenditure for various equipment; electricity purchase and sale cost represents the energy exchange cost between the distribution network and the upper-level grid; demand-side management cost represents the compensation expenditure when loads that can be reduced or shifted are called upon; and wind and solar curtailment penalty cost represents the opportunity loss and penalty cost when renewable energy is not integrated. , in, The annualized comprehensive cost, The annual investment cost of controllable components for power generation, grid, and energy storage. The annual operation and maintenance cost of controllable components for power generation, grid, load and storage. To cover the cost of purchasing and selling electricity, For demand-side management costs, The cost of penalties for abandoning wind and solar power.

[0041] Because different equipment has different investment cycles and service lives, in order to ensure the comparability of investment costs for various types of equipment on the same time scale, the annual value conversion method is used to convert one-time investments into annual costs: , in, This is the annual value conversion factor; It refers to the service life of the equipment; It is the discount rate.

[0042] , , , , , in, Indicates the location code of candidate access nodes or candidate connections in the distribution network; and These represent the unit capacity cost of photovoltaic and wind power, respectively. and These refer to the planned configuration capacity or maximum installed capacity of photovoltaic and wind power, respectively. and These represent the unit cost of energy storage and the unit cost of energy storage, respectively. This refers to the rated power of the energy storage system. This represents the maximum charging and discharging power of the energy storage system. The unit capacity cost of SOP; The rated capacity of the intelligent soft switch at the candidate contact location; This is the investment cost of a standard tie switch; Number of SOP ports; , It is a decision variable for the type of communication; when using a Standard Operating Procedure (SOP), , When using a standard tie switch, , ; , , , These are the annual operation and maintenance cost conversion factors for photovoltaic, wind power, energy storage, and SOP, respectively. The number of days in a year; Let n be the probability of scenario n; The power exchanged between the upstream power grid and the downstream power grid during time period t. The power transmitted to the upper-level power grid during time period t; Electricity purchase price, The electricity price; Let n be the probability of scenario n. The load reduction compensation factor is set to 1. It is the proportion of load that can be reduced; To reduce the active load during time period t; This is the load shift compensation coefficient. ; The target time period after load shifting; Representing the A movable load is composed of Time period shifted to The proportion of shift in time period; For the transferable load in Active load during a given time period; The penalty unit price for abandoning wind and solar power. and These refer to the power of curtailed solar and wind power.

[0043] Typical daily operating costs consist of daily electricity purchase and sales costs, daily demand-side management costs, and daily wind and solar curtailment penalty costs: , , , , in, , and These are the daily electricity purchase and sale costs of the distribution network, the daily demand-side management costs, and the daily wind and solar curtailment penalty costs.

[0044] The flexibility evaluation indicators include medium- and long-term configuration flexibility evaluation indicators and comprehensive operational flexibility evaluation indicators.

[0045] The medium- and long-term configuration flexibility evaluation index is used to reflect the adequacy of capacity after resource allocation, and consists of medium- and long-term power regulation flexibility and medium- and long-term grid structure regulation flexibility: , , , , , in, , These represent positive and negative adjustment indices, respectively, indicating the flexibility of power regulation. represent The maximum power reduction that can reduce the load; Representing the The maximum adjustable power of a load that can be shifted; , These represent the peak and trough values ​​of net load, respectively. and These represent the peak and trough net load values, taking into account future development and extreme scenarios, respectively. Indicates the first The maximum capacity of each contact point is taken as the maximum capacity of the SOP or contact switch, or 0. For medium- to long-term allocation flexibility indicators; It is a power regulation flexibility index of the distribution network, used to characterize the ability of resources such as energy storage systems, demand-side response and smart soft switching to regulate the fluctuations in the system's net load. It is an indicator of the grid regulation flexibility of the distribution network, used to characterize the support capability of grid-side equipment such as tie switches and smart soft switches for load transfer and power flow adjustment between feeders.

[0046] The comprehensive operational flexibility evaluation index is used to reflect the actual available adjustment capacity of the planning scheme under typical daily operating conditions, and consists of daily power adjustment flexibility and daily grid adjustment flexibility.

[0047] At the short-term operation level, the short-term power regulation capability of each controllable component is first defined. The short-term power regulation capability adopts the existing definition framework, and the DG term and energy storage energy constraint term are revised for consistency in combination with resource type and variable definition: , , , , in, , , They are the first Energy storage, load reduction and shifting in Power regulation capability during different time periods; , , These are the corresponding negative power regulation capabilities; and These represent the positive and negative power regulation capabilities during the SOP (Start of Production) period, respectively. and Representing the first Taiwan Energy Storage The charging and discharging power and state of charge during the time period; , and This represents the maximum depth of discharge, rated capacity, and charge / discharge efficiency of the energy storage. and Representing the A load that can be reduced and a load that can be shifted. The merits of the time period; and It refers to the percentage of load that can be reduced and its upper limit; and Representing the first A movable load is composed of Time period shifted to Time period, by The proportion of the shift from a time period to an uncertain time period; This represents the maximum delay in power supply for a load that can be shifted. This is the upper limit of the translation ratio; The active power injected by the SOP into the branch port is positive for inflow and negative for outflow.

[0048] Based on the typical daily time-series scheduling results, the operation layer calculates the daily power regulation flexibility and the daily grid regulation flexibility, and further obtains the comprehensive operation flexibility index: , , , , , , in, represent Power regulation flexibility index of time-of-use distribution network; and They are Net load of the system and maximum load of the feeders during the time period; This is the theoretical load factor verified by N-1, taken as 50%; This is the number of operating days and time periods, taken as 24; For feeder and the first Contact points Time period, maximum load capacity available for transfer; for The flexibility index of the grid structure during different time periods; A sign function to determine the power direction on both sides. This represents an indicator of overall operational flexibility. Indicators representing the daily power regulation flexibility of the distribution network; This is an indicator of the daily grid regulation flexibility of the distribution network. Compared with medium- and long-term configuration flexibility, comprehensive operational flexibility can further take into account operational factors such as energy storage state-of-charge timing constraints, actual active power transmission margin of smart soft switches, demand-side response call boundaries, node voltage limits, and branch capacity constraints. Therefore, adopting... As an evaluation metric for the flexibility dimension.

[0049] Low-carbon performance evaluation indicators include annualized comprehensive equivalent emissions and daily equivalent emissions.

[0050] Specifically, the low-carbon performance evaluation index is used to characterize the comprehensive equivalent emission level corresponding to the source-grid-load-storage coordinated planning scheme. This method focuses on the distribution network operation phase, attributing low-carbon performance mainly to the indirect emissions corresponding to the distribution network's purchase of electricity from the upper-level grid, and using a fixed emission factor to convert the purchased electricity volume. The direct emissions during the operation of distributed wind power and distributed photovoltaic power are approximately zero. Energy storage systems, smart soft switching, and demand-side response improve the local renewable energy absorption capacity and reduce the impact of external electricity purchases on the system's equivalent emissions.

[0051] The annualized comprehensive equivalent emissions are defined as the total annual emissions resulting from the weighted summation of equivalent emissions from electricity purchased from the upper-level power grid under each typical daily scenario: , in, This represents the annualized comprehensive equivalent emissions; The number of days in a year; Typical daytime scene The probability of; for The amount of electricity or equivalent power purchased by the distribution network from the upper-level power grid during a given time period; , , The pollutants are CO2 and NO. X And the emission coefficient of SO2 per unit of electricity.

[0052] Low-carbon accounting focuses on the operational phase and does not include emissions from the entire life cycle of equipment manufacturing, transportation, and decommissioning. Annualized comprehensive equivalent emissions are used as low-carbon targets at the planning level for optimization, while daily equivalent emissions are used as low-carbon evaluation indicators at the operational level. After fuzzy membership processing, these indicators are used in the comprehensive objective function at the operational level.

[0053] Step 140: Construct a two-layer optimization model for the coordinated planning of power distribution network sources, grids, loads and storage.

[0054] The two-layer optimization model includes a planning layer and an operation layer.

[0055] Specifically, the planning layer primarily focuses on the allocation schemes of three types of resources: power generation, grid, and storage. Power generation-side decision variables include the number and installed capacity of distributed wind and solar power at each candidate node; storage-side decision variables include the number, rated power, and rated capacity of energy storage systems at each candidate node; and grid-side decision variables include the selection of equipment type and capacity configuration for smart soft switches or ordinary interconnection switches at candidate interconnection locations. Demand-side response resources are not used as location and capacity determination variables in the planning layer; their access locations and available upper limits are pre-defined as scheduling boundaries in the operation layer.

[0056] The planning layer sets three types of objective functions: economy, flexibility, and low carbon emissions, expressed as follows: , and After unifying and normalizing the polarity of the three types of indicators mentioned above, the approximation degree of each candidate planning scheme to the positive and negative ideal solutions within the planning layer's objective space is calculated, yielding the multi-objective evaluation value of the planning layer. There is a certain degree of competition among the three types of objectives. Improving flexibility typically requires increasing the scale of energy storage or grid-side flexible equipment, thereby increasing investment costs; reducing equivalent emissions usually requires improving the local renewable energy consumption level, and may also place higher demands on energy storage capacity, demand response capability, and grid-side power flow regulation capability. Therefore, the planning layer adopts a multi-objective parallel optimization approach to generate a set of candidate planning schemes, and each candidate scheme is passed to the operation layer for verification during typical daily operations.

[0057] Based on the candidate schemes given in the planning layer, the operation layer performs time-series scheduling of various resources, including power sources, grids, loads, and storage, within a typical day. The decision variables of the operation layer of the two-level optimization model include at least the active power output of distributed wind power and distributed photovoltaic power, energy storage charging and discharging power, energy storage state of charge, active power transmission of smart soft switching, reduction ratio of loads that can be reduced, and transfer ratio of loads that can be shifted, for each time period.

[0058] The operational layer uses daily operating cost, short-term flexibility, and daily equivalent emissions as evaluation objectives. To unify operational indicators with different dimensions to the same evaluation scale, fuzzy membership functions for operating cost, short-term flexibility, and daily equivalent emissions are constructed respectively, forming the fuzzy comprehensive objective function for the operational layer: , in, For the fuzzy comprehensive target value, For the fuzzy membership degree of operating costs, Fuzzy membership degree for short-term flexibility The fuzzy membership degree of carbon emissions is represented by a value closer to 1, indicating higher satisfaction and a value closer to 0, indicating lower satisfaction. , , As weight, , and Greater than 0, . , and It is not limited to a fixed value and can be adjusted according to the investment constraints, low-carbon requirements and operational flexibility needs of the planning area.

[0059] The constraint set of the running layer of a two-level optimization model includes at least the following: (1) Power balance constraint, used to ensure that the power exchanged with the upper-level power grid, the output of distributed power sources, the charging and discharging power of energy storage, the active power transmission power of smart soft switching and the system load are balanced in each time period; (2) Network operation constraints, including node voltage upper and lower limits constraints, branch capacity constraints, and tie switch operation frequency constraints; (3) Distributed power output and curtailment constraints are used to limit the actual output of wind power and photovoltaic power to not exceed the maximum available output in the corresponding time period, and to set reasonable boundaries for the amount of wind and solar curtailment. (4) Energy storage operation constraints, including charging and discharging power constraints, upper and lower limits of state of charge constraints, charging and discharging mutual exclusion constraints, and state of charge constraints at the beginning and end of the day. (5) Intelligent soft switch operation constraints are used to constrain the active power transmission relationship at both ends of the intelligent soft switch and its rated capacity boundary. Only the active power transmission characteristics of the intelligent soft switch are considered, and reactive power adjustment variables are not introduced.

[0060] (6) Demand-side response constraints, including the upper limit of the proportion of load that can be reduced, the upper limit of the proportion of load that can be shifted, the time span constraint of load shifting, and the energy conservation constraint of load transfer.

[0061] In the two-layer optimization process, the planning layer generates candidate planning schemes and transmits the configuration results of distributed power sources, energy storage systems, and grid-side interconnection equipment to the operation layer. Under this configuration scheme, the operation layer performs typical daily time-series scheduling and outputs daily operating costs, overall operational flexibility, daily equivalent emissions, and operational constraint verification results. The operation layer's feedback results are used to evaluate the candidate planning schemes, ensuring that the judgment of the candidate planning schemes not only depends on static capacity configuration but also reflects the scheduling effect and constraint satisfaction during typical daily operation. Through this two-layer coupling structure, multi-year resource allocation decisions can be combined with typical daily operational adaptability verification, forming a source-grid-load-storage collaborative planning model that coordinates planning configuration and operational constraints.

[0062] Step 150: Through the planning layer of the two-layer optimization model, an improved harmony search algorithm is used to generate candidate planning schemes based on the input data.

[0063] Specifically, it includes: Based on the input data, an initial harmony memory is generated, where each harmony in the initial harmony memory corresponds to a candidate planning scheme. The candidate planning schemes include the configuration results of distributed wind power, distributed photovoltaic, energy storage systems, and grid-side interconnection equipment. Adaptive parameter scheduling strategy, harmony library structured reorganization strategy, and optimal solution approximation overbounded repair strategy are adopted to generate candidate planning schemes based on the initial harmony memory.

[0064] In some embodiments, the adaptive parameter scheduling strategy includes: adaptively updating the memory selection probability and the tone fine-tuning probability with the number of iterations, in order to balance early global search and later local development. , in, To select an upper limit for the probability of memory. The lower bound of the probability of selection for memory. This represents the upper limit of the probability of pitch fine-tuning. This is the lower bound of the probability of pitch fine-tuning. This represents the maximum number of iterations.

[0065] The aforementioned exponential scheduling is not an isolated parameter decay design, but a search rhythm control mechanism adapted to the two-layer nested solution structure. In the two-layer optimization, each time the planning layer generates a candidate planning scheme, it needs to call the runtime layer to complete a full temporal scheduling verification under all typical daily scenarios. The cost of a single evaluation is high, and the overhead of invalid iterations is significantly amplified by the nested structure. To address this, the memory selection probability remains high in the early stages of iteration, allowing new harmonies to preferentially inherit feasible configuration structures in the harmony memory that have already been verified by the runtime layer, thus increasing the proportion of effective evaluations in the early stages. As iteration progresses, the memory selection probability decays exponentially in a logarithmic proportion, while the pitch fine-tuning probability increases inversely, smoothly shifting the search focus from inheriting historical structures to fine-tuning neighborhoods. Simultaneously, the memory selection probability directly determines the trigger frequency of harmonyme library structured reorganization, and the later increase in the pitch fine-tuning probability increases the probability of out-of-bounds occurrence, thereby enhancing the effectiveness of the out-of-bounds repair strategy for approximating the optimal solution. Therefore, the improved strategy is uniformly driven by this scheduling mechanism and works in concert to form a holistic search control scheme for high-cost nested evaluation problems.

[0066] In other embodiments, the update basis for the memory selection probability and the tone fine-tuning probability is not limited to the number of iterations, but can also introduce the search state feedback quantity in the two-level iteration process as the scheduling independent variable.

[0067] For example, the percentage of times a new harmony successfully replaces the worst harmony within the most recent several generations is used as the memory bank update success rate. The memory bank update success rate is: , in, For the first The success rate of updating the memory bank in the generation; The statistical window length is the most recent iteration number selected. For the first The success indicator for each generation's update is set to 1 when the new harmony successfully replaces the worst harmony in the harmony memory, and 0 otherwise. This is satisfied in the early stages of the iteration. At that time, the statistics window is calculated based on the actual number of iterations generated. When the update success rate is lower than the set threshold, the probability of memory selection is lowered or the probability of tone fine-tuning is increased to enhance the ability to jump out of the search stagnation area.

[0068] Alternatively, the pass rate can be verified using the constraints of candidate planning schemes fed back from the runtime layer: , in, For the first The pass rate of constraint verification in the era; This is to count the total number of candidate planning schemes within the statistical window that have been verified by the operational layer constraints; For the first The constraint verification indicator for each candidate planning scheme is set to 1 if the candidate planning scheme satisfies the operational layer constraints in all typical daily scenarios, and 0 otherwise; the proportion of infeasible schemes is... When the proportion of infeasible solutions is too high, the probability of remembering the selection is increased to strengthen the inheritance of known feasible configuration structures.

[0069] The diversity measures, such as the dispersion of values ​​of variables in the harmony memory, can also be used in scheduling: , in, For the first A measure of the diversity of the harmonic memory bank; Let be the dimension of the decision vector for the planning layer; For the first The standard deviation of the dimensional components in the harmony memory bank; , The first The upper and lower bounds of the values ​​of the dimensional variables are defined; dimensions with values ​​in the range of zero are not included in the statistics. When the diversity measure is low, the probability of pitch fine-tuning is increased or the probability of memory selection is decreased to restore population diversity.

[0070] The three types of search status feedback quantities mentioned above can be used individually or in combination. When used in combination, they should be coordinated according to the principle of feasibility priority, that is, priority should be given to ensuring the output of feasible solutions, and then the exploration intensity of the search should be adjusted according to the update success rate and diversity measure.

[0071] In some embodiments, the harmonic library structured reorganization strategy includes: When a memory selection is triggered, harmony vectors are extracted from the harmony memory bank to form a harmony sub-bank, and new harmonies are formed using the diagonal elements of the harmony sub-bank.

[0072] Specifically, when triggering memory selection, instead of simply randomly extracting historical components independently for each dimension variable, a sub-library is constructed by extracting harmony vectors from the harmony memory bank, and new harmony vectors are formed using their diagonal elements. This approach enhances the ability to combine cross-dimensional information between different historical planning schemes and improves the structural diversity of new schemes.

[0073] The new harmony is: , in, The new harmony vectors generated after structured recombination are... Harmony sub-libraries extracted from harmony memory libraries This involves taking the diagonal elements of a matrix and forming a vector.

[0074] Unlike basic harmony search, which independently and randomly extracts historical components for each dimension of a variable during the memory selection phase, diagonal recombination ensures that each dimension of the new harmony is taken from distinct historical harmonies in the harmony sub-library. This mechanism eliminates the possibility of near-repeated solutions arising from multiple components originating from the same historical harmony, ensuring that each trigger of memory selection generates a new candidate solution with cross-scheme combination characteristics. For the source-grid-load-storage coordinated planning problem, the planning layer decision vector is structured in blocks according to distributed wind power, distributed photovoltaics, energy storage systems, and grid-side interconnection equipment. Diagonal recombination is equivalent to cross-assembling resource allocation structures among different historical planning schemes. The source-side configuration of the new scheme can inherit from a certain historical scheme, while the storage-side or grid-side configuration can inherit from other historical schemes, expanding the structural coverage of candidate planning schemes while maintaining the legality of each component value.

[0075] In other embodiments, the harmony sub-library can be reorganized into blocks diagonally according to the resource category of the decision variables. That is, the source-side variable block, storage-side variable block, and grid-side variable block are used as the smallest reorganization units. This allows variables that are coupled within the same resource category, such as the rated power and rated energy of the same energy storage node, to inherit from the same historical harmony as a whole, avoiding the mismatch of intra-block configuration that may be introduced by dimension-by-dimensional reorganization. The extraction of the harmony sub-library is not limited to equal probability random extraction. It can be extracted by applying an elite bias to the historical harmony according to fitness ranking, so that the diagonal reorganization can absorb more structural information of the excellent planning scheme.

[0076] In some embodiments, the out-of-bounds repair strategy for optimal solution approximation includes: When a new harmony has variables that go out of bounds, the out-of-bounds components of the new harmony are repaired using the current optimal harmony.

[0077] Specifically, when variables go out of bounds after pitch fine-tuning in a new harmony, instead of simply truncating or discarding them, the out-of-bounds components are repaired using the legal values ​​of the corresponding dimensions of the current optimal harmony, bringing the new solution back into the feasible region. , in, For New Harmony The Dimensional components; The current optimal solution The This strategy reduces the number of invalid solution evaluations and avoids excessive clustering of candidate planning schemes near the boundary, thereby improving the efficiency of feasible solution generation.

[0078] The targeted nature of this repair strategy stems from the mixed discrete characteristics of the planning-level decision variables and the cost of nested evaluation. On the one hand, planning-level variables include discrete variables such as the selection of access nodes and the number of device configurations. Conventional boundary truncation or random reset may generate illegal values ​​that do not belong to the allowed value set or lose search direction information. However, the corresponding dimension value of the current optimal harmony must belong to the allowed value set of that dimension variable. The repair result can directly enter the evaluation without additional projection processing. On the other hand, each candidate scheme evaluation requires a complete execution of a typical daily scheduling solution at the runtime layer. If out-of-bounds solutions are discarded and regenerated, the nested computational overhead will increase exponentially. On-site repair transforms the originally ineffective iteration into a directional search carrying the structural information of the optimal solution. This is equivalent to naturally introducing a guidance mechanism towards the current excellent region in the later stages of the iteration when the intensity of tone fine-tuning increases, which complements the stage division of adaptive parameter scheduling.

[0079] In other embodiments, to avoid the loss of diversity due to the long-term convergence of out-of-bounds components towards a single optimal solution, the repair source can be expanded from the current optimal harmony to a set of elite harmonies ranked high in fitness. Each time a repair is performed, a harmony is randomly selected from this set to provide a valid value for the corresponding dimension. Repair probabilities can also be introduced to balance the directionality and diversity of the repair. When a new harmonic component goes out of bounds, a... The random numbers are uniformly distributed, and the probability of using the current optimal solution component for targeted repair is used; otherwise, the out-of-bounds component is returned to the nearest legal boundary value, and the repair probability is adaptively increased with the iteration process. , in, For the first The probability of repairing the generation, , For its upper and lower limits and From the above formula, the early stage of iteration Smaller populations that have crossed the boundary are more likely to be brought back to the boundary to maintain population diversity; later... The enhancements include targeted repair and improved search convergence towards optimal regions, complementing the later-stage characteristic of increased probability for pitch fine-tuning. These two improvements can be used individually or in combination.

[0080] After generating new harmonics and processing their feasibility, if the new scheme is better than the worse schemes in the harmony memory, the harmony memory is updated; otherwise, the original scheme is retained. This process is repeated until the maximum number of iterations is reached or the convergence condition is met.

[0081] Upon reaching the termination condition, the improved harmony search algorithm does not simply output a single optimal harmony, but rather outputs the entire updated harmony memory as a set of candidate planning schemes. Since the memory update employs a mechanism of replacing the worst harmony with a new, superior one, the memory retains a set of high-performance and structurally diverse schemes accumulated throughout the search process at termination. The size of the harmony memory determines the upper limit of the number of candidate planning schemes. Harmonies that fail to pass all typical daily operational constraints at the operational layer are not included in the candidate planning scheme set. The annualized comprehensive cost, annualized comprehensive equivalent emissions, and operational flexibility of each candidate planning scheme are calculated by the operational layer during its fitness evaluation phase and saved with the scheme. These values ​​can be directly used for subsequent three-dimensional comprehensive decision-making without requiring repeated execution of the operational layer solution. In other embodiments, the top few harmonies can be extracted based on fitness sorting from the harmony memory, or harmony schemes with highly similar configuration results can be deduplicated before forming the candidate planning scheme set.

[0082] Step 160: Through the running layer of the two-layer optimization model, a genetic algorithm based on Logistic chaotic mapping is used to generate the running evaluation results of the candidate planning scheme.

[0083] In some embodiments, the genetic algorithm based on Logistic chaotic mapping introduces chaotic mutation and jump search mechanisms during the genetic evolution process to enhance the local search of individuals.

[0084] Specifically, for each candidate planning scheme generated by the planning layer, its equipment configuration results are input into the operation layer. The operation layer uses the fuzzy comprehensive objective function as the fitness evaluation basis and solves the typical daily source, grid, load and storage time sequence scheduling scheme under the constraints of power balance, node voltage, branch capacity, energy storage state of charge, SOP active power transmission and demand-side response.

[0085] The runtime solution employs a Genetic Algorithm (GA) based on the Logistic chaotic mapping. First, a chaotic sequence is generated using the Logistic mapping, with the iterative form as follows: , in, For chaotic variables, These are control parameters. To ensure the sequence exhibits strong chaotic characteristics, a value is typically set to... and set The non-special initial value. Since the chaotic variable obtained from the above equation is located at... The range of values ​​for chaotic variables is defined by the interval, while the decision variables at the operational level generally have physical upper and lower bounds. Therefore, it is necessary to linearly map the chaotic variables to the range of values ​​for each variable. The linear mapping form is as follows: , in, Indicates the first The mapping results of each gene , This is a transformation constant used to ensure... Transform within a specified range. For given upper and lower bounds. It is acceptable , This ensures that the mapped variables fall within the allowed range. If the variables are discrete or integer, rounding or piecewise mapping can be added after the mapping to satisfy the value set constraints.

[0086] Introducing a chaotic mechanism within the operational layer problem structure is specifically targeted. The operational layer decision variables cover distributed power output, energy storage charging and discharging power and state of charge, active power transmission of smart soft switches, and demand-side response ratios for typical intraday periods. These variables are high-dimensional and strongly coupled by constraints such as power balance, node voltage, branch capacity, and energy storage state of charge, resulting in a narrow and irregular feasible region. Individuals generated through completely random initialization are prone to falling into infeasible regions, leading to computational waste in punitive evaluations. The chaotic sequence generated by the Logistic mapping possesses ergodicity and determinism. After independent linear mapping of each dimension, it can expand within a variable space with different dimensions without clustering locally, providing a more representative initial population for searching feasible solutions under strong constraints. The same chaotic mechanism can be reused in the mutation phase, allowing the initialization and evolution processes to share a unified ergodic search characteristic. The chaotic sequence is completely reproducible under given initial values, facilitating engineering review of the planning scheme's operational verification results.

[0087] In the solution process, the algorithm parameters are first set, including the population size. With the maximum number of generations Subsequently, a chaotic sequence was generated using Logistic mapping, and an initial population was formed by performing interval mapping on each dimension of the variables. Compared to completely random initialization, chaotic initialization can provide more uniform search coverage with the same sample size, providing more representative initial candidate solutions for subsequent genetic evolution.

[0088] After initialization, the fitness of individuals in the population is evaluated. The fitness function is determined by the objective function of the runtime layer, and feasibility judgment and penalty processing are performed in conjunction with runtime constraints. Then, the genetic evolution cycle begins, and at the [missing information]th [missing information]th [missing information]... Generation, to the parent population Perform selection, crossover, and mutation operators to obtain the offspring population. In the genetic processing, chaotic sequences are introduced to participate in mutation operations. When an individual mutates, in addition to traditional random perturbation, new chaotic variables can be generated and mapped to the corresponding gene range to replace or perturb the original gene, thereby enhancing the ergodicity and leap of mutation and improving the ability to escape local optima.

[0089] In other embodiments, the location of chaotic mutation can be guided by the runtime constraint verification results, so that the mutation operator focuses on the key variables that constrain the feasibility of the scheme. Specifically, runtime constraint verification is performed on the current individual, and the occurrence of node voltage exceeding limits, branch capacity exceeding limits, and energy storage state of charge exceeding limits in each time period is statistically analyzed, and the results are then used as the first... The sum of the normalized exceedances of the three types of constraints within a time period is taken as the exceedance level for that time period. ; where, for a given constraint, the excess value is the portion of the actual value exceeding the upper allowable bound or falling below the lower allowable bound; if the constraint is not exceeded, it is taken as 0, and normalized according to its allowable range. , This indicates that none of the constraints exceeded the limits during that period.

[0090] During the mutation phase, the mutation probability of the associated gene loci at each time point is adaptively amplified according to their degree of exceeding the limit: , in, For the first The probability of mutation of the gene loci associated with the time period; The baseline probability of variation; This represents the upper limit of the mutation probability. ; This represents the maximum value of the degree of exceeding the limit in each time period. Therefore, the mutation operator focuses on key gene loci associated with the time periods exceeding the limit and preferentially applies chaotic perturbations to them, thereby reducing the number of ineffective evolutionary generations and improving the efficiency of producing feasible solutions under strong constraints. Furthermore, the generation of chaotic sequences is not limited to the Logistic map; other chaotic maps with ergodic properties, such as the Tent map, can also be used as alternatives.

[0091] To improve evolutionary stability and the retention of superior individuals, a parent-offspring merging and re-evaluation mechanism is adopted. The parent generation... with offspring Merging to form an intermediate population The average fitness of the population is calculated as shown in the following formula.

[0092] , Based on the merged evaluation, individuals can be updated or screened according to their fitness ratio. The corresponding update relationship can be expressed as shown in the following formula.

[0093] , in Represents an individual. Indicates the individual in the first... The corresponding allocation decision, Its fitness value, The above relationship reflects the evolutionary direction of strengthening the superior and suppressing the inferior, which helps to allocate more search resources to individuals with better fitness.

[0094] Building upon this, an elite preservation and local search reinforcement mechanism is further introduced. Specifically, in each generation, a subset of individuals with high fitness are selected for local search reinforcement. Let the fitness of these top-performing individuals be higher than the population average, and let the excess fitness be denoted as... It can be written as shown in the following formula.

[0095] , Furthermore, a new allocation decision form can be obtained by combining chaotic variables with individuals whose fitness is better than the average, as shown in the following formula.

[0096] , By strengthening and locally searching a small number of elite individuals, the ability to develop superior areas can be improved while maintaining population diversity, thereby improving later-stage convergence performance. If the maximum number of generations has not been reached, the next generation of evolution continues. Under the action of the mutation operator, the survival probability of an individual can be expressed as follows.

[0097] , in, The mutation probability, This represents the retention terms relevant to the individual. This expression describes the impact of variation on individual retention and works in conjunction with the aforementioned elite retention mechanisms to maintain the stability of the evolutionary process.

[0098] After the operation layer solution is completed, the typical daily operation results corresponding to the candidate planning scheme are output, including distributed power output, energy storage charging and discharging power, energy storage state of charge, SOP active transmission power, demand-side response, node voltage, branch power, daily operating cost, overall operational flexibility and daily equivalent emissions.

[0099] In some embodiments, to reduce the computational overhead of nested solutions, a mapping cache between candidate planning schemes and operational evaluation results can be established. Each candidate planning scheme generated during the planning layer iteration is compared with historical schemes that have already been verified in the cache before being sent to the operational layer. The configuration difference is measured by summing the absolute values ​​of the differences between the new candidate scheme and the historical scheme in terms of the configuration quantities of distributed wind power, distributed photovoltaic, energy storage systems, and grid-side interconnection equipment, normalized according to the allowable range of each configuration quantity. The smaller the value, the closer the equipment configurations of the two schemes are; zero indicates that the equipment configurations of the two schemes are completely identical. When the difference between the new candidate scheme and a certain historical scheme in the configuration of distributed wind power, distributed photovoltaic, energy storage systems, and grid-side interconnection equipment is less than a set threshold, the typical daily scheduling solution of that historical scheme is used as a seed individual in the initial population of the operational layer's genetic algorithm, achieving a hot start for the operational layer solution, allowing the operational layer to continue optimizing within the neighborhood of known high-quality scheduling solutions; when the configurations are completely identical, the cached operational evaluation results are directly reused. This mechanism takes advantage of the gradual convergence of candidate solutions and the recurring occurrence of solutions with similar configurations in the later stages of the planning layer iteration. It can reduce the overhead of repeated nested scheduling and solving without changing the basic process of the two-layer optimization model and each algorithm.

[0100] In other embodiments, an early termination mechanism for infeasible solutions can be set during the operational layer solution process. The degree of constraint violation is measured by the normalized sum of the excesses of an individual across all time periods and all operational constraints: for each constraint, the portion of its actual value exceeding the upper allowable bound or falling below the lower allowable bound is taken as the excess; if it does not exceed the bound, it is taken as zero. After normalization according to the allowable range of the constraint, the sum is calculated over all time periods and all constraints. The larger the value, the more the individual deviates from the feasible region; zero indicates that the individual meets all operational constraints. The operational constraints include power balance, node voltage, branch capacity, energy storage state of charge, SOP active power transmission, and demand-side response, etc. For a given candidate planning scheme, if no feasible individual that meets all operational constraints exists in the population for several consecutive generations during the genetic evolution process, or if the minimum value of the constraint violation degree is continuously higher than the set threshold and shows no decreasing trend, then the candidate planning scheme is determined to be infeasible under the current typical daily scenario, the current operational layer solution is terminated early, and an infeasibility marker and the main types of constraint violations are fed back to the planning layer. This mechanism avoids exhausting the maximum number of generations on candidate solutions that are not feasible, concentrates nested computing resources on candidate solutions that have the potential for improvement, and works in conjunction with the aforementioned implementation method that uses the constraint verification pass rate to participate in parameter scheduling.

[0101] The planning and execution layers can also be implemented using other heuristic algorithms, intelligent optimization algorithms, or mathematical programming solvers.

[0102] Step 170: Generate the optimal planning scheme based on the economic evaluation indicators, flexibility evaluation indicators, low-carbon evaluation indicators, and the operational evaluation results of candidate planning schemes.

[0103] Specifically, for each candidate planning scheme, the annualized comprehensive cost of the candidate planning scheme is calculated based on the operational evaluation results. Annualized carbon emissions and overall operational flexibility Among them, annualized comprehensive cost and annualized comprehensive equivalent emissions are inverse indicators, while comprehensive operational flexibility is a positive indicator.

[0104] Let the number of candidate planning schemes be . , No. The first scheme is in the The original evaluation value for each indicator , These correspond to annualized comprehensive cost, annualized carbon emissions, and overall operational flexibility, respectively. A reciprocal transformation is used to unify the polarity of the inverse indicators. Considering that the three indicators may have different dimensions and orders of magnitude, vector normalization is used for dimensionless processing, as shown in the following equation.

[0105] , Weight Determined by AHP. AHP treats economy, low carbon emissions, and flexibility as indicators at the same level of criteria. In the weighted normalization matrix, the maximum and minimum values ​​of each indicator are taken to form the positive ideal solution. and negative ideal solution The distances are calculated and evaluated. The three-dimensional Euclidean distances from each solution to the positive and negative ideal solutions are shown in the following formula.

[0106] , To further define relative proximity, see the following formula.

[0107] , Among them, proximity The larger the value, the closer the candidate programming solution is to the positive ideal solution and the farther it is from the negative ideal solution, resulting in better overall performance. Therefore, candidate programming solutions are categorized as follows: The candidate planning schemes are sorted in descending order, and the one with the highest similarity is selected as the optimal planning scheme. The output of the recommended planning scheme includes: the access nodes and configuration capacity of distributed wind power and distributed photovoltaic, the access nodes and configuration capacity of energy storage systems, the configuration location and capacity of smart soft switches or traditional tie switches, as well as the corresponding annualized comprehensive cost, annualized comprehensive equivalent emissions, and comprehensive operational flexibility indicators.

[0108] The optimal planning scheme is input into the operation layer for time-series scheduling and operational verification under typical daily scenarios. Verification includes power balance, node voltage, branch capacity, energy storage state of charge, demand-side response boundaries, and active power transmission constraints of smart soft switching. After operational verification, the output includes distributed generation output, energy storage charging and discharging power, energy storage state of charge, SOP active power transmission, demand-side response, node voltage, branch power, daily operating cost, daily equivalent emissions, power flexibility indicators, and grid flexibility indicators.

[0109] The optimal planning scheme can be output in the form of tables, node configuration lists, equipment capacity lists, or graphical interfaces.

[0110] The following section, based on experiments, explains the beneficial effects of the multi-objective distribution network source-grid-load-storage coordinated planning method.

[0111] The IEEE 33-bus distribution system was selected as the test object. The system has a typical radial distribution network structure, and the reference voltage is 10 kV. Figure 5 As shown in the figure. The wind, solar and load data are obtained by using hourly wind speed, light intensity and load data of a certain region throughout the year. After scene segmentation and clustering, time series curves for four typical daily 24-hour periods in spring, summer, autumn and winter are obtained. In some embodiments, the key equipment parameter settings are shown in Table 1.

[0112]

[0113] The source-side configuration targets include distributed wind power and distributed photovoltaic power; the storage-side configuration targets are energy storage systems; the grid-side configuration targets include smart soft switches and traditional tie switches; and the load-side considers two types of demand-side response resources: loads that can be reduced and loads that can be shifted. Based on the above system and input data, resource modeling, index construction, two-level optimization solution, three-dimensional comprehensive decision-making, and recommended scheme output are completed sequentially. To reflect the impact of distributed power output fluctuations and load time-series changes on planning configuration and operation scheduling, the per-unit curves of photovoltaic and wind power output and the per-unit curves of load on typical days in spring, summer, autumn, and winter are selected as input data, such as... Figure 6 and Figure 7 As shown.

[0114] The optimal planning scheme obtained after three-dimensional multi-objective integrated decision-making is shown in Table 2.

[0115]

[0116] The optimal planning scheme was input into the operation layer model, and time-series scheduling was performed under typical day scenarios in spring, summer, autumn and winter, respectively. The operation verification results for typical days in the four seasons are shown in Table 3.

[0117]

[0118] To further verify the comprehensive effect of the optimal planning scheme, four comparison scenarios were set up: Scenario 1 is a single-objective optimization scheme for economy; Scenario 2 is a dual-objective optimization scheme for economy and low carbon emissions; Scenario 3 is a dual-objective optimization scheme for economy and flexibility; and Scenario 4 is a three-objective synergistic optimization scheme for economy, low carbon emissions, and flexibility using this technical solution. Key results are shown in Table 4.

[0119]

[0120] As shown in Table 4, Scenario 1 has the lowest total cost but the highest comprehensive equivalent emissions and the lowest overall operational flexibility; Scenario 2 has the lowest comprehensive equivalent emissions but a relatively high total cost and limited improvement in overall operational flexibility; Scenario 3 has the highest overall operational flexibility but still has a relatively high comprehensive equivalent emissions. In contrast, Scenario 4 does not pursue an extreme value for a single indicator but achieves a more balanced result among total cost, comprehensive equivalent emissions, overall operational flexibility, and the level of renewable energy integration.

[0121] After calculating the distance between the positive and negative ideal solutions in the AHP-TOPSIS integrated decision-making, the TOPSIS proximity of scenario 4 is approximately 0.60, ranking the highest among the four scenarios. This indicates that the optimal planning scheme has the best comprehensive performance under the three-dimensional comprehensive evaluation.

[0122] The above experiments demonstrate that this method can achieve coordinated planning and operational verification of distributed renewable energy, energy storage systems, smart soft switches, and demand-side response resources in distribution networks. The recommended planning scheme not only provides clear equipment location and capacity determination results but also maintains good operational adaptability under typical daily operating conditions throughout the four seasons. Multi-scenario comparisons further show that single-objective or dual-objective optimization can easily lead to a planning scheme biased towards a particular performance dimension. In contrast, this method, through the coordinated evaluation of three-dimensional indicators—economic efficiency, low carbon footprint, and overall operational flexibility—can obtain a more balanced optimal planning scheme in a comprehensive sense.

[0123] Figure 9 This is a schematic diagram of a multi-objective distribution network source-grid-load-storage coordinated planning system according to some embodiments of this specification, such as... Figure 9 As shown, a multi-objective distribution network source-grid-load-storage coordinated planning system may include the following modules: The data acquisition module is used to acquire input data, which includes at least configuration data, source-load time-series data, distribution network operation parameters, and economic and low-carbon parameters. The unified modeling module is used to model distributed wind power, distributed photovoltaics, energy storage systems, smart soft switches, and demand-side response resources. The indicator construction module is used to construct economic evaluation indicators, flexibility evaluation indicators, and low-carbon evaluation indicators. The model building module is used to build a two-layer optimization model for the coordinated planning of power distribution network sources, grids, loads and storage. The two-layer optimization model includes a planning layer and an operation layer. The two-layer optimization module is used to generate candidate planning schemes based on input data through the planning layer of the two-layer optimization model and an improved harmony search algorithm. The two-level optimization module is also used to generate evaluation results of candidate planning schemes by using a genetic algorithm based on Logistic chaotic mapping through the running layer of the two-level optimization model. The planning generation module is used to generate the optimal planning scheme based on economic evaluation indicators, flexibility evaluation indicators, low-carbon evaluation indicators, and the operational evaluation results of candidate planning schemes.

[0124] The multi-objective distribution network source-grid-load-storage collaborative planning system can be used to execute the above-mentioned multi-objective distribution network source-grid-load-storage collaborative planning method, which will not be elaborated here.

[0125] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A multi-objective distribution network source-grid-load-storage coordinated planning method, characterized in that, include: Acquire input data, wherein the input data includes at least configuration data, source-load time-series data, distribution network operation parameters, and economic and low-carbon parameters; Modeling is performed on distributed wind power, distributed photovoltaics, energy storage systems, smart soft switching, and demand-side response resources; Construct economic evaluation indicators, flexibility evaluation indicators, and low-carbon evaluation indicators; A two-layer optimization model for the coordinated planning of power distribution network sources, grids, loads and storage is constructed, wherein the two-layer optimization model includes a planning layer and an operation layer; By using the planning layer of the two-layer optimization model and employing an improved harmony search algorithm, candidate planning schemes are generated based on the input data. By using the running layer of the two-layer optimization model, a genetic algorithm based on Logistic chaotic mapping is adopted to generate the running evaluation results of candidate planning schemes. The optimal planning scheme is generated based on the economic evaluation indicators, flexibility evaluation indicators, low-carbon evaluation indicators, and the operational evaluation results of candidate planning schemes.

2. The multi-objective distribution network source-grid-load-storage coordinated planning method according to claim 1, characterized in that, Through the planning layer of the two-layer optimization model, an improved harmony search algorithm is used to generate candidate planning schemes based on the input data, including: Based on the input data, an initial harmony memory library is generated, wherein each harmony in the initial harmony memory library corresponds to a candidate planning scheme, and the candidate planning scheme includes the configuration results of distributed wind power, distributed photovoltaic, energy storage system and grid-side interconnection equipment; Adaptive parameter scheduling strategy, harmony library structured reorganization strategy, and out-of-bounds repair strategy for optimal solution approximation are adopted to generate candidate planning schemes based on the initial harmony memory library.

3. The multi-objective distribution network source-grid-load-storage coordinated planning method according to claim 2, characterized in that, The adaptive parameter scheduling strategy includes: The memory selection probability and pitch fine-tuning probability are adaptively updated with the number of iterations.

4. The multi-objective distribution network source-grid-load-storage coordinated planning method according to claim 3, characterized in that, The harmonic library structured reorganization strategy includes: When memory selection is triggered, harmony vectors are extracted from the harmony memory bank to form a harmony sub-bank, and new harmony is formed using the diagonal elements of the harmony sub-bank. The out-of-bounds repair strategy for approximating the optimal solution includes: When a new harmony has variables that go out of bounds, the out-of-bounds components of the new harmony are repaired using the current optimal harmony.

5. The multi-objective distribution network source-grid-load-storage coordinated planning method according to claim 1, characterized in that, The decision variables of the operation layer of the two-layer optimization model include at least the active power output of distributed wind power and distributed photovoltaic power, energy storage charging and discharging power, energy storage state of charge, active power transmission of smart soft switching, load reduction ratio that can be reduced, and load transfer ratio that can be shifted in each time period. The constraint set of the operation layer of the two-layer optimization model includes at least power balance constraints, network operation constraints, distributed power output and curtailment constraints, energy storage operation constraints, smart soft switching operation constraints, and demand-side response constraints. The objective function of the running layer of the two-layer optimization model is: ， in, For the fuzzy comprehensive target value, For the fuzzy membership degree of operating costs, Fuzzy membership degree for short-term flexibility For fuzzy membership of carbon emissions, , and, As weight.

6. The multi-objective distribution network source-grid-load-storage coordinated planning method according to claim 5, characterized in that, The genetic algorithm based on Logistic chaotic mapping introduces chaotic mutation and jump search mechanisms during the genetic evolution process to enhance the local search of individuals.

7. The multi-objective distribution network source-grid-load-storage coordinated planning method according to any one of claims 1-6, characterized in that, The economic evaluation indicators include annualized comprehensive cost and typical daily operating cost; The annualized comprehensive cost consists of annual investment cost, annual operation and maintenance cost, electricity purchase and sale cost, demand-side management cost, and cost of penalties for wind and solar curtailment. The typical daily operating cost consists of daily electricity purchase and sales costs, daily demand-side management costs, and daily wind and solar curtailment penalty costs.

8. The multi-objective distribution network source-grid-load-storage coordinated planning method according to any one of claims 1-6, characterized in that, The flexibility evaluation indicators include medium- and long-term configuration flexibility evaluation indicators and comprehensive operational flexibility evaluation indicators; The evaluation index for medium- and long-term configuration flexibility consists of medium- and long-term power regulation flexibility and medium- and long-term grid structure regulation flexibility. The comprehensive operational flexibility evaluation index consists of daily power adjustment flexibility and daily grid structure adjustment flexibility.

9. The multi-objective distribution network source-grid-load-storage coordinated planning method according to any one of claims 1-6, characterized in that, The low-carbon performance evaluation indicators include annualized comprehensive equivalent emissions and daily equivalent emissions.

10. A multi-objective distribution network source-grid-load-storage coordinated planning system, characterized in that, The multi-objective distribution network source-grid-load-storage coordinated planning method according to any one of claims 1-9 includes: The data acquisition module is used to acquire input data, wherein the input data includes at least configuration data, source-load time-series data, distribution network operation parameters, and economic and low-carbon parameters; The unified modeling module is used to model distributed wind power, distributed photovoltaics, energy storage systems, smart soft switches, and demand-side response resources. The indicator construction module is used to construct economic evaluation indicators, flexibility evaluation indicators, and low-carbon evaluation indicators. The model building module is used to build a two-layer optimization model for the coordinated planning of power distribution network sources, grids, loads and storage, wherein the two-layer optimization model includes a planning layer and an operation layer; The two-layer optimization module is used to generate candidate planning schemes based on input data through the planning layer of the two-layer optimization model and an improved harmony search algorithm. The two-layer optimization module is also used to generate the evaluation results of candidate planning schemes by using a genetic algorithm based on Logistic chaotic mapping through the running layer of the two-layer optimization model. The planning generation module is used to generate the optimal planning scheme based on economic evaluation indicators, flexibility evaluation indicators, low-carbon evaluation indicators, and the operational evaluation results of candidate planning schemes.

Images