A method for collaborative optimization and configuration of power supply, load and storage in a mountainous power distribution network

Abstract

This invention relates to the field of distribution network optimization, and particularly to a method for coordinated optimization of power generation, load, and energy storage in mountainous distribution networks. The method includes: S1, constructing a basic dataset and a set of typical operating scenarios for coordinated optimization of power generation, load, and energy storage in mountainous distribution networks; S2, using the basic dataset and the set of typical operating scenarios constructed in S1, constructing a joint operation model of power generation, load, and energy storage in mountainous distribution networks to describe the power balance relationship between photovoltaic output, energy storage charging and discharging, grid power purchase and sales, and load response; S3, based on the joint operation model in S2 and the set of typical operating scenarios in S1, constructing a two-layer multi-objective optimization model for coordinated optimization of power generation, load, and energy storage; and S4, solving the two-layer multi-objective optimization model constructed in S3 to obtain the coordinated optimization configuration scheme. This method can achieve coordinated optimization of power generation, load, and energy storage in mountainous distribution networks, fully considering the topographical features of mountainous areas, the uncertainty of power generation and load, multi-objective coordination, and temporal coupling relationships, while also taking into account technical feasibility and economic rationality.

Description

Technical Field

[0001] This invention relates to the field of distribution network optimization, and in particular to a method for coordinated optimization of power source, load and storage configuration in mountainous distribution networks. Background Technology

[0002] As the global energy structure accelerates its transformation towards low-carbon and clean energy, the penetration rate of distributed photovoltaic, decentralized wind power, small-scale hydropower, and energy storage systems in power distribution networks continues to increase. For mountainous power distribution networks, factors such as large terrain undulations, long power supply radius, low load density, numerous line branches, weak end-point voltage support capacity, and limited transportation and construction conditions mean that traditional methods relying on unidirectional power supply from the upper-level grid and centralized capacity expansion often suffer from high construction costs, insufficient power supply reliability, and low equipment utilization. Therefore, developing coordinated planning of distributed new energy and energy storage systems, tailored to local renewable energy resource conditions in mountainous areas, has become an important technical approach to improving the power supply capacity, absorption capacity, and operational economy of mountainous power distribution networks.

[0003] However, the integration of distributed renewable energy in mountainous areas presents significant uncertainties and spatial variability. On the one hand, factors such as slope, aspect, altitude, shading, canyon wind tunnel effects, cloudy and foggy weather, and diurnal temperature variations lead to strong randomness, volatility, and spatiotemporal heterogeneity in solar irradiance, wind speed, and load demand. On the other hand, power distribution lines in mountainous areas are generally long and the grid structure is relatively weak. When a high proportion of distributed photovoltaic power is integrated, it can easily lead to problems such as reverse power flow, node voltage exceeding limits, line overload, increased curtailment, and increased grid losses. If renewable energy and energy storage capacity are configured based solely on a single typical day or deterministic forecast results, it is difficult to accurately reflect the actual operating conditions under various weather conditions, loads, and operating modes. This can result in risks such as capacity configuration deviations, insufficient economic efficiency, or failure to meet operational constraints in the planning results.

[0004] Energy storage systems can play a crucial role in mountainous power distribution networks by mitigating fluctuations in renewable energy output, increasing local photovoltaic (PV) absorption rates, reducing peak-to-valley differences, alleviating voltage exceedances, and enhancing the power supply guarantee for critical loads. However, energy storage systems are characterized by high investment costs, lifespans affected by charge / discharge depth, and dynamic changes in state of charge based on operational strategies. If the energy storage capacity is too small, it will be difficult to effectively absorb surplus PV power and support peak loads; if the capacity is too large, it will lead to investment redundancy and reduced utilization. Therefore, energy storage planning cannot be conducted in isolation but should be coordinated and optimized with the characteristics of distributed renewable energy output, load response capabilities, grid electricity purchase and sale prices, line voltage constraints, and construction conditions in mountainous areas.

[0005] Existing photovoltaic-storage planning methods for distribution networks typically focus on optimizing distributed photovoltaic capacity or energy storage capacity. Some methods consider time-of-use pricing, demand response, or node voltage constraints, but they still have shortcomings in mountainous application scenarios: First, the joint modeling of mountainous areas' illumination, load, and terrain conditions is insufficient, failing to effectively reflect the impact of factors such as slope, shading, road accessibility, and construction costs on the constructability of photovoltaic and energy storage candidate nodes. Second, the handling of source-load uncertainties often relies on a few typical scenarios or deterministic predictions, lacking the preservation of the temporal correlation between photovoltaic output and load demand, leading to deviations between the scenario set and the actual operating state. Third, there is a lack of unified coordination among source-side renewable energy configuration, load-side interruptible load response, energy storage charging and discharging operation, and grid-side voltage regulation, making it difficult to simultaneously consider multiple objectives such as local photovoltaic absorption rate, investment returns, life-cycle costs, node voltage deviation, and power supply reliability. Fourth, existing optimization models often ignore the impact of long feeders, weak grid structures, and construction penalty costs in mountainous areas on the planning scheme, resulting in insufficient feasibility of the planning results when implemented in engineering projects.

[0006] Furthermore, in mountainous distribution networks with a high proportion of distributed photovoltaic (PV) grid integration, there is a clear dynamic coupling relationship between power sources, loads, and energy storage. PV output fluctuates randomly due to sunlight and shading, load demand varies periodically due to factors such as residential production and daily life, seasonal temperature differences, and tourism and agriculture, and the state of charge of energy storage evolves continuously due to the charging and discharging strategies of different time periods. If these temporal coupling relationships are not incorporated into a unified model during the planning stage, and configuration is based solely on static capacity constraints or single-period power balance, problems may arise such as energy storage state of charge exceeding limits during continuous operation, insufficient protection of critical loads, or continuous voltage exceedances at nodes, even if the economic efficiency is optimal in typical scenarios.

[0007] The aforementioned problems mainly stem from the fact that the coordinated optimization of power generation, load, and energy storage in mountainous distribution networks involves complex systems engineering with multiple dimensions, objectives, and constraints. First, the complex and varied terrain of mountainous areas necessitates the development of refined modeling methods that accurately reflect spatial characteristics such as slope, shading, and accessibility, placing extremely high demands on data acquisition and model construction. Second, the uncertainty of power generation and load exhibits strong spatiotemporal correlation and nonlinear characteristics, requiring the construction of typical scenario sets that maintain temporal correlation, which involves large-scale data processing and probabilistic statistical modeling techniques. Third, multi-objective optimization requires seeking the optimal balance among multiple mutually constraining objectives such as local photovoltaic grid integration rate, life-cycle cost, investment return, voltage deviation, and curtailment, necessitating the design of efficient multi-objective solution algorithms. Finally, modeling the temporal coupling relationship between power generation, load, and energy storage requires considering the continuous evolution of the energy storage state of charge, the dynamic characteristics of load response, and the time-varying characteristics of network constraints, posing significant challenges to the complexity and solution difficulty of the optimization model.

[0008] Therefore, how to achieve coordinated and optimized allocation of power sources, loads, and storage in mountainous power distribution networks, fully considering the characteristics of mountainous terrain, the uncertainty of power sources and loads, multi-objective coordination, and temporal coupling relationships, while taking into account technical feasibility and economic rationality, has become an urgent problem to be solved. Summary of the Invention

[0009] To address the aforementioned shortcomings of existing technologies, the present invention aims to provide a method for the coordinated optimization of source, load, and storage configuration in mountainous distribution networks. This method can achieve coordinated optimization of source, load, and storage configuration in mountainous distribution networks, fully considering the characteristics of mountainous terrain, source and load uncertainties, multi-objective coordination, and temporal coupling relationships, while also taking into account technical feasibility and economic rationality.

[0010] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0011] A method for coordinated optimization of power generation, load, and energy storage in mountainous distribution networks includes the following steps:

[0012] S1. Construct a basic dataset and a set of typical operating scenarios for the coordinated optimization of power source, load and storage configuration in mountainous power distribution networks;

[0013] S2. Using the basic dataset and typical operating scenario set built in S1, construct a joint operation model of power generation, load and storage in mountainous distribution networks to describe the power balance relationship between photovoltaic output, energy storage charging and discharging, main grid power purchase and sales and load response;

[0014] S3, based on the joint operation model of S2 and the typical operation scenario set of S1, construct a two-layer multi-objective optimization model for source-load-storage collaborative optimization configuration;

[0015] The outer planning objective of the two-layer multi-objective optimization model includes at least two of the following: maximizing the local photovoltaic consumption rate, minimizing the net present value cost of the system's entire life cycle, and maximizing the investment return of the photovoltaic-storage combined system; the inner operation objective includes at least two of the following: minimizing the average node voltage deviation, minimizing curtailment of solar power, minimizing electricity purchase cost, minimizing interruptible load compensation cost, and minimizing grid loss.

[0016] The typical operating scenario set of S1 is configured as the basis for the two-layer multi-objective optimization model to consider the uncertainty of source and load, and the joint operating model of S2 is configured to define the inherent rules of source, load, and storage operation interaction and system reliability in the two-layer multi-objective optimization model.

[0017] S4. Solve the two-layer multi-objective optimization model constructed in S3 to obtain the collaborative optimization configuration scheme.

[0018] Compared with the prior art, the present invention has the following advantages:

[0019] 1. Enhance the ability to handle source-load uncertainties in mountainous power distribution networks. This method constructs a set of typical operating scenarios, fully considering the impact of mountainous terrain, meteorological conditions, and load characteristics on photovoltaic output and load demand, effectively capturing the spatiotemporal correlation and random fluctuation characteristics of source and load. Compared to existing technologies that only use a single typical day or deterministic prediction methods, this scheme can more realistically reflect the actual operating conditions of mountainous power distribution networks under multiple weather conditions, multiple loads, and multiple operating modes, significantly reducing the risk of deviation in planning results.

[0020] 2. Achieving coordinated and optimized configuration of multiple elements including power generation, load, and storage. By constructing a joint operation model of power generation, load, and storage, this method incorporates elements such as photovoltaic power output, energy storage charging and discharging, grid power purchase and sales, and load response into a unified framework, establishing power balance relationships and dynamic coupling mechanisms among these elements. Compared with existing technologies that optimize photovoltaic or energy storage capacity in isolation, this scheme can comprehensively consider the mutual influences between renewable energy configuration on the source side, interruptible load response on the load side, energy storage charging and discharging operation, and grid voltage regulation, achieving overall optimal configuration.

[0021] 3. Balancing Multi-Dimensional Planning and Operational Objectives. This method employs a two-layer, multi-objective optimization model. The outer planning layer considers economic indicators such as local photovoltaic grid integration rate, life-cycle cost, and investment returns. The inner operational layer considers technical indicators such as node voltage deviation, curtailment, electricity purchase cost, interruptible load compensation cost, and grid losses. Compared to existing technologies that often focus on a single objective or neglect certain important constraints, this approach seeks the optimal balance among multiple mutually constraining objectives, ensuring that the planning scheme is both economically feasible and technically reliable.

[0022] 4. Enhance the feasibility of power distribution network projects in mountainous areas. By incorporating factors such as mountainous terrain features, construction conditions, and grid structure into the optimization model, this method can fully consider actual constraints such as slope, obstruction, road accessibility, and construction costs, making the planning results closer to the actual project. Compared with existing technologies that ignore the special conditions of mountainous areas, resulting in insufficient feasibility of planning schemes, this scheme can effectively improve the implementability and adaptability of planning results.

[0023] 5. Ensuring the long-term operational stability of mountain power distribution networks. This method accurately characterizes the continuous evolution of the energy storage state of charge (SOC), as well as the temporal coupling relationship between source, load, and storage, through a joint operation model. This avoids problems such as SOC out-of-bounds behavior, insufficient protection of critical loads, or continuous voltage exceedances at nodes that may arise from configuring based solely on static capacity constraints or single-period power balance. Compared to the potential performance degradation that may occur in existing technologies during continuous operation, this solution ensures that mountain power distribution networks maintain a stable and reliable power supply capability during long-term operation.

[0024] In summary, this method can comprehensively improve the power supply reliability, renewable energy absorption capacity, and operational economy of mountain power distribution networks through source-load-storage synergistic optimization, providing effective technical support for the low-carbon transformation and high-quality development of mountain power distribution networks.

[0025] Preferably, in S1, the basic dataset constructed is:

[0026] ;

[0027] Where N represents the set of nodes in the mountainous power distribution network; L represents the set of branches; Indicates the branch impedance; and These represent the active and reactive loads of node i during time period t, respectively. This represents the node voltage of node i during time period t; Indicates the maximum capacity of the branch; and These represent the sets of candidate access nodes for photovoltaic and energy storage, respectively. This represents the light intensity at node i during time period t; Indicates time-of-use electricity pricing; Indicates the on-grid electricity price; and These represent the costs of photovoltaics and energy storage, respectively. The mountainous terrain constraint index represents node i.

[0028] The mountainous terrain constraints include altitude, slope, aspect, shading, road accessibility, construction and transportation conditions, and communication conditions.

[0029] The set of photovoltaic candidate access nodes The following steps were used to filter the results:

[0030] Determine whether a candidate node meets the constraints for mountainous area construction. If candidate node i meets the following conditions, then retain it in the set of photovoltaic candidate access nodes. middle:

[0031] ;

[0032] in, This represents the slope of the area where node i is located. Indicates the maximum allowed slope for access; Indicates the occlusion coefficient. Indicates the maximum allowable occlusion coefficient; Indicates the construction transportation distance or road accessibility index. This indicates the maximum construction transport distance or road accessibility threshold that is allowed for access; This indicates the additional costs of construction in mountainous areas. This indicates the maximum additional construction cost allowed for access in mountainous areas.

[0033] This approach significantly improves the engineering adaptability and planning feasibility of photovoltaic (PV) site selection. Existing technologies typically select PV access points based solely on electrical parameters or simple geographical information, neglecting the unique terrain limitations and construction feasibility constraints of mountainous areas. This can easily lead to planning schemes failing to be implemented during the actual construction phase due to excessively steep slopes, severe shading, impassable roads, or excessive costs. In contrast, this solution explicitly models the construction constraints in mountainous areas as quantifiable hard screening conditions and embeds them into the basic dataset construction process. This ensures that subsequent optimization and configuration are always based on a real set of candidate nodes, fundamentally avoiding the problem of theoretical planning that is difficult to implement and guaranteeing the engineering feasibility of the source-load-storage collaborative optimization results.

[0034] Preferably, the typical operating scenario set is obtained by sampling and scenario reduction based on time-series samples of photovoltaic power output and load demand;

[0035] The time-series sample of photovoltaic power output and load demand is constructed through the following steps:

[0036] Based on historical solar irradiance and load data, a photovoltaic power output model and a load demand model are established; wherein, the photovoltaic active power output of node i in time period t is... for:

[0037] ;

[0038] in, This represents the light intensity at node i during time period t. Indicates the conversion efficiency of photovoltaic modules. This represents the photovoltaic installation area or equivalent installed capacity of node i;

[0039] Load demand of node i in time period t for:

[0040] ;

[0041] in, This represents the baseline load of node i in time period t. This represents the amount of random load disturbance.

[0042] Combine photovoltaic output and load demand into a source-load random variable vector:

[0043] ;

[0044] in, Let t represent the source-load random variable vector for time period t, and n represent the number of distribution network nodes.

[0045] This setup effectively preserves the temporal correlation and joint fluctuation characteristics of source and load, enhancing the fidelity of the scene set in representing the actual operating conditions in mountainous areas. Existing technologies often employ independent sampling or model photovoltaic / load separately before stitching together the scene, which easily disrupts the inherent temporal correlation between sunlight and load, leading to scene distortion. In contrast, this solution uses historical time-series samples as a basis, modeling photovoltaic output and load demand as coupled variables for the same time period, and using random perturbation terms... It reflects the uncertainty of the load, so that the generated scene set reflects both deterministic trends and random fluctuations, and maintains the coordinated change pattern of the source load on the time axis. This provides a more realistic and representative input basis for subsequent two-layer optimization, and avoids configuration deviations and operational risks caused by scene mismatch.

[0046] Preferably, the process of performing scene reduction on the initial scene set includes:

[0047] Cluster the initial scene set to reduce its size; where the i-th cluster... Sum of squared intra-cluster distances for:

[0048] ;

[0049] Where x represents a scene sample; Represents the relationship between sample x and cluster center The Euclidean distance; Cluster Cluster centers;

[0050] When cluster Decomposed into two sub-clusters and At that time, the decrease in the sum of squared distances for:

[0051] ;

[0052] in, and They represent respectively by The decomposition yields two subclusters; Cluster The reduction in the sum of squared distances after decomposition;

[0053] When cluster Removed and merged into the nearest cluster At that time, the increase in the sum of squared distances for:

[0054] ;

[0055] in, This indicates the clusters to be removed; Cluster Cluster centers Indicates and The most recent cluster;

[0056] like Then execute cluster Decomposition and clustering The process involves merging clusters, repeatedly performing cluster decomposition, cluster merging, and cluster center updates until the cluster centers stabilize, resulting in a reduced set of typical operating scenarios. :

[0057] ;

[0058] in, This represents the k-th typical operating scenario;

[0059] Probability of occurrence of each typical operating scenario for:

[0060] ;

[0061] Where, N k M represents the number of original scenarios in the k-th cluster, and M represents the number of reduced typical running scenarios.

[0062] This setup enables the construction of a set of typical scenarios with high fidelity, low redundancy, and adaptability, providing a foundation for uncertainty modeling that combines computational efficiency and physical realism for the coordinated optimization of power distribution networks in mountainous areas.

[0063] Preferably, the joint operation model includes:

[0064] The photovoltaic power output model is used to calculate the active power output of photovoltaics in each time period and scenario based on the irradiance, photovoltaic installation capacity and photovoltaic module conversion efficiency in each typical operating scenario.

[0065] The load response model is used to classify node loads into critical loads, general loads, and interruptible loads, and to determine the actual load and the corresponding interruptible load compensation cost based on the interruptible load response ratio.

[0066] An energy storage operation model is used to describe the continuous transfer of the energy storage state of charge with charge and discharge power, and to define the charge and discharge power limits of energy storage and the mutual exclusion relationship between charge and discharge.

[0067] The energy management model is used to establish node-level active power balance equations that describe the balance between photovoltaic output, energy storage charging and discharging power, main grid power purchase and sales, load demand, and photovoltaic power reduction under each typical operating scenario.

[0068] A power supply reliability model is used to take the probability of load failure as the reliability index of the source-load-storage combined system and to establish corresponding reliability constraints.

[0069] This setup enables refined collaborative modeling of multiple entities in the power distribution network across mountainous areas across four dimensions: timing, power, energy, and reliability. It overcomes the operational distortions caused by fragmented or overly simplified models in traditional methods. Existing technologies often treat energy storage as an ideal power source, treat load response only as static reduction, or ignore the spatiotemporal coupling relationship between photovoltaic output and load. This makes it difficult to accurately depict the complex interactive behaviors in mountainous power grids caused by terrain obstruction, communication delays, and energy storage SOC continuity constraints. In contrast, this solution introduces interruptible load compensation costs, mutually exclusive energy storage charging and discharging relationships, node-level power balance equations, and load loss probability constraints. It integrates physical feasibility, economic incentives, and system resilience into the operational model, ensuring that the inner-layer optimization results reflect the dynamic characteristics of real equipment while supporting the rationality and robustness of outer-layer configuration decisions.

[0070] Preferably, the construction of the photovoltaic power output model includes:

[0071] The irradiance, photovoltaic installation capacity, and photovoltaic module efficiency of each candidate node are mapped to the photovoltaic active power output at each time period; for the photovoltaic output of node i at time period t under a typical operating scenario s. for:

[0072] ;

[0073] in, This represents the illumination intensity of node i in time period t and scene s; Indicates the conversion efficiency of photovoltaic modules; This represents the photovoltaic installation area or equivalent installed capacity of node i;

[0074] Total photovoltaic output of the system under time period t and scenario s for:

[0075] ;

[0076] in, This represents the set of selected photovoltaic candidate access nodes;

[0077] The construction of the load response model includes:

[0078] The node load is divided into critical load, general load, and interruptible load, and the response boundaries of each type of load are determined; whereby, the actual load of node i in time period t under scenario s is... for:

[0079] ;

[0080] in, Indicates general load; Indicates interruptible load; Indicates the proportion of interruptible load response;

[0081] Interruptible load response ratio satisfy:

[0082] ;

[0083] in, This indicates the upper limit of the interruptible load response ratio for node i;

[0084] The cost of interruptible load compensation is:

[0085] ;

[0086] in, This represents the interruptible load compensation cost under scenario s; T represents the total number of time periods. Represents the set of load nodes; This represents the unit interruption compensation price for time period t; Indicates the unit scheduling duration;

[0087] The construction of the energy storage operation model includes:

[0088] Establish an energy storage charge state transition model, charge / discharge power constraints, and charge / discharge mutual exclusion constraints; wherein, the energy storage charge state satisfies:

[0089] ;

[0090] in, and These represent the state of charge of node i's energy storage in time period t and time period t+1, scenario s, respectively; and These represent charging efficiency and discharging efficiency, respectively. and These represent charging power and discharging power, respectively. Indicates the rated capacity of energy storage; Indicates the unit scheduling duration;

[0091] The energy storage state of charge constraints are:

[0092] ;

[0093] in, and These represent the lower and upper limits of the energy storage's state of charge, respectively.

[0094] The energy storage charging and discharging power constraint is:

[0095] ;

[0096] ;

[0097] in, and These represent the charging state variable and the discharging state variable, respectively. and These represent the maximum charging power and the maximum discharging power, respectively.

[0098] The charge / discharge mutual exclusion constraint is:

[0099] .

[0100] Preferably, the construction of the energy management model for photovoltaics, energy storage, and the main grid includes:

[0101] Establish node-level active power balance relationships for each typical operating scenario; where the active power balance constraint of node i in time period t and scenario s is expressed as:

[0102] ;

[0103] in, Indicates photovoltaic power output; Indicates the energy storage discharge power; This represents the power purchased from the mainnet and allocated to node i; Indicates the actual load; Indicates the energy storage charging power; Indicates the reduction in photovoltaic power; Indicates internet access power;

[0104] The construction of the power supply reliability model includes:

[0105] The probability of load failure is used as a reliability indicator for the source-load-storage combined system; where the load failure power in time period t under scenario s is... for:

[0106] ;

[0107] in, Indicates the total system load demand; Indicates the total output of photovoltaic power; Indicates the total discharge power of the energy storage; Indicates the power purchased from the main grid;

[0108] Probability of load loss for:

[0109] ;

[0110] in, This represents a set of typical operating scenarios; Let represent the probability of scenario s occurring; T represents the total number of time periods.

[0111] And it meets the following reliability constraints:

[0112] ;

[0113] in, This indicates the preset reliability threshold.

[0114] Preferably, the decision variables of the two-level multi-objective optimization model are:

[0115]

[0116] in, Represents the vector of optimization decision variables; and These are 0-1 variables representing whether photovoltaic and energy storage are connected to node i, with a value of 1 indicating connection. This represents the photovoltaic installation area or equivalent installed capacity of node i; This represents the rated energy storage capacity of node i; This represents the rated energy storage power of node i; Indicates the proportion of interruptible load response; Indicates the proportion of electricity purchased from the main grid;

[0117] The construction of the outer planning objective includes:

[0118] The outer planning objectives are at least two of the following: maximizing the local photovoltaic (PV) grid integration rate, minimizing the system's total life-cycle net present value (NPV) cost, and maximizing the investment return of the PV-storage integrated system; among these, the local PV grid integration rate... for:

[0119] ;

[0120] in, This represents a set of typical operating scenarios; Let represent the probability of scenario s occurring; T represents the total number of time periods. This represents the photovoltaic power absorbed by local load; This represents the photovoltaic power absorbed by energy storage. Indicates the total output of photovoltaic power; Indicates the unit scheduling duration;

[0121] System lifecycle net present value cost for:

[0122] ;

[0123] in, and These represent the investment costs for photovoltaic (PV) and energy storage, respectively. Indicates operating and maintenance costs; Indicates the cost of replacing energy storage; This indicates the cost of purchasing electricity from the main grid; This indicates the cost of interruptible load compensation; This indicates revenue from surplus electricity sold to the grid; Represents residual value;

[0124] Investment returns of photovoltaic-storage integrated systems for:

[0125] ;

[0126] in, Indicates subsidy income; This indicates savings achieved by reducing electricity purchases from the main grid;

[0127] The construction of the inner runtime target includes:

[0128] Minimizing at least two of the following under typical operating scenarios: average node voltage deviation, curtailment of solar power, electricity purchase cost, interruptible load compensation cost, and grid loss; where the average node voltage deviation is the most important factor. for:

[0129] ;

[0130] in, This represents a set of typical operating scenarios; This represents the probability of scenario s occurring; T represents the number of nodes; T represents the total number of time periods; N represents the node set. This represents the voltage of node i in time period t and scenario s; Indicates the reference voltage;

[0131] Wasted light for:

[0132] ;

[0133] in, Represents the set of candidate photovoltaic access nodes; Indicates the power of light discarded; Indicates the unit scheduling duration;

[0134] Electricity purchase cost for:

[0135] ;

[0136] in, This represents the time-of-use electricity price for period t; Indicates the power purchased from the main grid;

[0137] Network loss is represented as:

[0138] ;

[0139] in, L represents the network loss under time period t and scenario s; L represents the set of branches. Represents the resistance of branch (i,j); and These represent the active power and reactive power of the branch, respectively.

[0140] This setup enables deep collaborative optimization between the planning and operation layers, avoiding suboptimal or even infeasible solutions caused by the two-stage separation of "planning first, then operation." Traditional methods typically determine photovoltaic / energy storage configuration schemes based on simplified operational assumptions, and then verify operational performance under fixed configurations. This is prone to causing the final scheme to fail in actual operation due to neglecting operational constraints. In contrast, this scheme uses the inner-layer operational objectives as explicit constraints or inner-layer sub-problem objectives for outer-layer optimization, and incorporates the probability weighting of typical scenarios into the outer-layer objective function. This allows configuration decisions to directly respond to operational feasibility and economic feedback, ensuring that the selected sites and capacities not only meet the stringent electrical and geographical constraints of mountainous areas but also achieve optimal comprehensive benefits throughout the entire lifecycle. It is particularly suitable for mountainous scenarios with complex terrain, weak grid structures, and strong source-load fluctuations, significantly improving the engineering feasibility and long-term economic viability of the planning scheme.

[0141] Preferably, the constraints of the two-layer multi-objective optimization model include power flow balance constraints, node voltage constraints, branch capacity constraints, photovoltaic and energy storage configuration capacity constraints, as well as the operation constraints and reliability constraints included in the joint operation model.

[0142] Preferably, in S4, the process of solving the two-level multi-objective optimization model includes:

[0143] The decision variables are encoded to generate candidate configuration schemes;

[0144] For each candidate configuration scheme, based on the joint operation model built by S2, under the typical operation scenario set built by S1, the corresponding joint operation state is simulated and calculated, and the objective function values ​​and constraint violation degrees of each item are evaluated.

[0145] The candidate configuration schemes are iteratively updated and globally searched using an intelligent optimization algorithm.

[0146] Pareto nondominated sorting is used to select the set of nondominated solutions that satisfy all constraints from all candidate configurations;

[0147] Based on the fuzzy membership method, a compromise optimal solution is selected from the set of non-dominated solutions;

[0148] The final collaborative optimization configuration scheme is determined based on the compromise optimal solution.

[0149] This approach effectively balances solution feasibility, Pareto front quality, and engineering practicality in high-dimensional, multi-objective, and strongly constrained mountain power distribution network optimization problems, avoiding decision bias or infeasibility risks caused by a single optimization criterion. Mountainous photovoltaic-storage configuration problems typically involve dozens of discrete-continuous mixed variables, multiple electrical constraints, and reliability requirements. Traditional single-objective optimization is prone to getting trapped in local optima or violating key constraints. This scheme, however, ensures physical feasibility for all candidate solutions through simulation evaluation and constraint violation quantification. Pareto non-dominated ranking preserves all efficient solutions under multi-objective trade-offs, and fuzzy membership comprehensively considers the importance of each objective and the balance of solutions, ensuring that the final solution is both on the Pareto front and conforms to engineering preferences, significantly improving the feasibility and robustness of the optimization results. Attached Figure Description

[0150] To make the objectives, technical solutions, and advantages of the invention clearer, the invention will now be described in further detail with reference to the accompanying drawings, wherein:

[0151] Figure 1 This is a flowchart of the method. Detailed Implementation

[0152] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0153] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0154] Example:

[0155] like Figure 1 As shown, this invention provides a method for coordinated optimization of power generation, load, and storage configuration in mountainous distribution networks, comprising the following steps:

[0156] S1. Construct a basic dataset and a set of typical operating scenarios for the coordinated optimization of power generation, load and storage in mountainous power distribution networks.

[0157] In practice, the basic dataset constructed is as follows:

[0158] ;

[0159] Where N represents the set of nodes in the mountainous power distribution network; L represents the set of branches; Indicates the branch impedance; and These represent the active and reactive loads of node i during time period t, respectively. This represents the node voltage of node i during time period t; Indicates the maximum capacity of the branch; and These represent the sets of candidate access nodes for photovoltaic and energy storage, respectively. This represents the light intensity at node i during time period t; Indicates time-of-use electricity pricing; Indicates the on-grid electricity price; and These represent the costs of photovoltaics and energy storage, respectively. The mountainous terrain constraint index represents node i.

[0160] By unifying network topology, electrical operation quantities, source-load time-series data, price and cost data, and mountainous terrain constraints into a unified basic dataset, a consistent data entry point can be provided for subsequent candidate node selection, operation scenario generation, and source-load-storage collaborative optimization. Compared to planning solely based on load and installed capacity, this scheme further considers constraints such as mountain slope, obstruction, construction and transportation conditions, and communication conditions, making the optimization results more aligned with the engineering construction conditions of mountainous power distribution networks.

[0161] The mountainous terrain constraints include altitude, slope, aspect, shading, road accessibility, construction and transportation conditions, and communication conditions.

[0162] The set of photovoltaic candidate access nodes The following steps were used to filter the results:

[0163] Determine whether a candidate node meets the constraints for mountainous area construction. If candidate node i meets the following conditions, then retain it in the set of photovoltaic candidate access nodes. middle:

[0164] ;

[0165] in, This represents the slope of the area where node i is located. Indicates the maximum allowed slope for access; Indicates the occlusion coefficient. Indicates the maximum allowable occlusion coefficient; Indicates the construction transportation distance or road accessibility index. This indicates the maximum construction transport distance or road accessibility threshold that is allowed for access; This indicates the additional costs of construction in mountainous areas. This indicates the maximum additional construction cost allowed for access in mountainous areas.

[0166] And based on the retained set of photovoltaic candidate access nodes and energy storage candidate access node set Calculate the penalty costs for construction in mountainous areas:

[0167] ;

[0168] in, This indicates the cost of construction penalties in mountainous areas. This represents the set of candidate energy storage nodes after filtering, where j represents the sequence number of the candidate energy storage node. and These represent the construction penalty coefficients for photovoltaic and energy storage candidate nodes in mountainous areas, respectively. This represents the investment cost corresponding to configuring photovoltaics at node i. This represents the investment cost corresponding to configuring energy storage at node j.

[0169] Traditional configuration methods typically select candidate nodes based solely on electrical access conditions or capacity requirements, easily resulting in solutions that are difficult to construct in mountainous environments, suffer from severe long-term obstruction, or have insufficient operation and maintenance communication conditions. This solution uses a constructibility screening process to first eliminate nodes that do not meet the requirements for slope, obstruction, road accessibility, and construction cost. Then, it uses the penalty cost for construction in mountainous areas to reflect the engineering difficulty of different nodes, ensuring that the selection of candidate nodes simultaneously possesses both electrical feasibility and engineering feasibility.

[0170] In practice, the typical operating scenario set is obtained by sampling and scenario reduction based on time-series samples of photovoltaic power output and load demand;

[0171] The time-series sample of photovoltaic power output and load demand is constructed through the following steps:

[0172] Based on historical solar irradiance and load data, a photovoltaic power output model and a load demand model are established; wherein, the photovoltaic active power output of node i in time period t is... for:

[0173] ;

[0174] in, This represents the light intensity at node i during time period t. Indicates the conversion efficiency of photovoltaic modules. This represents the photovoltaic installation area or equivalent installed capacity of node i;

[0175] Load demand of node i in time period t for:

[0176] ;

[0177] in, This represents the baseline load of node i in time period t. This represents the amount of random load disturbance.

[0178] Combine photovoltaic output and load demand into a source-load random variable vector:

[0179] ;

[0180] in, Let t represent the source-load random variable vector for time period t, and n represent the number of distribution network nodes.

[0181] In this way, by uniformly expressing photovoltaic power output and load demand as a source-load random variable vector, the combined impact of solar radiation fluctuations and load disturbances on system power balance can be described on the same time scale. Given the characteristics of mountainous power distribution networks—sunlight is significantly affected by mountain shading, sudden weather changes, and dispersed loads—this time-series sample construction method can provide representative source-load combinations for subsequent scenario sampling and reduction.

[0182] In practice, during the process of generating typical operating scenario sets, Latin hypercube sampling, Monte Carlo sampling, or methods based on historical sample resampling are used to generate initial scenario sets for photovoltaic power output and load demand.

[0183] When using Latin hypercube sampling, the probability interval corresponding to the m-th sample is:

[0184] ;

[0185] The light intensity sample and the load demand sample are represented as follows:

[0186] ;

[0187] ;

[0188] This forms the initial scene set:

[0189] ;

[0190] Each initial scene is represented as follows:

[0191] ;

[0192] in, This represents the probability value corresponding to the m-th sampling point, where m represents the sample number and M represents the initial number of scenes. This represents a random number that follows a uniform distribution in the range [0,1]. This represents the light intensity of the m-th sample during time period t. This represents the load demand of the m-th sample in time period t. This represents the photovoltaic output of the m-th sample in time period t. It represents the inverse function of the cumulative distribution function of light intensity. This represents the inverse function of the cumulative load demand distribution function. This represents the initial scene set. Let m represent the m-th initial scene, and T represent the total number of time periods.

[0193] The above sampling method can cover the main fluctuation ranges of light intensity and load demand with a limited number of samples. In particular, when using Latin hypercube sampling, each probability interval is sampled, which can reduce the problem of sample concentration or omission caused by random sampling and improve the initial scene's ability to cover the uncertainty of source load in mountainous areas.

[0194] In specific implementation, the process of reducing the number of scenes in the initial scene set includes:

[0195] Cluster the initial scene set to reduce its size; where the i-th cluster... Sum of squared intra-cluster distances for:

[0196] ;

[0197] Where x represents a scene sample; Represents the relationship between sample x and cluster center The Euclidean distance; Cluster Cluster centers;

[0198] When cluster Decomposed into two sub-clusters and At that time, the decrease in the sum of squared distances for:

[0199] ;

[0200] in, and They represent respectively by The decomposition yields two subclusters; Cluster The reduction in the sum of squared distances after decomposition;

[0201] When cluster Removed and merged into the nearest cluster At that time, the increase in the sum of squared distances for:

[0202] ;

[0203] in, This indicates the clusters to be removed; Cluster Cluster centers Indicates and The most recent cluster;

[0204] like Then execute cluster Decomposition and clustering The process involves merging clusters, repeatedly performing cluster decomposition, cluster merging, and cluster center updates until the cluster centers stabilize, resulting in a reduced set of typical operating scenarios. :

[0205] ;

[0206] in, This represents the k-th typical operating scenario;

[0207] Probability of occurrence of each typical operating scenario for:

[0208] ;

[0209] Where, N k M represents the number of original scenarios in the k-th cluster, and M represents the number of reduced typical running scenarios.

[0210] By combining cluster decomposition and cluster merging to reduce the number of scenes, representative source-load fluctuation characteristics of the initial scene set can be preserved while reducing the number of scenes. Compared with directly selecting a small number of historical typical days, this scheme can simultaneously take into account multiple operating conditions such as high illumination and low load, low illumination and high load, and load peak-valley variations, which helps to reduce the dependence of optimization results on a single scene.

[0211] S2 uses the basic dataset and typical operating scenario set built in S1 to construct a joint operation model of power generation, load and storage in mountainous distribution networks, which is used to describe the power balance relationship between photovoltaic output, energy storage charging and discharging, main grid power purchase and sales and load response.

[0212] In specific implementation, the joint operation model includes:

[0213] The photovoltaic power output model is used to calculate the active power output of photovoltaics in each time period and scenario based on the irradiance, photovoltaic installation capacity and photovoltaic module conversion efficiency in each typical operating scenario.

[0214] The load response model is used to classify node loads into critical loads, general loads, and interruptible loads, and to determine the actual load and the corresponding interruptible load compensation cost based on the interruptible load response ratio.

[0215] An energy storage operation model is used to describe the continuous transfer of the energy storage state of charge with charge and discharge power, and to define the charge and discharge power limits of energy storage and the mutual exclusion relationship between charge and discharge.

[0216] The energy management model is used to establish node-level active power balance equations that describe the balance between photovoltaic output, energy storage charging and discharging power, main grid power purchase and sales, load demand, and photovoltaic power reduction under each typical operating scenario.

[0217] A power supply reliability model is used to take the probability of load failure as the reliability index of the source-load-storage combined system and to establish corresponding reliability constraints.

[0218] The construction of the photovoltaic power output model includes:

[0219] The irradiance, photovoltaic installation capacity, and photovoltaic module efficiency of each candidate node are mapped to the photovoltaic active power output at each time period; for the photovoltaic output of node i at time period t under a typical operating scenario s. for:

[0220] ;

[0221] in, This represents the illumination intensity of node i in time period t and scene s; Indicates the conversion efficiency of photovoltaic modules; This represents the photovoltaic installation area or equivalent installed capacity of node i;

[0222] Total photovoltaic output of the system under time period t and scenario s for:

[0223] ;

[0224] in, This represents the set of selected photovoltaic candidate access nodes;

[0225] The construction of the load response model includes:

[0226] The node load is divided into critical load, general load, and interruptible load, and the response boundaries of each type of load are determined; whereby, the actual load of node i in time period t under scenario s is... for:

[0227] ;

[0228] in, Indicates general load; Indicates interruptible load; Indicates the proportion of interruptible load response;

[0229] Interruptible load response ratio satisfy:

[0230] ;

[0231] in, This indicates the upper limit of the interruptible load response ratio for node i;

[0232] The cost of interruptible load compensation is:

[0233] ;

[0234] in, This represents the interruptible load compensation cost under scenario s; T represents the total number of time periods. Represents the set of load nodes; This represents the unit interruption compensation price for time period t; Indicates the unit scheduling duration;

[0235] This load response modeling approach can ensure continuous power supply to critical loads while incorporating general and interruptible loads into the optimization and regulation scope. When photovoltaic output is insufficient, energy storage capacity is low, or grid purchase prices are high, peak power supply pressure can be reduced through constrained interruptible load response, and excessive load interruption can be avoided through cost compensation constraints.

[0236] The construction of the energy storage operation model includes:

[0237] Establish an energy storage charge state transition model, charge / discharge power constraints, and charge / discharge mutual exclusion constraints; wherein, the energy storage charge state satisfies:

[0238] ;

[0239] in, and These represent the state of charge of node i's energy storage in time period t and time period t+1, scenario s, respectively; and These represent charging efficiency and discharging efficiency, respectively. and These represent charging power and discharging power, respectively. Indicates the rated capacity of energy storage; Indicates the unit scheduling duration;

[0240] The energy storage state of charge constraints are:

[0241] ;

[0242] in, and These represent the lower and upper limits of the energy storage's state of charge, respectively.

[0243] The energy storage charging and discharging power constraint is:

[0244] ;

[0245] ;

[0246] in, and These represent the charging state variable and the discharging state variable, respectively. and These represent the maximum charging power and the maximum discharging power, respectively.

[0247] The charge / discharge mutual exclusion constraint is:

[0248] .

[0249] The construction of an energy management model integrating photovoltaics, energy storage, and the main grid includes:

[0250] The above model uses the continuity of state of charge, upper and lower power limits, and mutual exclusion constraints of charging and discharging to jointly limit the operation of energy storage, avoiding the problems of simultaneous charging and discharging or exceeding the state of charge limit at the same time. This allows energy storage to absorb surplus photovoltaic power and release power when photovoltaic power is insufficient or during peak load.

[0251] Establish node-level active power balance relationships for each typical operating scenario; where the active power balance constraint of node i in time period t and scenario s is expressed as:

[0252] ;

[0253] in, Indicates photovoltaic power output; Indicates the energy storage discharge power; This represents the power purchased from the mainnet and allocated to node i; Indicates the actual load; Indicates the energy storage charging power; Indicates the reduction in photovoltaic power; Indicates internet access power;

[0254] The energy management model prioritizes photovoltaic (PV) output to meet local load demand. When PV power is surplus, it prioritizes charging energy storage. Once the energy storage is fully charged, any remaining electricity is fed back to the grid or reduced. When PV power is insufficient, it prioritizes discharging from the energy storage to supply power. When energy storage is insufficient, it is supplemented by purchasing electricity from the main grid. This can increase the local consumption rate of PV power, reduce curtailment and electricity purchase costs, and improve the power supply adaptability of mountainous distribution networks under conditions of weak interconnection and long feeder lines.

[0255] The construction of the power supply reliability model includes:

[0256] The probability of load failure is used as a reliability indicator for the source-load-storage combined system; where the load failure power in time period t under scenario s is... for:

[0257] ;

[0258] in, Indicates the total system load demand; Indicates the total output of photovoltaic power; Indicates the total discharge power of the energy storage; Indicates the power purchased from the main grid;

[0259] Probability of load loss for:

[0260] ;

[0261] in, This represents a set of typical operating scenarios; Let represent the probability of scenario s occurring; T represents the total number of time periods.

[0262] And it meets the following reliability constraints:

[0263] ;

[0264] in, This indicates the preset reliability threshold.

[0265] This reliability constraint transforms the degree of power supply insufficiency in multiple scenarios into a unified evaluation index, ensuring that the configuration scheme is not only feasible under typical sunny or average load conditions, but also meets the preset power supply reliability requirements under adverse scenarios such as low light and high load.

[0266] S3, based on the joint operation model of S2 and the typical operation scenario set of S1, construct a two-layer multi-objective optimization model for source-load-storage collaborative optimization configuration;

[0267] The outer planning objective of the two-layer multi-objective optimization model includes at least two of the following: maximizing the local photovoltaic consumption rate, minimizing the net present value cost of the system's entire life cycle, and maximizing the investment return of the photovoltaic-storage combined system; the inner operation objective includes at least two of the following: minimizing the average node voltage deviation, minimizing curtailment of solar power, minimizing electricity purchase cost, minimizing interruptible load compensation cost, and minimizing grid loss.

[0268] The typical operating scenario set of S1 is configured as the basis for the two-layer multi-objective optimization model to consider the uncertainty of source and load. The joint operating model of S2 is configured to define the inherent rules of source, load, and storage operation interaction and system reliability in the two-layer multi-objective optimization model.

[0269] In practical implementation, the decision variables of the two-level multi-objective optimization model are:

[0270]

[0271] in, Represents the vector of optimization decision variables; and These are 0-1 variables representing whether photovoltaic and energy storage are connected to node i, with a value of 1 indicating connection. This represents the photovoltaic installation area or equivalent installed capacity of node i; This represents the rated energy storage capacity of node i; This represents the rated energy storage power of node i; Indicates the proportion of interruptible load response; Indicates the proportion of electricity purchased from the main grid;

[0272] The construction of the outer planning objective includes:

[0273] The outer planning objectives are at least two of the following: maximizing the local photovoltaic (PV) grid integration rate, minimizing the system's total life-cycle net present value (NPV) cost, and maximizing the investment return of the PV-storage integrated system; among these, the local PV grid integration rate... for:

[0274] ;

[0275] in, This represents a set of typical operating scenarios; Let represent the probability of scenario s occurring; T represents the total number of time periods. This represents the photovoltaic power absorbed by local load; This represents the photovoltaic power absorbed by energy storage. Indicates the total output of photovoltaic power; Indicates the unit scheduling duration;

[0276] System lifecycle net present value cost for:

[0277] ;

[0278] in, and These represent the investment costs for photovoltaic (PV) and energy storage, respectively. Indicates operating and maintenance costs; Indicates the cost of replacing energy storage; This indicates the cost of purchasing electricity from the main grid; This indicates the cost of interruptible load compensation; This indicates revenue from surplus electricity sold to the grid; Represents residual value;

[0279] Investment returns of photovoltaic-storage integrated systems for:

[0280] ;

[0281] in, Indicates subsidy income; This indicates savings achieved by reducing electricity purchases from the main grid;

[0282] The outer planning objectives simultaneously reflect renewable energy consumption, economic costs, and investment returns, ensuring that the configuration scheme no longer simply pursues the maximum installed capacity. For mountainous power distribution networks, this setup can prevent increased curtailment of solar power due to excessive photovoltaic capacity, and also avoid insufficient peak-valley regulation and reliability support capabilities due to inadequate energy storage configuration.

[0283] The construction of the inner runtime target includes:

[0284] Minimizing at least two of the following under typical operating scenarios: average node voltage deviation, curtailment of solar power, electricity purchase cost, interruptible load compensation cost, and grid loss; where the average node voltage deviation is the most important factor. for:

[0285] ;

[0286] in, This represents a set of typical operating scenarios; This represents the probability of scenario s occurring; T represents the number of nodes; T represents the total number of time periods; N represents the node set. This represents the voltage of node i in time period t and scenario s; Indicates the reference voltage;

[0287] Wasted light for:

[0288] ;

[0289] in, Represents the set of candidate photovoltaic access nodes; Indicates the power of light discarded; Indicates the unit scheduling duration;

[0290] Electricity purchase cost for:

[0291] ;

[0292] in, This represents the time-of-use electricity price for period t; Indicates the power purchased from the main grid;

[0293] Network loss is represented as:

[0294] ;

[0295] in, L represents the network loss under time period t and scenario s; L represents the set of branches. Represents the resistance of branch (i,j); and These represent the active power and reactive power of the branch, respectively.

[0296] In this way, the inner-layer operation model can further optimize the charging and discharging, power purchase and sale, curtailment and load response strategies for each time period under the given photovoltaic and energy storage configuration, making the configuration scheme operable at the intraday operation level.

[0297] In specific implementation, the constraints of the two-layer multi-objective optimization model include power flow balance constraints, node voltage constraints, branch capacity constraints, photovoltaic and energy storage configuration capacity constraints, as well as the operation constraints and reliability constraints included in the joint operation model.

[0298] The power flow balance constraints include active power balance constraints and reactive power balance constraints:

[0299] ;

[0300] ;

[0301] in, and These represent the injected active power and injected reactive power of node i in time period t and scenario s, respectively. and Let i and j represent the voltage magnitudes at node i and node j, respectively; N represents the set of nodes. and These represent the conductance and susceptance between node i and node j in the nodal admittance matrix, respectively. Indicates the phase angle difference of the node voltage;

[0302] The node voltage constraints, branch capacity constraints, and photovoltaic and energy storage configuration capacity constraints are as follows:

[0303] ;

[0304] ;

[0305] ;

[0306] ;

[0307] in, and These represent the lower and upper limits of the node voltage, respectively. and Let i and j represent the active power and reactive power of the branch (i,j), respectively. Indicates the maximum capacity of the branch; This represents the photovoltaic installation capacity of node i; Indicates the photovoltaic grid connection status; This indicates the maximum photovoltaic capacity that node i is allowed to configure; This represents the rated energy storage capacity of node i; Indicates the energy storage connection status; This indicates the maximum energy storage capacity that node i is allowed to configure.

[0308] The outer model of the two-layer multi-objective optimization model is:

[0309] ;

[0310] in, Represents the outer objective function vector; Indicates the outer configuration variable; Indicates configuration variables The corresponding local photovoltaic power consumption rate; Indicates configuration variables The corresponding net present value cost over the entire life cycle; Indicates configuration variables The corresponding investment returns for the photovoltaic-storage integrated system;

[0311] The inner model is:

[0312] ;

[0313] in, Represents the inner objective function vector; This represents the inner runtime variable; s represents the scenario. This indicates the cost of purchasing electricity from the main grid; This indicates the cost of interruptible load compensation; It represents the average node voltage deviation, used to measure the degree of deviation of the voltage of each node from the reference voltage in various typical operating scenarios, time periods, and periods. This refers to the amount of curtailed solar power, which is the amount of photovoltaic power that is not utilized under various typical operating scenarios due to limitations such as load absorption capacity, energy storage capacity, line constraints, or voltage constraints. This refers to network loss, which is the power loss generated by the distribution network lines during the transmission of active and reactive power.

[0314] And satisfy:

[0315] ;

[0316] in, Represents the set of equality constraints; This represents the set of inequality constraints.

[0317] By adopting the above two-layer model, the outer layer is responsible for determining the long-term configuration scheme of photovoltaics and energy storage, while the inner layer is responsible for verifying and optimizing the operating status under typical scenarios, which can avoid the problem of the planned capacity being out of sync with the actual operating strategy.

[0318] S4. Solve the two-layer multi-objective optimization model constructed in S3 to obtain the collaborative optimization configuration scheme.

[0319] In practice, the process of solving the two-level multi-objective optimization model includes:

[0320] The decision variables are encoded to generate candidate configuration schemes;

[0321] For each candidate configuration scheme, based on the joint operation model built by S2, under the typical operation scenario set built by S1, the corresponding joint operation state is simulated and calculated, and the objective function values ​​and constraint violation degrees of each item are evaluated.

[0322] The candidate configuration schemes are iteratively updated and globally searched using an intelligent optimization algorithm.

[0323] Pareto nondominated sorting is used to select the set of nondominated solutions that satisfy all constraints from all candidate configurations;

[0324] Based on the fuzzy membership method, a compromise optimal solution is selected from the set of non-dominated solutions;

[0325] The final collaborative optimization configuration scheme is determined based on the compromise optimal solution.

[0326] To facilitate better understanding, the solution process for S4 is explained as follows.

[0327] S401. Encode the photovoltaic grid connection status, photovoltaic installed capacity, energy storage grid connection status, rated energy storage capacity, energy storage power, interruptible load response ratio, and main grid power purchase ratio to form candidate configuration individuals; wherein, the q-th candidate configuration individual is represented as:

[0328] ;

[0329] S402. For each candidate configuration individual, solve the source-load-storage joint operation state under various typical operating scenarios, and calculate the objective function and constraint violation degree; wherein, the constraint violation degree of candidate configuration individual q is expressed as:

[0330] ;

[0331] In the formula, q represents the candidate configuration individual number. Let q represent the q-th candidate configuration individual. , , , , , and These represent the photovoltaic (PV) grid connection status, PV installed capacity, energy storage grid connection status, energy storage rated capacity, energy storage power, interruptible load response ratio, and grid power purchase ratio for the q-th candidate configuration individual, respectively. This represents the constraint violation degree of the q-th candidate configuration individual. This represents the constraint of the a-th inequality. Let represent the b-th equality constraint, where a and b represent the sequence numbers of the inequality constraint and the equality constraint, respectively.

[0332] In practice, a parallel two-quantum differential evolution algorithm or a harmony search algorithm is used to iteratively update the candidate configuration individuals; when using the parallel two-quantum differential evolution algorithm, the qubit encoding of the candidate configuration individuals is represented as follows:

[0333]

[0334] The quantum state position update is represented as:

[0335]

[0336] In the formula, This represents the qubit encoding of the i-th candidate configuration individual. This represents the quantum angle of the d-th dimension variable for the i-th candidate configuration individual, where d represents the dimension index of the decision variable. and These represent the positions of the i-th candidate configuration individual in the k-th and k+1-th generations, respectively. This represents the optimal location center in the k-th generation. This represents the control parameter, and u represents a random number in the interval (0,1).

[0337] Because of the mixed coupling relationship between discrete and continuous variables among photovoltaic access location, energy storage capacity, energy storage power, and interruptible load response ratio, traditional enumeration solutions are difficult to apply to scenarios with large-scale candidate nodes. This scheme achieves global search and local optimization of candidate configuration individuals through qubit encoding and position updates, which can improve the solution efficiency of complex multi-objective configuration problems.

[0338] In practice, Pareto undominated sorting is used to obtain a set of undominated solutions that satisfy the constraints; for any two candidate configurations... and The following conditions are met to determine Dominate :

[0339]

[0340] In the formula, and Represents any two candidate configuration individuals. This represents the Pareto dominance relation, where m represents the objective function index. Let m be the objective function. This means that the objective function holds true for all objective functions. This indicates that at least one objective function exists. Represents logical AND relation.

[0341] Based on the dominance relationship, the Pareto non-dominated solution set is obtained:

[0342]

[0343] In the formula, Denotes the set of candidate solutions. This represents the set of Pareto non-dominated solutions. Represents the set of candidate solutions Another candidate configuration individual, This indicates that it does not exist.

[0344] In practice, a compromise optimal solution is selected from the Pareto non-dominated solution set, and a source-load-storage coordinated configuration scheme is output; wherein, the fuzzy membership degree of the r-th Pareto solution with respect to the m-th objective function is expressed as:

[0345]

[0346] The comprehensive membership degree of the r-th Pareto solution is expressed as:

[0347]

[0348] The compromise optimal solution is expressed as:

[0349]

[0350] In the formula, Let represent the fuzzy membership degree of the r-th Pareto solution with respect to the m-th objective function. This represents the comprehensive membership degree of the r-th Pareto solution. This represents the r-th Pareto solution. Indicates the first A Pareto solution, This represents the Pareto solution summation index in the comprehensive membership normalization calculation. and Let represent the maximum and minimum values ​​of the m-th objective function in the Pareto non-dominated solution set, respectively. Indicates the number of objective functions. Let R represent the weight of the m-th objective function, and let R represent the number of Pareto solutions. This represents the compromise optimal solution. This represents the solution that maximizes the overall membership degree.

[0351] Based on the compromise optimal solution, output the photovoltaic access nodes and capacity, energy storage access nodes and capacity, energy storage charging and discharging plan, main grid power purchase and sales plan, load response plan, curtailment plan, and node voltage verification results.

[0352] Compared to configuration methods that only output a single economically optimal solution, this scheme preserves multiple compromise solutions between different objectives through Pareto non-dominated ranking and selects the configuration result with better overall performance through fuzzy membership, thus taking into account photovoltaic absorption, construction investment, operating costs, voltage quality, and power supply reliability. For scenarios with dispersed power sources and loads, long branch lines, and significant differences in terrain in mountainous distribution networks, this method can improve the adaptability of the configuration scheme to multiple operating scenarios and multiple engineering constraints.

[0353] In summary, by constructing a basic dataset, screening the constructibility of candidate nodes in mountainous areas, generating typical operating scenarios, modeling the joint operation of energy sources, loads, and storage, performing two-layer multi-objective optimization, and implementing Pareto trade-off decisions, we can obtain a coordinated optimization configuration scheme for energy sources, loads, and storage in mountainous distribution networks that takes into account local photovoltaic consumption, energy storage regulation capacity, load response capacity, node voltage security, and life-cycle economic efficiency. This will improve the carrying capacity, power supply reliability, and operating economy of mountainous distribution networks for distributed new energy access.

[0354] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit the technical solutions. Those skilled in the art should understand that any modifications or equivalent substitutions to the technical solutions of the present invention without departing from the spirit and scope of the present invention should be covered within the scope of the claims of the present invention.

Claims

1. A method for coordinated optimization of power generation, load, and storage configuration in mountainous distribution networks, characterized in that, Includes the following steps: S1. Construct a basic dataset and a set of typical operating scenarios for the coordinated optimization of power source, load and storage configuration in mountainous power distribution networks; S2. Using the basic dataset and typical operating scenario set built in S1, construct a joint operation model of power generation, load and storage in mountainous distribution networks to describe the power balance relationship between photovoltaic output, energy storage charging and discharging, main grid power purchase and sales and load response; S3, based on the joint operation model of S2 and the typical operation scenario set of S1, construct a two-layer multi-objective optimization model for source-load-storage collaborative optimization configuration; The outer planning objective of the two-layer multi-objective optimization model includes at least two of the following: maximizing the local photovoltaic consumption rate, minimizing the net present value cost of the system's entire life cycle, and maximizing the investment return of the photovoltaic-storage combined system; the inner operation objective includes at least two of the following: minimizing the average node voltage deviation, minimizing curtailment of solar power, minimizing electricity purchase cost, minimizing interruptible load compensation cost, and minimizing grid loss. The typical operating scenario set of S1 is configured as the basis for the two-layer multi-objective optimization model to consider the uncertainty of source and load, and the joint operating model of S2 is configured to define the inherent rules of source, load, and storage operation interaction and system reliability in the two-layer multi-objective optimization model. S4. Solve the two-layer multi-objective optimization model constructed in S3 to obtain the collaborative optimization configuration scheme.

2. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 1, characterized in that, In S1, the basic dataset constructed is as follows: ； Where N represents the set of nodes in the mountainous power distribution network; L represents the set of branches; Indicates the branch impedance; and These represent the active and reactive loads of node i during time period t, respectively. This represents the node voltage of node i during time period t; Indicates the maximum capacity of the branch; and These represent the sets of candidate access nodes for photovoltaic and energy storage, respectively. This represents the light intensity at node i during time period t; Indicates time-of-use electricity pricing; Indicates the on-grid electricity price; and These represent the costs of photovoltaics and energy storage, respectively. The mountainous terrain constraint index represents node i. The mountainous terrain constraints include altitude, slope, aspect, shading, road accessibility, construction and transportation conditions, and communication conditions. The set of photovoltaic candidate access nodes The following steps were used to filter the results: Determine whether a candidate node meets the constraints for mountainous area construction. If candidate node i meets the following conditions, then retain it in the set of photovoltaic candidate access nodes. middle: ； in, This represents the slope of the area where node i is located. Indicates the maximum allowed slope for access; Indicates the occlusion coefficient. Indicates the maximum allowable occlusion coefficient; Indicates the construction transportation distance or road accessibility index. This indicates the maximum construction transport distance or road accessibility threshold that is allowed for access; This indicates the additional costs of construction in mountainous areas. This indicates the maximum additional construction cost allowed for access in mountainous areas.

3. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 1, characterized in that, The typical operating scenario set is obtained by sampling and scenario reduction based on time-series samples of photovoltaic power output and load demand; The time-series sample of photovoltaic power output and load demand is constructed through the following steps: Based on historical solar irradiance and load data, a photovoltaic power output model and a load demand model are established; wherein, the photovoltaic active power output of node i in time period t is... for: ； in, This represents the light intensity at node i during time period t. Indicates the conversion efficiency of photovoltaic modules. This represents the photovoltaic installation area or equivalent installed capacity of node i; Load demand of node i in time period t for: ； in, This represents the baseline load of node i in time period t. This represents the amount of random load disturbance. Combine photovoltaic output and load demand into a source-load random variable vector: ； in, Let t represent the source-load random variable vector for time period t, and n represent the number of distribution network nodes.

4. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 1, characterized in that, The process of performing scene reduction on the initial scene set includes: Cluster the initial scene set to reduce its size; where the i-th cluster... Sum of squared intra-cluster distances for: ； Where x represents a scene sample; Represents the relationship between sample x and cluster center The Euclidean distance; Cluster Cluster centers; When cluster Decomposed into two sub-clusters and At that time, the decrease in the sum of squared distances for: ； in, and They represent respectively by The decomposition yields two subclusters; Cluster The reduction in the sum of squared distances after decomposition; When cluster Removed and merged into the nearest cluster At that time, the increase in the sum of squared distances for: ； in, This indicates the clusters to be removed; Cluster Cluster centers Indicates and The most recent cluster; like Then execute cluster Decomposition and clustering The process involves merging clusters, repeatedly performing cluster decomposition, cluster merging, and cluster center updates until the cluster centers stabilize, resulting in a reduced set of typical operating scenarios. : ； in, This represents the k-th typical operating scenario; Probability of occurrence of each typical operating scenario for: ； Where, N k M represents the number of original scenarios in the k-th cluster, and M represents the number of reduced typical running scenarios.

5. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 1, characterized in that, The joint operation model includes: The photovoltaic power output model is used to calculate the active power output of photovoltaics in each time period and scenario based on the irradiance, photovoltaic installation capacity and photovoltaic module conversion efficiency in each typical operating scenario. The load response model is used to classify node loads into critical loads, general loads, and interruptible loads, and to determine the actual load and the corresponding interruptible load compensation cost based on the interruptible load response ratio. An energy storage operation model is used to describe the continuous transfer of the energy storage state of charge with charge and discharge power, and to define the charge and discharge power limits of energy storage and the mutual exclusion relationship between charge and discharge. The energy management model is used to establish node-level active power balance equations that describe the balance between photovoltaic output, energy storage charging and discharging power, main grid power purchase and sales, load demand, and photovoltaic power reduction under each typical operating scenario. A power supply reliability model is used to take the probability of load failure as the reliability index of the source-load-storage combined system and to establish corresponding reliability constraints.

6. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 5, characterized in that, The construction of the photovoltaic power output model includes: The irradiance, photovoltaic installation capacity, and photovoltaic module efficiency of each candidate node are mapped to the photovoltaic active power output at each time period; for the photovoltaic output of node i at time period t under a typical operating scenario s. for: ； in, This represents the illumination intensity of node i in time period t and scene s; Indicates the conversion efficiency of photovoltaic modules; This represents the photovoltaic installation area or equivalent installed capacity of node i; Total photovoltaic output of the system under time period t and scenario s for: ； in, This represents the set of selected photovoltaic candidate access nodes; The construction of the load response model includes: The node load is divided into critical load, general load, and interruptible load, and the response boundaries of each type of load are determined; whereby, the actual load of node i in time period t under scenario s is... for: ； in, Indicates general load; Indicates interruptible load; Indicates the proportion of interruptible load response; Interruptible load response ratio satisfy: ； in, This indicates the upper limit of the interruptible load response ratio for node i; The cost of interruptible load compensation is: ； in, This represents the interruptible load compensation cost under scenario s; T represents the total number of time periods. Represents the set of load nodes; This represents the unit interruption compensation price for time period t; Indicates the unit scheduling duration; The construction of the energy storage operation model includes: Establish an energy storage charge state transition model, charge / discharge power constraints, and charge / discharge mutual exclusion constraints; wherein, the energy storage charge state satisfies: ； in, and These represent the state of charge of node i's energy storage in time period t and time period t+1, scenario s, respectively; and These represent charging efficiency and discharging efficiency, respectively. and These represent charging power and discharging power, respectively. Indicates the rated capacity of energy storage; Indicates the unit scheduling duration; The energy storage state of charge constraints are: ； in, and These represent the lower and upper limits of the energy storage's state of charge, respectively. The energy storage charging and discharging power constraint is: ； ； in, and These represent the charging state variable and the discharging state variable, respectively. and These represent the maximum charging power and the maximum discharging power, respectively. The charge / discharge mutual exclusion constraint is: 。 7. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 5, characterized in that, The construction of an energy management model integrating photovoltaics, energy storage, and the main grid includes: Establish node-level active power balance relationships for each typical operating scenario; where the active power balance constraint of node i in time period t and scenario s is expressed as: ； in, Indicates photovoltaic power output; Indicates the energy storage discharge power; This represents the power purchased from the mainnet and allocated to node i; Indicates the actual load; Indicates the energy storage charging power; Indicates the reduction in photovoltaic power; Indicates internet access power; The construction of the power supply reliability model includes: The probability of load failure is used as a reliability indicator for the source-load-storage combined system; where the load failure power in time period t under scenario s is... for: ； in, Indicates the total system load demand; Indicates the total output of photovoltaic power; Indicates the total discharge power of the energy storage; Indicates the power purchased from the main grid; Probability of load loss for: ； in, This represents a set of typical operating scenarios; Let represent the probability of scenario s occurring; T represents the total number of time periods. And it meets the following reliability constraints: ； in, This indicates the preset reliability threshold.

8. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 1, characterized in that, The decision variables for the two-level multi-objective optimization model are: in, Represents the vector of optimization decision variables; and These are 0-1 variables representing whether photovoltaic and energy storage are connected to node i, with a value of 1 indicating connection. This represents the photovoltaic installation area or equivalent installed capacity of node i; This represents the rated energy storage capacity of node i; This represents the rated energy storage power of node i; Indicates the proportion of interruptible load response; Indicates the proportion of electricity purchased from the main grid; The construction of the outer planning objective includes: The outer planning objectives are at least two of the following: maximizing the local photovoltaic (PV) grid integration rate, minimizing the system's total life-cycle net present value (NPV) cost, and maximizing the investment return of the PV-storage integrated system; among these, the local PV grid integration rate... for: ； in, This represents a set of typical operating scenarios; Let represent the probability of scenario s occurring; T represents the total number of time periods. This represents the photovoltaic power absorbed by local load; This represents the photovoltaic power absorbed by energy storage. Indicates the total output of photovoltaic power; Indicates the unit scheduling duration; System lifecycle net present value cost for: ； in, and These represent the investment costs for photovoltaic (PV) and energy storage, respectively. Indicates operating and maintenance costs; Indicates the cost of replacing energy storage; This indicates the cost of purchasing electricity from the main grid; This indicates the cost of interruptible load compensation; This indicates revenue from surplus electricity sold to the grid; Represents residual value; Investment returns of photovoltaic-storage integrated systems for: ； in, Indicates subsidy income; This indicates the savings achieved by reducing electricity purchases from the main grid; The construction of the inner runtime target includes: The inner-layer operational objective is to minimize at least two of the following under typical operating scenarios: average node voltage deviation, curtailment of solar power, electricity purchase cost, interruptible load compensation cost, and grid loss; where the average node voltage deviation is the most important factor. for: ； in, This represents a set of typical operating scenarios; This represents the probability of scenario s occurring; T represents the number of nodes; T represents the total number of time periods; N represents the node set. This represents the voltage of node i in time period t and scenario s; Indicates the reference voltage; Wasted light for: ； in, Represents the set of candidate photovoltaic access nodes; Indicates the power of light discarded; Indicates the unit scheduling duration; Electricity purchase cost for: ； in, This represents the time-of-use electricity price for period t; Indicates the power purchased from the main grid; Network loss is represented as: ； in, L represents the network loss under time period t and scenario s; L represents the set of branches. Represents the resistance of branch (i,j); and These represent the active power and reactive power of the branch, respectively.

9. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 1, characterized in that, The constraints of the two-layer multi-objective optimization model include power flow balance constraints, node voltage constraints, branch capacity constraints, photovoltaic and energy storage configuration capacity constraints, as well as the operation constraints and reliability constraints included in the joint operation model.

10. The method for coordinated optimization of power generation, load, and storage in mountainous distribution networks as described in claim 1, characterized in that, In S4, the process of solving the two-level multi-objective optimization model includes: The decision variables are encoded to generate candidate configuration schemes; For each candidate configuration scheme, based on the joint operation model built by S2, under the typical operation scenario set built by S1, the corresponding joint operation state is simulated and calculated, and the objective function values ​​and constraint violation degrees of each item are evaluated. The candidate configuration schemes are iteratively updated and globally searched using an intelligent optimization algorithm. Pareto nondominated sorting is used to select the set of nondominated solutions that satisfy all constraints from all candidate configurations; Based on the fuzzy membership method, a compromise optimal solution is selected from the set of non-dominated solutions; The final collaborative optimization configuration scheme is determined based on the compromise optimal solution.

Images