Port virtual power plant energy storage double-layer planning method considering demand response mechanism

Abstract

The invention provides a port virtual power plant energy storage double-layer planning method considering a demand response mechanism. The port virtual power plant energy storage double-layer planning method comprises the steps of 1, collecting operation data in a port area and topological structure information of an energy system; step 2, carrying out dimension reduction processing on the photovoltaic output and load data, and constructing a typical operation scene set with representativeness through a clustering method; 3, constructing a double-layer optimization model of the port virtual power plant; 4, operating scenes under different photovoltaic permeability are simulated, and the influence of the operating scenes on energy storage configuration and system flexibility is evaluated; 5, converting the port virtual power plant double-layer planning model into a single-layer optimization model by adopting an upper and lower layer association unification method, and calling a solver to solve the single-layer optimization model; and step 6, outputting a port virtual power plant energy storage planning scheme and related data. The invention provides a port virtual power plant energy storage double-layer planning method considering a demand response mechanism. The port virtual power plant energy storage double-layer planning method is used for improving the operation flexibility of a port multi-source energy system and the consumption level of renewable energy sources.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of virtual power plant scheduling and port comprehensive energy system planning, and particularly relates to a port virtual power plant energy storage double-layer planning method considering a demand response mechanism. BACKGROUND

[0002] With the continuous promotion of the "double carbon" strategy, as the core hub of the intersection of energy consumption and logistics, ports are accelerating the process of green and low-carbon transformation. In recent years, the penetration rate of renewable energy such as photovoltaic in the port energy system has gradually increased, and the "source-grid-load-storage" collaborative energy ecology has been gradually built in the port area. However, due to the complex composition of port load and the high uncertainty of renewable energy output, the existing energy planning and operation strategy has been difficult to meet the development needs of flexibility, economy and reliability.

[0003] The port energy system load presents significant heterogeneity. The shore power system needs to cope with the high power load impact caused by the uncertain berthing of ships, and the charging behavior of electric heavy trucks has strong impulsiveness and suddenness, while the port equipment and lighting load is relatively stable but has periodic regularity. The overlap and superposition of different types of loads in the time dimension make the overall power curve of the port area fluctuate sharply, which puts higher requirements on the energy dispatching system. At the same time, photovoltaic output is significantly affected by weather conditions, and the uncertainty of its output and the natural mismatch between the port rigid load make it difficult to fully release the utilization efficiency of renewable energy.

[0004] Under this background, energy storage system, as an important means to enhance system regulation ability and promote renewable energy consumption, has attracted widespread attention. Although the current port has initially carried out the planning and deployment of energy storage system, most methods still use static single-layer optimization model, which cannot effectively couple demand response resources and consider the long-term impact of dynamic evolution of photovoltaic penetration rate on the overall economy of the system. This traditional mode often leads to over-provisioning of energy storage systems, low operating efficiency, and difficulty in achieving precise coordination and value maximization of resources when facing high fluctuation and high heterogeneity of port energy systems.

[0005] In addition, most existing planning methods focus on capacity configuration itself, ignoring the flexibility value of energy storage systems in actual operation, and lack of systematic modeling and optimization of the internal relationship between "long-term planning-short-term operation". Although some schemes introduce uncertainty factors modeling, there are still certain limitations in scene generation, model structure and solving algorithm, which are difficult to support the high-quality development needs of port virtual power plants (VPP) in the future changing environment.

[0006] In summary, the energy storage planning problem of the port virtual power plant is not only a simple technical configuration problem, but also a complex systematic problem covering resource identification, mechanism design, scenario modeling and strategy coordination. This also puts higher requirements on the planning method: not only to have the modeling capability of high uncertainty and multi-objective, but also to be able to link the long-period capacity configuration and the short-period real-time scheduling strategy, to realize the dynamic optimization and global coordination among'source-load-storage'. SUMMARY

[0007] The present application proposes a port virtual power plant energy storage double-layer planning method considering demand response mechanism, which is used to improve the operation flexibility of port multi-source energy system and the consumption level of renewable energy, solve the current problems of port energy storage configuration redundancy, insufficient load response and operation scheduling fragmentation, and promote the deep integration of smart port and new power system.

[0008] In view of the complex characteristics of port load type diversity, power fluctuation and unstable photovoltaic output, the present application constructs a double-layer optimization framework based on typical operation scenarios, realizes the investment planning of energy storage capacity and power in the upper layer, and formulates the scheduling strategy considering the coordinated operation of demand response and energy storage in the lower layer, which can effectively cope with the problems of port power supply and demand imbalance, renewable output uncertainty and power market price fluctuation.

[0009] The present application is a port virtual power plant energy storage double-layer planning method considering demand response mechanism, and the specific steps are as follows:

[0010] Step 1. Collect photovoltaic output time series data, shore power system load, electric heavy truck charging load and other operation data in the port area, and topological structure information of the energy system;

[0011] Step 2. Perform dimensionality reduction processing on the photovoltaic output and load data, and construct a set of representative typical operation scenarios through clustering method;

[0012] Step 3. Construct a double-layer optimization model of the port virtual power plant, the upper layer takes minimizing the energy storage system investment and the system comprehensive operation cost as the target, the decision variables include the rated power, capacity and site selection strategy of the energy storage system; the lower layer takes minimizing the typical day operation cost as the target, formulates the scheduling strategy, and coordinates the energy storage charging and discharging, interruptible load management and price-type demand response mechanism;

[0013] In the demand response aspect, the interruptible load and elastic charging load modeling are introduced, the user satisfaction, price elasticity coefficient and interruption time constraint are considered, and the adjustable load resource response model facing the typical load characteristics of the port is constructed;

[0014] Step 4. Simulate the operation scenarios under different photovoltaic penetration rates, and evaluate the influence of the same on energy storage configuration and system flexibility;

[0015] Step 5. Use the unified upper and lower layer association method to convert the port virtual power plant two-level programming model into a single-level optimization model, and call the solver to solve it;

[0016] Step 6. Output the port virtual power plant energy storage planning scheme and related data.

[0017] Compared with the prior art, the beneficial effects are:

[0018] By constructing an integrated "planning-dispatching" two-layer optimization model, the system coordination capability and resource allocation efficiency of the port virtual power plant under the conditions of a high proportion of renewable energy access are significantly improved. Especially when facing complex load structures and uncertain output scenarios, the proposed model can fully tap the demand response potential, effectively reduce redundant energy storage investment, and improve the level of renewable energy absorption and system operation economy. Compared with existing technologies, this invention takes into account operational dynamics, planning foresight, and algorithm efficiency, providing an innovative solution for the intelligent and low-carbon operation of the new generation of port integrated energy systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a net load curve diagram of the port virtual power plant under different photovoltaic penetration rates of the present invention;

[0020] Figure 2 This is a flow chart of a two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to the present invention;

[0021] Figure 3 A hierarchical clustering dendrogram using a hierarchical clustering algorithm according to the present invention;

[0022] Figure 4 This is a diagram of the double-layer optimization model of the present invention;

[0023] Figure 5 The flow chart of the two-level programming model is iteratively solved by a hybrid optimization algorithm combining the particle swarm optimization algorithm and the second-order cone programming of the present invention. DETAILED DESCRIPTION

[0024] The following describes in detail a specific implementation of a two-tier planning method for energy storage of a port virtual power plant taking into account a demand response mechanism in conjunction with the accompanying drawings.

[0025] Flexibility analysis and flexibility resources of port virtual power plants

[0026] Distributed Generation (DG) penetration refers to the ratio of DG installed capacity to peak load. For port virtual power plants with high penetration, their net load curves also exhibit typical "duck-shaped" characteristics. With the increasing penetration of renewable energy sources such as photovoltaics, net load fluctuations intensify, affecting the regulation capacity and operational stability of virtual power plants, significantly increasing the demand for flexible resources.

[0027] In order to analyze the impact of different photovoltaic penetration rates on the operation of port virtual power plants, this paper takes the typical port load curve as the basis and gradually increases the photovoltaic penetration rate (that is, gradually connects distributed photovoltaics to the preset nodes of the virtual power plant system) to obtain the net load curves under different photovoltaic penetration rates, as shown in the following figure: Figure 1 shown.

[0028] from Figure 1 It can be seen that as the photovoltaic penetration rate in the port area increases, the net load curve presents the following characteristics:

[0029] 1) The net load trough of the port virtual power plant has gradually shifted from the late night hours to the midday hours, when photovoltaic power generation is high, creating a dual-trough pattern of "midday trough and nighttime peak." The higher the photovoltaic penetration rate, the more dramatic the net load fluctuations. If photovoltaic capacity continues to expand without effective load regulation or energy storage buffering mechanisms, the system could easily experience a negative net load, meaning that photovoltaic output exceeds the real-time load, leading to curtailment and affecting renewable energy utilization.

[0030] 2) With the increase in the proportion of photovoltaic access, the net load change rate has increased significantly, the difficulty of regulating the system at the hourly level has increased, and the operational flexibility of the port virtual power plant has been challenged, which is manifested in problems such as shortened regulation response time and reduced tolerance for load forecast errors.

[0031] 3) During the period of 8:00-3:00, the photovoltaic output increases rapidly and the net load of the system decreases rapidly. At this time, if the adjustment resources are insufficient or the response is not timely, the port virtual power plant may face a flexibility gap during this period; and during the period of 13:00-8:00, the photovoltaic output gradually decreases, and the port load increases, and the net load recovers rapidly, which is also likely to lead to the "load shedding" phenomenon.

[0032] The flexibility of port virtual power plants is primarily provided by a variety of regulatory resources, including energy storage systems distributed across various nodes, interruptible loads on the demand side, and the upstream main grid. In a multi-source, heterogeneous port energy system, the effective integration of flexible resources is crucial for achieving a high proportion of renewable energy consumption and ensuring stable system operation.

[0033] The calculation of the flexibility adjustment capability of energy storage devices, interruptible loads and main grid support can be expressed as:

[0034]

[0035] wherein, respectively represent the up-regulation and down-regulation ability of the energy storage system (ESS) at the port node i at time t; is the rated power of the energy storage device; is the actual output power at time t; respectively represent the minimum allowed power at this time; η is the charging and discharging efficiency; is the up-regulation and down-regulation ability of the interruptible load (CL) in the port virtual power plant at time t; represents the load state variable; is the interruptible load power of node i at time t; is the maximum up-regulation and down-regulation power provided by the upper master grid; is the maximum ramping ability of the master grid at time t; respectively represent the actual transmission power and the maximum allowed transmission power. In summary, the flexibility resources of the port virtual power plant are composed of energy storage charging and discharging, interruptible load switching, and the calling ability of the upper master grid, which jointly act to support the operation and regulation ability of the virtual power plant under complex load fluctuation and renewable energy output uncertainty.

[0036] As shown in Figure 2 , the port virtual power plant energy storage bi-level planning method considering demand response mechanism of the present application has the following specific steps:

[0037] Step 1. Collect the operation data of photovoltaic output time series data, shore power system load, electric heavy truck charging load, etc. in the port area, as well as the topological structure information of the energy system;

[0038] Step 2. Perform dimensionality reduction processing on the photovoltaic output and load data, and construct a representative set of typical operation scenarios through clustering method;

[0039] Step 3. Construct a bi-level optimization model of the port virtual power plant, the upper layer takes minimizing the investment of the energy storage system and the comprehensive operation cost of the system as the target, the decision variables include the rated power, capacity and site selection strategy of the energy storage system; the lower layer takes minimizing the daily operation cost as the target, formulates the dispatching strategy, and coordinates the energy storage charging and discharging, interruptible load management and price-type demand response mechanism;

[0040] At the demand response level, the interruptible load and flexible charging load modeling is introduced, considering the user satisfaction, price elasticity coefficient and interruption time constraint, a adjustable load resource response model facing the typical load characteristics of the port is constructed;

[0041] Step 4. Simulate the operation scenarios under different photovoltaic penetration rates, and evaluate the influence on the energy storage configuration and system flexibility;

[0042] Step 5. The port virtual power plant bi-level planning model is converted into a single-level optimization model using the upper and lower layer association unified method, and a solver is called to solve it;

[0043] Step 6. The port virtual power plant energy storage planning scheme and related data are output.

[0044] The specific method for constructing typical scenarios in step 2 is as follows:

[0045] The operation process of the port virtual power plant is affected by the fluctuation of renewable energy output and the change of multi-type load demand, with high uncertainty. In order to enhance the adaptability of the bi-level planning model to the actual operation characteristics, it is necessary to abstract the potential multiple operation scenarios in the historical data into representative typical scenarios for subsequent operation layer optimization modeling and flexibility evaluation.

[0046] Therefore, in the process of constructing typical scenarios, this paper first uses principal component analysis (PCA) to reduce the dimension of historical photovoltaic output and port load data, and extracts the main variation characteristics. Then, hierarchical clustering algorithm is applied to cluster the feature samples after dimension reduction, generating several representative typical scenarios. This method not only ensures the representativeness of scenario construction, but also effectively reduces the model solving dimension.

[0047] In the port virtual power plant, the dimension of historical photovoltaic and load data is high, and direct clustering processing may lead to large calculation amount and poor clustering quality. As a classic linear dimension reduction tool, principal component analysis can compress high-dimensional operation data and extract features by retaining the most representative principal component information in the original data.

[0048] Let the historical data matrix be X ∈ R m×n , where m is the number of samples, and n is the feature dimension. The PCA dimension reduction process is as follows: the sample matrix X is decentered to obtain the standardized matrix X * ; the covariance matrix R is calculated and eigenvalue decomposition is performed to obtain the eigenvalue λ k corresponding to the eigenvector v k ; the first p principal components are selected to form the transformation matrix W by sorting the eigenvalues from large to small; and the sample features after dimension reduction Y = XW are calculated.

[0049] Through the above dimension reduction operation, the redundant information in the original photovoltaic and load data can be effectively compressed, while the main variation trend is retained, laying a foundation for subsequent clustering analysis.

[0050] To further extract representative operating states, this paper adopts hierarchical clustering algorithm to classify the reduced samples. The hierarchical clustering tree diagram is shown in Figure 3 For the convenience of description, part of the graph is numbered without actual meaning. This method iteratively aggregates the distance between samples and does not depend on the preset cluster number, which can naturally reflect the similarity structure between samples.

[0051] The specific operation is as follows: each sample is regarded as an initial class, and a distance matrix is constructed; the distance d ij between any two samples is calculated ab(l) ; the two classes with the closest distance are merged, the sample categories are updated and the distance matrix is reconstructed; the iteration is repeated until the target class number or distance threshold is reached, and finally a clustering tree (as shown in Figure 3 ) is formed. Through this method, a typical operating scenario set can be effectively constructed while preserving the main load / output variation characteristics. Each scenario is represented by its corresponding occurrence probability p s , photovoltaic central output B PV,s and load center curve B L,s , which is described as follows:

[0052] Φ={p s ,[B PV,s ,B L,s ]∣s=1,2,...,N s}

[0053] Where Ns represents the total number of typical scenarios. The above scenario set will be used for response scheduling of various resources in the lower layer operating model and system flexibility evaluation, ensuring that the port virtual power plant has good operating adaptability under different uncertainty conditions.

[0054] A two-level optimization model of port virtual power plant is constructed. The upper layer minimizes the investment and comprehensive operating cost of the energy storage system, and the decision variables include the rated power, capacity and site selection strategy of the energy storage system. The lower layer minimizes the daily operating cost to develop a scheduling strategy, and coordinates the charging and discharging of energy storage, interruptible load management and price-based demand response mechanism.

[0055] The specific method of the adjustable load resource response model in step 3 facing the typical load characteristics of the port is as follows:

[0056] Demand response refers to the adjustment of electricity consumption by users through price information or incentive mechanisms. This paper considers interruptible load in price-based demand response and incentive-based demand response. Price-based demand response model Price-based demand response affects user electricity consumption through electricity price, which can shave peak load, promote photovoltaic power consumption, and improve the stability of distribution network. The degree of user response to electricity price is usually described by demand elasticity coefficient:

[0057]

[0058] In the formula:

[0059]

[0060] In the formula, e ii is the self-elasticity coefficient; e ij is the mutual elasticity coefficient; are the electricity prices at the i, j time before implementing demand response; c i , c j are the electricity prices at the i, j time after implementing demand response; q i are the electricity consumptions before and after the price response.

[0061] Divide a day into 24 hours, then the elasticity coefficient matrix E is expressed as:

[0062]

[0063] The electricity consumption change matrix after price change is expressed as:

[0064]

[0065] The user electricity consumption change after price change is:

[0066]

[0067] The user electricity consumption after price change is:

[0068]

[0069] Incentive-based demand response only considers interruptible load. Interruptible load is mainly aimed at large industrial users, which can reduce or interrupt part of the load in response to the signal request of dispatching department in the case of load peak or system failure. Its response characteristics are as follows.

[0070] 1) Interruptible capacity constraint

[0071]

[0072] In the formula, are the upper and lower limits of the interruptible load interruption capacity at node i.

[0073] 2) Interruption frequency constraint

[0074]

[0075] where v i,t denotes whether the interruptible load is interrupted at time t, N CL is the maximum interruption frequency of the interruptible load in a cycle.

[0076] 3) Maximum interruption time constraint

[0077]

[0078] where is the maximum interruption time of the interruptible load at node i.

[0079] The bi-level optimization model for constructing the port VPP in S3 is as follows:

[0080] Traditional energy storage planning for port energy systems often ignores the uncertainty of the operation and dispatching stage, and places too much emphasis on static configuration objectives, resulting in low utilization of energy storage systems, prolonged investment return periods, and even redundant configurations. Therefore, introducing a bi-level modeling method with multiple operation scenarios and flexible strategies in the planning stage is an effective means to improve the economic efficiency and coordination of the system.

[0081] The "planning-operation" bi-level optimization structure in the port VPP not only has a clear hierarchical target division, but also can systematically coordinate the coupling relationship between capacity configuration and operation strategy. The bi-level optimization model constructed in this paper is as follows: Figure 4 In the upper planning layer, the objective is to minimize the comprehensive cost of the port energy system, and the decision variable is the site selection and capacity configuration of the energy storage system. In the lower operation layer, based on the planning scheme, combined with renewable power output and load scenarios, the objective is to minimize the operation cost, and the multi-resource such as energy storage, interruptible load, demand response, and electricity price strategy is coordinated to optimize the specific operation strategy. Through information interaction between the two layers, the port VPP realizes coordinated operation under multiple scenarios.

[0082] The upper objective is to minimize the comprehensive investment and operation cost of the port VPP:

[0083] min C Total = C I + C O

[0084] where C I represents the energy storage investment cost, and C O represents the annual total operation cost.

[0085] Energy storage investment cost:

[0086]

[0087] Ω ESS Set of optional deployment nodes for energy storage; Energy storage capacity and power at node i; c e , c p Unit capacity / power investment cost.

[0088] Total operating cost:

[0089]

[0090] Ω Energy storage operating and maintenance cost; C Loss Internal line loss cost in port area; C CL Interruptible load incentive cost; C Grid Cost of purchasing electricity from the main grid; C FL Load response insufficient penalty; C Carbon Carbon emission penalty cost;

[0091] Energy storage configuration boundary constraints:

[0092]

[0093] Ω Charging and discharging power of the i-th energy storage device at the current time, Minimum charging and discharging power allowed for the i-th energy storage device, Maximum charging and discharging power allowed for the i-th energy storage device (positive value), Energy (electricity) state of the i-th energy storage device at the current time, Minimum allowed energy capacity of the i-th energy storage device (to avoid over-discharge), Maximum energy capacity of the i-th energy storage device (i.e., total capacity of the energy storage battery).

[0094] To simulate the impact of different photovoltaic access ratios on the operation strategy of the virtual power plant, the constraint relationship between the distributed photovoltaic installed capacity and the peak load of the port area is introduced:

[0095]

[0096] Ω PV Optional installation node for photovoltaic; α Photovoltaic installed capacity; α is the photovoltaic penetration rate; Load peak.

[0097] The lower operation layer aims to minimize the port virtual power plant operation cost, and establishes an optimization model based on the decision variables of the charging and discharging power of the energy storage system, interruptible load response, electricity price response, etc. The objective function is as follows:

[0098]

[0099] In the formula, is the operation and maintenance cost of the energy storage system; C Loss is the port line loss cost; C CL is the interruptible load compensation cost; C Grid is the electricity purchase cost from the upper-level power grid; C FL is the penalty cost caused by insufficient load response penalty.

[0100] To depict the charging and discharging process of the energy storage system, the following constraints must be met:

[0101]

[0102] SOC i,s,0 = SOC i,s,24 ,

[0103] SOC i,min ≤ SOC i,s,t ≤ SOC i,max .

[0104] Among them, represents the charging / discharging power of the energy storage at the scene s, time t, and port node i, is the charging and discharging power of the energy storage device at i at the current time; SOC i,s,t represents the state of charge of the energy storage at time t, and η is the charging and discharging efficiency; SOC i,max and SOC i,min are the upper and lower limits of the state of charge, respectively; SOC i,s,0 = SOC i,s,24 represents that the state of charge of the energy storage at the beginning and end of the day is the same (i.e. day-to-day cycle).

[0105] In the price-type demand response, users will adjust the power consumption mode according to the change of real-time electricity price, which can be modeled by a demand elasticity matrix to ensure that the total energy remains unchanged before and after the response and takes into account user satisfaction. Interruptible load mainly targets loads in the port area that have certain flexibility but do not affect core production, and meets the following interruption frequency and duration limits:

[0106]

[0107] Among them, v i,t represents whether the interruptible load is interrupted at time t, N CL is the maximum interruption times of the interruptible load, is the maximum interruption duration, and T represents the total number of discrete time steps in a complete scheduling cycle.

[0108] The upper main grid provides or absorbs power to the port virtual power plant, and must meet the following ramp and power upper and lower limits:

[0109]

[0110] Among them, P grid,s,t Instantaneous active power exchange between the port virtual power plant and the upper main grid at scene s and time t, Indicates the maximum climbing capacity of the upper power grid. is the maximum transmission power.

[0111] Photovoltaic output of each node in the port area Limit to ensure it is within the rated range:

[0112]

[0113] in, The theoretical maximum output of the photovoltaic array at node i under given environmental conditions is: Point i represents the installed photovoltaic capacity (or installed capacity percentage).

[0114] The DisFlow power flow model is used to calculate and constrain the power flow distribution within the port virtual power plant, mainly including the active power balance equation, reactive power balance equation, and voltage and current safety boundaries. The following formula is the second-order cone relaxation form of this model:

[0115] ∑ i∈AL(:,j) (P ij,s,t -r ij I ij,s,t )=Σ k∈AL(j,:) P jk,s,t +P j,s,t ,

[0116] Σ i∈AL(:,j) (Q ij,s,t -x ij I ij,s,t )=Σ k∈AL(j,:) Q jk,s,t +Q j,s,t ,

[0117]

[0118] Among them, i∈AL(:,j) represents all upstream branches (parent node set) pointing to node j, k∈AL(j,:) represents the downstream branches (child node set) from node j, P j,s,t , Q j,s,tP ij,s,t , Q ij,s,t Power flow of line (i, j), r ij , x ij Resistance and reactance of line; I ij,s,t , U i,s,t Line current square and node voltage square, respectively.

[0119] System current and node voltage need to meet the safety range:

[0120]

[0121] Where, I ij,s,t Current amplitude of line (i, j) at (s, t) (for subsequent line flow safety constraints), Rated maximum current of line (i, j), Lower / upper safety threshold of node j voltage square.

[0122] To realize the coordination of energy storage planning and operation strategy, binary variables are introduced in the model to represent whether node i is installed with energy storage, if The node energy storage power and capacity are both 0, and the charging and discharging power is also forced to be 0.

[0123] Finally, by overall processing of the upper and lower layer associated equations, the original double-layer model is converted into a single-layer mixed integer nonlinear programming problem, which can be solved in the way of linearization processing or directly calling the solver such as Gurobi:

[0124]

[0125] Where, F all (x), G all (x), H all (x), g all (x) and h all (x) represent the objective function and various types of equality and inequality constraints of the converted single-layer model, respectively. By this method, the optimal strategy of energy storage configuration and port virtual power plant daily operation can be obtained simultaneously in one solving process, which takes into account high proportion of renewable energy consumption and system comprehensive economy.

[0126] To verify the effectiveness and adaptability of the proposed demand response considering port virtual power plant energy storage bi-level planning method, a typical large-scale coastal container port is taken as the research object, and a representative system example is constructed for simulation analysis. The port has multiple types of loads such as shore power system, electric heavy truck charging station, port lighting, cold chain and mechanical equipment, and is equipped with distributed photovoltaic facilities, and plans to introduce energy storage system and load regulation mechanism, which is typical and practical.

[0127] The port energy system includes:

[0128] 1. Shore power system of ships: high-voltage shore power connection is adopted, with an average power of 2.5-5 MW per berthing ship, which is obviously intermittent;

[0129] 2. Electric heavy truck charging station: with rapid charging capability, showing sudden load fluctuation;

[0130] 3. Port infrastructure load: such as lighting, transmission machinery, cold chain equipment, etc., showing periodic fluctuation;

[0131] 4. Distributed photovoltaic: distributed on the roof of port office building and warehouse, with stable daytime output and great influence from climate;

[0132] 5. Upper main network interconnection capability: the port is interconnected with the municipal power grid in both directions, with a maximum power purchase of about 20 MW.

[0133] To depict the uncertainty characteristics of photovoltaic and load in the port system, three typical operating scenarios are generated, with three levels of photovoltaic penetration: low penetration (20%), medium penetration (40%) and high penetration (60%), to analyze the influence of different access levels on energy storage configuration and dispatching strategy. For scenarios with different photovoltaic penetration rates, the upper optimization model is used to optimize the capacity and site selection of the energy storage system. The simulation results show that:

[0134] When the photovoltaic penetration rate is 20%, the main grid is mainly relied on for power supply, and the energy storage is configured in the heavy truck charging station and the shore power hub, with a total capacity of about 3.2 MWh; when the photovoltaic penetration rate is 40%, the photovoltaic output is higher than the load demand in some periods, the energy storage configuration is significantly enhanced, with a total capacity of 5.6 MWh, which is used for peak shaving and smoothing of photovoltaic output; when the photovoltaic penetration rate is 60%, there is a serious risk of light abandonment, and the energy storage system capacity is significantly increased to 8.1 MWh, and is extended to the cold chain center and standby regulation area.

[0135] In addition, the rated power of the energy storage system is increased with the increase of photovoltaic fluctuation intensity to meet the demand of fast charging and discharging. In a typical scenario, based on the given upper-layer energy storage configuration, multi-scenario operation strategy optimization is performed, and the results in Table 1 show that in the low permeability scenario: the energy storage is more used for peak shaving and price arbitrage, and the adjustment time is concentrated in the price fluctuation period; in the medium permeability scenario: the energy storage participates in the absorption of excess photovoltaic power and cooperates with the shore power load balance, and the photovoltaic utilization rate is increased from 76.2% to 88.5%; in the high permeability scenario: the demand response mechanism is introduced, and part of the non-critical load (such as cold chain insulation and lighting) adjusts the operation time, and the photovoltaic utilization rate is further increased to 92.3%.

[0136] Table 1. Comparison of operation cost under different photovoltaic permeability

[0137]

[0138] The results of the example show that the double-layer optimization model considering demand response proposed in the present application can effectively coordinate the operation strategy of the port virtual power plant under different photovoltaic permeability, and realize the collaborative optimization of energy storage configuration and operation scheduling. Compared with static planning or strategies not considering the demand response mechanism, the system comprehensive cost can be reduced by 10%-18%, the photovoltaic utilization rate is increased by more than 15%, and it has good engineering application prospect.

[0139] To verify the superiority of the proposed upper and lower layer associated unified solution method, a hybrid optimization algorithm combining particle swarm algorithm and second-order cone programming widely used in this field is used to iteratively solve the double-layer planning model as a comparison, and the flow chart is shown in Figure 5 . The power distribution network costs and solution time under photovoltaic permeability of 60% are calculated by using the solution algorithm and the hybrid optimization algorithm respectively, and the results are shown in Table 2.

[0140] Table 2. Solution results of different algorithms

[0141]

[0142]

[0143] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not to limit the same. Those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced by equivalents, and these modifications or replacements are within the scope of protection of the claims.

Claims

1. A two-tier planning method for energy storage of a port virtual power plant considering demand response mechanism, characterized in that: The specific steps of the two-tier planning method for energy storage of the port virtual power plant are as follows: Step 1. Collect PV output time series data, shore power system load, electric heavy truck charging load operation data, and energy system topology information within the port area; Step 2: Perform dimensionality reduction on PV output and load data, and construct a representative set of typical operating scenarios using clustering methods; Step 3. Construct a two-tier optimization model for the port virtual power plant. The upper tier aims to minimize the energy storage system investment and the overall system operating costs. The decision variables include the energy storage system's rated power, capacity, and site selection strategy. The lower layer formulates dispatch strategies with the goal of minimizing typical daily operating costs, coordinating energy storage charging and discharging, interruptible load management, and price-based demand response mechanisms; At the demand response level, we introduce interruptible load and flexible charging load modeling, consider user satisfaction, electricity price elasticity coefficient, and interruption time constraints, and build an adjustable load resource response model for typical port load characteristics. Step 4. Simulate operating scenarios at different PV penetration rates to assess their impact on energy storage configuration and system flexibility. Step 5. Use the unified upper and lower layer association method to convert the port virtual power plant two-level programming model into a single-level optimization model, and call the solver to solve it; Step 6. Output the port virtual power plant energy storage planning scheme and related data.

2. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 1, characterized in that: The specific method for constructing a typical scenario in step 2 is: Firstly, principal component analysis is used to reduce the dimensionality of historical photovoltaic output and port load data and extract the main change characteristics. Then, a hierarchical clustering algorithm is applied to cluster the feature samples after dimensionality reduction to generate several representative typical scenarios.

3. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 2, characterized in that: The principal component analysis method is used to reduce the dimensionality of historical photovoltaic output and port load data and extract the main change characteristics, specifically: Assume that the historical data matrix is ​​X∈R m×n , where m is the number of samples and n is the feature dimension; the PCA dimensionality reduction process is as follows: decentralized the sample matrix X to obtain the standardized matrix X * ; Calculate the covariance matrix R and perform eigenvalue decomposition to obtain the eigenvalue λ k The corresponding eigenvector v k ; Sort by eigenvalue from large to small, select the first p principal components to form the transformation matrix W; calculate the sample feature Y = XW after dimensionality reduction; Through the above-mentioned dimensionality reduction operation, the redundant information in the original photovoltaic and load data can be effectively compressed while retaining its main change trends, laying the foundation for subsequent cluster analysis.

4. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 3, characterized in that: The specific method of applying the hierarchical clustering algorithm to cluster the feature samples after dimensionality reduction and generate several representative typical scenes is as follows: Treat each sample as an initial class and construct a distance matrix; calculate the distance d between any two samples. ij , and calculate the inter-class distance D based on this ab(l) ; Merge the two closest classes, update the sample categories and reconstruct the distance matrix; Repeat the iteration until the target number of classes or the distance threshold is reached, and finally form a clustering tree; Through this method, a set of typical operation scenarios can be effectively constructed while retaining the main load / output change characteristics; Each scenario is classified by its corresponding occurrence probability p s 、Photovoltaic center output B PV,s and load center curve B L,s Expressed as follows: Φ={p s ,[B PV,s ,B L,s ]∣s=1,2,...,N s } Among them, Ns represents the total number of typical scenarios; the above scenario set will be used for the response scheduling and system flexibility evaluation of various resources in the lower-level operation model to ensure that the port virtual power plant has good operational adaptability under different uncertainty situations.

5. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 4, characterized in that: The specific method of the adjustable load resource response model for typical port load characteristics in step 3 is: The degree of user response to electricity prices is usually described by the demand elasticity coefficient: Where: Where, e ii is the elastic coefficient; e ij is the mutual elastic coefficient; are the electricity prices at time i and time j before implementing demand response; c i 、c j are the electricity prices at time i and j after the implementation of demand response; q i are the electricity consumption before and after the electricity price response, respectively; If a day is divided into 24 hours, the elastic coefficient matrix E can be expressed as: The electricity quantity change matrix after the electricity price changes is expressed as: The change in user electricity consumption after the electricity price changes is: After the electricity price changes, the user's electricity consumption is:

6. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 5, characterized in that: Incentive demand response only considers interruptible loads, and its response characteristics are as follows: 1) Interruption capacity constraints Where, are the upper and lower limits of the interruption capacity of the interruptible load at node i, respectively; 2) Interruption number constraints Where, v i,t Indicates whether the interruptible load is interrupted at time t, N CL The maximum number of interruptions that can be made to the load in one cycle; 3) Maximum interruption time constraint Where, is the maximum interruption time of the interruptible load at node i.

7. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 6, characterized in that: The upper planning layer in S3 aims to minimize the comprehensive cost of the port energy system, and the decision variables are the location and capacity configuration of the energy storage system; Based on the planning scheme, the lower operation layer combines renewable output and load scenarios, takes minimum operating cost as the goal, coordinates and adjusts multiple resources such as energy storage, interruptible load, demand response, and electricity price strategy, and optimizes specific operating strategies; the two layers gradually converge through information exchange to realize the coordinated operation of the port virtual power plant in multiple scenarios.

8. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 7, characterized in that: The upper-level goal is to minimize the comprehensive investment and operating costs of the port virtual power plant: minC Total =C I +C O Among them, C I represents the energy storage investment cost, C O Indicates the total annual operating cost; Energy storage investment cost: Among them, Ω ESS Energy storage optional deployment node set; Energy storage capacity and power at node i; c e 、c p Unit capacity / power investment cost; Total operating cost: in, C is the energy storage operation and maintenance cost; Loss is the line loss cost within the port area; C CL is the interruptible load incentive cost; C Grid is the cost of purchasing electricity from the main grid; C FL Penalty for insufficient load response; C Carbon Penalty costs for carbon emissions; Energy storage configuration boundary constraints: in, The charging and discharging power of the i-th energy storage device at the current moment, The minimum charge and discharge power allowed by the i-th energy storage device, The maximum charge and discharge power allowed by the i-th energy storage device, The energy state of the i-th energy storage device at the current moment, The minimum allowable energy capacity of the i-th energy storage device, The maximum energy capacity of the i-th energy storage device; In order to simulate the impact of different photovoltaic access ratios on the virtual power plant operation strategy, the constraint relationship between distributed photovoltaic installed capacity and port area load peak is introduced: Among them, Ω PV Optional installation nodes for photovoltaics; is the photovoltaic installed capacity; α is the photovoltaic penetration rate; is the peak load.

9. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 8, characterized in that: The lower operation layer aims to minimize the operating costs of the port virtual power plant and establishes an optimization model based on decision variables such as the energy storage system's charging and discharging power, interruptible load response, and electricity price response. The objective function is as follows: Where, is the operation and maintenance cost of the energy storage system; C Loss is the line loss cost in the port area; C CL Compensation cost for interruptible load; C Grid The cost of purchasing electricity from the upper power grid; C FL Penalty costs for insufficient load response penalties; To characterize the charging and discharging process of the energy storage system, the following constraints must be met: SOCIETY i,s,0 =SOC i,s,24 , SOC i,min ≤SOC i,s,t ≤SOC i,max . in, represents the energy storage charging / discharging power at the port node i at scene s and time t, is the charge and discharge power of the energy storage device at i at the current moment; SOC i,s,t It represents the state of charge of the energy storage at time t, η is the charge and discharge efficiency; SOC i,max and SOC i,min They are the upper and lower limits of the state of charge; SOC i,s,0 =SOC i,s,24 This means that the state of charge of the energy storage is the same at the beginning and end of the day; Interruptible loads are mainly for loads within the port area that have a certain degree of flexibility but do not affect core production, and meet the following interruption frequency and duration restrictions: Among them, v i,t Indicates whether the interruptible load is interrupted at time t, N CL is the maximum number of interruptions of the interruptible load, is the maximum interruption duration, T represents the total number of discrete time steps in a complete scheduling cycle; The upper main grid provides or absorbs power to the port virtual power plant, and must meet the following ramp and power upper and lower limits: Among them, P grid,s,t Instantaneous active power exchange between the port virtual power plant and the upper main grid at scene s and time t, Indicates the maximum climbing capacity of the upper power grid. is the maximum transmission power; Photovoltaic output of each node in the port area Limit to ensure it is within the rated range: in, The theoretical maximum output of the photovoltaic array at node i under given environmental conditions is: Point i is the installed photovoltaic capacity; The DisFlow power flow model is used to calculate and constrain the power flow distribution within the port virtual power plant. It mainly includes the active power balance equation, the reactive power balance equation, and the voltage and current safety boundaries. The following formula is the second-order cone relaxation form of this model: Among them, i∈AL(:,j) represents all upstream branches pointing to node j, k∈AL(j,:) represents the downstream branches from node j, P j,s,t , Q j,s,t are the active and reactive injection powers of port node j at scenario s and time t, respectively; P ij,s,t , Q ij,s,t The power flow of line (i, j), r ij 、x ij is the line resistance and reactance; I ij,s,t 、U i,s,t represent the square of line current and node voltage respectively; Both system current and node voltage must meet the safety range: Among them, I ij,s,t The current amplitude of line (i, j) at (s, t), The rated maximum current of line (i,j), Lower / upper safety threshold of the square of node j voltage; In order to achieve the coordination between energy storage planning and operation strategy, binary variables are introduced Indicates whether node i is equipped with energy storage. Then the energy storage power and capacity of the node are both 0, and the charging and discharging power is also forced to be 0.

10. A two-tier planning method for energy storage of a port virtual power plant considering a demand response mechanism according to claim 9, characterized in that: By integrating the upper and lower layer correlation equations, the original two-layer model is converted into a single-layer mixed integer nonlinear programming problem, which can be solved by linearization or directly calling the Gurobi solver: Where, F all (x), G all (x), H all (x), g all (x) and h all (x) represents the objective function of the converted single-layer model and various equality and inequality constraints respectively; through this method, the optimal strategy for energy storage configuration and daily operation of the port virtual power plant can be obtained in a single solution process, taking into account the high proportion of renewable energy consumption and the comprehensive economic efficiency of the system.

Images