Distribution network distributed photovoltaic output prediction and consumption collaborative optimization method, system and device and storage medium

By constructing a multi-timescale collaborative optimization model and improving the gray wolf optimization algorithm, the problems of insufficient prediction accuracy of distributed photovoltaic power output and poor optimization effect of grid integration were solved, realizing deep coupling between photovoltaic power output and grid integration, and improving the operation economy and safety stability of the distribution network.

CN121123975APending Publication Date: 2025-12-12GUIZHOU POWER GRID CO LTD
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202511217710.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-08-28
Publication Date
2025-12-12

AI Technical Summary

Technical Problem

In existing technologies, the accuracy of distributed photovoltaic power output prediction is insufficient and the absorption optimization effect is poor, resulting in poor economic operation of the distribution network, safety problems caused by photovoltaic power output fluctuations, and a lack of multi-timescale collaborative optimization mechanism.

Method used

A multi-timescale collaborative optimization model integrating meteorological time series characteristics, load dynamic response, and energy storage regulation capabilities is constructed. By improving the gray wolf optimization algorithm and adopting a dynamic hierarchical search mechanism and adaptive constraint processing strategy, a deep coupling between photovoltaic power output prediction and consumption is achieved.

Benefits of technology

It improved the accuracy of photovoltaic output forecasting, enhanced the absorption capacity and operational economy of the distribution network, reduced curtailment losses, and ensured the safe and stable operation of the distribution network under high-proportion photovoltaic access.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121123975A_ABST
    Figure CN121123975A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of power distribution network consumption, in particular to a power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method, system and device and a storage medium. Aiming at the problems of insufficient prediction precision and poor absorption optimization effect caused by randomness, intermittency and volatility of distributed photovoltaic power generation, a multi-time-scale collaborative optimization model fusing meteorological time sequence characteristics, load dynamic response and energy storage adjustment capability is constructed. A photovoltaic output prediction parameter is determined through a dynamic hierarchical search mechanism of an improved grey wolf optimization algorithm, a dual-drive collaborative threshold is established to realize deep coupling of prediction and consumption, and multiple operation constraints are dynamically processed based on an adaptive constraint processing strategy. According to the method, the photovoltaic consumption capability and the operation economy of the power distribution network are improved, the safety problems of voltage out-of-limit, line overload and the like are solved, the light abandoning loss is reduced, technical support is provided for high-proportion photovoltaic access to the power distribution network, and collaborative optimization of prediction precision and consumption economy is realized.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution network consumption, in particular to a power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method, system, equipment and storage medium. BACKGROUND

[0002] With the promotion of the "double carbon" goal and the acceleration of the construction of new power systems, the penetration rate of distributed photovoltaic power generation in the distribution network continues to rise. Distributed photovoltaic power generation has significant randomness, intermittency and volatility characteristics, and its large-scale access to the distribution network brings unprecedented challenges to the safe and stable operation of the distribution network. The uncertainty of photovoltaic output and the spatio-temporal mismatch of load demand are increasingly prominent, and traditional distribution network operation control strategies are difficult to effectively cope with complex operating conditions in high-proportion photovoltaic access scenarios.

[0003] The main technical problems currently faced by the distribution network photovoltaic consumption include: insufficient photovoltaic power prediction accuracy leading to large scheduling plan deviations, affecting the economic operation of the power grid; photovoltaic output fluctuations causing safety problems such as voltage out-of-limit and line overload in the distribution network; low matching degree between energy storage configuration schemes and photovoltaic output characteristics, which cannot fully play the role of energy storage regulation; lack of collaborative optimization mechanism considering multi-time scale characteristics, with independent prediction and consumption strategies, and limited optimization effect.

[0004] Existing photovoltaic output prediction methods mostly use single-time scale prediction models, failing to fully consider the nonlinear impact of weather condition changes on photovoltaic output. At the same time, traditional consumption optimization algorithms are prone to local optimization when dealing with multi-constraint and multi-objective optimization problems, and lack dynamic adaptability to distribution network operation constraints. As a new group intelligence optimization algorithm, the grey wolf optimization algorithm has the advantages of strong global search ability and fast convergence speed, but it still has limitations such as single search mechanism and insufficient constraint handling capability in the application of distribution network photovoltaic consumption. SUMMARY

[0005] In view of the problems existing in the prior art, the present application is proposed.

[0006] Therefore, the technical problem solved by the present application is: the present application aims to solve the collaborative optimization problem of low photovoltaic output prediction accuracy and poor consumption optimization effect. By improving the grey wolf optimization algorithm, a multi-time scale collaborative optimization model is constructed, which integrates weather time series characteristics, load dynamic response and energy storage regulation capacity, to realize deep coupling of photovoltaic output prediction and consumption strategy. At the same time, a dynamic hierarchical search mechanism and an adaptive constraint handling strategy are designed to effectively solve the problem of traditional optimization algorithms easily falling into local optimization under complex constraint conditions, significantly improving the photovoltaic consumption capacity and operation economy of the distribution network.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] In a first aspect, the embodiments of the present application provide a power distribution network distributed photovoltaic output prediction and consumption coordination optimization method, which comprises: obtaining photovoltaic system operation parameters and constructing a prediction and consumption coordination model; fusing meteorological time series feature extraction, load dynamic response characteristics and energy storage flexible adjustment capacity through the prediction and consumption coordination model; establishing a multi-objective optimization strategy; and forming a prediction-consumption coupling system.

[0009] Based on the multi-objective optimization strategy, the photovoltaic output prediction parameters are determined through a dynamic hierarchical search mechanism, which is used for first-level prediction.

[0010] In combination with the first-level prediction, the dynamic characteristics of photovoltaic output and consumption capacity are analyzed to obtain a double-drive coordination threshold, and second-level consumption optimization is performed.

[0011] Based on an adaptive constraint processing strategy, a time-varying comprehensive optimization strategy is obtained in combination with system operation constraints to predict and control the photovoltaic consumption capacity of the power distribution network.

[0012] As a preferred scheme of the power distribution network distributed photovoltaic output prediction and consumption coordination optimization method, the meteorological time series feature extraction, load dynamic response characteristics and energy storage flexible adjustment capacity are fused through the coordination model to establish a multi-time scale optimization mechanism, which comprises: analyzing the operation state through the multi-time scale coordination model to obtain a power and load coupling distribution diagram under the influence of photovoltaic output; determining the space-time mismatch region of photovoltaic output and load demand, and the mismatch region is a key point for energy storage adjustment.

[0013] As a preferred scheme of the power distribution network distributed photovoltaic output prediction and consumption coordination optimization method, the formation of the prediction-consumption coupling system comprises: analyzing the operation state based on the coordination model, calculating the output prediction gradient model by setting the irradiance change rate, the load fluctuation coefficient, the energy storage response correction coefficient and combining the real-time operation parameters; and analyzing the operation state based on the coordination model, calculating the consumption capacity gradient model by combining the real-time constraints through the consumption capacity influence factor, the voltage related adjustment amount and the system safety margin.

[0014] As a preferred scheme of the power distribution network distributed photovoltaic output prediction and consumption coordination optimization method, based on the multi-objective optimization strategy, the photovoltaic output prediction parameters are determined through a dynamic hierarchical search mechanism, which comprises: based on the current photovoltaic output prediction value, superimposing the weighted sum of the output difference value in the historical time period and performing residual correction; introducing a coupled acceleration term of meteorology and load; and based on a convergence direction function, judging the change trend of the prediction accuracy to obtain the photovoltaic output prediction parameters.

[0015] As a preferred scheme of the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method, wherein: the dynamic characteristics of photovoltaic output and consumption capacity are analyzed to obtain a double-drive collaborative threshold, and a second-level consumption optimization is performed, including: obtaining a weight matrix of a key node through a leader weight distribution strategy; obtaining a consumption allowable capacity matrix according to a voltage-power characteristic curve;

[0016] Based on the influence of double-matrix quantization photovoltaic output change on the consumption capacity, an output correction term is obtained;

[0017] By setting the response rate coefficient combined with the load, the enhancement effect of energy storage configuration on the system regulation capacity is quantified to obtain an energy storage collaborative correction term;

[0018] Based on the two correction terms, a double-drive collaborative threshold is obtained combined with the reference reference capacity;

[0019] Based on the double-drive collaborative threshold, different early warning coefficients are set and compared with the actual running value to perform consumption capacity prediction and optimization.

[0020] As a preferred scheme of the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method, wherein: based on an adaptive constraint processing strategy, including: based on the actual consumption capacity under different operating conditions and the theoretical maximum consumption capacity under the corresponding operating conditions, combined with the coupling index of prediction accuracy and economy, the system collaborative benefit is obtained;

[0021] At the same time, an exponential decay function is introduced to obtain the influence loss of constraint violation on system operation;

[0022] The photovoltaic-energy storage collaborative configuration model is obtained by combining the collaborative benefit and the constraint loss.

[0023] As a preferred scheme of the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method, wherein: a time-varying comprehensive optimization strategy is obtained based on the system operation constraints, including: obtaining the potential optimization space of system operation through the photovoltaic-energy storage collaborative configuration model; the comprehensive optimization strategy is obtained by the ratio of the potential optimization space to the constraint evolution rate.

[0024] In the second aspect, the embodiment of the present application provides a power distribution network distributed photovoltaic output prediction and consumption collaborative optimization system, which comprises a data acquisition module, which acquires photovoltaic system operation parameters and constructs a prediction and consumption collaborative model, and fuses meteorological time sequence feature extraction, load dynamic response characteristics and energy storage flexible regulation capacity through the prediction and consumption collaborative model, establishes a multi-objective optimization strategy, and forms a prediction-consumption coupling system;

[0025] A prediction module, based on the multi-objective optimization strategy, determines the photovoltaic output prediction parameters through a dynamic hierarchical search mechanism, which is used for first-level prediction,

[0026] In combination with the first level prediction, the dynamic characteristics of the photovoltaic output and the accommodation capacity are analyzed to obtain a double driving coordination threshold, and a second level accommodation optimization is performed.

[0027] The optimization control module obtains a time-varying comprehensive optimization strategy based on a self-adaptive constraint processing strategy in combination with system operation constraints to predict and control the photovoltaic accommodation capacity of the distribution network.

[0028] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory and a processor, and the memory stores a computer program, wherein the computer program instructions are executed by the processor to implement the steps of the distributed photovoltaic output prediction and accommodation coordination optimization method of the distribution network according to the first aspect of the present application.

[0029] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, wherein the computer program instructions are executed by the processor to implement the steps of the distributed photovoltaic output prediction and accommodation coordination optimization method of the distribution network according to the first aspect of the present application.

[0030] The present application has the following beneficial effects: the present application constructs a multi-time scale coordination optimization model integrating meteorological time sequence characteristics, load dynamic response and energy storage adjustment capacity, realizes deep coupling of photovoltaic output prediction and accommodation strategy, and solves the problems of insufficient prediction accuracy and poor accommodation optimization effect of traditional methods. The early step accurately identifies the space-time mismatch area of photovoltaic output and load demand through the multi-time scale coordination model, lays a foundation for accurate positioning of the key points of the energy storage system; the middle step realizes adaptive optimization of photovoltaic output prediction parameters based on the dynamic hierarchical search mechanism of the improved grey wolf optimization algorithm, and at the same time establishes a double driving coordination threshold to effectively balance the prediction accuracy and accommodation capacity; the late step dynamically processes multiple operation constraints such as voltage safety and line current through the self-adaptive constraint processing strategy, ensures the safe and stable operation of the distribution network under high proportion of photovoltaic access. Overall, the photovoltaic accommodation capacity and operation economy of the distribution network are improved, the light loss is reduced, and the coordination optimization of prediction accuracy and accommodation economy is realized. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0032] Figure 1 The flowchart of the distributed photovoltaic output prediction and accommodation coordination optimization method for the distribution network;

[0033] Figure 2 A computer device diagram for the distributed photovoltaic output prediction and consumption coordination optimization method of the power distribution network;

[0034] Figure 3 A radiation intensity curve and load demand curve relationship diagram for the distributed photovoltaic output prediction and consumption coordination optimization method of the power distribution network;

[0035] Figure 4 A high time resolution photovoltaic output prediction and optimization comparison diagram for the distributed photovoltaic output prediction and consumption coordination optimization method of the power distribution network;

[0036] Figure 5 A comparison diagram of the traditional grey wolf algorithm and the optimized grey wolf algorithm for the distributed photovoltaic output prediction and consumption coordination optimization method of the power distribution network. DETAILED DESCRIPTION

[0037] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0038] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be practiced in other ways different from the description, and those skilled in the art can make similar extensions without departing from the concept of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0039] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an embodiment that is separate or selectively excluded from other embodiments.

[0040] Embodiment 1

[0041] Reference Figure 1 - Figure 2 For the first embodiment of the present application, the embodiment provides a distributed photovoltaic output prediction and consumption coordination optimization method of the power distribution network, comprising,

[0042] S100: Obtain photovoltaic system operation parameters and construct a prediction and consumption coordination model, fuse meteorological time series feature extraction, load dynamic response characteristics and energy storage flexible adjustment capacity through the prediction and consumption coordination model, establish a multi-objective optimization strategy, and form a prediction-consumption coupling system;

[0043] S200: Based on the multi-objective optimization strategy, determine the photovoltaic output prediction parameters through the dynamic hierarchical search mechanism, which is used for the first level prediction,

[0044] In combination with the first level prediction, the dynamic characteristics of the photovoltaic output and the accommodation capacity are analyzed to obtain a double driving coordination threshold for the second level accommodation optimization.

[0045] S300: Based on the adaptive constraint processing strategy, a time-varying comprehensive optimization strategy is obtained by combining system operation constraints to predict and control the photovoltaic accommodation capacity of the distribution network.

[0046] It should be noted that the distributed photovoltaic power generation of the distribution network has significant randomness, intermittency and volatility characteristics. During its operation, the photovoltaic output is greatly affected by meteorological factors such as irradiance, temperature, cloud cover and the like, which makes the photovoltaic power prediction accuracy insufficient. At the same time, there is a space-time mismatch between photovoltaic output and load demand. Under the operation environment of the distribution network with a high proportion of photovoltaic access, the traditional single prediction model cannot accurately capture the nonlinear variation characteristics of the photovoltaic output, the matching degree of the energy storage configuration scheme and the photovoltaic output characteristics is low, and the energy storage cannot fully play its regulating role, resulting in frequent safety problems such as voltage out-of-limit and line overload of the distribution network. At the same time, there is a lack of a coordinated optimization mechanism considering multi-time scale characteristics, and the prediction and accommodation strategies are independent of each other, so the optimization effect is limited. Therefore, it is very important to coordinate the photovoltaic output prediction and accommodation capacity of the distribution network.

[0047] Therefore, in view of the above problems of insufficient photovoltaic power prediction accuracy and poor accommodation optimization effect, through the steps of S100-S300, a multi-time scale coordinated optimization model is constructed, which integrates meteorological time series characteristics, load dynamic response and energy storage regulation capacity. The improved dynamic hierarchical search mechanism of the grey wolf optimization algorithm is used to accurately determine the photovoltaic output prediction parameters; a double driving coordination threshold mechanism is established to realize the deep coupling of photovoltaic output prediction and accommodation capacity; through an adaptive constraint processing strategy, multiple operation constraints such as voltage safety and line current carrying capacity are dynamically processed to ensure the safe and stable operation of the distribution network. At the same time, based on multi-objective optimization solution, the prediction accuracy and accommodation economy are coordinated and optimized to provide effective technical support for the distribution network with a high proportion of photovoltaic access.

[0048] Embodiment 2

[0049] Reference Figure 1 Figure 5 The second embodiment of the present application is provided.

[0050] In this embodiment, in step S100, the meteorological time series characteristic extraction, load dynamic response characteristics and energy storage flexible regulation capacity are integrated by the coordinated model to establish a multi-time scale optimization mechanism to form a prediction-accommodation coupling system, including the following steps A1-A4:

[0051] A1: Analyze the operating state through the multi-time scale coordinated model to obtain a power and load coupling distribution diagram under the influence of photovoltaic output;

[0052] Specifically, the construction of the multi-time scale collaborative model is based on the improved grey wolf optimization algorithm and a hierarchical optimization mechanism, which divides the time scale into four levels: ultra-short term (15 minutes), short term (1 hour), medium term (4 hours), and long term (24 hours). In the ultra-short term scale, the model mainly analyzes the immediate impact of rapid changes in irradiance and temperature on photovoltaic output; in the short term scale, the model focuses on the dynamic response characteristics of the load, including the periodic changes in industrial load and the changes in electricity usage habits of residential load; in the medium term scale, the model comprehensively analyzes the adjustment effect of the charging and discharging strategy of the energy storage system on power balance; in the long term scale, the model considers the matching relationship between the all-weather photovoltaic power curve and the load demand curve.

[0053] The specific process of analyzing the operating state through the multi-time scale collaborative model includes: first, collecting real-time operating data of each node in the distribution network, including photovoltaic power generation, load demand, node voltage, line current, and other parameters; second, based on the alpha, beta, and delta three-layer search mechanism of the improved grey wolf optimization algorithm, alpha wolves are responsible for global time scale optimization, beta wolves are responsible for local optimization of medium time scale, and delta wolves are responsible for fine adjustment of short time scale; third, through a dynamic weight distribution strategy, the optimization weights of each time scale are adjusted according to the current operating conditions, the short-term scale weight is increased when the photovoltaic output fluctuates sharply, and the long-term scale weight is increased when the load changes smoothly; finally, the optimization results of each time scale are weighted and fused to generate a comprehensive operating state analysis result.

[0054] The specific implementation process of obtaining the power and load coupling distribution graph under the influence of photovoltaic output includes: establishing a power-load coupling matrix, the matrix dimension is n x m (n is the number of time periods, m is the number of nodes), and the matrix elements represent the coupling degree of power and load at a specific time period and a specific node; the coupling coefficient is obtained by training historical operating data, and the corresponding node load influence coefficient is 0.1-0.3 for every 1 MW increase in photovoltaic output; based on real-time weather forecast data, predict the photovoltaic output changes in each time period; multiply the photovoltaic output prediction value by the coupling coefficient to obtain the load influence amount of each node; draw the power and load coupling distribution graph, with time as the horizontal coordinate and node number as the vertical coordinate, and the color depth representing the coupling strength.

[0055] A2: Determine the space-time mismatch area of photovoltaic output and load demand, and the mismatch area is the key point of energy storage adjustment.

[0056] Based on the power and load coupling distribution map obtained in step A1, the specific method for determining the time-space mismatch region of photovoltaic output and load demand includes: first, calculating the power surplus or deficiency of each time period and each node, the power surplus being defined as the difference between photovoltaic output and load demand, and the power deficiency being defined as the difference between load demand and photovoltaic output; second, setting a time-space mismatch threshold, when the power surplus or deficiency exceeds 20% of the node capacity, it is determined as a mismatch region; third, using a clustering algorithm to classify the mismatch region, and classifying the time-adjacent and space-close mismatch points into the same mismatch region; and finally, determining the priority of the mismatch region according to the mismatch degree and duration, the greater the mismatch degree and the longer the duration, the higher the priority of the region.

[0057] The detailed algorithm flow of the mismatch region identification is as follows: establishing a time-space mismatch matrix E, the matrix element E(i,j) representing the mismatch degree of time period i and node j; calculating the mismatch index:

[0058]

[0059] Wherein, P pv is the photovoltaic output, P load is the load demand, and P reted is the rated capacity of the node; setting the mismatch threshold θ=0.2, when E(i,j)>θ, it is marked as a mismatch point; using the DBSCAN clustering algorithm to cluster the mismatch points, the clustering radius being set to 2 (including time and space dimensions), and the minimum clustering point number being set to 3; calculating the mismatch severity index for each clustering region:

[0060] S=∑E(i,j)×T(i,j)

[0061] Wherein, T(i,j) is the mismatch duration; determining the energy storage adjustment key point according to the severity index sorting.

[0062] A3: Analyzing the operation state based on the collaborative model, calculating through setting the irradiance change rate, load fluctuation coefficient, and energy storage response correction coefficient combined with real-time operation parameters to obtain the output prediction gradient model;

[0063] The multi-time scale optimization model is established as follows:

[0064]

[0065] In the formula, E sale,t is the photovoltaic grid-connected power at time t (kWh), πt is the time-of-use electricity price (¥ / kWh), R air is the unit environmental benefit coefficient of power generation (¥ / kWh), Q pv,t is the actual photovoltaic power generation at time t (kWh). Wherein C inv is the present value of photovoltaic investment, Com For operation and maintenance costs, P loss,t The total active power loss of the system for the time period t is respectively:

[0066]

[0067] C om = k d · S pv · T

[0068]

[0069] In the formula, λ pv is the cost of the distributed photovoltaic system, S pv is the installed capacity of the photovoltaic system, a0 is the discount rate, a0 is the life. k d is the annual operation and maintenance rate per unit capacity, and Ω is the branch set.

[0070] A4: Based on the collaborative model analysis of the operating state, the consumption capacity influence factor, the voltage related adjustment amount, and the system safety margin are combined with real-time constraints to calculate the consumption capacity gradient model.

[0071] Constraint condition modeling

[0072] 1) Voltage safety constraint

[0073]

[0074] In the formula, U b,min and U b,max are the lower limit and the upper limit.

[0075] 2) Line current carrying capacity constraint

[0076]

[0077] In the formula, I l,t is the current through the lth branch in the time period t, and l is a set of lines in the distribution network system. I l,max is the maximum allowable current.

[0078] 3) Photovoltaic capacity constraint

[0079] S pv,m,min ≤ S m ≤ S pv,m,max

[0080] In the formula, S m is the apparent power of the mth distributed PV. Among them, S pv,m,min and S pv,m,max are the lower limit and the upper limit of the apparent power of the mth PV.

[0081] 4) Energy storage collaboration constraint

[0082]

[0083] wherein E max is the maximum value of the distributed photovoltaic configuration energy storage, P charge is the charging capacity of the configuration energy storage, η charge is the charging efficiency. P discharge is the discharging capacity of the configuration energy storage in the time period t, ηdischargeis the discharging efficiency, SOC t is the dynamic equation of the energy storage SOC in the time period t, i.e., the amount of change in the state of charge of the energy storage.

[0084] In the embodiment, in step S200, the photovoltaic output prediction parameter is determined based on a multi-objective optimization strategy through a dynamic hierarchical search mechanism, the dynamic characteristics of the photovoltaic output and the accommodation capacity are analyzed, the double-drive coordination threshold is obtained, and the second-level accommodation optimization is performed, including the following steps B1-B9:

[0085] B1: based on the current photovoltaic output prediction value, superimposing the weighted sum of the output difference values in the historical time period, and performing residual correction;

[0086] Specifically, the acquisition of the current photovoltaic output prediction value is based on the output prediction gradient model established in step A3, and is calculated in combination with real-time meteorological data and system operation state. The selection of the historical time period considers the periodic characteristics of the photovoltaic output, including three levels of daily cycle (24 hours), weekly cycle (7 days) and seasonal cycle (90 days). The output difference value is calculated by using a first-order difference method, i.e., the difference value of adjacent time instants, which is used to capture the trend information of the output change.

[0087] The weight distribution of the weighted sum is determined based on the time distance and the correlation. The closer the time is, the greater the weight is, and the stronger the correlation is, the greater the weight is. The weight calculation formula is:

[0088] w(i) = exp(-λ×Δt(t))×corr(i)

[0089] wherein w(i) is the weight of the i-th historical time period, λ is the attenuation coefficient, Δt(t) is the time interval, and corr(i) is the correlation coefficient.

[0090] The residual correction adopts an adaptive filtering method, and an error correction model is established by analyzing the statistical characteristics of the prediction error. When the prediction error presents a systematic deviation, the correction parameter is automatically adjusted; when the prediction error is a random fluctuation, a smoothing filter is used to reduce the noise influence. The residual correction formula is:

[0091] P corrected = P predicted + α × error trend + β × errorsmooth

[0092] where P corrected is the corrected prediction value, P predicted is the original prediction value, error trend is the error trend term, error smooth is the error smoothing term, and a, b are correction coefficients.

[0093] B2: Introducing the coupling acceleration term of meteorology and load;

[0094] The coupling acceleration term of meteorology and load considers the influence of meteorological condition change on load demand, as well as the feedback effect of load change on photovoltaic system operation. The coupling relationship includes three aspects of temperature-load coupling, irradiance-load coupling and humidity-load coupling.

[0095] Temperature-load coupling is mainly reflected in the change of cooling and heating load. Air conditioning load increases significantly in summer when the temperature is high, and heating load increases accordingly in winter when the temperature is low. The temperature-load coupling coefficient is obtained by regression analysis of historical data. In summer, the load increases by about 3%-5% when the temperature increases by 1°C; in winter, the load increases by about 2%-3% when the temperature decreases by 1°C.

[0096] Irradiance-load coupling is mainly reflected in the change of industrial electricity and commercial electricity. High irradiance period often corresponds to high temperature weather, resulting in increased cooling load; at the same time, strong irradiance is conducive to distributed photovoltaic power generation, which may affect the electricity consumption behavior of users. The irradiance-load coupling coefficient is about 0.1%-0.2% / 100W / m 2 .

[0097] Humidity-load coupling mainly affects the electricity consumption of dehumidification equipment and ventilation equipment. Dehumidification load increases in high humidity weather, and humidification load increases in low humidity weather. The influence of humidity on load is relatively small, and the coupling coefficient is about 0.05%-0.1% / 1%RH.

[0098] The calculation formula of the coupling acceleration term of meteorology and load is:

[0099]

[0100] where γ1, γ2, γ3 are coupling acceleration coefficients, are the change rates of temperature, irradiance and humidity, respectively, are the sensitivity coefficients of load to each meteorological factor, respectively.

[0101] B3: Based on the convergence direction function, the change trend of prediction accuracy is judged to obtain the photovoltaic output prediction parameters.

[0102] The convergence direction function is designed based on the convergence characteristics of the improved grey wolf optimization algorithm, and is used to determine the direction and step size of the optimization of the prediction model parameters. The function includes three main components: convergence speed evaluation, convergence direction judgment, and parameter adjustment strategy.

[0103] The convergence speed evaluation determines the rate of change of the objective function value in consecutive iterations. When the rate of change is less than a set threshold, the convergence speed is considered to be too slow, and the search step size needs to be increased. When the rate of change is greater than a set threshold, the convergence speed is considered to be too fast, and the optimal solution may be missed, and the search step size needs to be reduced.

[0104] The convergence direction judgment is based on gradient information and historical search experience. By analyzing the change trend of the objective function in the parameter space, the most favorable search direction is determined. When the gradient direction is consistent with the historical optimal direction, the search weight of this direction is increased. When the gradient direction is opposite to the historical optimal direction, the search weight of this direction is reduced.

[0105] The parameter adjustment strategy adopts an adaptive adjustment mechanism, which dynamically adjusts the model parameters according to the current prediction accuracy and convergence state. When the prediction accuracy continues to improve, the current parameter setting is maintained. When the prediction accuracy stagnates or decreases, the parameter adjustment is triggered.

[0106] The mathematical expression of the convergence direction function is:

[0107] D i = δ1 x g c + δ2 x d h + δ3 x P f

[0108] where δ1, δ2, and δ3 are weight coefficients, g c is the current gradient direction, d h is the historical optimal direction, and P f is the performance feedback term.

[0109] It should be noted that the preferred dynamic hierarchical search mechanism is because its core logic is divided into three parts: one is the adaptive weight adjustment, which dynamically adjusts the search weight of each level parameter based on the current prediction bias, captures the change law of photovoltaic output; the meteorological-load coupling term is introduced to correct the prediction model parameters, and the prediction accuracy is improved; the convergence direction function is used to judge the effectiveness of the search direction, such as accelerating the convergence or expanding the search range, enhancing the response ability to meteorological mutations, and realizing the multi-factor coupling prediction of future multi-period photovoltaic output.

[0110] B4: Obtain the weight matrix of the key nodes through the leader weight distribution strategy;

[0111] Specifically, the leader weight allocation strategy is realized based on the social hierarchy mechanism of improved grey wolf optimization algorithm, and the nodes of the power distribution network are divided into three levels of importance: alpha, beta and delta. Alpha nodes are key hub nodes, usually including main transformer outlet ends and important load access points, etc. The failure of these nodes will have a serious impact on the system. Beta nodes are important support nodes, usually including branch line branch points and large user access points, etc. Delta nodes are general nodes, mainly end load nodes. The multi-index comprehensive evaluation method is used to identify key nodes, including four dimensions of node load importance, network topology importance, electrical distance importance and fault impact importance. The node load importance is determined based on the capacity and type of the access load, with the first-level load weight being 1.0, the second-level load weight being 0.7, and the third-level load weight being 0.4. The network topology importance is determined based on the connectivity and betweenness centrality of the node, with the connectivity of the node being high. The electrical distance importance is determined based on the electrical distance from the node to the power supply point, with the closer the distance, the greater the weight. The fault impact importance is determined based on the outage range and outage load after the node failure. The weight matrix is constructed using the analytic hierarchy process (AHP), and the weight coefficients of each index are determined through expert scoring and consistency check. The comprehensive weight calculation formula is:

[0112]

[0113] wherein W total is the node comprehensive weight, is the weight coefficient of each index, W load , W topo , W elec , and W fault are the load importance, topology importance, electrical distance importance, and fault impact importance weights, respectively. The dynamic updating mechanism of the weight matrix considers the changes in system operation state, and automatically updates the weight allocation when the system operation mode changes. The update frequency is set to once an hour, or immediately updated when important events occur (such as device commissioning, large load changes).

[0114] B5: Obtain the accommodation allowable capacity matrix according to the voltage-power characteristic curve;

[0115] Specifically, the voltage-power characteristic curve describes the relationship between the voltage of each node of the power distribution network and the photovoltaic access power, and is an important basis for determining the allowable capacity of consumption. The characteristic curve is obtained through power flow calculation and sensitivity analysis. The voltage distribution of each node under different photovoltaic access powers is calculated, and a voltage-power relationship diagram is drawn. The determination of the allowable capacity of consumption is based on the voltage quality constraint. According to the national standard GB / T12325-2008, the voltage of the node of the power distribution network should be maintained within the range of +7% to -7% of the rated voltage. When the photovoltaic access power increases and causes the node voltage to exceed the upper limit, the power value is the upper limit of the allowable capacity of consumption of the node; when the photovoltaic access power is insufficient and causes the node voltage to be lower than the lower limit, the minimum photovoltaic output constraint needs to be considered. The calculation of the allowable capacity of consumption matrix uses the step-by-step incremental method, starting from zero power and gradually increasing the photovoltaic access power, with each increment being 1% of the rated capacity. The corresponding node voltage distribution is calculated. When the voltage of any node exceeds the limit, the power increase is stopped and the current value is recorded as the allowable capacity of consumption. The matrix dimension is n x m (n is the number of nodes and m is the number of time periods), and the matrix element represents the allowable capacity of consumption of a specific node at a specific time. The calculation of the allowable capacity of consumption also needs to consider other constraint conditions, including line current carrying capacity constraint, transformer capacity constraint and system stability constraint. The line current carrying capacity constraint is determined based on the thermal stability limit of the line, and when the line current reaches 95% of the rated value, the photovoltaic access is limited; the transformer capacity constraint considers the overload capacity and voltage regulation capacity of the transformer; the system stability constraint is based on small disturbance stability analysis to ensure that the system can maintain stable operation under small disturbance.

[0116] B6: Based on the influence of the double-matrix quantized photovoltaic output change on the consumption capacity, an output correction term is obtained;

[0117] Specifically, the double-matrix refers to the key node weight matrix obtained in step B4 and the allowable capacity of consumption matrix obtained in step B5. The output correction term quantifies the influence of the photovoltaic output change on the consumption capacity of each node through matrix operation, and the operation method uses weighted matrix multiplication. The calculation formula of the output correction term is:

[0118] Δp correction = W x C x Δp pv

[0119] Δp correction , W is the key node weight matrix, C is the allowable capacity of consumption matrix, and Δp pvis the photovoltaic output change vector. The physical meaning of matrix operation is that the weight matrix reflects the contribution of each node to the overall accommodation capacity of the system, and the important node has a large weight and has a significant impact on the system; the accommodation allowable capacity matrix reflects the accommodation potential of each node, and the node with large capacity can accommodate more photovoltaic power; the photovoltaic output change vector reflects the fluctuation of photovoltaic power generation in actual operation. Through the multiplication operation of the three matrices, the quantitative results of the influence of photovoltaic output change on the overall accommodation capacity of the system are obtained. The calculation of the output correction term also considers the time correlation, and the influence degree of photovoltaic output change on the accommodation capacity is different in different time periods. During the day, the increase of photovoltaic output has a significant impact on the voltage rise; at night, the change of photovoltaic output has a smaller impact on the system. The time correlation coefficient is obtained by statistical analysis of historical operation data, and the coefficient is 1.0 during the day (8:00-18:00), 0.7 during the transition period (6:00-8:00, 18:00-20:00), and 0.3 during the night (20:00-6:00).

[0120] B7: By setting the response rate coefficient of the combined load, the enhancement effect of the energy storage configuration on the system regulation capacity is quantified to obtain the energy storage coordination correction term;

[0121] Specifically, the energy storage coordination modifier considers the role of the energy storage system in improving the photovoltaic power consumption capacity of the distribution network, including power regulation capacity, capacity support capacity, and response speed advantage. The response rate coefficient reflects the response speed of the energy storage system to changes in load and photovoltaic output, and is a key parameter for quantifying the regulation effect of energy storage. The calculation of the load response rate is based on the time constant of load change. The time constant of industrial load is generally 15-30 minutes, the time constant of commercial load is generally 5-15 minutes, and the time constant of residential load is generally 30-60 minutes. The response rate coefficient is defined as the ratio of the response time of the energy storage system to the response time of the load. The response time of the lithium battery energy storage system is generally milliseconds, so the response rate coefficient can reach 1000 or more. The enhancement effect of energy storage configuration on system regulation capacity mainly reflects in three aspects: first, power regulation capacity, the energy storage system can change the charging and discharging power within seconds to quickly smooth the photovoltaic output fluctuation; second, capacity support capacity, the energy storage system can discharge to supplement the power supply capacity during load peak, and charge to consume excess power during photovoltaic output peak; third, response speed advantage, the response speed of the energy storage system is much faster than that of traditional regulation equipment, which can effectively cope with rapidly changing operating conditions. The available state of charge range considers the operating constraints of the energy storage system. In order to protect the battery life, the lithium battery energy storage system generally operates in the 20%-90% SOC range, and the available range is 70%. The round-trip efficiency considers the energy loss in the charging and discharging process of the energy storage system. The round-trip efficiency of the lithium battery energy storage system is generally 85%-95%. The energy storage coordination modifier also needs to consider the power allocation strategy of the energy storage system, including on-demand allocation, uniform allocation, and priority allocation. On-demand allocation dynamically adjusts the energy storage power allocation according to the consumption demand of each node; uniform allocation evenly distributes the energy storage power to each node; priority allocation determines the priority of energy storage power allocation according to the importance of the node.

[0122] B8: Based on the two modifiers, the double-drive coordination threshold is obtained in combination with the benchmark reference capacity;

[0123] Specifically, the double-drive coordination threshold is the core parameter of the optimal control of photovoltaic power consumption in the distribution network. By comprehensively considering the output modifier and the energy storage coordination modifier, and combining the benchmark reference capacity, the optimal operating threshold of the system is determined. The benchmark reference capacity is determined based on the design capacity and safety margin of the distribution network, and generally takes 80%-90% of the design capacity as the reference value. The calculation of the double-drive coordination threshold uses a weighted combination method, and the weight distribution is dynamically adjusted according to the current operating state. When the photovoltaic output fluctuation is large, the weight of the output modifier is increased; when the regulation capacity of the energy storage system is sufficient, the weight of the energy storage coordination modifier is increased; when the system operation is close to the limit, the weight of the benchmark reference capacity is increased. The mathematical expression of the double-drive coordination threshold is:

[0124]

[0125] where TH coop is the dual-drive coordination threshold, ΔP cor is the reference capacity, is the energy storage coordination correction amount, and α, β, and γ are weight coefficients, and α + β + γ = 1. The dynamic adjustment of the weight coefficients is based on system operation state evaluation, and the evaluation indexes include photovoltaic output fluctuation rate, energy storage available capacity, system safety margin, etc. The photovoltaic output fluctuation rate is calculated by sliding variance, and the fluctuation rate is high, and β weight is increased; the energy storage available capacity is evaluated by the SOC state, and the available capacity is high, and γ weight is increased; the system safety margin is evaluated by N-1 check, and the safety margin is low, and α weight is increased. The dual-drive coordination threshold also needs to consider the influence of the time scale, including real-time threshold, short-term threshold and long-term threshold three levels. The real-time threshold is used for second-level adjustment control, responding to the rapid change of photovoltaic output; the short-term threshold is used for minute-level scheduling control, balancing photovoltaic power generation and load demand; the long-term threshold is used for hour-level planning control, optimizing the energy storage charging and discharging strategy.

[0126] B9: Based on the dual-drive coordination threshold, different early warning coefficients are set and compared with the actual operation value to predict and optimize the accommodation capacity.

[0127] Specifically, the setting of the early warning coefficient is based on risk assessment theory, and the system operation risk is divided into four levels of normal, attention, warning and danger, corresponding to different early warning coefficients. The normal state corresponds to the early warning coefficient 1.0, indicating that the system is running within the safe range; the attention state corresponds to the early warning coefficient 0.8, indicating that the system is close to the operation limit but still within the controllable range; the warning state corresponds to the early warning coefficient 0.6, indicating that the system operation has risks and measures need to be taken; the dangerous state corresponds to the early warning coefficient 0.4, indicating that the system operation is seriously out of limit and needs emergency treatment.

[0128] The comparison of the actual operation value and the dual-drive coordination threshold adopts real-time monitoring mode, and the monitoring frequency is once every minute. The judgment logic of the comparison result is: when the actual operation value is less than the threshold value x 1.0, the system is in normal state; when the actual operation value is between the threshold value x 0.8 and the threshold value x 1.0, the system is in attention state; when the actual operation value is between the threshold value x 0.6 and the threshold value x 0.8, the system is in warning state; when the actual operation value is greater than the threshold value x 0.6, the system is in danger state.

[0129] The accommodation capacity prediction is based on the early warning state and historical trend analysis, and the time series prediction method is used to predict the accommodation capacity change in the next 2-4 hours. The prediction model adopts ARIMA model, and the model parameters are determined by fitting historical data. When the prediction result shows that the accommodation capacity will be insufficient, the optimization measures are started in advance.

[0130] The consumption capacity optimization measures include three aspects of energy storage scheduling optimization, photovoltaic output control and load demand response. The energy storage scheduling optimization increases the charging power at the peak of photovoltaic output and increases the discharging power at the peak of load by adjusting the charging and discharging strategy of the energy storage; the photovoltaic output control limits the photovoltaic output by adjusting the power factor and active power output of the photovoltaic inverter when necessary; the load demand response guides users to adjust the power consumption behavior through price signals or direct control.

[0131] It should be pointed out that the construction form of the above-mentioned dual-drive cooperative threshold model quantifies the comprehensive influence of photovoltaic output change and energy storage cooperation on system consumption capacity through dynamic weight allocation, thereby replacing the traditional static threshold method, which can solve the threshold failure problem caused by the change of operating conditions; the fine management of the system operating state is realized through the early warning grading mechanism, and the risk change of the system operation can be captured in the dynamic evaluation of the consumption capacity, and the early warning coefficient can automatically adjust the response intensity of the system through the form of grading setting.

[0132] For example, in B9, the specific implementation of comparing different early warning coefficients with actual operating values based on the dual-drive cooperative threshold can be represented as:

[0133] If the actual output of photovoltaic exceeds 100% of the cooperative threshold, a dangerous early warning is triggered, and emergency power limiting measures are immediately started, the energy storage system is fully charged, and part of the photovoltaic power generation unit is cut off if necessary;

[0134] If the actual output of photovoltaic exceeds 80% but is lower than 100% of the cooperative threshold, a warning early warning is triggered, the energy storage charging and load demand response are started, and the power factor of the photovoltaic inverter is adjusted;

[0135] If the actual output of photovoltaic exceeds 60% but is lower than 80% of the cooperative threshold, a notice early warning is triggered, the energy storage scheduling strategy is optimized, and the adjustment margin is reserved;

[0136] If the actual output of photovoltaic is lower than 60% of the cooperative threshold, it is judged as a normal operating state, the current operating strategy is maintained, and the system state change is monitored.

[0137] In the embodiment, based on the adaptive constraint processing strategy in step S300, the time-varying comprehensive optimization strategy is obtained in combination with the system operating constraints, including the following steps C1-C4:

[0138] C1: Based on the actual consumption capacity under different operating conditions and the theoretical maximum consumption capacity under the corresponding operating conditions, the coupling index of prediction accuracy and economy is obtained to obtain the system cooperative benefit;

[0139] Specifically, the definition of different operating conditions is based on three dimensions of photovoltaic output level, load demand level and meteorological conditions, forming a multi-condition matrix. Photovoltaic output level is divided into high output (80% rated value), medium output (40%-80% rated value) and low output (<40% rated value); load demand level is divided into high load (80% peak value), medium load (50%-80% peak value) and low load (<50% peak value); meteorological conditions are divided into sunny, cloudy and rainy, forming 27 typical operating conditions.

[0140] The actual consumption capacity is determined by real-time power flow calculation, considering various constraint conditions under the current system operating state, including voltage constraint, thermal stability constraint, static stability constraint, etc. The calculation method uses continuous power flow technology, gradually increases the photovoltaic injection power until the system reaches the stability limit, and records the photovoltaic power at this time as the actual consumption capacity.

[0141] The theoretical maximum consumption capacity is determined based on the design parameters and rated capacity of the system, without considering various limiting factors in actual operation. The calculation method is the weighted sum of the rated consumption capacity of each node, and the weight is determined according to the importance of the node. The theoretical maximum consumption capacity represents the ideal consumption capacity of the system, and is the benchmark for evaluating the consumption efficiency of the system.

[0142] The coupling index of prediction accuracy and economy reflects the degree of influence of prediction accuracy on the economic operation of the system. Prediction accuracy is quantified by mean absolute percentage error (MAPE), and economy is quantified by unit consumption cost. The coupling index calculation formula is:

[0143] C idx =(1-MAPE)×C eic

[0144] Where C eic is economic efficiency, defined as the ratio of theoretical minimum cost to actual operating cost.

[0145] The calculation of system synergy benefit considers the capacity utilization efficiency and economic operation efficiency, and the calculation formula is:

[0146]

[0147] Where W fac is the condition weight factor, determined according to the frequency and importance of the condition.

[0148] C2: The exponential decay function is introduced to obtain the influence loss of constraint violation on system operation;

[0149] Specifically, the impact loss of constraint violation includes four categories: voltage excursion loss, line overload loss, frequency deviation loss and insufficient stability margin loss. The selection of the exponential decay function is based on the cumulative effect of constraint violation. The impact of slight violation is relatively small, and the impact of serious violation increases sharply.

[0150] The frequency deviation loss and the insufficient stability margin loss adopt similar exponential function forms, and the parameters are determined through historical accident analysis and expert experience. The total constraint violation loss is the weighted sum of the losses, and the weights are determined according to the severity and probability of the loss.

[0151] C3: Combine the synergy benefit and the constraint loss to obtain a photovoltaic-energy storage collaborative configuration model.

[0152] Specifically, the photovoltaic-energy storage collaborative configuration model is a comprehensive optimization objective function, aiming to maximize the system synergy benefit while minimizing the constraint violation loss. The mathematical expression of the model is:

[0153]

[0154] Wherein, Penalty is the penalty coefficient, used to balance the weight relationship between benefit and loss.

[0155] The solution of the collaborative configuration model adopts the improved grey wolf optimization algorithm, and the optimization variables include the capacity configuration, power configuration, installation location and operation strategy of the energy storage system. The capacity configuration variable represents the rated capacity of the energy storage system at each node; the power configuration variable represents the rated power of the energy storage system at each node; the installation location variable is a discrete variable, representing the candidate installation node of the energy storage system; the operation strategy variable represents the charge and discharge plan of the energy storage system.

[0156] The model constraint conditions include energy storage system technology constraints, power distribution network operation constraints and economic constraints. The technology constraints include energy storage power constraints, capacity constraints, state of charge constraints and cycle life constraints; the operation constraints include voltage constraints, line current constraints and power balance constraints; the economic constraints include investment cost constraints and operation cost constraints.

[0157] Wolf individual coding design:

[0158] Each grey wolf individual X i represents a set of photovoltaic-energy storage configuration schemes, and each grey wolf individual X i The coding includes prediction model parameters and energy storage configuration parameters:

[0159] X i =[ω1,ω2,...,ω t ,E max ,η charge ,η discharge ,τ peak ]

[0160]

[0161] where ω t is the network weight of the prediction model parameter, ΔG t is the time period rate of change of irradiance and temperature, G max and G min are the maximum and minimum values of irradiance, T max and T min are the maximum and minimum values of temperature in time period t. L t is the load fluctuation in time t, L max and L min are the maximum and minimum values of load fluctuation in time t, L avg is the average value of load fluctuation in time t. is the maximum value of distributed PV configuration energy storage. η charge is the charging efficiency, η discharge is the discharging efficiency, is the time-of-use electricity price threshold.

[0162] 2) Dynamic fitness function

[0163] F total = ω1F1+ ω2F2+ ρΣmax(0, g j (X i )) 2

[0164]

[0165] where F total represents the total objective function of the multi-time scale optimization model, F1and F2represent the objective function of each time period in the iteration, respectively. g i (X i) is the constraint violation value such as voltage out-of-limit amount, line overload amount, etc., and ρ is the adaptive penalty coefficient, which increases exponentially with the number of iterations.

[0166] 3) Improved hierarchical search mechanism

[0167] The improved hierarchical search mechanism includes dynamic weight allocation of leaders and Levy flight disturbance enhancement. The dynamic weight of the leader approaches 0 as the time period in the early iteration, which encourages the group to explore the development ability of the leader wolf in the later time period. The Levy flight disturbance is more serious, the disturbance step is smaller, in order to avoid invalid search.

[0168]

[0169] where ω α represents the proportion of dynamic weight of the leader in the early iteration, ω βThis indicates the proportion of leader development capabilities enhanced in the later stages of the iteration. Among them, violation(X) i) To constrain the total number of violations, the perturbation step size adaptively decreases as the degree of violation increases.

[0170] 4) Voltage over-limit feedback mechanism

[0171] After each location update, a three-phase unbalanced power flow calculation is invoked. If a node voltage exceeds the limit, a secondary penalty is applied, and capacity correction is triggered for the node exceeding the limit.

[0172] F total ←F total +100·Σ(U i -U lim ) 2

[0173]

[0174] In the formula, For the adjusted capacity, S pv,k This is the capacity before adjustment. U lim U is the voltage limit threshold value. i Voltage updated for position.

[0175] C4: The potential optimization space of the system operation is obtained through the photovoltaic-energy storage collaborative configuration model; the comprehensive optimization strategy is obtained through the ratio of the potential optimization space to the constraint evolution rate.

[0176] Specifically, the potential optimization space of system operation is defined as the gap between the current operating state and the optimal operating state, including three dimensions: the space for improving absorption capacity, the space for improving economic benefits, and the space for increasing operational safety margin.

[0177] The constraint evolution rate reflects how quickly system constraints change, including load growth rate, equipment aging rate, and environmental change rate. The calculation of the evolution rate is based on historical trend analysis and future forecasts, using linear regression or exponential smoothing methods to determine the changing trend.

[0178] The comprehensive optimization strategy adopts a multi-timescale hierarchical optimization method, including three levels: long-term planning strategy, medium-term scheduling strategy, and short-term control strategy. The long-term planning strategy mainly targets the capacity configuration and site selection planning of the energy storage system, with a planning period of 5 to 10 years; the medium-term scheduling strategy mainly targets the operation plan and maintenance arrangement of the energy storage system, with a scheduling period of 1 month to 1 year; and the short-term control strategy mainly targets the real-time charging and discharging control of the energy storage system, with a control period of minutes to hours.

[0179] In summary, in terms of multi-time scale collaborative model construction, by dividing the time scale into four levels of ultra-short, short, medium and long term, and combining the improved layered search mechanism of grey wolf optimization algorithm, the fine management of photovoltaic power output prediction and consumption optimization under different time scales is realized, and the problem that the traditional single time scale method cannot adapt to complex operation conditions is solved. In terms of space-time mismatch identification technology, by establishing the power-load coupling matrix and using the DBSCAN clustering algorithm, the accurate identification and hierarchical management of mismatched areas are realized.

[0180] In terms of prediction parameter optimization, embodiment 2 introduces LSTM neural network and CNN deep learning model, which performs well in dealing with rapid changes of meteorological conditions through feature extraction of multi-dimensional time series data. In terms of double-driven collaborative threshold construction, through the leader weight distribution strategy and voltage-power characteristic curve analysis, a dynamic adjustment threshold mechanism is established, which can adjust the optimization strategy in real time according to the system operation state, avoiding the limitations of static threshold method.

[0181] Embodiment 3

[0182] Reference Figure 1 - Figure 5 , which is the third embodiment of the present application.

[0183] When evaluating the high proportion of photovoltaic consumption capacity of a certain 10kV actual distribution network, comprehensive analysis needs to be carried out combining multi-dimensional data and system characteristics. The simplified topological structure of distribution network adopts radial network, containing 15 nodes and 5 feeder lines, which can clearly identify the spatial distribution of photovoltaic access point and key load node, laying the foundation for subsequent power flow analysis and constraint modeling. Figure 3 The 24-hour irradiance curve and load demand curve shown in FIG. 1 reveal the typical daily operation characteristics: the photovoltaic output peaks at 4.8MW from 12:00 to 14:00, and the load peak of 4.5MW from 18:00 to 20:00 forms a space-time mismatch, highlighting the necessity of energy storage configuration. Based on this, the total installed capacity of photovoltaic is set to 5MW, and the irradiance mutation period is marked as the boundary for algorithm robustness test.

[0184] Comparison of high time resolution (15-minute interval) photovoltaic output prediction and optimization Figure 4In the middle, the improved grey wolf optimization algorithm shows significant dynamic response ability and multi-objective synergy advantage: during the noon 12:30 irradiance peak, the algorithm accurately predicts the power of 920kW, and absorbs 130kW of excess power through energy storage charging, stabilizes the actual consumption power at 850kW, and increases the energy storage SOC from 75% to 82%, effectively inhibiting the voltage rise; At 18:30, the low point of photovoltaic output, the algorithm drives the energy storage to discharge 200kW to make up for the load gap, and the consumption power increases to 350kW, and the SOC decreases from 45% to 38%. The algorithm verifies the comprehensive performance of economic efficiency, safety and prediction accuracy in complex scenarios.

[0185] Figure 5 The comparative diagram of the traditional grey wolf algorithm and the optimized grey wolf algorithm is shown. With the increase of the number of iterations, the traditional grey wolf algorithm continues to steeply decline, and reaches the platform period, i.e. local optimum, after 40 iterations. The grey wolf optimization algorithm rapidly decreases in the early stage, and reaches the stable optimal value after 20 iterations, without significant oscillation.

[0186] Table 1 performance quantification comparison diagram

[0187] Indicator Conventional GWO Improved GWO Convergence iteration number 20 40 Final objective function value 0.58 0.46 Voltage qualification rate (%) 92.1 99.3 Calculation time-consuming (second / time) 12.3 9.8

[0188] The optimized grey wolf algorithm as shown in Table 1 reduces the number of iterations by 50% when reaching the target function value, and reduces the calculation time by 20.3%. The voltage qualification rate is improved from 92.1% to 99.3%, verifying the effectiveness of the out-of-limit repair mechanism.

[0189] The improved GWO algorithm is suitable for power distribution network distributed photovoltaic output prediction and consumption coordination optimization method. The test scene completely depicts the coupling influence of irradiance mutation, load fluctuation and grid constraint, provides a real data verification environment for the dynamic layered search, weather related disturbance and other innovative mechanisms of the improved grey wolf algorithm, and fully proves its engineering applicability in complex power distribution network. The main engineering benefits are as follows: 1) accelerate decision response: in the real-time scheduling scene, 20 iterations can generate a reliable strategy (the traditional method needs 40 iterations), which meets the minute-level optimization demand; 2) reduce resource consumption: the calculation time is reduced by 20%, which is suitable for edge computing device deployment; 3) weather related disturbance mechanism makes the algorithm still maintain stable convergence in the irradiance mutation scene; 4) dynamic weight allocation synchronously optimizes prediction accuracy and economy, avoiding strategy oscillation caused by target conflict.

[0190] Example 4

[0191] The above is a schematic scheme of a power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method. It should be noted that the technical scheme of the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization system belongs to the same concept as the technical scheme of the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method described above. The technical details of the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization system in this embodiment are not described in detail, and can be referred to the description of the technical scheme of the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method described above.

[0192] The embodiment also provides a power distribution network distributed photovoltaic output prediction and consumption collaborative optimization system, comprising:

[0193] The data acquisition module acquires photovoltaic system operation parameters and constructs a prediction and consumption collaborative model, fuses meteorological time series feature extraction, load dynamic response characteristics and energy storage flexible adjustment capability through the prediction and consumption collaborative model, establishes a multi-objective optimization strategy, and forms a prediction-consumption coupling system;

[0194] The prediction module determines photovoltaic output prediction parameters through a dynamic hierarchical search mechanism based on the multi-objective optimization strategy, and is used for first-level prediction,

[0195] In combination with the first-level prediction, the dynamic characteristics of photovoltaic output and consumption capacity are analyzed to obtain a double-drive collaborative threshold, and second-level consumption optimization is performed;

[0196] The optimization control module obtains a time-varying comprehensive optimization strategy based on an adaptive constraint processing strategy in combination with system operation constraints, and predicts and controls the photovoltaic consumption capacity of the power distribution network.

[0197] The embodiment also provides an electronic device suitable for power distribution network distributed photovoltaic output prediction and consumption collaborative optimization, comprising a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method proposed in the above embodiment.

[0198] The embodiment also provides a storage medium having a computer program stored thereon, which is executed by a processor to realize the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method proposed in the above embodiment.

[0199] The storage medium proposed in the embodiment belongs to the same inventive concept as the power distribution network distributed photovoltaic output prediction and consumption collaborative optimization method proposed in the above embodiment. The technical details not described in detail in the embodiment can be referred to the above embodiment, and the embodiment has the same beneficial effects as the above embodiment.

[0200] Those skilled in the art can clearly understand the present application by the above description of the embodiments, and the present application can be realized by software and necessary general hardware, and of course, can also be realized by hardware. Based on such understanding, the technical solutions of the present application or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a floppy disk, a read-only memory (ROM), a random access memory (RAM), a FLASH, a hard disk, or an optical disc, and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods of various embodiments of the present application.

[0201] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application, and although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and all should be covered in the scope of the claims of the present application.

Claims

1. A power distribution network distributed photovoltaic output prediction and consumption coordination optimization method, characterized in that: The method comprises the following steps: acquiring operation parameters of a photovoltaic system and constructing a prediction and consumption coordination model; fusing meteorological time series feature extraction, load dynamic response characteristics and energy storage flexible adjustment capacity through the prediction and consumption coordination model; establishing a multi-objective optimization strategy; and forming a prediction-consumption coupling system. Based on the multi-objective optimization strategy, photovoltaic output prediction parameters are determined through a dynamic hierarchical search mechanism, which are used for first-level prediction. In combination with the first-level prediction, dynamic characteristics of photovoltaic output and consumption capacity are analyzed to obtain a double-drive coordination threshold for second-level consumption optimization. Based on an adaptive constraint processing strategy, a time-varying comprehensive optimization strategy is obtained in combination with system operation constraints to predict and control photovoltaic consumption capacity of a distribution network.

2. The power distribution network distributed photovoltaic output prediction and coordinated optimization method of power consumption according to claim 1, characterized in that: Through the coordination model, meteorological time series feature extraction, load dynamic response characteristics and energy storage flexible adjustment capacity are fused to establish a multi-time scale optimization mechanism, which comprises the following steps: analyzing the operation state through the multi-time scale coordination model to obtain a power and load coupling distribution diagram under the influence of photovoltaic output; and determining a space-time mismatch region of photovoltaic output and load demand, which is a key point for energy storage adjustment.

3. The power distribution network distributed photovoltaic output prediction and coordinated optimization method of power consumption according to claim 2, characterized in that: The formation of the prediction-consumption coupling system comprises the following steps: analyzing the operation state based on the coordination model; calculating an output prediction gradient model by setting an irradiance change rate, a load fluctuation coefficient, an energy storage response correction coefficient and combining real-time operation parameters. The operation state is analyzed based on the coordination model, and a consumption capacity gradient model is obtained by calculating a consumption capacity influence factor, a voltage-related adjustment amount and a system safety margin in combination with real-time constraints.

4. The power distribution network distributed photovoltaic output prediction and coordinated optimization method of power consumption according to claim 3, characterized in that: Based on the multi-objective optimization strategy, photovoltaic output prediction parameters are determined through a dynamic hierarchical search mechanism, which comprises the following steps: based on a current photovoltaic output prediction value, a weighted sum of output difference values in a historical time period is superimposed, and residual correction is performed; and a coupling acceleration term of meteorology and load is introduced. Based on a convergence direction function, the change trend of prediction accuracy is judged to obtain photovoltaic output prediction parameters.

5. The power distribution grid distributed photovoltaic output prediction and coordinated optimization method of claim 4, wherein: The dynamic characteristics of photovoltaic output and consumption capacity are analyzed to obtain a double-drive coordination threshold for second-level consumption optimization, which comprises the following steps: a weight matrix of a key node is obtained through a leader weight distribution strategy; and a consumption allowable capacity matrix is obtained according to a voltage-power characteristic curve; Based on the double-matrix, the influence of photovoltaic output change on consumption capacity is quantified to obtain an output correction term. By setting a response rate coefficient in combination with load, the enhancement effect of energy storage configuration on system adjustment capacity is quantified to obtain an energy storage coordination correction term. Based on the two correction terms, a double-drive coordination threshold is obtained in combination with a benchmark reference capacity. Based on the double-drive coordination threshold, different early warning coefficients are set and compared with actual operation values to predict and optimize consumption capacity.

6. The power distribution grid distributed photovoltaic output prediction and coordinated optimization method of claim 5, wherein: Based on the adaptive constraint processing strategy, the system coordination benefit is obtained in combination with the coupling index of prediction accuracy and economy based on the actual consumption capacity under different operation conditions and the theoretical maximum consumption capacity under the corresponding operation conditions. At the same time, an exponential decay function is introduced to obtain the influence loss of constraint violation on system operation. The photovoltaic-energy storage coordination configuration model is obtained by summing the coordination benefit and the constraint loss.

7. The power distribution grid distributed photovoltaic output prediction and coordinated optimization method of claim 6, wherein: The time-varying comprehensive optimization strategy is obtained in combination with system operation constraints, including obtaining a potential optimization space of system operation through a photovoltaic-energy storage collaborative configuration model; and obtaining the comprehensive optimization strategy through a ratio of the potential optimization space to a constraint evolution rate.

8. A power distribution network distributed photovoltaic output prediction and consumption coordination optimization system based on the power distribution network distributed photovoltaic output prediction and consumption coordination optimization method of any one of claims 1-7, characterized in that: The data acquisition module acquires photovoltaic system operation parameters and constructs a prediction and consumption collaboration model, fuses meteorological time sequence feature extraction, load dynamic response characteristics and energy storage flexible adjustment capacity through the prediction and consumption collaboration model, establishes a multi-objective optimization strategy, and forms a prediction-consumption coupling system. The prediction module determines photovoltaic output prediction parameters through a dynamic hierarchical search mechanism based on the multi-objective optimization strategy, and is used for first-level prediction. In combination with the first-level prediction, dynamic characteristics of photovoltaic output and consumption capacity are analyzed to obtain a double-drive collaborative threshold, and second-level consumption optimization is performed. The optimization control module obtains a time-varying comprehensive optimization strategy in combination with system operation constraints based on an adaptive constraint processing strategy, and predicts and controls photovoltaic consumption capacity of the distribution network. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to implement the steps of the distributed photovoltaic output prediction and consumption collaboration optimization method of the distribution network according to any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to implement the steps of the distributed photovoltaic output prediction and consumption collaboration optimization method of the distribution network according to any one of claims 1-7.

Citation Information

Cited By

  • Urban distributed energy storage configuration method, system and platform

    CN121906586A

  • Power distribution network operation optimization method and system based on partition electricity price

    CN122114558A

  • Distributed photovoltaic output control method, device, equipment, medium and program product

    CN122118972A