A hierarchical collaborative operation control method for a distribution network of distributed photovoltaic in-situ consumption

Abstract

The application discloses a kind of distributed photovoltaic in-situ consumption power distribution network hierarchical collaborative operation control method, it is related to electrical engineering technical field, including for the power consumption problem of a large number of distributed photovoltaic and flexible resource in power distribution area, based on day-ahead-intra-day centralized optimization strategy and hierarchical-zoning coordinated control, flexible resource and distributed photovoltaic in substation area are globally scheduled, and the optimization result is applied to real-time control stage, from top to bottom for the distribution of optimized scheduling instruction is allocated, the optimal operation of in-situ consumption of distributed photovoltaic in area is realized.Therefore, using the above-mentioned distributed photovoltaic in-situ consumption power distribution network hierarchical collaborative operation control method, new type power system under complex operation mode can be accurately analyzed, the operation scheme of flexible resource and distributed power supply is optimized by collaborative optimization, the photovoltaic consumption capacity of power distribution network is improved, and the operation economy of new type power system is improved.

Description

A hierarchical and coordinated operation control method for distributed photovoltaic power grid local consumption Technical Field

[0001] This invention relates to the field of electrical engineering technology, and in particular to a hierarchical and coordinated operation control method for distributed photovoltaic power distribution networks to achieve local consumption. Background Technology

[0002] With increasing investment in clean energy sources such as wind power and distributed photovoltaics, low-carbon development is gradually becoming the mainstream trend in energy development. The explosive growth of distributed photovoltaics has placed significant pressure on the existing power grid structure, and traditional low-voltage distribution network operation and control methods are unable to cope with the large-scale local consumption of distributed photovoltaic power. The advancement of new power system construction has added substantial flexibility resources to the distribution network, but it faces the problem of low utilization rates and cannot effectively participate in grid optimization and dispatch. With the promotion of new power system construction and the development of cloud-edge collaborative technology, the control and operation modes of the distribution network have undergone significant changes, shifting from unidirectional tiered transmission, integrated control of the large power grid, source-load responsiveness, and real-time balancing to AC / DC hybrid interconnection, source-grid-load-storage collaborative interaction and flexible conversion, and resource mutual assistance. Furthermore, distributed generation, microgrids, and smart devices possess observable, measurable, and controllable capabilities, enhancing the active support and regulation capabilities of the new power system, which is conducive to the flexible regulation and energy consumption of the distribution network. Based on the above technological developments and equipment iterations, the highly digitalized, intelligent, and networked new power system can realize intelligent coordinated control of massive distributed power generation and consumption objects, achieve friendly interaction among multiple production and consumption units, and further promote the coordinated dispatch of various flexible resources and the local consumption of distributed photovoltaic power.

[0003] Currently, research on the operation and control of distributed photovoltaic power generation for local consumption is mainly divided into local adaptive control and cloud-based centralized control. Local adaptive control is based on the measurement information of the equipment side and has the advantages of fast response speed and low investment cost. However, the lack of information communication between equipment makes it impossible to coordinate control and achieve global optimization operation. Cloud-based centralized control can coordinate the scheduling of controllable resources in each transformer area. However, due to the limitations of investment economy and data transmission, the communication of low-voltage distribution networks is weak, the real-time measurement capability is insufficient, and the time consumed by centralized optimization calculation and the layer-by-layer distribution of instructions is difficult to meet the real-time control requirements.

[0004] Therefore, there is an urgent need for a method to globally schedule flexible resources and distributed photovoltaics within a substation area, and to apply it to the real-time control stage, allocating optimized scheduling instructions from top to bottom to achieve optimal operation for local consumption of distributed photovoltaics in the substation area. Summary of the Invention

[0005] The purpose of this invention is to provide a hierarchical collaborative operation control method for distribution networks that enables local consumption of distributed photovoltaic power. This method can be applied to new power systems with different structures, different equipment, and different operating modes. It can accurately analyze new power systems with distributed photovoltaic power and various flexible resources under complex operating modes. By collaboratively optimizing the operation schemes of flexible resources and distributed power sources, it can improve the photovoltaic consumption capacity of the distribution network and enhance the operational economy of the new power system.

[0006] To achieve the above objectives, this invention provides a hierarchical coordinated operation control method for distributed photovoltaic power grids to achieve local consumption, comprising the following steps:

[0007] S1. Determine the topology of the target distribution network, clarify the configuration parameters and operating data of distributed photovoltaic (PV) systems, and clarify the operating parameters of flexible resource equipment participating in distributed PV consumption; determine the expected weather conditions and related data for the day-ahead dispatch plan, and predict the power output of distributed PV systems and the energy supply and consumption behavior of flexible resources.

[0008] S2. Based on the operating parameters and forecast information of the flexible resource equipment participating in distributed photovoltaic consumption, establish a simplified mathematical model of electricity consumption for each flexible resource equipment. Based on the operating constraints of the distribution network, determine the day-ahead medium- and long-term economic and technical aggregation feasible domain of all flexible resource equipment in the distribution area, that is, the regulation capability of flexible resources.

[0009] S3. Based on the regulation capability of flexible resources and the power prediction data of distributed photovoltaic power generation, establish a day-ahead operation optimization model: with the goal of minimizing network loss and node voltage fluctuation of low-voltage distribution network, and with the distributed photovoltaic curtailment target as a penalty item;

[0010] S4. Based on the day-ahead operation optimization model, and based on the ultra-short-term forecast results of distributed photovoltaic power generation and the intraday adjustment results of various flexibility resources, calculate and generate the intraday ultra-short-term economic and technical feasible domain; with the goal of minimizing the network loss of the low-voltage distribution network and the deviation from the day-ahead optimized voltage, and with the distributed photovoltaic curtailment target as a penalty, establish an intraday rolling optimization model to continuously optimize the intraday operation and control plan of various equipment.

[0011] S5. Based on the optimization scheme obtained in S4, decompose the control instructions from top to bottom and generate specific collaborative control schemes; the substation layer allocates aggregated power dispatch instructions to each feeder layer based on the minimum operating cost and backfeed power; the feeder layer optimizes the flexible resource clusters of its subordinate transformer areas according to the aggregated power dispatch instructions to obtain the optimal dispatch scheme in response to the aggregated power dispatch instructions.

[0012] Furthermore, S1 includes:

[0013] S11. Based on the distribution network topology and the distribution of equipment clusters in the distribution area, determine the spatial location and equipment type and number of each distribution area in the distribution network; based on the construction planning data and the electricity consumption information collection system, establish a set of technical parameter data for the feeder layer, distribution area layer and flexibility resources;

[0014] S12. Based on medium- and long-term meteorological forecast data and historical data of various equipment, predict the power output of distributed photovoltaic systems and the energy supply and consumption behavior of flexible resources.

[0015] Furthermore, in S1 or S2, flexible resource devices participating in distributed photovoltaic power consumption include air conditioners, electric vehicles, electric water heaters, large cold storage facilities, and energy storage systems.

[0016] Furthermore, S2 adopts a linearized power flow model based on reactive power and voltage, and the distribution network operation constraints include node power balance constraints, line power flow constraints, and grid security constraints.

[0017] Furthermore, in S2, the day-ahead medium- and long-term economic and technical feasible domains of all flexible resource equipment in the distribution area are determined, including solving the economic and technical feasible domain aggregation model of various flexible resources using an improved asymptotic vertex enumeration algorithm.

[0018] The improved asymptotic vertex enumeration algorithm includes calculating only the external normal vector of the hyperplane containing the newly added vertex in each search round, and eliminating directions that overlap with those in previous searches; and using the convex hull area in each round... The increment ratio is less than the given value As a condition for terminating the search.

[0019] Furthermore, S3 includes:

[0020] S31. Establish a prediction model for photovoltaic power generation, and combine meteorological information to predict the power generation and data of each distributed photovoltaic system;

[0021] S32. Establish a day-ahead operation optimization model by combining photovoltaic forecast information and the regulation capabilities of flexible resources;

[0022] The objective function expression of the currently running optimization model is as follows:

[0023] ;

[0024] In the formula, express Timetable The transmission current value on, The objective function of the current optimized model is... , Let these represent the weights of the two sub-objectives, network loss target and voltage fluctuation, respectively, satisfying the following conditions: ; , These represent the normalization coefficients for the two sub-targets: network loss target and voltage fluctuation, respectively. This indicates the total number of steps optimized up to date. Indicates the line The resistance values ​​between; This indicates the step size of the current optimization model; This represents the set of all nodes in the transformer area; , They represent The target voltage amplitude and rated voltage amplitude at each time point; This represents the penalty coefficient for the active power control penalty term in distributed photovoltaic systems. This represents the set of distributed photovoltaic grid-connected nodes in a transformer area with controllable active power. , Don't mean Time of the first Predicted and dispatched active power values ​​of distributed photovoltaic power with individual nodes connected to the grid.

[0025] Furthermore, in S3, the day-ahead operation optimization model adopts a second-order cone power flow model, taking the tap position of the on-load tap-changing distribution transformer, the number of parallel capacitors switched on and off, and the active / reactive power of distributed photovoltaic operation in each time period of the next day as the objectives. The constraints include node power balance constraints, line power flow constraints, grid security constraints, distributed photovoltaic operation constraints, parallel capacitor constraints, and distribution transformer tap constraints.

[0026] Furthermore, S4 includes:

[0027] S41. Within the day, forecast the ultra-short-term operating data of distributed photovoltaic and flexible resources in each transformer area, aggregate them to generate the corresponding economic and technical feasible domain, and establish the relevant data set.

[0028] S42. Combining ultra-short-term forecast data and day-ahead optimization schemes, establish an intraday rolling optimization model to perform rolling optimization of the active and reactive power of distributed photovoltaic and flexible resources. The constraints include node power balance constraints, line power flow constraints, grid security constraints, and distributed photovoltaic operation constraints.

[0029] The objective function expression for the intraday rolling optimization model is as follows:

[0030] ;

[0031] In the formula, This represents the total number of steps in the intraday rolling optimization model; This indicates the step size of the current optimization model; express Time Node The operating voltage calculated in the recently optimized model.

[0032] Furthermore, S5 includes:

[0033] S51. Based on the intraday optimization scheduling results of S4, establish a multi-feeder collaborative optimization scheduling model at the substation level to reasonably allocate the aggregated power of each feeder.

[0034] S52. Establish a flexible resource optimization scheduling model for different transformer areas at the feeder layer to obtain a solution that meets technical and economic constraints, and distribute instructions to the flexible resource control terminal of each transformer area.

[0035] Furthermore, in S51, the multi-feeder collaborative optimization scheduling model at the substation level is expressed as follows:

[0036] ;

[0037] In the formula, For feeder The cost of regulating polymerization power, For feeder Aggregated power scheduling instructions, , feeders The power and operating cost of the aggregation port; For feeder Corresponding flexible resource optimization and scheduling schemes; All constraints for system operation.

[0038] Therefore, the present invention employs the above-mentioned hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption, which has the following technical effects:

[0039] (1) The present invention adopts a day-ahead-intraday centralized optimization strategy: In the day-ahead stage, the equipment, lines, capacity and prediction parameters of the substation layer, feeder layer and distribution area layer are combined to characterize the economic and technical feasible domain of multiple flexible resources in the distribution area based on the improved vertex search feasible domain aggregation method, and form an equivalent aggregation model to provide boundary feasible information such as power, energy and cost for the formulation of the optimized scheduling scheme. Then, the substation master station formulates the intraday hour-level control strategy based on the boundary of optimized control and combined with the predicted power generation capacity of distributed photovoltaic. In the intraday stage, the existing communication conditions and hardware devices of the low-voltage distribution network are fully utilized to collect various operating data of local equipment for short-term prediction and economic and technical feasible domain modeling, optimize the intraday control scheme in real time, and apply the results to various flexible resources to further improve the local absorption capacity of distributed photovoltaic.

[0040] (2) This invention establishes a hierarchical-regional distribution network coordination control framework, which is divided into substation layer, feeder layer and distribution area layer. The three control layers realize the hierarchical projection feasible domain through cloud-edge collaboration, which can reduce the aggregation computing cost and protect the privacy of each user. The cloud master station is located in the substation layer. This layer can perform coordinated optimization scheduling of energy consumption based on the aggregation results uploaded by each feeder layer, and can promptly exchange energy with the external power grid when its own capacity cannot achieve full absorption. Each feeder layer needs to allocate power according to the scheduling instructions and determine a multi-flexible resource optimization scheduling scheme that meets the power scheduling instructions of the cloud master station to achieve quantitative local photovoltaic consumption. The distribution area layer is mostly a distributed flexible resource cluster formed based on geographical location. It mostly adopts local edge computing to classify and aggregate resources, and uploads the aggregation results to the cloud for unified control without transmitting sensitive operating parameters.

[0041] The technical solution of the present invention will be further described in detail below through embodiments. Detailed Implementation

[0042] The present invention will be explained in more detail through the following embodiments. The purpose of disclosing the present invention is to protect all changes and modifications within the scope of the present invention. The present invention is not limited to the following embodiments.

[0043] This invention provides a hierarchical collaborative operation control method for distributed photovoltaic (PV) power grids to achieve local consumption. It employs a day-ahead and intraday centralized optimization strategy: During the day-ahead phase, considering equipment, lines, capacity, and predicted parameters at the substation, feeder, and distribution area levels, an improved vertex search-based feasible region aggregation method is used to characterize the economic and technical feasible regions of various flexible resources within the distribution area, forming an equivalent aggregation model. This model provides boundary feasible information such as power, energy, and cost for the formulation of optimized scheduling schemes. The substation master station then formulates intraday hourly-level control strategies based on the optimized control boundaries and the predicted power generation capacity of the distributed PV power grid.

[0044] During the intraday phase, existing communication conditions and hardware devices of the low-voltage distribution network are fully utilized to collect various operating data from local equipment for short-term forecasting and economically and technically feasible domain modeling. Intraday control schemes are optimized in real time, and the results are applied to various flexible resources to further enhance the local absorption capacity of distributed photovoltaic power. In addition, a hierarchical-regional distribution network coordinated control framework is established, divided into substation, feeder, and distribution area layers. These three control layers achieve hierarchical projection feasible domains through cloud-edge collaboration, reducing aggregation computing costs while protecting user privacy. The cloud master station is located at the substation layer, which can perform coordinated energy optimization scheduling based on the aggregation data uploaded by each feeder layer, and promptly exchange energy with the external grid when its own capacity cannot achieve full absorption. Each feeder layer needs to allocate power according to scheduling instructions, determining a multi-flexible resource optimization scheduling scheme that satisfies the power scheduling instructions of the cloud master station to achieve quantitative local photovoltaic absorption. The distribution area layer mainly consists of distributed flexible resource clusters based on geographical location, which mostly use local edge computing for resource classification and aggregation, and upload the aggregation results to the cloud for unified control without transmitting sensitive operating parameters. Specifically, the steps are as follows:

[0045] S1. Determine the topology of the target distribution network, clarify the configuration parameters and operating data of distributed photovoltaics, clarify the operating parameters of flexible resource equipment such as air conditioners, electric vehicles, electric water heaters, large cold storage, and energy storage that participate in the consumption of distributed photovoltaics; determine the expected weather conditions and related data for the day-ahead dispatch plan.

[0046] S11. Based on the distribution network topology and the distribution of equipment clusters in the distribution area, clarify the spatial location and equipment type and number of each distribution area in the distribution network; based on the construction planning data and the electricity consumption information collection system, establish a set of technical parameter data for the feeder layer, distribution area layer and flexibility resources.

[0047] S12. Combining medium- and long-term meteorological forecast data and historical data from various types of equipment, predict the power output of distributed photovoltaic systems and the energy supply and consumption behavior of flexible resources, and generate the data change curves required for S2.

[0048] S2. Based on the equipment data and forecast information determined in S1, establish a simplified mathematical model for the electricity consumption of each flexible resource; under the premise of considering the power flow constraints and safety constraints of the distribution network, determine the day-ahead medium- and long-term economic and technical aggregation feasible domain of all flexible resources in the distribution area, so as to provide feasible information for participating in the day-ahead operation and control plan of the distribution network.

[0049] S21. Taking air conditioners, electric vehicles, electric water heaters, large cold storage facilities, and energy storage systems as examples, as flexible resources participating in distributed photovoltaic power consumption, and based on the operating characteristics of various types of equipment, establish corresponding simplified mathematical models to clarify their flexible and adjustable power range.

[0050] (1) First-order equivalent thermal parameter model of air conditioner:

[0051] ;

[0052] In the formula, , They are respectively Indoor and outdoor temperatures at any given time, in °C; This refers to the energy efficiency ratio of the air conditioner. for The power consumption of the air conditioner at all times; Equivalent thermal resistance, in °C / W; This is the equivalent heat capacity, expressed in J / ℃. For time intervals.

[0053] The process of air conditioning participating in the local consumption of distributed photovoltaic power is mainly constrained by its adjustability and must meet the user's temperature adjustment range for high comfort. In this embodiment, the user demand is set as follows: when the outdoor temperature is greater than 30°C, the air conditioning demand is 24~28°C; when the outdoor temperature is less than 30°C, the air conditioning demand is 20~26°C. In addition, the air conditioning adjustment capacity of a single user is limited. This embodiment is based on smart buildings and adjusts the central air conditioning system shared by all users in the residential complex. By aggregating the air conditioning systems of multiple buildings in the community, the corresponding adjustment capacity range is obtained. The air conditioning parameters and start-up times required for aggregation are all obtained by S1.

[0054] (2) Simplified model of electric vehicle charging power:

[0055] ;

[0056] In the formula, , They are respectively , The electric vehicle's state of charge at any given time is a 0-1 variable; Improve the charging and discharging efficiency of electric vehicles; The rated charging and discharging power of electric vehicles; for The charging and discharging state of the electric vehicle at any given time is a 0-1 variable; This refers to the battery capacity of an electric vehicle, measured in kWh.

[0057] The simplified charging and discharging model of a single electric vehicle has a relatively small adjustable range. This embodiment combines the electric vehicle charging station as an aggregation device to aggregate the feasible domains of all charging vehicles within the station, thereby deriving the overall external control capability of the charging station. Data such as the charging and discharging behavior, time periods, number of vehicles, and energy levels of electric vehicles within the station are all obtained from S1.

[0058] (3) Simplified mathematical model of electricity consumption for electric water heaters:

[0059] ;

[0060] In the formula, , They are respectively , The temperature of the electric water heater at all times; The heating coefficient of the electric water heater; for The power consumption of the electric water heater at any time; The rate at which the water temperature naturally decreases; This refers to the volume of the electric water heater; for The water filling speed of the electric water heater at all times; This refers to the water filling temperature of the electric water heater.

[0061] The electric water heater involved in this embodiment is a large-scale building heating management system with centralized heating and distributed heating capabilities. By aggregating the hot water supply systems of each building in the area into feasible domains, the corresponding adjustment range is obtained.

[0062] (4) Simplified model of large cold storage:

[0063] Large commercial cold storage facilities have constant loads and clear electricity demands, while their operating temperatures have a certain range of adjustment. Given that the refrigeration process of cold storage is similar to that of air conditioning, a first-order equivalent thermal parameter model for air conditioning is adopted to simplify the mathematical model of cold storage electricity consumption.

[0064] (5) Simplified model of charging and discharging power of energy storage system:

[0065] The operating behavior of energy storage systems is consistent with that of electric vehicles (EVs), the difference being that energy storage charges during off-peak hours and discharges during peak hours, using charge / discharge control to smooth out peak demand and generate economic benefits. Therefore, the mathematical model of an energy storage system is similar to the simplified model of an EV's charge / discharge power, and will not be elaborated further here. In terms of charge / discharge behavior, compared to the complexity of EVs, energy storage systems are more constant, discharging during the evening peak and charging at midday. Relevant data on charge / discharge periods and power can be obtained through S1.

[0066] In summary, the constraints on the adjustability of various flexible resources are as follows:

[0067] ;

[0068] In the formula, for Real-time power consumption of various flexible resources within the distribution network area; , They are respectively The lower and upper limits of the power adjustability of various flexible resources within the distribution network area at all times, under the condition of meeting the user-side demand constraints.

[0069] S22. Considering the operational constraints of the distribution network, aggregate the flexibility resources of each transformer area to establish an economically and technically feasible domain.

[0070] (1) To maintain the convexity of the distribution network model, this embodiment adopts a linearized power flow model that considers reactive power and voltage. Distribution network operation constraints include:

[0071] The node power balance constraint is:

[0072] ;

[0073] In the formula, , They are respectively Timetable Active power and reactive power flowing through; , They are respectively Timetable The resistance and reactance values ​​between them; , They are respectively Time Node The effort that contributes merit and the effort that contributes without merit; , They are respectively Time Node The load consists of active and reactive power; For nodes The set of child nodes; , They are respectively Timetable Active power and reactive power flowing through.

[0074] The power flow constraints of the line are:

[0075] ;

[0076] In the formula, , These represent the voltages of the corresponding nodes.

[0077] The power grid security constraints are:

[0078] ;

[0079] In the formula, , They are respectively Timetable Upper and lower limits of the transmission current; , They are respectively The upper and lower limits of safe operation of the node voltage at any given time.

[0080] (2) The economic and technical feasible domain of flexible resources refers to the range of technical and economic aggregate feasibility of flexible resources at the interaction nodes of the distribution area, considering their flexible adjustability while meeting the electricity demand of users and the safety operation constraints of the distribution network. Compared with most existing power feasible domains, the economic and technical feasible domain considers the aggregate equivalent characteristics of power and cost, provides feasible information when participating in the operation and control of the power grid, and meets the needs of system technical constraints and economic efficiency.

[0081] In the flexible resource economic and technical feasible region, projection variables P and C are defined for equivalent projection aggregation, representing the adjustable power and control cost at the equivalent aggregation port, respectively. Projecting the feasible region using projection variables P and C yields the feasible region with external characteristics of technical and economic aggregation, i.e., the economic and technical feasible region. , is represented as:

[0082] ;

[0083] In the formula, , These are the coefficient matrix and constant vector of the economically and technically feasible region, respectively. This refers to the operating scheme, i.e., variables other than the projection variables. At a given power... Below, the system's control costs With Operation Plan Regarding the most economical control costs It must lie on a hyperplane within the economically and technically feasible region. Therefore, the hyperplane containing the operating point with the lowest cost within the adjustable power range is defined as the optimal economic operating hyperplane. The expression is as follows:

[0084] ;

[0085] In the formula, , These are the coefficient matrix and constant vector of the hyperplane, respectively.

[0086] The economic and technical feasible region aggregation method not only eliminates internal variables in the entire system model but also considers system operation constraints and costs. By extracting externally equivalent technical and economic characteristics, it determines the optimal economic operation hyperplane information within the feasible region, providing concise and observable boundary information for higher-level coordination and optimization. This significantly reduces the number of optimization variables and constraints in the system model, thereby lowering computational complexity.

[0087] S23. An improved asymptotic vertex enumeration algorithm is used to solve the economic and technical feasible domain aggregation model for various flexible resources.

[0088] (1) The progressive vertex enumeration (PVE) algorithm is a vertex search convex hull projection algorithm. The essence of the PVE algorithm is to approximate from the inside out, and only search a portion of the vertices to obtain a conservative approximation of the convex hull. The external normal vector of the hyperplane of the convex hull is used as the search direction, and new vertices are enumerated in each round of iteration until the accuracy requirement is met. However, the PVE algorithm traverses all hyperplanes in each round of search, which has the problem of repeated searching in some directions and low search efficiency. In view of the above problems, the following improvements are made in this embodiment:

[0089] First, each round of the search process only calculates the external normal vector of the hyperplane where the newly added vertex is located, and removes directions that are repeated with those in the previous search, ensuring that the search directions in each round are not redundant and avoiding repeated searches in some directions, which can effectively improve the search speed.

[0090] Second, based on the area of ​​each round of the convex hull. The increment ratio is less than the given value As a search termination condition, the approximation degree of the convex hull in each round is clearly quantified, requiring only the increment ratio calculation in each round, thus balancing the overall search speed and accuracy.

[0091] (2) Regarding the projection variable as Given the economically and technically feasible region, the vertex search method for projection can be equivalently transformed into the following optimization problem:

[0092] ;

[0093] In the formula, For the search vector, For constraint variables, As constraints. Characterization The convex hull of the economically and technically feasible region needs at least Each vertex represents the convex hull of the economically and technically feasible region.

[0094] ;

[0095] In the formula, The economic and technically feasible region is represented by the vertex representation. For convex hull calculation algorithms, this embodiment adopts the Quick Hull convex hull algorithm, which can quickly solve for the convex hull and calculate its area; To construct all vertices of the convex hull; The number of all vertices; The coefficients are convex combination coefficients; For the first One vertex.

[0096] After characterizing the economically and technically feasible region using vertex representations, it is transformed into a hyperplane representation. Constraints equivalent to several hyperplanes are aggregated to extract externally equivalent technical and economic characteristics, thus determining the optimal economic operation hyperplane information within the economically and technically feasible region. At this point, the hyperplane representation is as follows:

[0097] ;

[0098] In the formula, The economic and technically feasible region is represented by a hyperplane. for The set of real numbers of dimension ; , These are the coefficient matrix and constant vector of the economic and technical feasible region, respectively.

[0099] S3. Based on the information on the adjustability of flexible resources (obtained from S2) and the power prediction data of distributed photovoltaic power generation, establish a day-ahead operation optimization model: with the goal of minimizing the network loss and node voltage fluctuation of the low-voltage distribution network, and with the distributed photovoltaic curtailment target as a penalty.

[0100] S31. Establish a prediction model for photovoltaic power generation, and combine meteorological information to predict the power generation and data of each distributed photovoltaic system.

[0101] With the large-scale integration of distributed photovoltaic (PV) power into low-voltage distribution networks, accurate power flow calculations have become crucial for assessing grid operation, optimizing energy storage planning, and ensuring stable grid operation. PV power generation is directly affected by changes in meteorological factors such as sunlight, cloud cover, and temperature; single power variations typically occur within 10 minutes or even on the order of seconds. To accurately assess the impact of distributed PV on distribution network power flow, more refined and dynamic calculation models are needed. These models should be able to consider the time-varying nature of PV output, which can be achieved by combining historical meteorological data with physical characteristic models of PV panels. Time series analysis or machine learning algorithms can be used to predict PV output at different time scales, thus providing input data for the calculations. The PV output prediction model can be expressed as:

[0102] ;

[0103] In the formula, for Solar power output at all times This is a function describing the relationship between photovoltaic output and light intensity, temperature, and photovoltaic panel parameters. , They are respectively Light intensity and temperature at any given time This is a set of parameters for photovoltaic panels.

[0104] The model must include a detailed model of the inverter, considering its reactive power compensation capability, voltage regulation function, and low voltage ride-through characteristics. The inverter output can be expressed as:

[0105] ;

[0106] In the formula, , Inverter Active and reactive power output at all times; This is a function that describes the relationship between inverter output, photovoltaic output, grid voltage, and inverter parameters. This refers to the voltage at the grid connection point. This is the set of control parameters for the inverter.

[0107] Based on the above model, the solution for the operation control scheme and power flow distribution of the distribution network can take into account the dynamic changes of photovoltaic output and inverter control, as well as their impact on the voltage, current and power distribution of the grid.

[0108] S32. By combining photovoltaic forecast information and data on the regulation and control capabilities of flexible resources, establish a day-ahead operation optimization model.

[0109] During the day-ahead decision-making phase, relying on the computing power of the distribution cloud master station or the electricity consumption information collection master station, a centralized decision-making mode is adopted with a time step of 1 hour. Based on the day-ahead forecast of the distributed photovoltaic power generation in the distribution area and the power regulation capability of flexible resource users, with network losses, voltage fluctuations, and photovoltaic curtailment of the low-voltage distribution network as targets, the on-load tap changer (OLTC) tap positions, the number of parallel capacitors switched on and off, and the active / reactive power of distributed photovoltaic operation are optimized for each time period of the next 24 hours. The operation instructions are then sent to the integrated terminal of each distribution area as reference instructions for intraday rolling optimization.

[0110] The current optimization model aims to minimize network losses and node voltage fluctuations in low-voltage distribution networks, while incorporating the target of distributed photovoltaic curtailment as a penalty term. The objective function expression is as follows:

[0111] ;

[0112] In the formula, express Timetable The transmission current value on, The objective function of the current optimized model is... , These represent the weights of the two sub-targets: network loss target and voltage fluctuation, respectively. These weights can be determined based on the actual control requirements of the target transformer area to meet [the requirements]. ; , These represent the normalization coefficients for the two sub-targets, network loss target and voltage fluctuation, respectively, and are taken as the reciprocal of the index values ​​under the initial state. This represents the total number of steps optimized so far, which is 24 here; Indicates the line The resistance values ​​between; This indicates the step size of the current optimization model, which is 1h here; Represents the set of all nodes in the transformer area. , They represent The target voltage amplitude and rated voltage amplitude at each time point; This represents the penalty coefficient for the active power control penalty term in distributed photovoltaic systems. This represents the set of distributed photovoltaic grid-connected nodes in a transformer area with controllable active power. , Don't mean Time of the first Predicted and dispatched active power values ​​of distributed photovoltaic power with individual nodes connected to the grid.

[0113] The recently optimized model is based on the Distflow power flow model. The main constraints include node power balance constraints, line power flow constraints, grid security constraints, distributed photovoltaic operation constraints, parallel capacitor constraints, and distribution transformer tap constraints. Among these, node power balance constraints, line power flow constraints, and grid security constraints are already addressed above. Other constraints are as follows:

[0114] (1) Operational constraints of distributed photovoltaic systems:

[0115] ;

[0116] In the formula, Represents a node Maximum regulation ratio for grid-connected distributed photovoltaic systems; Represents a node Maximum power factor angle of grid-connected distributed photovoltaic systems; express Time Node Reactive power dispatch values ​​for grid-connected distributed photovoltaic systems; Represents a node The capacity of grid-connected distributed photovoltaic inverters.

[0117] (2) Distribution transformer tap constraints:

[0118] ;

[0119] In the formula, express Time Node The rated voltage amplitude; Represents a node Distribution transformer tap adjustment step size; Indicates the operating position of the node distribution transformer at any given time; This indicates the maximum range that the distribution transformer at the node can be adjusted up or down.

[0120] (3) Parallel capacitor constraint:

[0121] ;

[0122] In the formula, express Time Node The reactive power output of a single parallel capacitor; Represents a node The number of parallel capacitors connected; , These represent the upper and lower limits of the number of parallel capacitors connected to the node, respectively.

[0123] S33. The day-to-day optimization model is simplified using the second-order cone relaxation method and then imported into the mathematical optimization solver for solution, as follows:

[0124] The current optimization model is a mixed integer nonconvex nonlinear optimization model, which uses a second-order cone relaxation method, and sets respectively... , Then the power flow constraint of the line can be equivalent to:

[0125] ;

[0126] The operational constraints of distributed photovoltaic systems can be equivalently represented as:

[0127] .

[0128] After variable substitution, the second term in the objective function of the day-ahead optimization model is linearized using an incremental linearization method. Let... Introducing interval variables :

[0129] ;

[0130] In the formula, This represents the number of sub-intervals in the linearized model; express In the The values ​​of each segmented sub-interval; express In the The values ​​of each segmented sub-interval.

[0131] in,

[0132] ;

[0133] In the formula, It is a binary variable, guaranteeing Continuous values ​​within each segmented sub-interval.

[0134] Through the above process, the mixed-integer nonconvex nonlinear optimization model is transformed into a mixed-integer second-order cone optimization model, which is then solved intensively using the CPLEX solver.

[0135] S4. Based on the day-ahead operation optimization model, and based on the ultra-short-term forecast results of distributed photovoltaic power generation and the intraday adjustment results of various flexibility resources, calculate and generate the intraday ultra-short-term economic and technical feasible domain; with the goal of minimizing the network loss of the low-voltage distribution network and the deviation from the day-ahead optimized voltage, and with the distributed photovoltaic curtailment target as a penalty term, establish an intraday rolling optimization model to continuously optimize the operation and control plan of various equipment within the day.

[0136] S41. Within the day, forecast the ultra-short-term operating data of distributed photovoltaic and flexible resources in each transformer area, aggregate and generate the corresponding economic and technical feasible domain, and establish a related dataset to provide short-term high-precision data input for S42.

[0137] S42. Combining ultra-short-term forecast data and day-ahead optimization schemes, an intraday rolling optimization model is established. Specifically, based on the day-ahead optimization model, the intraday rolling optimization model, at each time point, uses the ultra-short-term forecast results of distributed photovoltaic power generation for the next four 15-minute periods, the boundary information of the regulation capability of flexible resources, etc., with the goal of minimizing the network loss of the low-voltage distribution network and the deviation from the day-ahead optimized voltage, and also considers the distributed photovoltaic curtailment target as a penalty term, to perform rolling optimization on the active and reactive power of distributed photovoltaic and flexible resources. The objective function of the intraday rolling optimization model is as follows: The expression is as follows:

[0138] ;

[0139] In the formula, This represents the total number of steps in the intraday rolling optimization model, which is set to 4 here. This represents the step size of the intraday rolling optimization model, which is set to 15 minutes here; express Time Node The operating voltage calculated in the recently optimized model.

[0140] The constraints of the intraday rolling optimization model mainly include node power balance constraints, line power flow constraints, grid security constraints, and distributed photovoltaic operation constraints, which are the same as those of S3.

[0141] Both the day-to-day optimization model and the intraday rolling optimization model are mixed-integer non-convex nonlinear optimization models, and both can be solved using the second-order cone relaxation method: First, equivalent processing is performed on the line power flow constraints, photovoltaic operation constraints, etc.; then, the objective function is linearized using the incremental linearization method, transforming the mixed-integer non-convex nonlinear optimization model into a mixed-integer second-order cone optimization model; finally, the model is imported into the mathematical solver for solving.

[0142] S5. Combining the optimization scheme generated in S4, the control instructions are decomposed from top to bottom to generate specific collaborative control schemes. The substation level considers minimizing operating costs and backfeed power, and allocates aggregated power dispatch instructions to each feeder level; the feeder level optimizes the flexible resource clusters of its subordinate transformer areas according to the dispatch instructions to obtain the optimal dispatch scheme in response to the aggregated power dispatch instructions.

[0143] S51. Based on the intraday optimization scheduling results of S4, establish a multi-feeder collaborative optimization scheduling model at the substation level to rationally allocate the aggregated power of each feeder.

[0144] Considering the output characteristics and flexible resource adjustability of distributed photovoltaic power, and simultaneously taking into account operational economics, the diverse and flexible resources on the user side are aggregated in a partitioned and hierarchical manner, taking into account the entire set of constraints of the distribution network. The economically and technically feasible region is obtained through equivalent projection aggregation. .

[0145] For the multi-feeder collaborative optimization scheduling problem in the substation domain layer, aggregated boundary information obtained from projection is used. This paper describes a substation-level collaborative optimization problem that optimizes the scheduling of various feeders to meet the local photovoltaic (PV) consumption demand. The goal is to achieve complete local PV consumption at both the transformer substation and feeder levels. The substation-level multi-feeder collaborative optimization scheduling problem aims to minimize the total operating cost of the substation while satisfying the requirement for complete local PV consumption. The specific mathematical model for obtaining the aggregated power scheduling command for each feeder is as follows:

[0146] ;

[0147] In the formula, Total operating cost; For feeder The cost of regulating polymerization power; The total power of the substation meets the local consumption requirements of the two-level distributed photovoltaic power generation of the distribution area feeder. For the feeder set; For feeder Aggregated power scheduling instructions; For feeder Maximum transmission capacity; For feeder The economic and technically feasible region of the projection equivalent.

[0148] The constraint that the total aggregated power of the station area must be greater than or equal to zero means that the distributed photovoltaic power within the station area's power supply region must be fully absorbed, and no power must be fed back to the upper-level grid. Solving the above model yields the aggregated power dispatch instructions for each feeder. and By using the feasible region aggregation and equivalence method, the high-dimensional operating space of each feeder is reduced to a two-dimensional economic and technical feasible region, which greatly reduces the number of constraints and variables in the collaborative optimization problem and lowers the complexity of the mathematical optimization problem.

[0149] S52. Establish a flexible resource optimization scheduling model for multiple distribution areas at the feeder layer, find a solution that meets technical and economic constraints, and distribute instructions to the flexible resource control terminals of each distribution area.

[0150] The feeder layer needs to determine a flexible resource optimization scheduling scheme that satisfies the aggregated power dispatch command, i.e., the problem of resolving the aggregation of dispatch commands involving multiple flexible resources for local photovoltaic consumption. According to the theory of economically and technically feasible region projection, there must exist an achievable operating scheme for the power operating point within the feasible region. Since the dispatch command is solved based on information from the economically and technically feasible region, it must be within the feasible region. Therefore, there must exist a flexible resource dispatch scheme that satisfies technical and economic constraints. The specific mathematical model is as follows:

[0151] ;

[0152] In the formula, , feeders The power and operating cost of the aggregation port; For feeder Corresponding flexible resource optimization and scheduling schemes; Let these be all the constraints for the system's operation. By solving the above optimization problem, we can achieve... and Satisfy scheduling instructions and Solve for this. This allows for the creation of a scheduling scheme for diverse and flexible resources within the feeder that responds to aggregated power scheduling commands.

[0153] Therefore, this invention adopts the above-mentioned hierarchical collaborative operation control method for distribution network local consumption of distributed photovoltaic power. Addressing the problem of local consumption of power in distribution substations containing a large number of distributed photovoltaic and flexible resources, it uses a day-ahead-intraday centralized optimization strategy and hierarchical-regional coordinated control as a basis to globally schedule flexible resources and distributed photovoltaic power within the substation area. The optimization results are then applied to the real-time control stage, distributing optimized scheduling commands from top to bottom to achieve optimal operation for local consumption of distributed photovoltaic power in the substation area.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A hierarchical collaborative operation control method for distributed photovoltaic power grids for local consumption, characterized in that, Includes the following steps: S1. Determine the topology of the target distribution network, clarify the configuration parameters and operating data of distributed photovoltaic (PV) systems, and clarify the operating parameters of flexible resource devices participating in PV consumption; determine the expected weather conditions and related data for the day-ahead dispatch plan, and predict the power output of distributed PV and the energy supply and consumption behavior of flexible resources; S2. Based on the operating parameters and forecast information of flexible resource devices participating in PV consumption, establish a simplified mathematical model for the power consumption of each flexible resource device, and based on the distribution network operation constraints, determine the day-ahead medium- and long-term economic and technical aggregation feasible domain of all flexible resource devices in the distribution area, i.e., the regulation capability of flexible resources; S3. Based on the regulation capability of flexible resources and the forecast data of distributed PV power generation, establish a day-ahead operation optimization model: with the goal of minimizing network losses and node voltage fluctuations in the low-voltage distribution network, and using the distributed PV curtailment target as a penalty term; S4. Based on the day-ahead operational optimization model, and using the ultra-short-term forecast results of distributed photovoltaic power generation and the intraday adjustment results of various flexible resources, calculate and generate the intraday ultra-short-term economic and technical feasible domain. With the goal of minimizing network losses in the low-voltage distribution network and the deviation from the day-ahead optimized voltage, and using the distributed photovoltaic curtailment target as a penalty, establish an intraday rolling optimization model to continuously optimize the intraday operation and control plans for various equipment. S5. Based on the optimization scheme obtained in S4, decompose the control instructions from top to bottom to generate specific collaborative control schemes. At the substation level, based on minimizing operating costs and backfeed power, allocate aggregated power dispatch instructions to each feeder level. The feeder level optimizes the control of the flexible resource clusters of its subordinate transformer areas according to the aggregated power dispatch instructions to obtain the optimal dispatch scheme in response to the aggregated power dispatch instructions.

2. The method for hierarchical and coordinated operation control of a distribution network for local consumption of distributed photovoltaic power, as described in claim 1, is characterized in that, S1 includes: S11. Based on the distribution network topology and the distribution of equipment clusters in the distribution area, determine the spatial location and equipment type and number of each distribution area in the distribution network; Based on construction planning data and electricity consumption information collection system, establish a set of technical parameter data for feeder layer, transformer area layer and flexible resources; S12, combine medium and long-term meteorological forecast data and historical data of various equipment to predict the power output of distributed photovoltaic and the energy supply and consumption behavior of flexible resources.

3. The hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption according to claim 1, characterized in that, In S1 or S2, flexible resource equipment participating in distributed photovoltaic power consumption includes air conditioners, electric vehicles, electric water heaters, large cold storage facilities, and energy storage systems.

4. The hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption according to claim 1, characterized in that, S2 adopts a linearized power flow model based on reactive power and voltage. The distribution network operation constraints include node power balance constraints, line power flow constraints, and grid security constraints.

5. The hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption according to claim 1, characterized in that, In S2, the day-ahead and long-term economic and technical feasible regions for all flexible resource equipment within the distribution area are determined. This includes using an improved asymptotic vertex enumeration algorithm to solve the economic and technical feasible region aggregation model for various flexible resources. The improved asymptotic vertex enumeration algorithm calculates only the external normal vector of the hyperplane containing the newly added vertex in each search round, and removes directions that overlap with previous searches. The area of ​​the convex hull in each round is used as the basis for the calculation. The increment ratio is less than the given value As a condition for terminating the search.

6. The hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption according to claim 1, characterized in that, S3 This includes: S31, establishing a photovoltaic power generation prediction model and predicting the power generation and data of each distributed photovoltaic system by combining meteorological information; S32, establishing a day-ahead operation optimization model by combining photovoltaic prediction information and the regulation capability of flexible resources; wherein, the objective function expression of the day-ahead operation optimization model is as follows: In the formula, express Timetable The transmission current value on, The objective function of the current optimized model is... 、 Let these represent the weights of the two sub-objectives, network loss target and voltage fluctuation, respectively, satisfying the following conditions: ； 、 These represent the normalization coefficients for the two sub-targets: network loss target and voltage fluctuation, respectively. This indicates the total number of steps optimized up to date. Indicates the line The resistance values ​​between; This indicates the step size of the current optimization model; This represents the set of all nodes in the transformer area; 、 They represent The target voltage amplitude and rated voltage amplitude at each time point; This represents the penalty coefficient for the active power control penalty term in distributed photovoltaic systems. This represents the set of distributed photovoltaic grid-connected nodes in a transformer area with controllable active power. 、 They represent Time of the first Predicted and dispatched active power values ​​of distributed photovoltaic power with individual nodes connected to the grid.

7. The hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption according to claim 1, characterized in that, In S3, the day-ahead operation optimization model adopts a second-order cone power flow model, taking the tap position of the on-load tap-changing distribution transformer, the number of parallel capacitors switched on and off, and the active / reactive power of distributed photovoltaic operation in each time period of the next day as the objectives. The constraints include node power balance constraints, line power flow constraints, grid security constraints, distributed photovoltaic operation constraints, parallel capacitor constraints, and distribution transformer tap constraints.

8. The hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption according to claim 6, characterized in that, S4 includes: S41, forecasting the ultra-short-term operating data of distributed photovoltaic (PV) and flexible resources in each transformer area within the day, aggregating and generating the corresponding economically and technically feasible domains, and establishing relevant datasets; S42, combining the ultra-short-term forecast data and the day-ahead optimization scheme, establishing an intraday rolling optimization model to perform rolling optimization of the active and reactive power of distributed PV and flexible resources, with constraints including node power balance constraints, line power flow constraints, grid security constraints, and distributed PV operation constraints; the objective function expression of the intraday rolling optimization model is as follows: In the formula, This represents the total number of steps in the intraday rolling optimization model; This indicates the step size of the intraday rolling optimization model; express Time Node The operating voltage calculated in the recently optimized model.

9. The hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption according to claim 1, characterized in that, S5 This includes: S51, combining the intraday optimization scheduling results of S4, establishing a multi-feeder collaborative optimization scheduling model at the substation level, and rationally allocating the aggregated power of each feeder; S52, establishing a flexible resource optimization scheduling model for different distribution areas at the feeder level, obtaining a scheme that meets technical and economic constraints, and distributing instructions to the flexible resource control terminals of each distribution area.

10. A hierarchical collaborative operation control method for distributed photovoltaic power grid local consumption according to claim 9, characterized in that, In S51, the multi-feeder collaborative optimization scheduling model at the substation level is expressed as follows: In the formula, For feeder The cost of regulating polymerization power, For feeder Aggregated power scheduling instructions, 、 feeders The power and operating cost of the aggregation port; For feeder Corresponding flexible resource optimization and scheduling schemes; All constraints for system operation.