A source-grid-load-storage hierarchical rolling scheduling method for active power distribution network

Abstract

The application discloses a source-grid-load-storage hierarchical rolling scheduling method for an active power distribution network, obtains operation basic data of the active power distribution network; according to the operation basic data, a BP neural network optimized by a genetic algorithm is used to predict distributed renewable energy output and load demand, to obtain day-ahead prediction data and intra-day rolling prediction data; according to the day-ahead prediction data, a day-ahead source-grid-load-storage coordinated scheduling model is constructed to generate a day-ahead scheduling plan; according to the intra-day rolling prediction data and the day-ahead scheduling plan, an intra-day source-grid-load-storage coordinated scheduling model is constructed to generate an intra-day scheduling plan; based on a power flow relationship of the active power distribution network, nonlinear power flow constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intra-day source-grid-load-storage coordinated scheduling model are second-order cone relaxed to obtain a second-order cone programming scheduling model; and the second-order cone programming scheduling model is solved to obtain source-grid-load-storage hierarchical rolling regulation and control results of the active power distribution network.

Description

Technical Field

[0001] This invention belongs to the field of power grid dispatching technology, specifically relating to a hierarchical rolling dispatching method for source-grid-load-storage systems oriented towards active distribution networks. Background Technology

[0002] Renewable energy sources such as wind and solar power are intermittent, uncertain, and volatile. With the development of active distribution networks, a large number of distributed renewable energy sources are being integrated into the distribution network, transforming it from a traditional unidirectional power receiving network into an active network with multi-source access, multi-energy interaction, and multi-device regulation capabilities. In this operational scenario, renewable energy output, load demand, energy storage status, and network power flow all exhibit significant temporal variations. If these uncertainties cannot be effectively predicted and coordinated, problems such as wind and solar power curtailment, node voltage exceeding limits, branch power flow exceeding limits, insufficient energy storage utilization, and increased distribution network operating costs can easily occur. Therefore, how to improve the renewable energy absorption capacity in active distribution networks while ensuring the safe and economical operation of the distribution network has become an urgent technical problem to be solved.

[0003] Existing technologies have proposed various methods for coordinated and optimized scheduling of power generation, grid, load, and storage. For example, CN119965974A discloses a method that establishes objective functions and optimization variables at different time scales by dividing the scheduling into day-ahead optimization scheduling, intraday rolling optimization scheduling, and real-time feedback correction, and then combines network constraints, wind and solar power operation constraints, and demand response constraints for optimized scheduling. Other technologies use convex relaxation, piecewise linearization, and the Big M method to transform scheduling models containing nonlinear terms and integer variables into mixed-integer second-order cone programming problems to improve the efficiency of optimization solutions.

[0004] However, existing multi-timescale source-grid-load-storage coordinated scheduling methods still have the following shortcomings: First, existing methods usually only predict wind power, photovoltaic output and load according to a fixed time resolution, and directly use the prediction results for subsequent optimization. They lack a prediction accuracy improvement mechanism for the random fluctuation characteristics of renewable energy and the uncertainty characteristics of load, which makes it easy for prediction errors to be transmitted between day-ahead plans, intraday rolling plans and real-time control, thereby affecting the reliability of scheduling results.

[0005] Secondly, while existing methods propose an overall framework for coordinated optimization of power generation, grid, load, and storage, their modeling of various flexible resources in active distribution networks remains insufficiently refined. They often focus on resources such as wind power, solar power, energy storage, or demand response, failing to fully reflect the coupling and regulation relationships between the source, grid, load, and energy storage sides. Particularly on the grid side, some existing technologies primarily focus on conventional power flow constraints and voltage constraints, neglecting issues such as dynamic reconfiguration of the distribution network, the number of segmented switching operations, maintaining the radial topology, and adjusting the network structure according to changes in renewable energy and load. This makes it difficult to fully leverage the role of network topology regulation in renewable energy absorption and operational economics.

[0006] Third, in existing technologies, intraday rolling scheduling and real-time feedback correction usually aim to minimize operating costs or deviations. Although they can improve short-term operating conditions, they are prone to inconsistencies between control targets and new energy consumption targets when renewable energy output fluctuates rapidly. This makes it difficult for resources such as energy storage, flexible loads, reactive power compensation equipment, and network reconfiguration to form a coordinated regulation effect oriented towards new energy consumption.

[0007] Fourth, the constraints in existing methods mostly adopt general network constraints, wind and solar operation constraints, and demand response constraints. The constraint system is relatively general and does not fully consider the number of actions, capacity boundaries, charging and discharging mutual exclusion, regulation conservation, and equipment life limits of various regulating resources such as on-load tap-changing transformers, group switching capacitors, static var compensators, controllable loads, and energy storage devices. This can easily lead to insufficient executability of the optimization results in actual distribution network operation.

[0008] Fifth, the active distribution network optimization scheduling model itself has non-convex, nonlinear, and mixed-integer characteristics. Although existing heuristic optimization algorithms are good at handling complex constraints, they are prone to getting trapped in local optima and have long solution times. Existing scheduling methods based on second-order cone programming can improve solution efficiency, but some methods handle nonlinear power flow, discrete equipment switching, capacity circular constraints, and the coupling relationship between equipment status and continuous operating variables in a rather general way, making it difficult to balance model accuracy, solution efficiency, and the executability of scheduling results.

[0009] Therefore, existing active distribution network source-grid-load-storage scheduling technology still has technical problems such as insufficient prediction accuracy, insufficient collaborative modeling of multiple types of flexible resources, insufficient utilization of network-side dynamic adjustment capabilities, weak matching between intraday rolling scheduling and new energy consumption targets, and difficulty in efficiently and reliably solving nonlinear optimization models. Summary of the Invention

[0010] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a hierarchical rolling scheduling method for source-grid-load-storage in active distribution networks.

[0011] The objective of this invention can be achieved through the following technical solutions: This invention provides a hierarchical rolling scheduling method for source-grid-load-storage systems in active distribution networks, comprising the following steps: S1. Obtain basic operational data of the active distribution network; S2. Based on the aforementioned operational data, a BP neural network optimized by a genetic algorithm is used to predict the output and load demand of distributed renewable energy, thereby obtaining day-ahead forecast data and intraday rolling forecast data. S3. Based on the day-ahead forecast data, construct a day-ahead source-grid-load-storage coordinated scheduling model and generate a day-ahead scheduling plan; S4. Based on the intraday rolling forecast data and the day-ahead scheduling plan, construct an intraday source-grid-load-storage coordinated scheduling model and generate an intraday scheduling plan; S5. Based on the power flow relationship of the active distribution network, the nonlinear power flow constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model are relaxed by second-order cone relaxation, and the nonlinear terms in the model are linearized to obtain the second-order cone planning scheduling model. S6. Solve the second-order cone programming scheduling model to obtain the active distribution network source-grid-load-storage hierarchical rolling control results, and generate distributed renewable energy output commands, network reconfiguration commands, controllable load adjustment commands, energy storage charging and discharging commands, and flexible regulation equipment control commands based on the active distribution network source-grid-load-storage hierarchical rolling control results.

[0012] Furthermore, the operational foundation data includes distributed renewable energy data, load data, network topology data, energy storage device data, and flexible regulation device data; the distributed renewable energy data includes historical output data of distributed renewable energy, access node data, available output data, and operational limitation data. The load data includes historical load data for each node, load characteristic data required for prediction, and controllable load adjustment capacity data. The network topology data includes node data, branch data, root node data, sectional switch status data, and network operation safety limit data of the active distribution network. The energy storage device data includes access node data, state of charge data, capacity limitation data, charge / discharge power limitation data, and charge / discharge status data. The data from the flexible regulation equipment includes on-load tap-changing transformer data, group switching capacitor data, and static var compensator data.

[0013] Furthermore, S2 specifically includes: S21. Based on the aforementioned operational data, construct a BP neural network for predicting the output and load demand of distributed renewable energy, and determine the input layer, hidden layer, and output layer of the BP neural network, wherein the input layer is used to input distributed renewable energy data and load data, and the output layer is used to output the predicted output value and the predicted load demand value of distributed renewable energy. S22. The network parameters of the BP neural network are encoded with real numbers to obtain the initial population of the genetic algorithm. The network parameters include the connection weights between the input layer and the hidden layer, the connection weights between the hidden layer and the output layer, the hidden layer threshold, the output layer threshold, and the neuron smoothing factor. S23. Assign each individual in the initial population as a set of candidate network parameters to the BP neural network, and calculate the fitness value of the corresponding individual based on the prediction output error of the BP neural network, wherein the prediction output error satisfies: The fitness value satisfies: in, Indicates the prediction output error. Indicates the number of training samples. Indicates the training sample number. Indicates the first The actual output value corresponding to each training sample. Indicates the first The predicted output value corresponding to each training sample. This represents the fitness value of an individual. This indicates a positive number used to avoid a denominator of zero; S24. Based on the fitness value, perform a selection operation on the individuals in the initial population, retain the individuals with higher fitness values ​​as parent individuals, and perform real crossover and mutation operations on the parent individuals to obtain an updated population; S25. Determine whether the updated population meets the preset stopping condition. If not, treat the updated population as a new population and repeat S23 to S25. If it meets the condition, output the individual with the highest fitness value as the optimal individual. S26. Decode the optimal individual to obtain the connection weights, thresholds, and smoothing factors of the optimized BP neural network, and assign the connection weights, thresholds, and smoothing factors to the BP neural network to obtain the BP neural network optimized by the genetic algorithm; S27. Input the basic operational data into the BP neural network optimized by the genetic algorithm to predict the output and load demand of distributed renewable energy under the day-ahead time scale and the output and load demand of distributed renewable energy under the intraday rolling time scale, so as to obtain the day-ahead prediction data and the intraday rolling prediction data.

[0014] Furthermore, the real-number crossover operation includes: selecting two individuals from the parent individuals to participate in the crossover, and recombining the two individuals using real-number encoding based on the crossover coefficient to obtain two child individuals, as shown in the formula: in, and They represent the first The encoding vectors of the two parent individuals that participate in the real crossover operation in each genetic iteration. and Let represent the encoding vectors of the two offspring individuals generated after the real number crossover operation. Indicates the cross coefficient. This represents a function that takes the minimum value. The mutation operation includes: perturbing the coding gene of the individual to be mutated in the updated population, and determining the perturbation direction and amplitude of the coding gene based on a random factor, using the following formula: in, in, Indicates the first Among the individuals, the first The gene value of each coding gene. This indicates the upper bound of the gene encoding the gene. This indicates the lower bound of the gene encoding the gene. This represents a variable-length function that varies with the number of iterations. Indicates the current iteration number. Indicates the maximum number of iterations. This represents the random factor used to determine the direction of mutation. This represents a random number located in the interval [0,1].

[0015] Furthermore, S3 specifically includes: S31. Based on the day-ahead forecast data, determine the forecast values ​​of available output of distributed renewable energy and the forecast values ​​of load demand of each node for each scheduling period within the day-ahead scheduling cycle. S32. Based on the predicted available output of distributed renewable energy, the predicted load demand of each node, and the basic operational data, determine the decision variables of the day-ahead source-grid-load-storage coordinated scheduling model. The decision variables of the day-ahead source-grid-load-storage coordinated scheduling model include the actual output of distributed renewable energy, the abandoned power of distributed renewable energy, the power exchanged with the upstream grid, the status of the branch switches of the distribution network, the tap position of the on-load tap changer, the number of switching groups of the grouped capacitors, the reactive power output of the static var compensator, the controllable load regulation power, the charging and discharging power of the energy storage equipment, and the state of charge of the energy storage equipment. S33. With the goal of minimizing the operating cost of the active distribution network during the day-ahead scheduling cycle, construct the objective function of the day-ahead source-grid-load-storage coordinated scheduling model. The objective function of the day-ahead source-grid-load-storage coordinated scheduling model is expressed as follows: Among them, the cost of flexible resource scheduling satisfies: in, This represents the total operating cost of the active distribution network during the current day's dispatch cycle; This indicates the total number of scheduling periods within the current day's scheduling cycle; Indicates the first The cost of curtailed distributed renewable energy during each scheduling period; Indicates the first The cost of purchasing electricity from the upper-level power grid for each dispatch period; Indicates the first Network loss cost per scheduling period; Indicates the first The cost of flexible resource scheduling for each scheduling period; Indicates the first Equipment operating costs for each scheduling period; Indicates the first Distributed renewable energy reactive power regulation cost per scheduling period; Indicates the first The static var compensator adjustment cost for each scheduling period; Indicates the first Cost of switching capacitors in groups for each scheduling period; Indicates the first Controllable load adjustment cost for each scheduling period; Indicates the first Operating costs of energy storage equipment during each scheduling period; S34. The constraints of distributed renewable energy operation, network topology operation, controllable load operation, energy storage equipment operation, flexible regulation equipment operation, and active distribution network power flow constraints are used as the constraints of the day-ahead source-grid-load-storage coordinated scheduling model, so that the decision variables participate in day-ahead optimization under the condition of meeting the requirements of safe operation of the active distribution network. S35. Based on the objective function and the constraints, optimize and solve the day-ahead source-grid-load-storage coordinated scheduling model to generate the day-ahead scheduling plan. The day-ahead scheduling plan includes a distributed renewable energy output plan, a distributed renewable energy curtailment plan, a network reconfiguration plan, a controllable load adjustment plan, an energy storage equipment charging and discharging plan, and a flexible adjustment equipment operation plan.

[0016] Furthermore, the step of constructing an intraday source-grid-load-storage coordinated scheduling model based on the intraday rolling forecast data and the day-ahead scheduling plan, and generating an intraday scheduling plan, specifically includes: S41. Based on the intraday rolling forecast data, determine the set of rolling optimization periods corresponding to the current intraday rolling moment. The intraday rolling forecast data includes the ultra-short-term available output forecast of distributed renewable energy and the ultra-short-term load demand forecast. The set of rolling optimization periods is represented as follows: in, Indicates the current intraday rolling time. The corresponding set of rolling optimization periods, This indicates the number of intraday scheduling periods included in the rolling optimization time period set; S42. Extract the day-ahead scheduling plan segment corresponding to the rolling optimization period set from the day-ahead scheduling plan, and use the day-ahead scheduling plan segment as the scheduling benchmark of the intraday source-grid-load-storage coordinated scheduling model; S43. Based on the intraday rolling forecast data, the day-ahead scheduling plan segment, and the current operating status of the active distribution network, determine the intraday decision variables of the intraday source-grid-load-storage coordinated scheduling model. The intraday decision variables include the actual output of distributed renewable energy, the abandoned power of distributed renewable energy, the controllable load adjustment power, the charging and discharging power of energy storage equipment, the power of energy storage equipment, the number of switching groups of grouped capacitors, the reactive power output of static var compensators, the tap position of on-load tap changers, and the status of branch switches in the distribution network. S44. Taking the maximization of distributed renewable energy consumption within the rolling optimization time period set as the objective, construct the objective function of the intraday source-grid-load-storage coordinated scheduling model. The objective function of the intraday source-grid-load-storage coordinated scheduling model is expressed as follows: in, This represents the amount of distributed renewable energy consumed within the set of rolling optimization periods. This represents the set of nodes connected to distributed renewable energy sources. Indicates the first Daily scheduling time period nodes The actual absorption capacity of distributed renewable energy, Indicates the length of the intraday scheduling period; S45. Based on the day-ahead scheduling plan segment and the basic operational data, construct the operational cost constraint of the intraday source-grid-load-storage coordinated scheduling model, so that the operational cost within the current rolling optimization period set is not higher than the sum of the operational cost corresponding to the day-ahead scheduling plan segment and the preset cost margin; S46. Construct the constraints of the intraday source-grid-load-storage coordinated scheduling model. The constraints include intraday operation constraints of distributed renewable energy, intraday adjustment constraints of controllable load, intraday operation constraints of energy storage equipment, intraday switching constraints of grouped switching capacitors, intraday output constraints of static var compensators, and intraday operation constraints of network topology. S47. Based on the objective function and the constraints, perform rolling optimization on the intraday source-grid-load-storage coordination scheduling model to obtain the intraday rolling optimization results within the current rolling optimization time period set; S48. The control quantity corresponding to the current intraday rolling moment in the intraday rolling optimization result is used as the actual intraday scheduling plan to be executed, and the updated intraday rolling prediction data is obtained again at the next intraday rolling moment to re-execute the intraday scheduling period that has not been executed.

[0017] Furthermore, the constraints of the intraday source-grid-load-storage coordinated scheduling model include: Distributed renewable energy intraday operating constraints, which include actual output constraints, curtailment constraints, capacity constraints, and ramping constraints, are expressed as follows: in, Indicates the first Daily scheduling time period nodes The actual absorption capacity of distributed renewable energy, Indicates the first Daily scheduling time period nodes Forecasted output of distributed renewable energy sources. Indicates the first Daily scheduling time period nodes Distributed renewable energy curtailment power, Indicates the first Daily scheduling time period nodes Distributed renewable energy reactive power, Represents a node The capacity of distributed renewable energy, Represents a node The ramp-up limit for distributed renewable energy; Controllable load intraday adjustment constraints, which include controllable load adjustment capacity constraints and electricity consumption conservation constraints during the rolling optimization period, are expressed as follows: in, Indicates the first Daily scheduling time period nodes The controllable load active power after adjustment Indicates the first Daily scheduling time period nodes Adjusting the active power of the preload, Represents a node The maximum adjustment ratio of the controllable load. Indicates the length of the intraday scheduling period; The daily operating constraints of energy storage devices include recursive constraints on energy storage capacity, energy storage capacity constraints, mutual exclusion constraints on charge / discharge states, and constraints on charge / discharge power, expressed as follows: in, Indicates the first Energy storage equipment during the daily dispatch period The amount of electricity, Indicates the energy storage equipment during the dispatch period of the next day. The amount of electricity, Indicates the first Energy storage equipment during the daily dispatch period The charging power, Indicates the first Energy storage equipment during the daily dispatch period The discharge power, Indicates energy storage devices Charging efficiency, Indicates energy storage devices The discharge efficiency, and These represent energy storage devices. The lower and upper limits of battery capacity. and These represent energy storage devices. In the Charging and discharging status variables for each daily scheduling period. and These represent energy storage devices. The lower limit and upper limit of charging power, and These represent energy storage devices. The lower limit and upper limit of discharge power; The intraday switching constraint for grouped capacitors is expressed as follows: In the formula, Indicates the first Daily scheduling time period nodes The reactive power compensation of the grouped switching capacitors. Indicates the first Daily scheduling time period nodes The number of switching groups of the capacitor bank. Represents a node The reactive power compensation for each group of capacitors switched on and off. Represents a node The maximum number of capacitor groups allowed to be switched on at the designated location. Represents a node The maximum number of operations allowed for grouped switching of capacitors within the rolling optimization time period set; The daily output constraint of the static var compensator is expressed as: in, Indicates the first Daily scheduling time period nodes The reactive power output of the static var compensator. and Representing nodes respectively The lower limit and upper limit of reactive power output of the static var compensator; Intraday network topology operational constraints, including branch switch operation count constraints and radial topology constraints, are expressed as follows: in, Indicates the first Branch lines during the daily dispatch period The switch state variables, Indicates the branch line during the dispatch period of the previous day. The switch state variables, Indicates the first Branch lines during the daily dispatch period The switching action variable, Indicates a branch The maximum number of switching operations allowed within the rolling optimization time period set. Indicates the set of active distribution network branches. Indicates the total number of active distribution network nodes. This indicates the number of root nodes in the active distribution network.

[0018] Furthermore, based on the power flow relationship of the active distribution network, second-order cone relaxation is applied to the nonlinear power flow constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model, specifically including: To address the active distribution network power flow constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model, node voltage square variables and branch current square variables are introduced to replace the node voltage amplitude and branch current amplitude. Based on the squared variables of the node voltages, the squared variables of the branch currents, the active power of the branches, and the reactive power of the branches, the nonlinear branch power relationships in the power flow constraints of the active distribution network are transformed into second-order cone constraints, which are expressed as follows: in, Represents the L2 norm, Indicates the first Branches in each scheduling period Active power transmitted upstream, Indicates the first Branches in each scheduling period reactive power transmitted upstream, Indicates the first Branches in each scheduling period The squared variable of the branch current, Indicates the first The first node of the branch during each scheduling period The squared variable of the node voltage; The second-order cone constraint, together with the branch active power balance constraint, the branch reactive power balance constraint, the node voltage drop constraint, and the active distribution network operation safety constraint, are used as the relaxed power flow constraint and are used to construct the second-order cone planning and scheduling model.

[0019] Furthermore, the linearization of the nonlinear terms in the model to obtain the second-order cone programming scheduling model specifically includes: The nonlinear terms in the day-ahead source-grid-load-storage coordination and scheduling model and the intraday source-grid-load-storage coordination and scheduling model are identified. The nonlinear terms include nonlinear terms generated by multiplying equipment state variables with continuous operation variables, nonlinear terms generated by absolute value operations, and square nonlinear terms generated by distributed renewable energy capacity constraints. For the nonlinear terms generated by multiplying equipment state variables and continuous operating variables, the Big M method is used to establish a linear relationship between equipment state variables and continuous operating variables, so that the corresponding continuous operating variables are zero when the equipment is in an unused state, and the corresponding continuous operating variables take values ​​within the allowable range when the equipment is in an used state. For nonlinear terms generated by absolute value operations, auxiliary variables are introduced for equivalent linearization, such that the auxiliary variables are not less than the variable to be linearized and its opposite, and the auxiliary variables are substituted into the corresponding cost term or action number term. To address the squared nonlinear term in the capacity constraint of distributed renewable energy, a piecewise linearization approach is adopted. The capacity arc of distributed renewable energy is divided into multiple linear segments, and the capacity arc constraint is replaced by a linear inequality formed by each linear segment. This linear inequality is expressed as: in, Indicates the first Each scheduling period node Active power of distributed renewable energy sources Indicates the first Each scheduling period node Reactive power of distributed renewable energy sources, , and Representing nodes respectively The first arc of distributed renewable energy capacity The linearization coefficients of each linear segment. This represents the total number of linear segments constraining the capacity of distributed renewable energy sources. The power flow constraints after second-order cone relaxation, the nonlinear terms after linearization, and the remaining linear constraints in the day-ahead source-grid-load-storage coordination scheduling model and the intraday source-grid-load-storage coordination scheduling model are combined to obtain the second-order cone planning scheduling model.

[0020] Furthermore, S6 specifically includes: The objective function, constraints, and decision variables formed by the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model after second-order cone relaxation and linearization are input into the second-order cone programming solver for solving, so as to obtain the optimal scheduling solution that satisfies the active distribution network operation safety constraints. Based on the optimal scheduling solution, extract the actual output of distributed renewable energy, the abandoned power of distributed renewable energy, the status of distribution network branch switches, the controllable load regulation power, the charging and discharging power of energy storage devices, the power of energy storage devices, the number of switching groups of grouped capacitors, the reactive power output of static var compensators, and the tap position of on-load tap changers from the day-ahead scheduling plan and the intraday scheduling plan. The execution period of the day-ahead scheduling plan is updated according to the day-ahead scheduling plan. The day-ahead scheduling plan is used as the actual execution plan for scheduling periods that have entered the scope of the day-ahead rolling optimization execution, while the day-ahead scheduling plan is retained as the subsequent scheduling benchmark for scheduling periods that have not entered the scope of the day-ahead rolling optimization execution. The distributed renewable energy output command is generated based on the actual output and curtailment of distributed renewable energy in the actual execution plan. The network reconfiguration instruction is generated based on the status of the distribution network branch switches in the actual execution plan; The controllable load adjustment command is generated based on the controllable load adjustment power in the actual execution plan; The energy storage charging and discharging command is generated based on the energy storage device's charging and discharging power and energy storage device's power in the actual execution plan. Based on the number of capacitor switching groups, the reactive power output of the static var compensator, and the tap position of the on-load tap changer in the actual execution plan, control commands for the flexible regulating equipment are generated.

[0021] Compared with the prior art, the present invention has the following advantages: (1) Existing active distribution network dispatching methods typically use conventional forecasting results directly as optimization inputs, which are difficult to adapt to the random fluctuations in wind power and photovoltaic output and load demand, and are prone to causing large deviations between day-ahead plans and intraday operating conditions. This invention obtains day-ahead forecasting data and intraday rolling forecasting data by acquiring basic operating data and using a BP neural network optimized by a genetic algorithm to predict distributed renewable energy output and load demand. This improves the reliability of forecasting inputs, reduces the transmission of forecasting errors to the dispatching model, and reduces wind and solar curtailment and ineffective regulation caused by forecasting deviations.

[0022] (2) Existing source-grid-load-storage scheduling methods mostly focus on regulating some resources, with insufficient coordination among the source side, grid side, load side, and energy storage side, which easily leads to local optima. This invention constructs a day-ahead source-grid-load-storage coordinated scheduling model and an intraday source-grid-load-storage coordinated scheduling model, which integrates distributed renewable energy, network topology, controllable load, energy storage equipment, and flexible regulation equipment into the regulation process, enabling multiple types of resources to participate in optimization in a coordinated manner, thereby improving the overall regulation capability of the active distribution network and improving the consumption of new energy and the economic efficiency of operation.

[0023] (3) Although existing multi-timescale scheduling methods divide the day-ahead and intraday phases, the intraday phase lacks the ability to dynamically correct the day-ahead plan and is difficult to adapt to short-term forecast changes in a timely manner. This invention generates a day-ahead scheduling plan based on the day-ahead forecast data, and constructs an intraday source-grid-load-storage coordinated scheduling model based on the intraday rolling forecast data and the day-ahead scheduling plan. It performs rolling optimization on the unexecuted periods, thereby enhancing the scheduling plan's ability to track short-term fluctuations and improving the continuity of regulation and operational adaptability.

[0024] (4) Existing intraday scheduling methods often aim to minimize operating costs or deviations, which can easily overlook short-term renewable energy consumption needs. In the intraday source-grid-load-storage coordinated scheduling model, this invention aims to maximize the consumption of distributed renewable energy within the rolling optimization period, and generates intraday scheduling plans by combining operating cost constraints and source-grid-load-storage operating constraints. This improves the actual consumption power of renewable energy while controlling operating costs and reducing the risk of power curtailment.

[0025] (5) Existing scheduling methods do not make sufficient use of the dynamic adjustment capabilities of the network side, and it is usually difficult to alleviate local power flow congestion and voltage overruns through changes in network structure. This invention incorporates the branch switch status of the distribution network and network topology operation constraints into the scheduling model, and generates network reconfiguration instructions based on the optimization results. This enables the power flow path to be adjusted according to load distribution and changes in new energy output, thereby reducing the risk of local overruns and network losses and improving the spatial coordination capability of the distribution network.

[0026] (6) Existing flexible regulation equipment modeling often only considers the output range and ignores the number of equipment actions and operating boundaries, which can easily lead to the optimization results being difficult to implement in practice. This invention incorporates on-load tap-changing transformers, group-switched capacitors and static var compensators into the flexible regulation equipment data and scheduling constraints, and generates corresponding control commands, thereby reducing frequent switching and over-limit regulation, and improving the executability of the regulation results and the stability of equipment operation.

[0027] (7) Existing energy storage scheduling easily overlooks state of charge, mutual exclusion of charging and discharging, and continuity of power across time periods, resulting in insufficient executability of energy storage plans. This invention sets relevant constraints on energy storage charging and discharging power and energy storage capacity in the day-ahead scheduling plan and intraday scheduling plan, and generates energy storage charging and discharging instructions, so that energy storage can absorb power when new energy sources are in surplus and release power when the load is at its peak, thereby improving the continuity of energy storage participation in the coordinated regulation of source, grid, load and storage.

[0028] (8) Existing controllable load scheduling, if it only focuses on load reduction, is likely to affect the energy demand of users and the acceptability of actual response. This invention incorporates the controllable load adjustment capability into the basic operational data and sets controllable load operation constraints in the scheduling model. Based on the optimization results, controllable load adjustment instructions are generated, thereby achieving the transfer of electricity consumption periods while maintaining the load adjustment boundary, alleviating peak load pressure and increasing the space for new energy consumption.

[0029] (9) Existing active distribution network optimization models typically have non-convex, nonlinear, and mixed-integer characteristics, making direct solutions computationally intensive and yielding unstable results. This invention performs second-order cone relaxation on the nonlinear power flow constraints and linearizes the nonlinear terms in the model to obtain a second-order cone planning scheduling model, thereby improving the solvability and stability of the model and enabling efficient solutions to complex source-grid-load-storage hierarchical rolling regulation problems.

[0030] (10) Existing model conversion methods do not handle capacity constraints, equipment state variables and continuous operation variables with sufficient precision, which can easily lead to oversimplification of the model or difficulty in solving it. This invention uses second-order cone relaxation, piecewise linearization, the Big M method and auxiliary variables to handle related nonlinear terms, so that power flow constraints, capacity boundaries and equipment switching states can be expressed in a unified model, thereby taking into account the model accuracy, solution efficiency and the actual executability of scheduling results. Attached Figure Description

[0031] Figure 1 This is a flowchart of a hierarchical rolling control method for source-grid-load-storage systems in an active distribution network according to an embodiment of the present invention. Figure 2 This is a diagram of the BP neural network topology according to an embodiment of the present invention; Figure 3 The flowchart below shows the BP neural network improved by genetic algorithm according to an embodiment of the present invention. Figure 4 This diagram illustrates how a genetic algorithm improves the predictive ability of a BP neural network according to an embodiment of the present invention. Figure 5 This is a schematic diagram illustrating the multi-timescale optimization of an embodiment of the present invention; Figure 6 This is a graph showing the amount of electricity wasted according to an embodiment of the present invention; Figure 7This is a diagram showing the system node voltage and current distribution before scheduling in an embodiment of the present invention. Figure 8 This is a schematic diagram of the system node voltage and current after scheduling according to an embodiment of the present invention; Figure 9 This is a spatial and temporal distribution diagram of the ESS power output according to an embodiment of the present invention; Figure 10 This is a spatial and temporal distribution diagram of the CB power output according to an embodiment of the present invention; Figure 11 This is a spatial and temporal distribution diagram of the SVC output power according to an embodiment of the present invention; Figure 12 This is a day-ahead dispatch diagram of the distribution network according to an embodiment of the present invention; Figure 13 This is a daily dispatch diagram of the distribution network according to an embodiment of the present invention; Figure 14 This is a diagram showing the changes in the power distribution network topology according to an embodiment of the present invention. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0033] Example 1: like Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 As shown in the figure, this embodiment provides a hierarchical rolling scheduling method for source-grid-load-storage systems in an active distribution network, characterized by the following steps: S1. Obtain basic operational data of the active distribution network; In one specific implementation, the basic operational data can be collected by the distribution automation system, distributed power source monitoring device, energy storage management device, load management platform, and flexible regulation equipment control terminal, or it can be obtained by calling historical operation databases and scheduling plan databases. The purpose of obtaining the basic operational data is to uniformly describe the operating status of the source side, grid side, load side, energy storage side, and flexible regulation equipment in the active distribution network, so that subsequent prediction models, day-ahead scheduling models, and intraday rolling control models have consistent data inputs.

[0034] The operational foundational data includes distributed renewable energy data, load data, network topology data, energy storage device data, and flexible regulation device data. Distributed renewable energy data includes historical output data, access node data, available output data, and operational constraint data, used to determine the predicted input, adjustable range, and actual absorption boundary of distributed renewable energy. Load data includes historical load data for each node, load forecasting characteristics data, and controllable load regulation capacity data, used to distinguish between uncontrollable load demand and load resources that can participate in regulation.

[0035] Network topology data includes node data, branch data, root node data, sectionalizer switch status data, and network operation safety limit data for the active distribution network. This data is used to determine the network boundaries required for distribution network power flow calculations, network reconfiguration, and safety verification. Energy storage device data includes access node data, state of charge data, capacity limit data, charge / discharge power limit data, and charge / discharge status data. This data is used to constrain the charge / discharge capacity and power changes of energy storage devices during subsequent regulation. Flexible regulation device data includes on-load tap-changing transformer data, grouped capacitor switching data, and static var compensator data. This data is used to determine voltage regulation and reactive power compensation capabilities.

[0036] After obtaining the above data, it can be organized according to node number, device number, and time identifier, so that the data of various devices can correspond to the nodes, branches, and scheduling periods in the active distribution network. This processing can reduce the problems of time inconsistency and unclear device correspondence between different data sources, and provide a complete data foundation for subsequent distributed renewable energy output forecasting, load demand forecasting, and layered rolling regulation of source-grid-load-storage systems.

[0037] S2. Based on the basic operational data, a BP neural network optimized by a genetic algorithm is used to predict the output and load demand of distributed renewable energy, and obtain day-ahead forecast data and intraday rolling forecast data. In one specific implementation, a prediction sample set is constructed based on operational baseline data. This prediction sample set may include historical output of distributed renewable energy, available output, meteorological correlation characteristics, historical load curves, date type, time period information, and controllable load change characteristics. The prediction sample set is divided into day-ahead prediction samples and intraday rolling prediction samples according to the prediction time scale. The day-ahead prediction samples are used to obtain predicted values ​​for distributed renewable energy output and load demand within the day-ahead scheduling cycle, while the intraday rolling prediction samples are used to obtain predicted values ​​for distributed renewable energy output and load demand within the rolling optimization period.

[0038] like Figure 2 , Figure 3 , Figure 4As shown, in this embodiment, a BP neural network is constructed to predict the output and load demand of distributed renewable energy. The BP neural network includes an input layer, a hidden layer, and an output layer. The input layer takes into account distributed renewable energy data and load data, the hidden layer extracts the nonlinear relationships in the input data, and the output layer outputs the predicted output and load demand values ​​of distributed renewable energy. Since wind power, photovoltaic power, and load are volatile, randomly assigning the initial weights and thresholds of the BP neural network can easily cause the training results to fall into local optima. Therefore, a genetic algorithm is used to optimize the network parameters of the BP neural network before training.

[0039] In this embodiment, the network parameters of the BP neural network are encoded with real numbers to obtain the initial population of the genetic algorithm. The network parameters include the connection weights between the input layer and the hidden layer, the connection weights between the hidden layer and the output layer, the hidden layer threshold, the output layer threshold, and the neuron smoothing factor. Each individual in the initial population corresponds to a set of candidate network parameters. After decoding each individual, the parameters are assigned to the BP neural network, and the fitness value is calculated based on the prediction output error.

[0040] The predicted output error satisfies: Fitness values ​​satisfy: in, Indicates the prediction output error. Indicates the number of training samples. Indicates the training sample number. Indicates the first The actual output value corresponding to each training sample. Indicates the first The predicted output value corresponding to each training sample. This represents the fitness value of an individual. This indicates a positive number used to avoid a denominator of zero; the fitness value is inversely related to the prediction output error. The smaller the prediction output error, the higher the fitness value of the corresponding individual, and the more likely the network parameters contained in that individual are to be preserved. This setting allows the genetic algorithm to continue searching around the network parameter region with the smaller prediction error, reducing the impact of the randomness of the initial values ​​during BP neural network training on prediction accuracy.

[0041] In this embodiment, individuals in the initial population are selected based on their fitness values, retaining those with higher fitness values ​​as parents. Subsequently, real crossover and mutation operations are performed on the parent individuals to obtain an updated population.

[0042] The real-number crossover operation includes: selecting two individuals from the parent generation to participate in the crossover, and recombining these two individuals using real-number encoding based on the crossover coefficient to obtain two offspring individuals. The formula is as follows: in, and They represent the first The encoding vectors of the two parent individuals that participate in the real crossover operation in each genetic iteration. and Let represent the encoding vectors of the two offspring individuals generated after the real number crossover operation. Indicates the cross coefficient. This represents the minimum value function; the crossover coefficients decrease with the number of genetic iterations, giving the early crossover process a large search range, while the later crossover process gradually stabilizes. This balances global search and local convergence, preventing the network parameters from fixing too early in a locally optimal region.

[0043] The mutation operation includes: perturbing the coding genes of the individuals to be mutated in the updated population, and determining the direction and magnitude of the perturbation based on a random factor, using the following formula: in, in, Indicates the first Among the individuals, the first The gene value of each coding gene. This indicates the upper bound of the gene encoding the gene. This indicates the lower bound of the gene encoding the gene. This represents a variable-length function that varies with the number of iterations. Indicates the current iteration number. Indicates the maximum number of iterations. This represents the random factor used to determine the direction of mutation. This represents a random number located in the interval [0,1].

[0044] The aforementioned variable-length function decreases with the number of iterations, giving the genetic algorithm a strong ability to escape local optima in the early stages and reducing perturbation amplitude in the later stages, thus preventing significant damage to network parameters that have already approached optimality. The combination of crossover and mutation operations can improve the stability of parameter optimization in the BP neural network.

[0045] In this embodiment, it is determined whether the updated population meets a preset stopping condition. The preset stopping condition may be reaching the maximum number of iterations, the change in fitness value being less than a set threshold over multiple consecutive generations, or the fitness value reaching a set target. If the updated population does not meet the preset stopping condition, the updated population is treated as a new population, and fitness calculation, selection, crossover, and mutation continue; if the updated population meets the preset stopping condition, the individual with the highest fitness value is output as the optimal individual.

[0046] In this embodiment, the optimal individual is decoded to obtain the optimized BP neural network connection weights, thresholds, and smoothing factors. These values ​​are then assigned to the BP neural network to obtain a genetic algorithm-optimized BP neural network. Subsequently, the basic operating data is input into the genetic algorithm-optimized BP neural network to predict the output and load demand of distributed renewable energy on the day-ahead timescale and on the intraday rolling timescale, respectively, obtaining day-ahead and intraday rolling forecast data. This process improves the matching degree between the predicted data and the actual operating status, providing more reliable input for subsequent day-ahead scheduling and intraday rolling control.

[0047] S3. Based on the day-ahead forecast data, construct a day-ahead source-grid-load-storage coordinated scheduling model and generate a day-ahead scheduling plan; In one specific implementation, such as Figure 5 As shown, the day-ahead forecast data includes the predicted available output of distributed renewable energy and the predicted load demand of each node for each scheduling period within the day-ahead scheduling cycle. The day-ahead scheduling cycle can be the next day, and the scheduling periods can be divided into 1-hour segments or adjusted according to the accuracy requirements of active distribution network scheduling. By using the day-ahead forecast data, the changing trends of renewable energy output and load demand within the future scheduling cycle can be obtained in advance, providing a benchmark for subsequent source-grid-load-storage coordinated optimization.

[0048] In this embodiment, the decision variables for the day-ahead source-grid-load-storage coordinated dispatch model are determined based on the predicted available output of distributed renewable energy, the predicted load demand of each node, and operational baseline data. These decision variables include the actual output of distributed renewable energy, the curtailed power of distributed renewable energy, the power exchanged with the upstream grid, the status of branch switches in the distribution network, the tap position of on-load tap changers, the number of switching capacitor banks, the reactive power output of the static var compensator, the controllable load regulation power, the charging and discharging power of energy storage devices, and the state of charge of energy storage devices. These decision variables correspond to the source side, grid side, load side, energy storage side, and flexible regulation devices, respectively, enabling the day-ahead dispatch model to simultaneously consider renewable energy consumption, network topology adjustment, load transfer, energy storage charging and discharging, and reactive power and voltage regulation.

[0049] In this embodiment, the objective function of the day-ahead source-grid-load-storage coordinated scheduling model is constructed with the goal of minimizing the operating cost of the active distribution network within the day-ahead scheduling cycle. The objective function of the day-ahead source-grid-load-storage coordinated scheduling model is expressed as follows: Among them, the cost of flexible resource scheduling satisfies: in, This represents the total operating cost of the active distribution network during the current day's dispatch cycle; This indicates the total number of scheduling periods within the current day's scheduling cycle; Indicates the first The cost of curtailed distributed renewable energy during each scheduling period; Indicates the first The cost of purchasing electricity from the upper-level power grid for each dispatch period; Indicates the first Network loss cost per scheduling period; Indicates the first The cost of flexible resource scheduling for each scheduling period; Indicates the first Equipment operating costs for each scheduling period; Indicates the first Distributed renewable energy reactive power regulation cost per scheduling period; Indicates the first The static var compensator adjustment cost for each scheduling period; Indicates the first Cost of switching capacitors in groups for each scheduling period; Indicates the first Controllable load adjustment cost for each scheduling period; Indicates the first Operating costs of energy storage equipment during each scheduling period; The aforementioned objective function integrates the costs of curtailment, power purchase, grid loss, flexible resource dispatch, and equipment operation into the day-ahead optimization objective. Curtailment costs constrain renewable energy curtailment behavior; power purchase and grid loss costs reflect the economic efficiency of distribution network operation; flexible resource dispatch costs measure the regulatory costs associated with distributed renewable energy reactive power regulation, SVC regulation, CB switching, demand response, and energy storage operation; and equipment operation costs suppress excessive operation of switches, tap changers, and switching equipment. This cost combination avoids the problem of excessively frequent equipment operation or high operating costs due to solely pursuing renewable energy consumption, and also avoids sacrificing renewable energy consumption by simply reducing operating costs.

[0050] In this embodiment, the constraints of the day-ahead source-grid-load-storage coordinated dispatch model include distributed renewable energy operation constraints, network topology operation constraints, controllable load operation constraints, energy storage device operation constraints, flexible regulation device operation constraints, and active distribution network power flow constraints. Distributed renewable energy operation constraints limit the actual output of new energy sources, curtailed power, power factor, capacity, and ramp-up variations; network topology operation constraints limit branch switch states, number of switch operations, and radial operating structure; controllable load operation constraints limit the adjustable range of load and the balance of electricity consumption within the dispatch cycle; energy storage device operation constraints limit the state of charge, charging and discharging power, and mutual exclusion of charging and discharging states; flexible regulation device operation constraints limit the adjustment range of on-load tap-changing transformers, grouped switching capacitors, and static var compensators; and active distribution network power flow constraints ensure that node voltage, branch current, and power flow meet safe operation requirements.

[0051] In this embodiment, the day-ahead source-grid-load-storage coordinated scheduling model is optimized and solved according to the objective function and constraints to generate a day-ahead scheduling plan. The day-ahead scheduling plan includes distributed renewable energy output plans, distributed renewable energy curtailment plans, network reconfiguration plans, controllable load adjustment plans, energy storage device charging and discharging plans, and flexible regulation device operation plans. As the basic plan for the future scheduling cycle of the active distribution network, the day-ahead scheduling plan is used to coordinate various regulation resources in advance and provide a scheduling benchmark for subsequent intraday rolling control.

[0052] S4. Based on the intraday rolling forecast data and the day-ahead scheduling plan, construct an intraday source-grid-load-storage coordinated scheduling model and generate an intraday scheduling plan; In one specific implementation, intraday rolling forecast data is used to correct the unexecuted portions of the day-ahead dispatch plan. Because wind power, solar power, and load can change short-term during the day due to weather and user behavior, directly executing the day-ahead dispatch plan can easily lead to power curtailment, power flow exceeding limits, or energy storage regulation deviations. Therefore, a rolling optimization approach is adopted during the intraday phase. At each intraday rolling moment, the latest forecast data is reacquired, and only the control variables corresponding to the current rolling moment are executed.

[0053] Based on intraday rolling forecast data, the set of rolling optimization periods corresponding to the current intraday rolling moment is determined. The intraday rolling forecast data includes the ultra-short-term available output forecast of distributed renewable energy and the ultra-short-term load demand forecast. The set of rolling optimization periods is represented as follows: in, Indicates the current intraday rolling time. The corresponding set of rolling optimization periods, This indicates the number of intraday scheduling periods included in the rolling optimization time period set; Extracting day-ahead scheduling plan segments corresponding to the rolling optimization period set from the day-ahead scheduling plan, and using these day-ahead scheduling plan segments as the scheduling benchmark for the intraday source-grid-load-storage coordinated scheduling model. This approach ensures that intraday scheduling is not completely replanned, but rather finely revised based on the day-ahead plan, thereby maintaining the continuity of the scheduling plan and reducing frequent changes in equipment status.

[0054] Based on intraday rolling forecast data, day-ahead dispatch plan segments, and the current operating status of the active distribution network, intraday decision variables for the intraday source-grid-load-storage coordinated dispatch model are determined. Intraday decision variables include actual output of distributed renewable energy, curtailed power of distributed renewable energy, controllable load regulation power, charging and discharging power of energy storage devices, energy storage device capacity, number of switching capacitor banks, reactive power output of static var compensators, tap position of on-load tap changers, and the status of branch switches in the distribution network. By simultaneously setting these variables, the intraday model can coordinate choices among renewable energy absorption, voltage regulation, network reconfiguration, load transfer, and energy storage charging and discharging.

[0055] With the goal of maximizing the absorption of distributed renewable energy within the rolling optimization time period set, an objective function for the intraday source-grid-load-storage coordinated scheduling model is constructed. The objective function of the intraday source-grid-load-storage coordinated scheduling model is expressed as follows: in, This represents the amount of distributed renewable energy consumed within the rolling optimization period set. This represents the set of nodes connected to distributed renewable energy sources. Indicates the first Daily scheduling time period nodes The actual absorption capacity of distributed renewable energy, Indicates the length of the intraday scheduling period; To avoid excessively high operating costs due to solely pursuing renewable energy consumption, operating cost constraints can be constructed based on day-ahead scheduling plan segments and basic operational data. These constraints ensure that the operating cost within the current rolling optimization period set does not exceed the sum of the operating cost corresponding to the day-ahead scheduling plan segment and the preset cost margin, expressed as: in, This represents the intraday operating cost within the current rolling optimization time period set. This indicates the operating cost corresponding to the current day's scheduling plan segment. This indicates the allowable cost margin. This constraint is used to limit the cost increase brought about by intraday adjustments, ensuring that intraday regulation maintains economic efficiency while improving the capacity for renewable energy absorption.

[0056] The constraints of the intraday source-grid-load-storage coordination and scheduling model include intraday operating constraints of distributed renewable energy, intraday adjustment constraints of controllable load, intraday operating constraints of energy storage equipment, intraday switching constraints of grouped capacitors, intraday output constraints of static var compensators, intraday tap position constraints of on-load tap changers, and intraday operating constraints of network topology.

[0057] Distributed renewable energy intraday operating constraints include actual output constraints, curtailment constraints, capacity constraints, and ramping constraints, expressed as follows: in, Indicates the first Daily scheduling time period nodes The actual absorption capacity of distributed renewable energy, Indicates the first Daily scheduling time period nodes Forecasted output of distributed renewable energy sources. Indicates the first Daily scheduling time period nodes Distributed renewable energy curtailment power, Indicates the first Daily scheduling time period nodes Distributed renewable energy reactive power, Represents a node The capacity of distributed renewable energy, Represents a node The ramp-up limits for distributed renewable energy are defined as follows: the first formula is used to distinguish between available output, actual power consumption, and abandoned power; the second formula is used to prevent the combined output of active and reactive power from exceeding the equipment capacity; and the third formula is used to limit sudden changes in output between adjacent time periods to avoid excessive impact of dispatch instructions on power generation equipment.

[0058] Controllable load intraday adjustment constraints include controllable load adjustment capacity constraints and electricity consumption conservation constraints within the rolling optimization period, expressed as: in, Indicates the first Daily scheduling time period nodes The controllable load active power after adjustment Indicates the first Daily scheduling time period nodes Adjusting the active power of the preload, Represents a node The maximum adjustment ratio of the controllable load. This indicates the length of the intraday scheduling period; the first formula limits the adjustment range of controllable load, and the second formula ensures that the total electricity consumption remains unchanged during the rolling optimization period, so that load regulation reflects the shift of electricity consumption periods rather than simply reducing users' electricity demand.

[0059] Daily operating constraints for energy storage devices include recursive constraints on energy storage capacity, energy storage capacity constraints, mutual exclusion constraints on charge / discharge states, and constraints on charge / discharge power, expressed as follows: in, Indicates the first Energy storage equipment during the daily dispatch period The amount of electricity, Indicates the energy storage equipment during the dispatch period of the next day. The amount of electricity, Indicates the first Energy storage equipment during the daily dispatch period The charging power, Indicates the first Energy storage equipment during the daily dispatch period The discharge power, Indicates energy storage devices Charging efficiency, Indicates energy storage devices The discharge efficiency, and These represent energy storage devices. The lower and upper limits of battery capacity. and These represent energy storage devices. In the Charging and discharging status variables for each daily scheduling period. and These represent energy storage devices. The lower limit and upper limit of charging power, and These represent energy storage devices. The lower limit and upper limit of discharge power; The intraday switching constraint for grouped capacitors is expressed as follows: In the formula, Indicates the first Daily scheduling time period nodes The reactive power compensation of the grouped switching capacitors. Indicates the first Daily scheduling time period nodes The number of switching groups of the capacitor bank. Represents a node The reactive power compensation for each group of capacitors switched on and off. Represents a node The maximum number of capacitor groups allowed to be switched on at the designated location. Represents a node The maximum number of operations allowed for grouped switching of capacitors within the rolling optimization time period set; The daily output constraint of the static var compensator is expressed as follows: in, Indicates the first Daily scheduling time period nodes The reactive power output of the static var compensator. and Representing nodes respectively The lower and upper limits of reactive power output of the static var compensator (SVC) are defined; this constraint is used to limit the range of reactive power that the SVC can provide or absorb, so that the intraday voltage regulation remains within the equipment's capacity.

[0060] The intraday tap position constraint of an on-load tap-changing transformer is expressed as follows: in, Indicates the first During the daily dispatch period, there is a voltage-changing transformer. The tap position, and These represent on-load tap-changing transformers. The lower and upper limits of the tap position, Indicates an on-load tap-changing transformer The maximum number of tap adjustments allowed within the rolling optimization period set. This constraint is used to align voltage regulation with the equipment's operational lifespan and prevent frequent changes in tap position during rolling control.

[0061] Intraday network topology operational constraints include branch switch operation count constraints and radial topology constraints, expressed as: in, Indicates the first Branch lines during the daily dispatch period The switch state variables, Indicates the branch line during the dispatch period of the previous day. The switch state variables, Indicates the first Branch lines during the daily dispatch period The switching action variable, Indicates a branch The maximum number of switching operations allowed within the rolling optimization time period set. Indicates the set of active distribution network branches. Indicates the total number of active distribution network nodes. This represents the number of root nodes in the active distribution network. The constraint on the number of switching operations is used to avoid frequent network reconfiguration, while the constraint on the number of closed branches is used to maintain the radial operating structure of the distribution network, ensuring that network topology adjustments do not disrupt the safe operation of the active distribution network.

[0062] Based on the objective function and the aforementioned constraints, a rolling optimization solution is performed on the intraday source-grid-load-storage coordinated scheduling model to obtain the intraday rolling optimization results within the current rolling optimization period set. In the intraday rolling optimization results, only the control quantity corresponding to the current intraday rolling moment is used as the actual intraday scheduling plan; the remaining unexecuted periods are used as reference results. Upon entering the next intraday rolling moment, updated intraday rolling forecast data is re-acquired, and the unexecuted intraday scheduling periods are re-optimized. This rolling execution method can enhance the adaptability of the active distribution network to short-term fluctuations in new energy sources and loads while maintaining the continuity of regulation.

[0063] S5. Based on the power flow relationship of the active distribution network, the nonlinear power flow constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model are relaxed by second-order cone relaxation, and the nonlinear terms in the model are linearized to obtain the second-order cone planning scheduling model. In one specific implementation, both the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model include active distribution network power flow constraints. Because the power flow constraints involve nonlinear relationships between voltage, current, active power, and reactive power, direct solution would result in a non-convex nonlinear optimization problem in the scheduling model, making it difficult to solve. Therefore, when constructing the second-order cone programming scheduling model, the nonlinear branch power relationships in the power flow constraints are first relaxed.

[0064] Specifically, the square variables of node voltage and branch current are introduced to replace the magnitudes of node voltage and branch current: in, Indicates the first Each scheduling period node The squared variable of the node voltage, Indicates the first Each scheduling period node The node voltage amplitude, Indicates the first Branches in each scheduling period The squared variable of the branch current, Indicates the first Branches in each scheduling period The amplitude of the branch current.

[0065] By using squared variables, the nonlinear relationship between branch power, voltage, and current can be transformed into a form suitable for second-order cone relaxation. Furthermore, based on the squared variables of node voltage, branch current, branch active power, and branch reactive power, the nonlinear branch power relationship is transformed into a second-order cone constraint, which is expressed as: in, Represents the L2 norm, Indicates the first Branches in each scheduling period Active power transmitted upstream, Indicates the first Branches in each scheduling period reactive power transmitted upstream, Indicates the first Branches in each scheduling period The squared variable of the branch current, Indicates the first The first node of the branch during each scheduling period The squared variable of the node voltage; After the second-order cone relaxation is completed, the second-order cone constraint, along with the branch active power balance constraint, branch reactive power balance constraint, node voltage drop constraint, and active distribution network operation safety constraint, are used as the relaxed power flow constraints. The operation safety constraints may include the upper and lower limits of node voltage, the upper and lower limits of branch current, and the power limit for interaction with the upper-level grid, to ensure that both the day-ahead dispatch plan and the intraday dispatch plan meet the requirements for safe operation of the distribution network.

[0066] In this embodiment, nonlinear terms in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model are also identified. These nonlinear terms mainly include those generated by multiplying equipment state variables with continuous operating variables, those generated by absolute value operations, and squared nonlinear terms generated by distributed renewable energy capacity constraints.

[0067] For the nonlinear terms generated by multiplying equipment state variables and continuous operating variables, the Big M method is used to establish a linear relationship between the equipment state variables and continuous operating variables. After this processing, when the equipment is in an inactive state, the corresponding continuous operating variable is restricted to zero; when the equipment is in an active state, the corresponding continuous operating variable can take values ​​within an allowable range. This processing is applicable to the relationship between branch switch status and branch power, energy storage charging / discharging status and charging / discharging power, and the switching status of grouped capacitors and reactive power compensation, thus avoiding mismatches between equipment status and operating quantities.

[0068] For nonlinear terms generated by absolute value operations, auxiliary variables are introduced for equivalent linearization. These auxiliary variables are made no less than the variable to be linearized and its opposite. The auxiliary variables are then substituted into the corresponding cost or action count terms. This transforms the absolute value expressions of switching action counts, tap changer adjustments, switching operations, and deviation costs into linear constraints, facilitating their integration into the second-order cone programming scheduling model.

[0069] For the squared nonlinear term in the capacity constraint of distributed renewable energy, a piecewise linearization method is adopted. Specifically, the capacity arc of distributed renewable energy is divided into multiple linear segments, and the capacity arc constraint is replaced by a linear inequality formed by each linear segment. The linear inequality is expressed as: in, Indicates the first Each scheduling period node Active power of distributed renewable energy sources Indicates the first Each scheduling period node Reactive power of distributed renewable energy sources, , and Representing nodes respectively The first arc of distributed renewable energy capacity The linearization coefficients of each linear segment. This represents the total number of linear segments constraining the capacity of distributed renewable energy sources. By combining the power flow constraints after second-order cone relaxation, the nonlinear terms after linearization, and the remaining linear constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model, a second-order cone programming scheduling model is obtained. This model can express the operational constraints of multiple resources such as sources, grids, loads, and storage in the active distribution network, and can also be stably solved using a second-order cone programming solver, thus providing a computable optimization model for the subsequent generation of hierarchical rolling control results.

[0070] S6. Solve the second-order cone programming scheduling model to obtain the results of the active distribution network source-grid-load-storage hierarchical rolling regulation. Based on the results of the active distribution network source-grid-load-storage hierarchical rolling regulation, generate distributed renewable energy output commands, network reconfiguration commands, controllable load adjustment commands, energy storage charging and discharging commands, and flexible regulation equipment control commands. In one specific implementation, the objective function, constraints, and decision variables of the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model, after undergoing second-order cone relaxation and linearization, are input into a second-order cone programming solver for solving, yielding the optimal scheduling solution that satisfies the operational safety constraints of the active distribution network. The second-order cone programming solver can be an optimization solver capable of handling continuous variables, integer variables, and second-order cone constraints. Through the aforementioned second-order cone relaxation and linearization, the original scheduling problem, which involved nonlinear power flow relationships, discrete equipment states, and multi-time-period coupling relationships, is transformed into a planning model that is easier to solve, thereby improving the solution efficiency and the stability of the scheduling results.

[0071] In this embodiment, based on the optimal scheduling solution, the following variables are extracted from the day-ahead and intraday scheduling plans: actual output of distributed renewable energy, abandoned power of distributed renewable energy, status of distribution network branch switches, controllable load regulation power, charging and discharging power of energy storage devices, energy storage device power, number of switching groups of grouped capacitors, reactive power output of static var compensators, and tap position of on-load tap changers. These variables correspond to the actual execution quantities of the source side, grid side, load side, energy storage side, and flexible regulation equipment, respectively, enabling the optimization results to be converted into control content that can be recognized by the field equipment.

[0072] In this embodiment, the execution period of the day-ahead scheduling plan is updated based on the intraday scheduling plan. For scheduling periods that have already entered the intraday rolling optimization execution range, the intraday scheduling plan is used as the actual execution plan; for scheduling periods that have not yet entered the intraday rolling optimization execution range, the day-ahead scheduling plan is retained as the subsequent scheduling benchmark. This avoids scheduling discontinuity caused by intraday scheduling completely replacing the day-ahead plan, while using the latest intraday forecast data to correct the control quantities to be executed, making the control process both planned and adaptable in real time.

[0073] In this embodiment, a distributed renewable energy output instruction is generated based on the actual output and curtailment power of distributed renewable energy in the actual execution plan. This instruction is used to determine the active power output target and necessary curtailment control amount of each distributed renewable energy access node in the current execution period, so that the output of new energy sources matches the absorption capacity of the active distribution network, and reduces disorderly curtailment caused by local power flow constraints or voltage constraints.

[0074] Based on the actual distribution network branch switch status in the implementation plan, a network reconfiguration command is generated. This command controls the opening and closing status of corresponding sectional switches, ensuring the distribution network operates according to the optimized topology. This command allows for adjustment of power flow paths when load distribution changes or when renewable energy output is concentrated, mitigating local branch overload and voltage exceedance issues.

[0075] Based on the controllable load adjustment power in the actual execution plan, controllable load adjustment instructions are generated. These instructions are used to control adjustable loads to adjust power or shift time periods within permissible ranges, enabling the load side to participate in regulation in accordance with changes in renewable energy output and network operating status. This process can increase the capacity for renewable energy consumption and reduce localized peak loads without disrupting the overall electricity demand of users.

[0076] Based on the actual charging and discharging power and energy capacity of the energy storage devices in the execution plan, energy storage charging and discharging commands are generated. These commands include the charging status, discharging status, standby status, and corresponding charging and discharging power of the energy storage devices. Through these commands, energy storage devices can charge when there is a surplus of renewable energy and discharge during peak load periods or when there is a localized power shortage, thereby smoothing out fluctuations in distributed renewable energy output and improving the continuity of regulation.

[0077] Based on the actual implementation plan, including the number of capacitor banks to be switched, the reactive power output of the static var compensator (SVC), and the tap positions of the on-load tap-changing transformer (OTCT), control commands for the flexible regulation equipment are generated. These commands include capacitor bank switching commands, SVC reactive power regulation commands, and OCT tap position regulation commands. Through these commands, node voltage and reactive power flow can be coordinated and regulated, reducing voltage deviations caused by renewable energy integration and load fluctuations.

[0078] In this embodiment, after generating various control commands, a pre-execution check can be performed. The check includes verifying whether node voltages exceed limits, branch currents exceed limits, whether the network topology meets radial operation requirements, whether the energy storage capacity is within allowable limits, and whether the number of actions of flexible adjustment equipment meets limits. If the check passes, the various control commands are sent to the corresponding equipment; if the check fails, the process returns to the intraday source-grid-load-storage coordinated scheduling model to revise relevant constraints or control quantities. This process reduces inconsistencies between optimization results and on-site execution conditions, improving the executability and security of the active distribution network's layered rolling control of source, grid, load, and storage.

[0079] Example 2: The parts not mentioned in this embodiment are the same as in embodiment 1. This embodiment adopts the following... Figure 5 shown This embodiment uses, as follows: Figure 6The IEEE 33-node distribution system shown verifies the hierarchical rolling control method of source-grid-load-storage for active distribution networks. The voltage reference value of the IEEE 33-node distribution system is 12.66 kV, and the capacity reference value is 10 MVA. Grouped switching capacitors (CB) are connected to nodes 21 and 32; energy storage devices (ES) are connected to nodes 3, 8, and 16; photovoltaic (PV) power generation units (PV) are connected to nodes 6 and 24; wind power generation units (WT) are connected to nodes 16 and 29; and static var compensators (SVCs) are connected to nodes 5, 15, and 31.

[0080] To verify the control effect of the source-grid-load-storage hierarchical rolling control method, this embodiment sets up three operating scenarios. Scenario 1 adopts the source-grid-load-storage hierarchical rolling control method of the present invention; Scenario 2 adopts the optimized scheduling method under a single time scale; Scenario 3 adopts the scheduling method without considering the dynamic reconfiguration of the distribution network and the renewable energy consumption target. By comparing the three scenarios, the impact of hierarchical rolling control, network reconfiguration, and renewable energy consumption target on the operating cost of the active distribution network and the renewable energy consumption capacity can be analyzed.

[0081] The renewable energy configuration in the IEEE 33-node system is shown in Table 1.

[0082] Table 1. Renewable Energy Allocation The parameters of the energy storage equipment are shown in Table 2.

[0083] Table 2 ESS Parameters The parameters for group switching capacitors are shown in Table 3.

[0084] Table 3 CB Parameters The parameters of the static var compensator are shown in Table 4.

[0085] Table 4 SVC Parameters The parameters of the on-load tap-changing transformer are shown in Table 5.

[0086] Table 5 OLTC Parameters For a 33-node distribution network system, this embodiment employs the three scenarios described above for dispatching. The renewable energy absorption rate and operating costs of the distribution network are shown in Table 6, and the wind curtailment and voltage conditions for each time period are as follows. Figure 7 As shown.

[0087] Table 6 Operating Costs and Utilization Rate From Table 6 and Figure 7 It can be seen that Scenario 1 corresponds to the highest renewable energy absorption rate and the lowest operating cost. Compared with Scenario 2, Scenario 1 increases the renewable energy absorption rate by 5.35% and reduces operating costs by 7.96%; compared with Scenario 3, Scenario 1 increases the renewable energy absorption rate by 24.39% and reduces operating costs by 18.28%. These results indicate that, under the combined effect of day-ahead dispatching and intraday rolling control, the source side, grid side, load side, energy storage side, and flexible regulation equipment can form a coordinated regulation relationship, thereby improving renewable energy absorption capacity and reducing the operating costs of active distribution networks.

[0088] The economic cost breakdown for the three scenarios is shown in Table 7.

[0089] Table 7. Economic Costs of the Three Methods As shown in Table 7, compared with Scenario 2, the cost of curtailment in Scenario 1 is reduced by RMB 0.156 million; compared with Scenario 3, the cost of curtailment in Scenario 1 is reduced by RMB 0.69 million. Although Scenario 1 introduces flexible resources such as energy storage, controllable loads, reactive power compensation equipment, and network reconfiguration to participate in regulation, the flexibility and operating costs are still reduced by RMB 0.214 million compared with Scenario 3. This is because the tiered rolling regulation can coordinate the adjustment actions of different resources based on the day-ahead forecast data and the intraday rolling forecast data, reducing ineffective regulation and passive curtailment, and making the use of flexible resources more focused on the needs of renewable energy consumption and network security operation.

[0090] For the IEEE 33-node system, the distribution network scheduling operation results obtained using the method in this embodiment are as follows: Figures 7 to 13 As shown. Figure 7 and Figure 8 The voltage and current distributions before and after the scheduling are displayed respectively. Figure 7 and Figure 8 It can be seen that before the dispatch, due to the uneven distribution of distributed renewable energy access locations and load distribution, the distribution network suffered from overvoltage and large voltage fluctuations. After the method of this embodiment was used for regulation, the distribution of node voltage and branch current was improved, and the problems of large overvoltage and voltage deviation were alleviated.

[0091] Figure 9 Output the results to ES. Figure 10 Output the result for CB. Figure 11 This is the output of SVC. Combined with... Figure 7 and Figure 8The voltage and current distribution data shows that energy storage devices, group-switched capacitors, and static var compensators can make corresponding adjustments at nodes and periods where the distribution network voltage fluctuates significantly or is too high or too low. At nodes or periods with high load, the distribution network is prone to voltage drops and increased network losses. In this case, the reactive power compensation devices provide reactive power support, and the energy storage devices discharge to alleviate insufficient active power and low voltage. At nodes or periods with low load and high renewable energy output, the distribution network is prone to voltage rises. In this case, the static var compensators operate in a state of absorbing capacitive reactive power, and the energy storage devices charge to promote renewable energy absorption and improve voltage levels.

[0092] Figure 12 and Figure 13 These are the daily scheduling status and the intraday scheduling status, respectively. Figure 12 and Figure 13 It can be seen that intraday rolling control does not completely replace the day-ahead scheduling plan, but rather makes refined adjustments based on intraday rolling forecast data, making the control plan during non-execution periods closer to the actual operating state, thereby promoting the high-quality and efficient consumption of new energy. (Summary) Figures 7 to 13 It can be seen that before and after the interaction and regulation of power generation, grid, load and storage, the system voltage, current, energy storage output, reactive power compensation and intraday dispatch results have all been improved.

[0093] This embodiment also verifies the effect of dynamic reconfiguration of the distribution network. The network topology after dynamic reconfiguration of the distribution network is as follows: Figure 14 As shown in Table 8, the reconstruction results are as follows.

[0094] Table 8 Results of Reconstruction Depend on Figure 14 As shown in Table 8, after network reconfiguration, the total network loss of the distribution network in one day is 1925.13 kWh; without network reconfiguration, the total network loss is 2258.72 kWh. Compared with the unreconfigured scenario, the total network loss is reduced by 333.59 kWh after dynamic reconfiguration, and the reconfigured network still maintains a radial operating structure. Dynamic reconfiguration, by changing the branch switch status, can transfer some load to branches with lower load levels during peak load periods, and can also guide renewable energy power to other load areas when renewable energy output is high but local absorption capacity is insufficient, thereby achieving spatiotemporal coordination between power generation, grid, load, and storage.

[0095] This embodiment demonstrates that the source-grid-load-storage hierarchical rolling control method for active distribution networks can comprehensively utilize the regulation capabilities of distributed renewable energy, network topology, controllable loads, energy storage devices, and flexible regulation devices. The day-ahead scheduling plan provides the basic operating scheme, while intraday rolling control corrects unexecuted periods based on the latest forecast data. The second-order cone programming scheduling model improves the solution efficiency and stability of complex control models. Simulation results show that this method can improve the renewable energy absorption rate, reduce operating costs and curtailment costs, and improve operating indicators such as node voltage, branch current, and network losses, while meeting the safety operation requirements of active distribution networks.

[0096] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A hierarchical rolling dispatching method for source-grid-load-storage systems in an active distribution network, characterized in that, Includes the following steps: S1. Obtain basic operational data of the active distribution network; S2. Based on the aforementioned operational data, a BP neural network optimized by a genetic algorithm is used to predict the output and load demand of distributed renewable energy, thereby obtaining day-ahead forecast data and intraday rolling forecast data. S3. Based on the day-ahead forecast data, construct a day-ahead source-grid-load-storage coordinated scheduling model and generate a day-ahead scheduling plan; S4. Based on the intraday rolling forecast data and the day-ahead scheduling plan, construct an intraday source-grid-load-storage coordinated scheduling model and generate an intraday scheduling plan; S5. Based on the power flow relationship of the active distribution network, the nonlinear power flow constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model are relaxed by second-order cone relaxation, and the nonlinear terms in the model are linearized to obtain the second-order cone planning scheduling model. S6. Solve the second-order cone programming scheduling model to obtain the active distribution network source-grid-load-storage hierarchical rolling control results, and generate distributed renewable energy output commands, network reconfiguration commands, controllable load adjustment commands, energy storage charging and discharging commands, and flexible regulation equipment control commands based on the active distribution network source-grid-load-storage hierarchical rolling control results.

2. The hierarchical rolling dispatching method for source-grid-load-storage systems in an active distribution network according to claim 1, characterized in that, The basic operational data includes distributed renewable energy data, load data, network topology data, energy storage device data, and flexible regulation device data; the distributed renewable energy data includes historical output data, access node data, available output data, and operational limitation data of distributed renewable energy. The load data includes historical load data for each node, load characteristic data required for prediction, and controllable load adjustment capacity data. The network topology data includes node data, branch data, root node data, sectional switch status data, and network operation safety limit data of the active distribution network. The energy storage device data includes access node data, state of charge data, capacity limitation data, charge / discharge power limitation data, and charge / discharge status data. The data from the flexible regulation equipment includes on-load tap-changing transformer data, group switching capacitor data, and static var compensator data.

3. The hierarchical rolling dispatching method for source-grid-load-storage systems in an active distribution network according to claim 1, characterized in that, S2 specifically includes: S21. Based on the aforementioned operational data, construct a BP neural network for predicting the output and load demand of distributed renewable energy, and determine the input layer, hidden layer, and output layer of the BP neural network, wherein the input layer is used to input distributed renewable energy data and load data, and the output layer is used to output the predicted output value and the predicted load demand value of distributed renewable energy. S22. The network parameters of the BP neural network are encoded with real numbers to obtain the initial population of the genetic algorithm. The network parameters include the connection weights between the input layer and the hidden layer, the connection weights between the hidden layer and the output layer, the hidden layer threshold, the output layer threshold, and the neuron smoothing factor. S23. Assign each individual in the initial population as a set of candidate network parameters to the BP neural network, and calculate the fitness value of the corresponding individual based on the prediction output error of the BP neural network, wherein the prediction output error satisfies: The fitness value satisfies: in, Indicates the prediction output error. Indicates the number of training samples. Indicates the training sample number. Indicates the first The actual output value corresponding to each training sample. Indicates the first The predicted output value corresponding to each training sample. This represents the fitness value of an individual. This indicates a positive number used to avoid a denominator of zero; S24. Based on the fitness value, perform a selection operation on the individuals in the initial population, retain the individuals with higher fitness values ​​as parent individuals, and perform real crossover and mutation operations on the parent individuals to obtain an updated population; S25. Determine whether the updated population meets the preset stopping condition. If not, treat the updated population as a new population and repeat S23 to S25. If it meets the condition, output the individual with the highest fitness value as the optimal individual. S26. Decode the optimal individual to obtain the connection weights, thresholds, and smoothing factors of the optimized BP neural network, and assign the connection weights, thresholds, and smoothing factors to the BP neural network to obtain the BP neural network optimized by the genetic algorithm; S27. Input the basic operational data into the BP neural network optimized by the genetic algorithm to predict the output and load demand of distributed renewable energy under the day-ahead time scale and the output and load demand of distributed renewable energy under the intraday rolling time scale, so as to obtain the day-ahead prediction data and the intraday rolling prediction data.

4. A hierarchical rolling dispatching method for source-grid-load-storage systems in an active distribution network according to claim 3, characterized in that, The real-number crossover operation includes: selecting two individuals from the parent individuals to participate in the crossover, and recombining the two individuals using real-number encoding based on the crossover coefficient to obtain two child individuals, as shown in the formula: in, and They represent the first The encoding vectors of the two parent individuals that participate in the real crossover operation in each genetic iteration. and Let represent the encoding vectors of the two offspring individuals generated after the real number crossover operation. Indicates the cross coefficient. This represents a function that takes the minimum value. The mutation operation includes: perturbing the coding gene of the individual to be mutated in the updated population, and determining the perturbation direction and amplitude of the coding gene based on a random factor, using the following formula: in, in, Indicates the first Among the individuals, the first The gene value of each coding gene. This indicates the upper bound of the gene encoding the gene. This indicates the lower bound of the gene encoding the gene. This represents a variable-length function that varies with the number of iterations. Indicates the current iteration number. Indicates the maximum number of iterations. This represents the random factor used to determine the direction of mutation. This represents a random number located in the interval [0,1].

5. A hierarchical rolling dispatching method for source-grid-load-storage systems oriented towards active distribution networks according to claim 1, characterized in that, S3 specifically includes: S31. Based on the day-ahead forecast data, determine the forecast values ​​of available output of distributed renewable energy and the forecast values ​​of load demand of each node for each scheduling period within the day-ahead scheduling cycle. S32. Based on the predicted available output of distributed renewable energy, the predicted load demand of each node, and the basic operational data, determine the decision variables of the day-ahead source-grid-load-storage coordinated scheduling model. The decision variables of the day-ahead source-grid-load-storage coordinated scheduling model include the actual output of distributed renewable energy, the abandoned power of distributed renewable energy, the power exchanged with the upstream grid, the status of the branch switches of the distribution network, the tap position of the on-load tap changer, the number of switching groups of the grouped capacitors, the reactive power output of the static var compensator, the controllable load regulation power, the charging and discharging power of the energy storage equipment, and the state of charge of the energy storage equipment. S33. With the goal of minimizing the operating cost of the active distribution network during the day-ahead scheduling cycle, construct the objective function of the day-ahead source-grid-load-storage coordinated scheduling model. The objective function of the day-ahead source-grid-load-storage coordinated scheduling model is expressed as follows: Among them, the cost of flexible resource scheduling satisfies: in, This represents the total operating cost of the active distribution network during the current day's dispatch cycle; This indicates the total number of scheduling periods within the current day's scheduling cycle; Indicates the first The cost of curtailed distributed renewable energy during each scheduling period; Indicates the first The cost of purchasing electricity from the upper-level power grid for each dispatch period; Indicates the first Network loss cost per scheduling period; Indicates the first The cost of flexible resource scheduling for each scheduling period; Indicates the first Equipment operating costs for each scheduling period; Indicates the first Distributed renewable energy reactive power regulation cost per scheduling period; Indicates the first The static var compensator adjustment cost for each scheduling period; Indicates the first Cost of switching capacitors in groups for each scheduling period; Indicates the first Controllable load adjustment cost for each scheduling period; Indicates the first Operating costs of energy storage equipment during each scheduling period; S34. The constraints of distributed renewable energy operation, network topology operation, controllable load operation, energy storage equipment operation, flexible regulation equipment operation, and active distribution network power flow constraints are used as the constraints of the day-ahead source-grid-load-storage coordinated scheduling model, so that the decision variables participate in day-ahead optimization under the condition of meeting the requirements of safe operation of the active distribution network. S35. Based on the objective function and the constraints, optimize and solve the day-ahead source-grid-load-storage coordinated scheduling model to generate the day-ahead scheduling plan. The day-ahead scheduling plan includes a distributed renewable energy output plan, a distributed renewable energy curtailment plan, a network reconfiguration plan, a controllable load adjustment plan, an energy storage equipment charging and discharging plan, and a flexible adjustment equipment operation plan.

6. A hierarchical rolling dispatching method for source-grid-load-storage systems oriented towards active distribution networks according to claim 1, characterized in that, The step of constructing an intraday source-grid-load-storage coordinated scheduling model based on the intraday rolling forecast data and the day-ahead scheduling plan, and generating an intraday scheduling plan, specifically includes: S41. Based on the intraday rolling forecast data, determine the set of rolling optimization periods corresponding to the current intraday rolling moment. The intraday rolling forecast data includes the ultra-short-term available output forecast of distributed renewable energy and the ultra-short-term load demand forecast. The set of rolling optimization periods is represented as follows: in, Indicates the current intraday rolling time. The corresponding set of rolling optimization periods, This indicates the number of intraday scheduling periods included in the rolling optimization time period set; S42. Extract the day-ahead scheduling plan segment corresponding to the rolling optimization period set from the day-ahead scheduling plan, and use the day-ahead scheduling plan segment as the scheduling benchmark of the intraday source-grid-load-storage coordinated scheduling model; S43. Based on the intraday rolling forecast data, the day-ahead scheduling plan segment, and the current operating status of the active distribution network, determine the intraday decision variables of the intraday source-grid-load-storage coordinated scheduling model. The intraday decision variables include the actual output of distributed renewable energy, the abandoned power of distributed renewable energy, the controllable load adjustment power, the charging and discharging power of energy storage equipment, the power of energy storage equipment, the number of switching groups of grouped capacitors, the reactive power output of static var compensators, the tap position of on-load tap changers, and the status of branch switches in the distribution network. S44. Taking the maximization of distributed renewable energy consumption within the rolling optimization time period set as the objective, construct the objective function of the intraday source-grid-load-storage coordinated scheduling model. The objective function of the intraday source-grid-load-storage coordinated scheduling model is expressed as follows: in, This represents the amount of distributed renewable energy consumed within the set of rolling optimization periods. This represents the set of nodes connected to distributed renewable energy sources. Indicates the first Daily scheduling time period nodes The actual absorption capacity of distributed renewable energy, Indicates the length of the intraday scheduling period; S45. Based on the day-ahead scheduling plan segment and the basic operational data, construct the operational cost constraint of the intraday source-grid-load-storage coordinated scheduling model, so that the operational cost within the current rolling optimization period set is not higher than the sum of the operational cost corresponding to the day-ahead scheduling plan segment and the preset cost margin; S46. Construct the constraints of the intraday source-grid-load-storage coordinated scheduling model. The constraints include intraday operation constraints of distributed renewable energy, intraday adjustment constraints of controllable load, intraday operation constraints of energy storage equipment, intraday switching constraints of grouped switching capacitors, intraday output constraints of static var compensators, and intraday operation constraints of network topology. S47. Based on the objective function and the constraints, perform rolling optimization on the intraday source-grid-load-storage coordination scheduling model to obtain the intraday rolling optimization results within the current rolling optimization time period set; S48. The control quantity corresponding to the current intraday rolling moment in the intraday rolling optimization result is used as the actual intraday scheduling plan to be executed, and the updated intraday rolling prediction data is obtained again at the next intraday rolling moment to re-execute the intraday scheduling period that has not been executed.

7. A hierarchical rolling dispatching method for source-grid-load-storage systems oriented towards active distribution networks according to claim 6, characterized in that, The constraints of the intraday source-grid-load-storage coordinated scheduling model include: Distributed renewable energy intraday operating constraints, which include actual output constraints, curtailment constraints, capacity constraints, and ramping constraints, are expressed as follows: in, Indicates the first Daily scheduling time period nodes The actual absorption capacity of distributed renewable energy, Indicates the first Daily scheduling time period nodes Forecasted output of distributed renewable energy sources. Indicates the first Daily scheduling time period nodes Distributed renewable energy curtailment power, Indicates the first Daily scheduling time period nodes Distributed renewable energy reactive power, Represents a node The capacity of distributed renewable energy, Represents a node The ramp-up limit for distributed renewable energy; Controllable load intraday adjustment constraints, which include controllable load adjustment capacity constraints and electricity consumption conservation constraints during the rolling optimization period, are expressed as follows: in, Indicates the first Daily scheduling time period nodes The controllable load active power after adjustment Indicates the first Daily scheduling time period nodes Adjusting the active power of the preload, Represents a node The maximum adjustment ratio of the controllable load. Indicates the length of the intraday scheduling period; The daily operating constraints of energy storage devices include recursive constraints on energy storage capacity, energy storage capacity constraints, mutual exclusion constraints on charge / discharge states, and constraints on charge / discharge power, expressed as follows: in, Indicates the first Energy storage equipment during the daily dispatch period The amount of electricity, Indicates the energy storage equipment during the dispatch period of the next day. The amount of electricity, Indicates the first Energy storage equipment during the daily dispatch period The charging power, Indicates the first Energy storage equipment during the daily dispatch period The discharge power, Indicates energy storage devices Charging efficiency, Indicates energy storage devices The discharge efficiency, and These represent energy storage devices. The lower and upper limits of battery capacity. and These represent energy storage devices. In the Charging and discharging status variables for each daily scheduling period. and These represent energy storage devices. The lower limit and upper limit of charging power, and These represent energy storage devices. The lower limit and upper limit of discharge power; The intraday switching constraint for grouped capacitors is expressed as follows: In the formula, Indicates the first Daily scheduling time period nodes The reactive power compensation of the grouped switching capacitors. Indicates the first Daily scheduling time period nodes The number of switching groups of the capacitor bank. Represents a node The reactive power compensation for each group of capacitors switched on and off. Represents a node The maximum number of capacitor groups allowed to be switched on at the designated location. Represents a node The maximum number of operations allowed for grouped switching of capacitors within the rolling optimization time period set; The daily output constraint of the static var compensator is expressed as: in, Indicates the first Daily scheduling time period nodes The reactive power output of the static var compensator. and Representing nodes respectively The lower limit and upper limit of reactive power output of the static var compensator; Intraday network topology operational constraints, including branch switch operation count constraints and radial topology constraints, are expressed as follows: in, Indicates the first Branch lines during the daily dispatch period The switch state variables, Indicates the branch line during the dispatch period of the previous day. The switch state variables, Indicates the first Branch lines during the daily dispatch period The switching action variable, Indicates a branch The maximum number of switching operations allowed within the rolling optimization time period set. Indicates the set of active distribution network branches. Indicates the total number of active distribution network nodes. This indicates the number of root nodes in the active distribution network.

8. A hierarchical rolling dispatching method for source-grid-load-storage systems in an active distribution network according to claim 1, characterized in that, The second-order cone relaxation of the nonlinear power flow constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model, based on the power flow relationship of the active distribution network, specifically includes: To address the active distribution network power flow constraints in the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model, node voltage square variables and branch current square variables are introduced to replace the node voltage amplitude and branch current amplitude. Based on the squared variables of the node voltages, the squared variables of the branch currents, the active power of the branches, and the reactive power of the branches, the nonlinear branch power relationships in the power flow constraints of the active distribution network are transformed into second-order cone constraints, which are expressed as follows: in, Represents the L2 norm, Indicates the first Branches in each scheduling period Active power transmitted upstream, Indicates the first Branches in each scheduling period reactive power transmitted upstream, Indicates the first Branches in each scheduling period The squared variable of the branch current, Indicates the first The first node of the branch during each scheduling period The squared variable of the node voltage; The second-order cone constraint, together with the branch active power balance constraint, the branch reactive power balance constraint, the node voltage drop constraint, and the active distribution network operation safety constraint, are used as the relaxed power flow constraint and are used to construct the second-order cone planning and scheduling model.

9. A hierarchical rolling dispatching method for source-grid-load-storage systems in an active distribution network according to claim 1, characterized in that, The linearization of the nonlinear terms in the model yields a second-order cone programming scheduling model, specifically including: The nonlinear terms in the day-ahead source-grid-load-storage coordination and scheduling model and the intraday source-grid-load-storage coordination and scheduling model are identified. The nonlinear terms include nonlinear terms generated by multiplying equipment state variables with continuous operation variables, nonlinear terms generated by absolute value operations, and square nonlinear terms generated by distributed renewable energy capacity constraints. For the nonlinear terms generated by multiplying equipment state variables and continuous operating variables, the Big M method is used to establish a linear relationship between equipment state variables and continuous operating variables, so that the corresponding continuous operating variables are zero when the equipment is in an unused state, and the corresponding continuous operating variables take values ​​within the allowable range when the equipment is in an used state. For nonlinear terms generated by absolute value operations, auxiliary variables are introduced for equivalent linearization, such that the auxiliary variables are not less than the variable to be linearized and its opposite, and the auxiliary variables are substituted into the corresponding cost term or action number term. To address the squared nonlinear term in the capacity constraint of distributed renewable energy, a piecewise linearization approach is adopted. The capacity arc of distributed renewable energy is divided into multiple linear segments, and the capacity arc constraint is replaced by a linear inequality formed by each linear segment. This linear inequality is expressed as: in, Indicates the first Each scheduling period node Active power of distributed renewable energy sources Indicates the first Each scheduling period node Reactive power of distributed renewable energy sources, , and Representing nodes respectively The first arc of distributed renewable energy capacity The linearization coefficients of each linear segment. This represents the total number of linear segments constraining the capacity of distributed renewable energy sources. The power flow constraints after second-order cone relaxation, the nonlinear terms after linearization, and the remaining linear constraints in the day-ahead source-grid-load-storage coordination scheduling model and the intraday source-grid-load-storage coordination scheduling model are combined to obtain the second-order cone planning scheduling model.

10. A hierarchical rolling dispatching method for source-grid-load-storage systems oriented towards active distribution networks according to claim 1, characterized in that, S6 specifically includes: The objective function, constraints, and decision variables formed by the day-ahead source-grid-load-storage coordinated scheduling model and the intraday source-grid-load-storage coordinated scheduling model after second-order cone relaxation and linearization are input into the second-order cone programming solver for solving, so as to obtain the optimal scheduling solution that satisfies the active distribution network operation safety constraints. Based on the optimal scheduling solution, extract the actual output of distributed renewable energy, the abandoned power of distributed renewable energy, the status of distribution network branch switches, the controllable load regulation power, the charging and discharging power of energy storage devices, the power of energy storage devices, the number of switching groups of grouped capacitors, the reactive power output of static var compensators, and the tap position of on-load tap changers from the day-ahead scheduling plan and the intraday scheduling plan. The execution period of the day-ahead scheduling plan is updated according to the day-ahead scheduling plan. The day-ahead scheduling plan is used as the actual execution plan for scheduling periods that have entered the scope of the day-ahead rolling optimization execution, while the day-ahead scheduling plan is retained as the subsequent scheduling benchmark for scheduling periods that have not entered the scope of the day-ahead rolling optimization execution. The distributed renewable energy output command is generated based on the actual output and curtailment of distributed renewable energy in the actual execution plan. The network reconfiguration instruction is generated based on the status of the distribution network branch switches in the actual execution plan; The controllable load adjustment command is generated based on the controllable load adjustment power in the actual execution plan; The energy storage charging and discharging command is generated based on the energy storage device's charging and discharging power and energy storage device's power in the actual execution plan. Based on the number of capacitor switching groups, the reactive power output of the static var compensator, and the tap position of the on-load tap changer in the actual execution plan, control commands for the flexible regulating equipment are generated.

Images