Intelligent micro-grid optimal scheduling method and system based on improved particle swarm algorithm

Abstract

The application discloses an intelligent micro-grid optimization scheduling method and system based on an improved particle swarm algorithm. The method performs energy storage state forward recursion on candidate scheduling particles in time sequence to form a cross-period state chain. The first abnormal period that first breaks the continuous executability is located in the cross-period state chain to generate a preceding feasible segment and a subsequent gap segment. A particle repair boundary is generated according to the preceding feasible segment and the subsequent gap segment, the energy storage charging and discharging path is adjusted, and the power difference is distributed to obtain a repaired scheduling particle. After recalculation of the repaired scheduling particle, the recalculation particle that meets the constraint replaces the individual guide position or the group guide position. The application can avoid the problem that the preceding feasible and subsequent unexecutable candidate scheduling particles appear due to the cross-period transmission error of the energy storage state, avoid the unexecutable particle error from participating in the particle swarm search guide, and improve the continuous executability and search convergence stability of the candidate scheduling scheme.

Description

Technical Field

[0001] This invention relates to the field of smart microgrid scheduling technology, and in particular to a smart microgrid optimization scheduling method and system based on an improved particle swarm optimization algorithm. Background Technology

[0002] Smart microgrids typically include distributed renewable energy generation units, energy storage units, controllable power sources, loads, and grid connection and interaction units with the public power grid. Because the output of renewable energy sources such as wind and solar power is volatile, and load demand also varies over time, microgrid dispatching needs to coordinate the charging and discharging of energy storage, the output of controllable power sources, and the power purchased and sold from the grid during continuous dispatch periods to maintain power balance while also considering operational economics.

[0003] Existing microgrid optimization scheduling methods often employ particle swarm optimization (PSO), genetic algorithms, mixed-integer programming, or other intelligent optimization algorithms to establish scheduling models, incorporating factors such as energy storage state of charge, controllable power output limits, ramping constraints, grid-connected power boundaries, and operating costs into the objective function or constraints. For PSO, inertia weights, learning factors, fitness functions, and penalty terms are typically set to gradually bring candidate solutions closer to a scheduling scheme that satisfies both economic efficiency and constraints during the search process.

[0004] However, in multi-period continuous scheduling, the state of charge of energy storage exhibits a clear temporal transmission relationship, with the charging and discharging actions of the previous period directly affecting the schedulable space of the next period. Existing methods often focus more on the overall fitness or whether a single constraint is exceeded. When a candidate scheduling solution is low-cost in an earlier period but becomes unexecutable in subsequent periods due to insufficient energy storage margin, limited ramp-up of controllable power sources, or grid-connected power boundary constraints, this candidate solution may still adversely affect the particle search direction, leading to problems such as power gaps in subsequent periods, energy storage state exceeding limits, or convergence to an unexecutable region in the scheduling results.

[0005] Therefore, there is an urgent need for an optimized scheduling method suitable for continuous scheduling of smart microgrids to address the problem that candidate scheduling solutions are locally feasible but subsequently unexecutable during cross-time period state transfer, thereby improving the continuous executability and search convergence stability of candidate scheduling schemes. Summary of the Invention

[0006] The technical problem to be solved by this invention is: in the continuous scheduling of smart microgrids, how to identify the first unacceptable abnormal period of a candidate scheduling particle, and to repair and recalculate the candidate scheduling particle based on the feasible paths before the abnormal period and the power difference from the abnormal period onwards, so as to avoid the erroneous participation of particles that were feasible in the preceding order but unexecutable in the subsequent particle swarm search guidance.

[0007] To address the aforementioned technical problems, this invention provides a smart microgrid optimization scheduling method based on an improved particle swarm optimization algorithm, comprising: The operation prediction and operation constraints of the smart microgrid during the continuous scheduling period are obtained. Based on the operation prediction and operation constraints, multiple candidate scheduling particles are generated. Each candidate scheduling particle carries the energy storage charging and discharging path, the controllable power output path and the grid-connected interaction path during the continuous scheduling period. The candidate scheduling particles are recursively pushed forward to their energy storage charge state along the time sequence, and the recursively obtained charge state is bound to the scheduling path of the corresponding time period to form a cross-time period state chain. In the cross-time state chain, locate the first abnormal period that first breaks the continuity of executability, retain the preceding feasible segments that can be continuously passed before the first abnormal period, and form subsequent gap segments based on the power difference that has not been accepted since the first abnormal period. A particle repair boundary is generated based on the preceding feasible fragment and the subsequent gap fragment. The energy storage charging and discharging path is adjusted according to the particle repair boundary. The remaining power difference after adjustment is distributed to the controllable power output path and the grid-connected interaction path in the positive and negative directions to obtain the repair scheduling particle. The energy storage charge state forward recursion is re-executed for the repaired scheduling particle. When the recalculated cross-time state chain satisfies the operating constraints, the individual or group guidance position of the corresponding candidate scheduling particle is replaced by the repaired scheduling particle, and the candidate scheduling scheme is output.

[0008] Furthermore, the step of performing forward recursion of the energy storage charge state for the candidate scheduling particles along the time sequence, and binding the recursively obtained charge state with the scheduling path of the corresponding time period to form a cross-time period state chain includes: Read the initial charge state of a continuous scheduling period and use the initial charge state as the starting point for the recursion of the first scheduling period; The charging and discharging path of the energy storage is used to generate the charging state of each scheduling period in sequence, and the charging state of the current scheduling period is used as the starting point for the next scheduling period. The charge state of each scheduling period is bound to the energy storage charging and discharging path, controllable power output path and grid connection interaction path of the corresponding scheduling period, so that the scheduling paths within the same candidate scheduling particle form a continuous state transmission relationship in chronological order.

[0009] Further, the step of generating a particle repair boundary based on the preceding feasible fragment and the subsequent gap fragment, adjusting the energy storage charging and discharging path according to the particle repair boundary, and distributing the remaining power difference after adjustment to the controllable power output path and the grid-connected interaction path in positive and negative directions to obtain repair scheduling particles includes: Based on the charge status of each scheduling period in the preceding feasible segments, determine the reserved power range that the energy storage can retain before the first abnormal period. Based on the power balance difference of each scheduling period in the subsequent gap segment, determine the range of compensation power that energy storage needs to participate in or take over from the first abnormal period. The reserved power range and the compensation power range form the particle repair boundary, and the charge and discharge amount in the energy storage charge and discharge path before the first abnormal period is compressed or transferred according to the particle repair boundary; The power difference within the continuous scheduling period is re-determined based on the compressed or transferred energy storage charging and discharging path, and the power difference is allocated according to the positive and negative directions of the power difference, the ramp margin of the controllable power output path, and the power purchase and sale margin of the grid-connected interactive path to form the repair scheduling particle.

[0010] Further, the step of locating the first abnormal period that first disrupts continuous executability in the cross-time-period state chain, retaining the preceding feasible segments that can be continuously transmitted before the first abnormal period, and forming subsequent gap segments based on the power difference that has not been accepted since the first abnormal period includes: The charge carrying capacity and power carrying capacity of each scheduling period are written sequentially along the cross-time period state chain. Establish a state succession relationship between adjacent scheduling periods, and transfer the charge and power acceptance margins of the previous scheduling period to the next scheduling period. When the charge or power capacity carried over to a certain scheduling period is insufficient to support the scheduling path of that scheduling period, that scheduling period is designated as the first abnormal period. Using the first abnormal period as a boundary, the scheduling paths that have been continuously accepted before the first abnormal period are written into the preceding feasible segment, and the power difference that has not been accepted since the first abnormal period is written into the subsequent gap segment.

[0011] Furthermore, the step of writing the power difference that has not been received since the first abnormal period into the subsequent gap segment includes: Starting from the first abnormal period, the power difference that was not carried by the energy storage charging and discharging path is accumulated in chronological order to form a power sequence to be compensated; Align the power sequence to be compensated with the adjustable output portion of the controllable power supply output path by time period to obtain the controllable power supply receiving sequence. The portion of the power sequence to be compensated that is not covered by the controllable power supply takeover sequence is aligned with the time period of the power available for purchase or sale in the grid connection interaction path to obtain the grid connection takeover sequence. The difference between the power sequence to be compensated, the controllable power supply sequence, and the grid-connected sequence is written into the subsequent gap segment.

[0012] Further, the step of forming the particle repair boundary by the reserved power range and the compensation power range, and compressing or transferring the charge / discharge amount in the energy storage charge / discharge path prior to the first abnormal period according to the particle repair boundary, includes: The charge retention boundary before the first abnormal period is generated according to the reserved power range, the charging acceptance boundary after the first abnormal period is generated according to the compensation power range, and the particle repair boundary is formed by the charge retention boundary and the charging acceptance boundary. Based on the charge retention boundary, trace back the discharge period before the first abnormal period, and reduce the amount of discharge that causes the energy storage state of charge to fall below the charge retention boundary; Based on the charging acceptance boundary, backtrack the charging period before the first abnormal period, and transfer the charging amount that prematurely occupied the upper limit of charge and made the energy storage unable to accept charging or the power surplus in the subsequent scheduling period; The reduced discharge amount and the transferred charge amount are rewritten into the energy storage charging and discharging path to obtain the energy storage path after boundary repair.

[0013] Further, the power difference within the continuous scheduling period is re-determined based on the compressed or transferred energy storage charging and discharging path, and the power difference is allocated according to the positive or negative direction of the power difference, the ramp-up margin of the controllable power output path, and the power purchase and sale margin of the grid-connected interactive path to form the repair scheduling particle, including: Based on the energy storage path that has undergone boundary repair, the change in energy storage output for each scheduling period is regenerated. The change in energy storage output is backfilled into the power balance relationship of the corresponding scheduling period to obtain the remaining power difference; The remaining power difference is written into the controllable power output path in chronological order according to the positive and negative directions of the remaining power difference, and a controllable power compensation path is continuously formed according to the upward or downward output changes between adjacent scheduling periods. The remaining power difference that the controllable power compensation path does not accept is written into the grid connection interaction path according to the direction of power purchase or power sale, thus forming a grid connection compensation path. The repair scheduling particle is formed by combining the energy storage path that has undergone boundary repair, the controllable power supply compensation path, and the grid-connected compensation path.

[0014] Further, the step of re-executing the forward recursion of the energy storage charge state for the repaired scheduling particle, and replacing the individual or group guidance position of the corresponding candidate scheduling particle with the repaired scheduling particle when the recalculated cross-time period state chain satisfies the operational constraints, includes: Using the energy storage charging and discharging path of the repaired scheduling particles as input, the repaired cross-time period state chain is regenerated; The repaired cross-time state chain, along with the controllable power supply compensation path and the grid-connected compensation path, is synchronously written into the recalculation particle; While the recalculated particle maintains its state during the continuous scheduling period, the recalculated particle is written into the individual guidance position of the corresponding candidate scheduling particle. Select a complex particle for swarm search from among multiple complex particles, and write the complex particle for swarm search to the swarm guide position.

[0015] Furthermore, the output candidate scheduling scheme includes: The path offset of the candidate scheduling particle is generated based on the recalculated particle written to the individual guidance position and the recalculated particle written to the group guidance position. The path offset is limited to the scheduling time period corresponding to the particle repair boundary, so that the subsequent updates of candidate scheduling particles converge along the repaired energy storage charging and discharging path, controllable power supply compensation path and grid-connected compensation path. Update the scheduling path of the candidate scheduling particles in the continuous scheduling period according to the path offset to generate multiple constrained scheduling schemes. Output candidate scheduling schemes for smart microgrid scheduling from the plurality of constrained scheduling schemes.

[0016] This invention also provides a smart microgrid optimization scheduling system based on an improved particle swarm optimization algorithm, comprising: The particle generation module is used to obtain the operation prediction and operation constraints of the smart microgrid during the continuous scheduling period, and generate multiple candidate scheduling particles based on the operation prediction and operation constraints. Each candidate scheduling particle carries the energy storage charging and discharging path, the controllable power output path and the grid-connected interaction path during the continuous scheduling period. The state chain generation module is used to perform forward recursion of the energy storage charge state of the candidate scheduling particles in chronological order, and bind the recursed charge state with the scheduling path of the corresponding time period to form a cross-time period state chain. The fragment generation module is used to locate the first abnormal period that first breaks the continuous executability in the cross-time state chain, retain the preceding feasible fragments that can be continuously passed before the first abnormal period, and form subsequent gap fragments based on the power difference that has not been accepted since the first abnormal period. The particle repair module is used to generate a particle repair boundary based on the preceding feasible fragment and the subsequent gap fragment, adjust the energy storage charging and discharging path according to the particle repair boundary, and distribute the remaining power difference after adjustment to the controllable power output path and the grid-connected interaction path in the positive and negative directions to obtain repair scheduling particles. The guidance update module is used to re-execute the forward recursion of the energy storage charge state for the repair scheduling particle. When the recalculated cross-time period state chain satisfies the operation constraints, the individual guidance position or group guidance position of the corresponding candidate scheduling particle is replaced by the repair scheduling particle, and the candidate scheduling scheme is output.

[0017] The beneficial effects of this invention are as follows: This invention incorporates the cross-period transfer relationship of energy storage state of charge into the evaluation process of candidate scheduling particles. By forming a cross-period state chain, the candidate scheduling particles no longer participate in the search based solely on overall fitness or single-period constraints, and can promptly discover scheduling paths that are feasible in the preceding order but not executable in the subsequent order.

[0018] This invention locates the first abnormal period in a cross-period state chain and generates preceding feasible segments and subsequent gap segments accordingly, so that the feasible parts of unexecutable candidate particles can be retained and utilized, reducing search disturbances caused by overall discarding or random reset, and improving search efficiency.

[0019] This invention generates particle repair boundaries based on preceding feasible segments and subsequent missing segments, giving the repair of energy storage charging and discharging paths a clear direction. It can also form repair scheduling particles together with the controllable power supply ramp-up margin and the grid-connected power purchase and sale margin, ensuring that the repaired scheduling scheme is physically executable.

[0020] This invention re-executes the forward recursion of the energy storage and charged state for the repaired scheduled particles, and only allows particles that meet the running constraints after recalculation to replace the individual or group guiding positions, thereby reducing the risk of low-cost but non-executable particles misleading the group's search direction and improving the convergence stability of the algorithm. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating an intelligent microgrid optimization scheduling method based on an improved particle swarm optimization algorithm, provided as an embodiment of the present invention.

[0023] Figure 2This is a schematic diagram illustrating the forward recursion of the energy storage charge state in an intelligent microgrid optimization scheduling method based on an improved particle swarm optimization algorithm, provided as an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram of particle repair boundary and energy storage path adjustment in an intelligent microgrid optimization scheduling method based on an improved particle swarm optimization algorithm, provided as an embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram of the structure of an intelligent microgrid optimization scheduling system based on an improved particle swarm optimization algorithm, provided as an embodiment of the present invention. Detailed Implementation

[0026] The present invention will be further described in detail below with reference to specific embodiments.

[0027] Example 1: This embodiment provides an optimized scheduling method for smart microgrids based on an improved particle swarm optimization algorithm. This method is applied to smart microgrids comprising photovoltaic power generation units, wind power generation units, battery energy storage systems, gas turbines, and grid-connected interaction units. The microgrid needs to coordinate the output of each unit during continuous scheduling periods to maintain power balance and reduce operating costs.

[0028] like Figure 1 As shown, the optimized scheduling method in this embodiment includes the following steps: S1: Obtain the predicted operation and constraints of the smart microgrid during the continuous scheduling period, and generate multiple candidate scheduling particles based on the predicted operation and constraints. Each candidate scheduling particle carries the energy storage charging and discharging path, the controllable power output path, and the grid-connected interaction path during the continuous scheduling period.

[0029] This step obtains the operational forecasts and constraints of the smart microgrid during continuous scheduling periods. Operational forecasts include load forecasts, photovoltaic (PV) output forecasts, wind power output forecasts, and time-of-use pricing. Operational constraints include upper and lower limits of energy storage state of charge, energy storage charge / discharge power boundaries, controllable power source output boundaries, controllable power source ramp-up boundaries, grid-connected interactive power boundaries, and power balance constraints.

[0030] In this embodiment, the continuous scheduling period is set to 24 hours, with each hour constituting a scheduling period, i.e., K=24. For each scheduling period k (k=1, 2, ..., 24), the microgrid needs to determine the energy storage charging and discharging power, the controllable power output, and the grid-connected interaction power.

[0031] Multiple candidate scheduling particles are generated based on operational predictions and constraints. Each candidate scheduling particle carries the energy storage charging / discharging path, the controllable power supply output path, and the grid connection interaction path within a continuous scheduling period. The position of a candidate scheduling particle can be represented as: ; in, This represents the energy storage charging power of particle i during the k-th scheduling period. Indicates the energy storage discharge power. Indicates the output of a controllable power source. This represents the grid-connected power output; a positive value indicates purchased electricity, while a negative value indicates sold electricity.

[0032] During the initialization phase, the positions of candidate scheduling particles are randomly generated within the boundaries of each scheduling path. For example, the energy storage charging power and energy storage discharging power are randomly selected between 0 and their corresponding maximum power, and mutual exclusion constraints restrict energy storage from being in charging and discharging states simultaneously within the same scheduling period; the controllable power output is randomly selected between the minimum and maximum output, and the grid-connected interactive power is randomly selected between the minimum and maximum interactive power. The velocities of candidate scheduling particles are randomly generated within a preset speed range.

[0033] In one specific implementation, the number of particles can be set according to the scheduling window length, computing resources, and convergence requirements. Since this embodiment uses separate encoding for energy storage charging power and energy storage discharging power, and each time period includes four types of variables: energy storage charging, energy storage discharging, controllable power output, and grid-connected interaction, the search space dimension is... .

[0034] S2: Perform forward recursion of the energy storage charge state for the candidate scheduling particles in chronological order, and bind the recursed charge state with the scheduling path of the corresponding time period to form a cross-time period state chain.

[0035] First, read the initial state of charge for the continuous scheduling period. The initial state of charge is used as the starting point for the recursion of the first scheduling period. The initial state of charge can be determined based on the actual state of charge of the energy storage system before the start of the scheduling window, or set to a preset value, for example... That is, the state of charge is 50%. For example... Figure 2 As shown, the initial state of charge serves as the starting point for the first scheduling period, and the state of charge received at the end of the current scheduling period continues to serve as the starting point for the next scheduling period, so that the energy storage state of charge in the continuous scheduling period forms a traceable forward recursion chain along the time direction.

[0036] Then, the state of charge (SOC) for each scheduling period is generated sequentially according to the energy storage charging and discharging path. The recursive relationship of the energy storage SOC is as follows: ; in, This represents the state of charge of the energy storage at the end of the previous scheduling period. This represents the state of charge of the energy storage at the end of the current scheduling period. For energy storage charging efficiency, For energy storage and discharge efficiency, For the rated capacity of energy storage, The length of a single scheduling period (e.g., 1 hour).

[0037] In one specific implementation, the rated energy storage capacity It can be set to Charging efficiency Discharge efficiency Energy storage state of charge limit Upper limit of energy storage state of charge These parameters are determined based on the technical specifications of the actual energy storage equipment.

[0038] Based on the above recursive relationship, calculate sequentially. State of charge of energy storage at the end of each time period The current load status during the scheduling period. As the starting point for the next scheduling period .

[0039] Finally, the charge state of each scheduling period is bound to the corresponding energy storage charging / discharging path, controllable power output path, and grid connection interaction path, so that the scheduling paths within the same candidate scheduling particle form a continuous state transmission relationship in chronological order. Specifically, for each scheduling period... Record the following information in the cross-time state chain: State of charge ; Energy storage charging and discharging power and ; Controllable power output ; Grid-connected interactive power ; Charge capacity margin indicates the remaining space for energy storage to reach its upper or lower limit in terms of state of charge. Power margin indicates the remaining regulation capability that the controllable power supply and grid-connected interaction unit can provide.

[0040] The cross-period state chain explicitly expresses the temporal transmission relationship of the energy storage's state of charge, making the schedulable space of the later period constrained by the charging and discharging actions of the previous period. If, during the recursive process, a certain period's... Exceeding If the power balance relationship cannot be maintained within the specified range, or if the subsequent time period scheduling path cannot be assumed by it, then proceed to the next step to locate the first abnormal time period.

[0041] S3: Locate the first abnormal period in the cross-period state chain that first disrupts continuous executability, retain the preceding feasible segments that can be continuously transmitted before the first abnormal period, and form subsequent gap segments based on the power difference that has not been accepted since the first abnormal period.

[0042] This step locates the first abnormal period in the cross-period state chain that first disrupts continuous executability, preserves the preceding feasible segments that can be continuously passed before the first abnormal period, and forms subsequent gap segments based on the power difference that has not been accepted since the first abnormal period.

[0043] Specifically, this step includes the following sub-steps: S3.1: Write the charge capacity and power capacity of each scheduling period sequentially along the cross-time period state chain.

[0044] Charge capacity margin represents the remaining space between the energy storage's state of charge and its operational boundary. When near When the energy storage charging capacity decreases; when near At that time, the energy storage discharge margin decreases. The charge-carrying margin can be expressed as: ; ; Power margin represents the remaining regulation capacity provided by the controllable power source and grid-connected interaction unit. For controllable power sources, the power margin is limited by ramp-up constraints. The output variation of the controllable power source between adjacent time periods must meet the following requirements: ; in, To control the power supply, reduce the ramp rate. The ramp rate is adjusted to increase the controllable power supply. In one specific implementation, if the controllable power supply is a gas turbine, It can be set to , It can be set to .

[0045] For grid-connected interconnection units, the power carrying capacity is limited by the grid-connected power boundary: ; In one specific implementation, It can be set to (Maximum power output) It can be set to (Maximum power purchase capacity).

[0046] S3.2: Establish a state transfer relationship between adjacent scheduling periods, and transfer the charge transfer margin and power transfer margin of the previous scheduling period to the next scheduling period.

[0047] Specifically, for scheduling periods Its available energy storage dispatch space is limited by time period State of charge at the end Constraints; its available controllable power dispatch space is limited by time period. Controllable power output Together with the ramp constraint, it restricts the state. The state succession relationship enables the continuous transmission of physical constraints in the state chain across time periods.

[0048] S3.3: When the charge carrying capacity or power carrying capacity transferred to a certain scheduling period is insufficient to continue supporting the scheduling path of that scheduling period, the scheduling period shall be determined as the first abnormal period.

[0049] For example, if a candidate scheduling particle prematurely discharges its stored energy in a preceding time slot, it will cause the time slot to... State of charge at the end of the period Approaching the lower limit During the period During peak load periods, energy storage can no longer provide discharge power, and controllable power sources are constrained by ramp-up limitations and cannot instantly fill the power gap, while grid-connected power is also nearing its limit. At this time, the time period... The power balance relationship cannot be satisfied, and the system will [period] [time period]. The first abnormal period was identified. .

[0050] The criteria for determining the first abnormal period include, but are not limited to: (1) The energy storage state of charge exceeds the operating boundary: or ; (2) The controllable power supply ramp-up constraint cannot be met: or ; (3) Grid-connected power exceeds the boundary: or ; (4) The power balance relationship cannot be satisfied: .

[0051] S3.4: Using the first abnormal period as the boundary, write the scheduling paths that have been continuously accepted before the first abnormal period into the preceding feasible segment, and write the power difference that has not been accepted since the first abnormal period into the subsequent gap segment.

[0052] The preceding feasible fragments retain the portions of candidate scheduling particles that already conform to the cross-time period state transmission relationship. For the first anomalous time period... Candidate scheduling particles, with preceding feasible segments containing time periods To the time period The scheduling path and corresponding energy storage state of charge; subsequent gap segments from the time period Start recording unaccepted power differences. Preceding feasible fragments prevents the entire feasible information in candidate particles from being discarded due to subsequent local unexecutability.

[0053] The generation of subsequent gap segments includes the following sub-steps: (1) The power difference that is not carried by the energy storage charging and discharging path is accumulated along the time sequence from the first abnormal period to form the power sequence to be compensated.

[0054] The power sequence to be compensated represents the power difference that needs to be taken over by other regulating resources in each time period after energy storage can no longer provide regulation capacity. For each time period... ( The power to be compensated can be expressed as: ; in, For time period The power difference to be compensated. When When this occurs, it indicates a power shortage, requiring energy storage discharge, controllable power supply adjustment, or grid connection for power purchase; when When this occurs, it indicates a power surplus, requiring energy storage charging, controllable power supply downgrading, or grid-connected power sales.

[0055] (2) Align the power sequence to be compensated with the adjustable output part of the controllable power supply output path to obtain the controllable power supply receiving sequence.

[0056] The adjustable output of a controllable power supply is limited by both ramping constraints and upper and lower output limits. (For time periods...) The maximum adjustable output power and maximum adjustable output power of the controllable power supply are as follows: ; ; in, This is the maximum output of the controllable power supply. This is the minimum output of a controllable power supply. Not greater than The adjustable output power. The controllable power supply receiving sequence represents the portion of the power sequence that the controllable power supply can receive from the power sequence to be compensated.

[0057] (3) Align the portion of the power sequence to be compensated that is not covered by the controllable power supply takeover sequence with the power available for purchase or sale in the grid-connected interaction path to obtain the grid-connected takeover sequence.

[0058] The interactive power of a grid-connected interactive unit is limited by the grid-connected power boundary. For time periods... The purchasable and saleable power of the grid-connected interactive unit are as follows: ; ; in, This indicates the positive balance of electricity purchases. Indicates not greater than The surplus of electricity sold. The grid connection acceptance sequence represents the portion of the power sequence that the grid-connected interaction unit can accept from the power sequence to be compensated.

[0059] (4) Write the difference between the power sequence to be compensated, the controllable power supply sequence and the grid connection sequence that has not yet been accepted into the subsequent gap segment.

[0060] Subsequent gap segments record the power gap or surplus that still exists when neither the controllable power source nor the grid-connected interaction unit can fully absorb the load. This power difference needs to be compensated for or absorbed by adjusting the energy storage charging and discharging paths of the preceding periods. Subsequent gap segments provide a clear repair direction for the generation of particle repair boundaries.

[0061] S4: Generate a particle repair boundary based on the preceding feasible segment and the subsequent gap segment, adjust the energy storage charging and discharging path according to the particle repair boundary, and distribute the remaining power difference after adjustment to the controllable power output path and the grid-connected interaction path in the positive and negative directions to obtain the repair scheduling particle.

[0062] This step generates a particle repair boundary based on the preceding feasible fragments and the subsequent missing fragments. The energy storage charging and discharging paths are adjusted according to the particle repair boundary, and the remaining power difference after adjustment is distributed to the controllable power output path and the grid-connected interaction path in the positive and negative directions to obtain the repair scheduling particles.

[0063] Specifically, this step includes the following sub-steps: S4.1: Based on the charge status of each scheduling period in the preceding feasible segments, determine the reserved power range that the energy storage can retain before the first abnormal period.

[0064] The reserved capacity range represents the state of charge range that needs to be reserved for energy storage during the preceding period to avoid unavailability of energy storage during the initial abnormal period and thereafter. The lower bound of the reserved capacity range is determined based on the maximum cumulative discharge demand formed by the positive power gap in the subsequent gap segment.

[0065] Let the time period in the subsequent gap segment be... ( The power difference to be compensated is Then the lower bound of the reserved power range is: ; The upper limit of the reserved power range is usually the upper limit of the energy storage state of charge. Reserved power range The preceding time period must not be depleted of energy storage prematurely, among which From arrive Any scheduling period between.

[0066] S4.2: Based on the power difference of each scheduling period in the subsequent gap segment, determine the compensation power range that energy storage needs to participate in after the first abnormal period.

[0067] The compensation power range indicates the range of energy required for energy storage to provide discharge or accept charging in subsequent periods. The compensation power range can be determined based on the positive and negative portions of the power difference in each period within the subsequent gap segment: ; ; in, The total amount of electricity that needs to be compensated for by energy storage discharge. This represents the total amount of electricity that needs to be supplied for energy storage charging. The compensation ranges are as follows: and It is used to determine the subsequent discharge capacity and charging capacity of energy storage.

[0068] S4.3: The particle repair boundary is formed by the reserved power range and the compensation power range, and the charge and discharge amount in the energy storage charge and discharge path before the first abnormal period is compressed or transferred according to the particle repair boundary.

[0069] The particle repair boundary includes the charge retention boundary and the charge acceptance boundary. The charge retention boundary is used to limit excessive energy discharge before the first abnormal period, while the charge acceptance boundary is used to prevent the energy storage from approaching the upper charge limit too early in the preceding period, which would result in the inability to continue accepting charging or power surplus.

[0070] like Figure 3 As shown, the charge retention boundary is used to impose a lower limit constraint on the excessive discharge path before the first abnormal period, and the charging acceptance boundary is used to impose an upper limit constraint on the charging path that prematurely occupies the upper limit of charge; after the original energy storage path of the candidate scheduling particle is reduced in discharge and transferred in charging, a repaired path that satisfies the particle repair boundary is formed.

[0071] The charge retention boundary is generated based on the reserved charge interval. This applies to time periods within the preceding feasible segments. ( If the recursive state of charge is obtained Below If the energy storage discharge during that period is too large, it needs to be reduced.

[0072] The charging acceptance boundary is generated based on the charging acceptance demand within the compensation capacity range. This applies to time periods within the preceding feasible segments. ( If the state of charge formed after energy storage charging approaches the target value too early... If the power surplus or charging demand cannot be met in subsequent periods, the energy storage charging capacity of that period needs to be transferred to a period closer to the first abnormal period and still meeting the state of charge constraints.

[0073] The specific method for adjusting the energy storage charging and discharging path according to the particle repair boundary is as follows: (1) Based on the discharge period before the first abnormal period, the discharge amount that causes the energy storage state of charge to fall below the charge maintenance boundary is reduced.

[0074] For time period ( ),like The discharge amount during that period Reduce to: ; Reduced discharge volume ensures the duration of the period The state of charge at the end is not lower than .

[0075] (2) Based on the charging acceptance boundary, trace back the charging period before the first abnormal period and transfer the charging amount that was prematurely occupied by the upper limit of charge.

[0076] For time period ( If the charging action during that period results in the energy storage state of charge... If the power level is too high, making it impossible to continue absorbing the power surplus or charging demand in subsequent periods, some of the charging will be transferred to periods closer to the initial abnormal period that still have charging capacity. ( The transferred charge ensures that the energy storage retains both the necessary discharge capacity and the charge space to absorb subsequent surplus when the first abnormal period arrives. The target transfer period is determined by the remaining charge margin, the power balance relationship of the corresponding period, and the subsequent negative power difference; if there is no target transfer period that meets the above conditions, the charge transfer will not be performed and the process will enter the unrepairable particle determination stage.

[0077] (3) The reduced discharge amount and the transferred charge amount are rewritten into the energy storage charging and discharging path to obtain the energy storage path after boundary repair.

[0078] S4.4: Based on the compressed or transferred energy storage charging and discharging paths, redetermine the power difference within the continuous scheduling period, and allocate the power difference according to the positive and negative directions of the power difference, the ramp margin of the controllable power output path, and the power purchase and sale margin of the grid-connected interactive path to form repair scheduling particles.

[0079] Based on the energy storage path after boundary repair, the energy storage output change for each scheduling period is regenerated. For each time period... The change in energy storage output is: ; in, and The restored energy storage discharge power and charging power, and The energy storage discharge power and charging power before repair.

[0080] The change in energy storage output is then backfilled into the power balance relationship for the corresponding scheduling period to obtain the remaining power difference: ; The remaining power difference represents the amount of power adjustment that needs to be undertaken by the controllable power source and grid-connected interaction unit after the energy storage path is repaired; when A positive value indicates that supply still needs to be increased. A negative value indicates that the surplus still needs to be absorbed or the supply needs to be reduced.

[0081] The remaining power difference is written into the controllable power supply output path in chronological order. For time periods... The compensation power of the controllable power supply is determined according to the positive or negative direction of the remaining power difference: like ,but ;like ,but ; in, For time period The maximum adjustable output of the controllable power supply For time period The maximum adjustable output of the controllable power supply and its value is not greater than The controllable power compensation path ensures that the output variation between adjacent time periods meets the ramping constraint.

[0082] The remaining power difference not covered by the controllable power compensation path is written into the grid connection interaction path according to the direction of power purchase or power sale, thus forming the grid connection compensation path: ; The grid-connected compensation path ensures that the grid-connected interactive power remains within the boundary range. If If the grid-connected interactive power exceeds the boundary, the remaining upward or downward adjustment space of the controllable power supply is checked first, and then the previous steps are returned to adjust the energy storage path. If the power balance, ramp-up constraints, and grid-connected boundary cannot be met after a preset number of energy storage path adjustments, the candidate scheduling particle is marked as an unrepairable particle and prohibited from entering the individual or group guidance position. The preset number of times is configured by the scheduling system or determined based on the scheduling window length, single-round calculation time, and real-time scheduling cycle.

[0083] The repair scheduling particle is formed by combining the energy storage path, the controllable power source compensation path, and the grid-connected compensation path after boundary repair. ; The repair scheduling particle comprehensively considers the cross-time state transfer of energy storage, the ramp-up constraints of controllable power sources, and the power boundary of grid-connected interaction, thus transforming the originally unexecutable candidate scheduling particles into executable scheduling schemes.

[0084] S5: Re-execute the forward recursion of the energy storage charge state for the repaired scheduling particle. When the recalculated cross-time period state chain satisfies the operating constraints, replace the individual or group guidance position of the corresponding candidate scheduling particle with the repaired scheduling particle, and output the candidate scheduling scheme.

[0085] Specifically, this step includes the following sub-steps: S5.1: Using the energy storage charging and discharging path of the repaired scheduling particles as input, regenerate the repaired cross-time period state chain.

[0086] Following the forward recursive method for the state of charge of energy storage described in step S2, the state of charge of energy storage for each time period is recalculated for the repair scheduling particles. The recalculation process verifies whether the repaired energy storage path meets the state of charge constraints. .

[0087] S5.2: Synchronously write the repaired cross-time state chain, controllable power supply compensation path, and grid-connected compensation path into the recalculation particle.

[0088] The complex particle simultaneously carries information on the continuity of energy storage state, the continuity of controllable power supply ramp-up, and the grid-connected interaction boundary. For each time period... Complex particles need to satisfy: (1) Energy storage state of charge constraints: ; (2) Controllable power supply ramp-up constraint: ; (3) Grid-connected interactive power constraints: ; (4) Power balance constraints: .

[0089] S5.3: When the recalculated particle maintains state continuity along the continuous scheduling period, the recalculated particle is written into the individual guidance position of the corresponding candidate scheduling particle.

[0090] Individual guiding position Represents particles The optimal position experienced during the search process. In traditional particle swarm optimization, the individual guiding position is determined by comparing fitness values. In this invention, a recalculated particle is considered as a candidate for an individual guiding position only if it satisfies the requirement of state continuity across time periods.

[0091] Specifically, for candidate scheduling particles If the repaired scheduling particles satisfy all operational constraints after recalculation, and the fitness value... Fitness value better than the current individual guidance position Then update the individual guide position: ; This update strategy ensures that the individual guidance position always points to an executable scheduling scheme, avoiding low-cost but non-executable particles from misleading the search direction.

[0092] S5.4: Select a complex particle for swarm search from among multiple complex particles and write the complex particle for swarm search to the swarm guide position.

[0093] Group guidance position This represents the globally optimal position found by the entire particle swarm during the search process. In this invention, the swarm-guided position is selected from all complexly computed particles that satisfy the constraints, choosing the particle with the smallest fitness value: ; By recalculating after repair and guiding position substitution, the particle swarm search process is made to converge along the executable region, thereby improving the search stability of the algorithm.

[0094] S5.5: Based on the recalculated particles written to the individual guide position and the recalculated particles written to the group guide position, generate the path offset of the candidate scheduling particles and output the candidate scheduling scheme.

[0095] In the velocity update formula of the particle swarm optimization algorithm, the individual guide position and the swarm guide position are used to calculate the particle's path offset. This invention employs a particle swarm update method with dynamic inertia weights: ; ; ; in, For the first Inertia weights in the next iteration Initial values ​​for inertia weights (e.g.) ), The inertial weight termination value (e.g.) ), The maximum number of iterations, For individual learning factors (e.g.) ), For group learning factors (e.g.) ), and for Random numbers within the interval. The aforementioned inertia weights and learning factors can be determined by the particle swarm optimization algorithm configuration parameters, historical scheduling results, or the convergence requirements of the scheduling system.

[0096] Path offset and Constrained by the particle repair boundary, subsequent updates of candidate scheduling particles converge along the repaired energy storage charging / discharging path, controllable power supply compensation path, and grid-connected compensation path. Specifically, the path offset is limited to the scheduling time period corresponding to the particle repair boundary, preventing particles from re-entering the unexecutable region during the update process.

[0097] The scheduling paths of candidate scheduling particles are updated according to the path offset within a continuous scheduling period, generating multiple constrained scheduling schemes. The algorithm terminates iteration when the maximum number of iterations is reached, the fitness change is less than a preset convergence threshold, or the output scheme stabilizes after multiple consecutive rounds. The convergence threshold and the number of stable rounds can be determined based on the normalization scale of the fitness function, the scheduling period, and the configuration of computing resources.

[0098] From multiple constrained scheduling schemes, candidate scheduling schemes for smart microgrid scheduling are output. Each candidate scheduling scheme includes the energy storage charging and discharging power, controllable power output, grid-connected interactive power, and the corresponding energy storage state of charge for each scheduling period. These candidate scheduling schemes can be confirmed, issued, or further verified for safety by the scheduling system.

[0099] In this embodiment, the fitness function comprehensively considers the cost of purchasing and selling electricity, the operating cost of controllable power sources, the operating cost of energy storage, and the cost of constraint repair. The fitness function is expressed as: ; in, For the first The electricity purchase price for each dispatch period, For the first Electricity sales price for each dispatch period, This refers to the purchased portion of the grid-connected interactive power. This refers to the electricity sales portion of the grid-connected interactive power. The operating cost coefficient for controllable power supplies. This is the cost coefficient for energy storage operations. The repair cost is the cost formed after the candidate scheduling particles are fragmented and repaired.

[0100] In one specific implementation, the time-of-use electricity price, the retail electricity price, the controllable power source operating cost coefficient, and the energy storage operation cost coefficient can be determined by the electricity market price, equipment operating costs, or dispatch system configuration. Repair costs The cost can be determined by normalizing the repair magnitude, for example, by calculating the proportion of energy storage path adjustments to the total dispatch volume. Specifically, the repair cost... It can be determined by the weighted normalization of energy storage charging and discharging adjustment, controllable power supply compensation, and grid connection compensation, with each weight configured by the dispatching system.

[0101] The goal of the fitness function is to minimize the operating cost of the microgrid. By combining cross-time state chains and particle-repaired boundaries, this invention ensures that the optimal fitness scheduling scheme also meets the executability requirements.

[0102] Example 2: This embodiment provides an intelligent microgrid optimization scheduling system based on an improved particle swarm optimization algorithm, used to execute the optimization scheduling method described in Embodiment 1. Figure 4 As shown, the system includes the following modules: The particle generation module M1 is used to obtain the operation prediction and operation constraints of the smart microgrid during the continuous scheduling period. Based on the operation prediction and operation constraints, it generates multiple candidate scheduling particles. Each candidate scheduling particle carries the energy storage charging and discharging path, the controllable power output path, and the grid connection interaction path during the continuous scheduling period.

[0103] The particle generation module M1 obtains load forecast values, photovoltaic output forecast values, wind power output forecast values, and time-of-use electricity prices from the data acquisition system. It reads the upper and lower limits of energy storage charge state, energy storage charging and discharging power boundaries, controllable power output boundaries, controllable power ramping boundaries, and grid-connected interactive power boundaries from the configuration database, and initializes the position and velocity of candidate scheduling particles based on the above data.

[0104] The state chain generation module M2 is used to perform forward recursion of the energy storage charge state of candidate scheduling particles along the time sequence, and bind the recursed charge state with the scheduling path of the corresponding time period to form a cross-time period state chain.

[0105] The state chain generation module M2 calculates the received charge state for each scheduling period sequentially according to the recursive formula for energy storage charge state, and writes the received charge state, charge receiving margin, and power receiving margin into the cross-time period state chain data structure. The cross-time period state chain data structure can be implemented using an array or a linked list, with each node corresponding to a scheduling period, and the nodes establishing successive connection relationships through pointers or indexes.

[0106] The segment generation module M3 is used to locate the first abnormal period in the cross-period state chain that first breaks the continuity of executability, retain the preceding feasible segments that can be continuously passed before the first abnormal period, and form subsequent gap segments based on the power difference that has not been accepted since the first abnormal period.

[0107] The segment generation module M3 traverses the cross-time period state chain, checking whether the charge and power carrying capacity of each scheduling period is sufficient to support the scheduling path of subsequent scheduling periods. When the first abnormal time period is detected, the segment generation module M3 splits the cross-time period state chain into preceding feasible segments and subsequent gap segments, and stores them into independent data structures respectively.

[0108] The particle repair module M4 is used to generate particle repair boundaries based on the preceding feasible fragments and the subsequent missing fragments, adjust the energy storage charging and discharging paths according to the particle repair boundaries, and distribute the remaining power difference after adjustment to the controllable power output path and the grid-connected interaction path in the positive and negative directions to obtain repair scheduling particles.

[0109] The particle repair module M4 includes a boundary calculation unit, an energy storage path adjustment unit, and a power allocation unit. The boundary calculation unit calculates the reserved capacity range and the compensation capacity range based on the preceding feasible segments and subsequent gap segments, and generates the charge retention boundary and the charging acceptance boundary. The energy storage path adjustment unit traces back the energy storage charging and discharging periods in the preceding feasible segments according to the charge retention boundary and the charging acceptance boundary, performing discharge reduction and charging transfer operations. The power allocation unit calculates the power difference based on the adjusted energy storage path and allocates the power difference to the controllable power output path and the grid-connected interaction path.

[0110] The guide update module M5 is used to re-execute the forward recursion of the energy storage charge state for the repair scheduling particles. When the recalculated cross-time state chain meets the operating constraints, the individual or group guide position of the corresponding candidate scheduling particle is replaced by the repair scheduling particle, and the candidate scheduling scheme is output.

[0111] The guidance update module M5 includes a recalculation unit, a guidance position update unit, and a particle update unit. The recalculation unit re-executes the forward recursion of the energy storage charge state for the repaired scheduling particles and verifies whether the recalculated cross-time period state chain satisfies the energy storage charge state constraints, controllable power supply ramp-up constraints, grid-connected interactive power constraints, and power balance constraints. The guidance position update unit writes the recalculated particles that satisfy the constraints into individual guidance positions or group guidance positions and selects the globally optimal particle based on the fitness value. The particle update unit updates the particle velocity and position based on the dynamic inertia weight, individual learning factor, and group learning factor, generating candidate scheduling particles for the next iteration.

[0112] The modules exchange information via data interfaces. Particle generation module M1 passes candidate scheduling particles to state chain generation module M2, state chain generation module M2 passes cross-time state chains to fragment generation module M3, fragment generation module M3 passes preceding feasible fragments and subsequent gap fragments to particle repair module M4, particle repair module M4 passes repaired scheduling particles to guide update module M5, and guide update module M5 feeds back the updated candidate scheduling particles to particle generation module M1 for the next iteration.

[0113] This system can be deployed on the dispatch server of a microgrid energy management system, running as software modules. Each module can be implemented using programming languages ​​such as C++, Python, or Java, and communicates with other modules via shared memory or message queues. The system's operating cycle can be set from 15 minutes to 1 hour, depending on the actual operating needs of the microgrid and available computing resources.

[0114] This invention explicitly embeds the continuous transmission relationship of energy storage charge state into the particle swarm search process through a cross-time period state chain, enabling the executability of candidate scheduling particles to be checked and repaired along the time direction. Through the initial abnormal time period location and fragmentation processing, this invention can retain the preceding feasible fragments in the candidate scheduling particles and form a repair direction based on the subsequent missing fragments. Compared with directly penalizing or discarding inexecutable particles as a whole, this can reduce the loss of feasible scheduling information.

[0115] By using particle boundary repair, energy storage path repair, controllable power supply compensation, and grid connection compensation, this invention enables candidate scheduling particles to participate in individual or group guidance position updates after recalculation, thereby reducing the adverse effects of non-executable particles on the search direction and improving the search stability and convergence efficiency of the algorithm.

[0116] The optimized scheduling method and system provided by this invention can effectively handle the cross-time period state transfer problem in the continuous scheduling of smart microgrids, generate a scheduling scheme that takes into account both economy and feasibility, and are applicable to smart microgrid scenarios that include new energy power generation, energy storage, controllable power sources and grid-connected interaction.

[0117] In other embodiments, the controllable power source is not limited to a gas turbine, but can also be a diesel generator, fuel cell, or other adjustable power source. If a gas turbine is used, the start-up and shutdown costs, fuel consumption costs, and ramping constraints can be further elaborated in the specification; if other controllable power sources are used, their operating boundaries and ramping capabilities can be replaced using the same logic.

[0118] In other implementations, the grid connection interaction path may include a power purchase path and a power sales path. For time-of-use pricing scenarios, peak, flat, and valley prices can be introduced into the fitness function, allowing candidate scheduling particles to perform economic optimization while satisfying executability. The power purchase price and power sales price can be dynamically adjusted based on real-time electricity market prices.

[0119] In other implementations, when subsequent gap segments cannot be repaired jointly by the energy storage path, the controllable power output path, and the grid-connected interaction path, the corresponding candidate scheduling particles can be marked as unrepairable particles, preventing them from participating in individual or group guidance position updates. Unrepairable particles can be reinitialized in the next iteration or introduced with a new search direction through mutation operations.

[0120] In other embodiments, when the wind and solar forecast deviation is large, the operational forecast can be reacquired when the rolling scheduling window is updated, and a new candidate scheduling particle swarm can be generated based on the candidate scheduling schemes output in the previous round, making the present invention applicable to day-ahead, intraday, or real-time rolling scheduling. The rolling scheduling window can be set to 24 hours, 4 hours, or other time lengths, and the scheduling period length can be set to 1 hour, 15 minutes, or other time granularities.

[0121] In other implementations, the dynamic inertia weight can be linearly decreasing or adaptively adjusted based on the distribution of the repaired and scheduled particles. For example, when the particle swarm is relatively concentrated, the inertia weight can be increased to expand the search range; when the particle swarm is relatively dispersed, the inertia weight can be decreased to accelerate the convergence speed. Regardless of the inertia weight adjustment method used, particles participating in the update of individual or group guided positions should be prioritized for recalculation after repair.

[0122] In other implementations, the fitness function can further consider carbon emission costs, environmental benefits, or other optimization objectives. For multi-objective optimization scenarios, weighted summation, Pareto front, or other multi-objective optimization methods can be used to balance economic efficiency, environmental impact, and reliability.

[0123] The above provides a detailed description of the intelligent microgrid optimization scheduling method and system based on an improved particle swarm optimization algorithm provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make various improvements and modifications to this invention without departing from its principles, and these improvements and modifications also fall within the protection scope of this invention.

Claims

1. A smart microgrid optimization scheduling method based on an improved particle swarm optimization algorithm, characterized in that, include: The operation prediction and operation constraints of the smart microgrid during the continuous scheduling period are obtained. Based on the operation prediction and operation constraints, multiple candidate scheduling particles are generated. Each candidate scheduling particle carries the energy storage charging and discharging path, the controllable power output path and the grid-connected interaction path during the continuous scheduling period. The candidate scheduling particles are recursively pushed forward to their energy storage charge state along the time sequence, and the recursively obtained charge state is bound to the scheduling path of the corresponding time period to form a cross-time period state chain. In the cross-time state chain, locate the first abnormal period that first breaks the continuity of executability, retain the preceding feasible segments that can be continuously passed before the first abnormal period, and form subsequent gap segments based on the power difference that has not been accepted since the first abnormal period. A particle repair boundary is generated based on the preceding feasible fragment and the subsequent gap fragment. The energy storage charging and discharging path is adjusted according to the particle repair boundary. The remaining power difference after adjustment is distributed to the controllable power output path and the grid-connected interaction path in the positive and negative directions to obtain the repair scheduling particle. The energy storage charge state forward recursion is re-executed for the repaired scheduling particle. When the recalculated cross-time state chain satisfies the operating constraints, the individual or group guidance position of the corresponding candidate scheduling particle is replaced by the repaired scheduling particle, and the candidate scheduling scheme is output.

2. The intelligent microgrid optimization scheduling method based on the improved particle swarm optimization algorithm according to claim 1, characterized in that, The step of performing forward recursion of the energy storage charge state for the candidate scheduling particles in chronological order, and binding the recursed charge state with the scheduling path of the corresponding time period to form a cross-time period state chain includes: Read the initial charge state of a continuous scheduling period and use the initial charge state as the starting point for the recursion of the first scheduling period; The charging and discharging path of the energy storage is used to generate the charging state of each scheduling period in sequence, and the charging state of the current scheduling period is used as the starting point for the next scheduling period. The charge state of each scheduling period is bound to the energy storage charging and discharging path, controllable power output path and grid connection interaction path of the corresponding scheduling period, so that the scheduling paths within the same candidate scheduling particle form a continuous state transmission relationship in chronological order.

3. The intelligent microgrid optimization scheduling method based on the improved particle swarm optimization algorithm according to claim 1, characterized in that, The process of generating a particle repair boundary based on the preceding feasible fragment and the subsequent gap fragment, adjusting the energy storage charging and discharging path according to the particle repair boundary, and distributing the remaining power difference after adjustment to the controllable power output path and the grid-connected interaction path in positive and negative directions to obtain repair scheduling particles includes: Based on the charge status of each scheduling period in the preceding feasible segments, determine the reserved power range that the energy storage can retain before the first abnormal period. Based on the power balance difference of each scheduling period in the subsequent gap segment, determine the range of compensation power that energy storage needs to participate in or take over from the first abnormal period. The reserved power range and the compensation power range form the particle repair boundary, and the charge and discharge amount in the energy storage charge and discharge path before the first abnormal period is compressed or transferred according to the particle repair boundary; The power difference within the continuous scheduling period is re-determined based on the compressed or transferred energy storage charging and discharging path, and the power difference is allocated according to the positive and negative directions of the power difference, the ramp margin of the controllable power output path, and the power purchase and sale margin of the grid-connected interactive path to form the repair scheduling particle.

4. The intelligent microgrid optimization scheduling method based on the improved particle swarm optimization algorithm according to claim 1, characterized in that, The process of locating the first abnormal period that first disrupts continuous executability in the cross-time state chain, retaining the preceding feasible segments that can be continuously transmitted before the first abnormal period, and forming subsequent gap segments based on the power difference that has not been accepted since the first abnormal period includes: The charge carrying capacity and power carrying capacity of each scheduling period are written sequentially along the cross-time period state chain. Establish a state succession relationship between adjacent scheduling periods, and transfer the charge and power acceptance margins of the previous scheduling period to the next scheduling period. When the charge or power capacity carried over to a certain scheduling period is insufficient to support the scheduling path of that scheduling period, that scheduling period is designated as the first abnormal period. Using the first abnormal period as a boundary, the scheduling paths that have been continuously accepted before the first abnormal period are written into the preceding feasible segment, and the power difference that has not been accepted since the first abnormal period is written into the subsequent gap segment.

5. The intelligent microgrid optimization scheduling method based on the improved particle swarm optimization algorithm according to claim 4, characterized in that, The step of writing the power difference that has not been received since the first abnormal period into the subsequent gap segment includes: Starting from the first abnormal period, the power difference that was not carried by the energy storage charging and discharging path is accumulated in chronological order to form a power sequence to be compensated; Align the power sequence to be compensated with the adjustable output portion of the controllable power supply output path by time period to obtain the controllable power supply receiving sequence. The portion of the power sequence to be compensated that is not covered by the controllable power supply takeover sequence is aligned with the time period of the power available for purchase or sale in the grid connection interaction path to obtain the grid connection takeover sequence. The difference between the power sequence to be compensated, the controllable power supply sequence, and the grid-connected sequence is written into the subsequent gap segment.

6. The intelligent microgrid optimization scheduling method based on the improved particle swarm optimization algorithm according to claim 3, characterized in that, The step of forming the particle repair boundary by the reserved power range and the compensation power range, and compressing or transferring the charge and discharge amounts in the energy storage charge and discharge path prior to the first abnormal period according to the particle repair boundary, includes: The charge retention boundary before the first abnormal period is generated according to the reserved power range, the charging acceptance boundary after the first abnormal period is generated according to the compensation power range, and the particle repair boundary is formed by the charge retention boundary and the charging acceptance boundary. Based on the charge retention boundary, trace back the discharge period before the first abnormal period, and reduce the amount of discharge that causes the energy storage state of charge to fall below the charge retention boundary; Based on the charging acceptance boundary, backtrack the charging period before the first abnormal period, and transfer the charging amount that prematurely occupied the upper limit of charge and made the energy storage unable to accept charging or the power surplus in the subsequent scheduling period; The reduced discharge amount and the transferred charge amount are rewritten into the energy storage charging and discharging path to obtain the energy storage path after boundary repair.

7. The intelligent microgrid optimization scheduling method based on the improved particle swarm optimization algorithm according to claim 6, characterized in that, The power difference within the continuous scheduling period is re-determined based on the compressed or transferred energy storage charging and discharging path, and the power difference is allocated according to the positive or negative direction of the power difference, the ramp-up margin of the controllable power output path, and the power purchase and sale margin of the grid-connected interactive path to form the repair scheduling particle, including: Based on the energy storage path that has undergone boundary repair, the change in energy storage output for each scheduling period is regenerated. The change in energy storage output is backfilled into the power balance relationship of the corresponding scheduling period to obtain the remaining power difference; The remaining power difference is written into the controllable power output path in chronological order according to the positive and negative directions of the remaining power difference, and a controllable power compensation path is continuously formed according to the upward or downward output changes between adjacent scheduling periods. The remaining power difference that the controllable power compensation path does not accept is written into the grid connection interaction path according to the direction of power purchase or power sale, thus forming a grid connection compensation path. The repair scheduling particle is formed by combining the energy storage path that has undergone boundary repair, the controllable power supply compensation path, and the grid-connected compensation path.

8. The intelligent microgrid optimization scheduling method based on the improved particle swarm optimization algorithm according to claim 7, characterized in that, The step of re-executing the forward recursion of the energy storage charge state for the repaired scheduling particle, and replacing the individual or group guidance position of the corresponding candidate scheduling particle with the repaired scheduling particle when the recalculated cross-time-period state chain satisfies the operational constraints, includes: Using the energy storage charging and discharging path of the repaired scheduling particles as input, the repaired cross-time period state chain is regenerated; The repaired cross-time state chain, along with the controllable power supply compensation path and the grid-connected compensation path, is synchronously written into the recalculation particle; While the recalculated particle maintains its state during the continuous scheduling period, the recalculated particle is written into the individual guidance position of the corresponding candidate scheduling particle. Select a complex particle for swarm search from among multiple complex particles, and write the complex particle for swarm search to the swarm guide position.

9. The intelligent microgrid optimization scheduling method based on the improved particle swarm optimization algorithm according to claim 8, characterized in that, The output candidate scheduling scheme includes: The path offset of the candidate scheduling particle is generated based on the recalculated particle written to the individual guidance position and the recalculated particle written to the group guidance position. The path offset is limited to the scheduling time period corresponding to the particle repair boundary, so that the subsequent updates of candidate scheduling particles converge along the repaired energy storage charging and discharging path, controllable power supply compensation path and grid-connected compensation path. Update the scheduling path of the candidate scheduling particles in the continuous scheduling period according to the path offset to generate multiple constrained scheduling schemes. Output candidate scheduling schemes for smart microgrid scheduling from the plurality of constrained scheduling schemes.

10. A smart microgrid optimization scheduling system based on an improved particle swarm optimization algorithm, characterized in that, include: The particle generation module is used to obtain the operation prediction and operation constraints of the smart microgrid during the continuous scheduling period, and generate multiple candidate scheduling particles based on the operation prediction and operation constraints. Each candidate scheduling particle carries the energy storage charging and discharging path, the controllable power output path and the grid-connected interaction path during the continuous scheduling period. The state chain generation module is used to perform forward recursion of the energy storage charge state of the candidate scheduling particles in chronological order, and bind the recursed charge state with the scheduling path of the corresponding time period to form a cross-time period state chain. The fragment generation module is used to locate the first abnormal period that first breaks the continuous executability in the cross-time state chain, retain the preceding feasible fragments that can be continuously passed before the first abnormal period, and form subsequent gap fragments based on the power difference that has not been accepted since the first abnormal period. The particle repair module is used to generate a particle repair boundary based on the preceding feasible fragment and the subsequent gap fragment, adjust the energy storage charging and discharging path according to the particle repair boundary, and distribute the remaining power difference after adjustment to the controllable power output path and the grid-connected interaction path in the positive and negative directions to obtain repair scheduling particles. The guidance update module is used to re-execute the forward recursion of the energy storage charge state for the repair scheduling particle. When the recalculated cross-time period state chain satisfies the operation constraints, the individual guidance position or group guidance position of the corresponding candidate scheduling particle is replaced by the repair scheduling particle, and the candidate scheduling scheme is output.

Images