A warehouse park light storage and charging intelligent flexible carbon reduction method and system

Abstract

The present application relates to the technical field of light storage and charging management, and particularly relates to a warehouse park light storage and charging intelligent flexible carbon reduction method and system, comprising: based on the distribution of the warehouse park power distribution network topology and energy-using equipment, dividing the target control unit bound with the power distribution link; synchronously collecting the full-link multi-source space-time operation data, obtaining the standardized time series feature set through cleaning and multi-scale feature coding; based on the feature set, calibrating the physical safety compliance threshold matrix and the multi-objective benchmark set of carbon reduction benefits; generating the initial scheduling scheme with the threshold matrix as the hard constraint, obtaining the safety compliance scheme through full-time period verification and iterative correction; in the feasible region, completing multi-objective optimization with the benchmark set as the target, outputting the optimal scheme and issuing the execution. Through the present application, the warehouse park light storage and charging system can maximize the release of carbon reduction benefits and cooperatively prolong the equipment life and smooth the power grid peak-valley difference under the premise of ensuring the full-time period full-link physical safety compliance.

Description

Technical Field

[0001] This invention relates to the technical field of photovoltaic, energy storage, and charging management, and in particular to a smart and flexible carbon reduction method and system for photovoltaic, energy storage, and charging in warehouse parks. Background Technology

[0002] As energy-consuming nodes in the logistics industry chain, warehousing and logistics parks have the typical characteristics of large potential for rooftop distributed photovoltaic installations, concentrated load of charging piles for new energy trucks, rigid and stable energy consumption of cold chain warehousing, and abundant adjustable load resources. They are the main scenarios for the implementation of integrated photovoltaic, energy storage and charging systems and the application of park-level carbon reduction technologies.

[0003] Existing carbon reduction methods generally adopt a centralized multi-objective optimization scheduling model. This model uses the physical safety of the power distribution system in the storage park as a hard constraint and the optimal carbon reduction benefit as the core optimization objective. It couples the safety constraints and the carbon reduction optimization objective into the same multi-objective optimization mathematical model for solution, thereby obtaining a photovoltaic-storage-charging coordinated scheduling scheme that balances operational safety and carbon reduction needs. However, the existing coupled solution scheduling model cannot simultaneously guarantee the all-time safety compliance of the photovoltaic-storage-charging scheduling scheme in the storage park and maximize carbon reduction benefits within the safety feasible domain. For example, in the coupled solution process, in pursuit of better carbon reduction benefits, the optimization results are prone to exceeding the safety constraints at the power distribution link level during local periods with large source-load fluctuations, failing to guarantee all-time, all-link safety compliance. Furthermore, to avoid safety exceedances and ensure the safety of the entire link operation, a large conservative safety margin is often required, which significantly compresses the feasible space for carbon reduction optimization, making it difficult to maximize carbon reduction benefits within the safety boundary. Summary of the Invention

[0004] This invention provides a smart and flexible carbon reduction method and system for photovoltaic, energy storage and charging facilities in warehouse parks, which can effectively solve the problems in the background technology.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A smart and flexible carbon reduction method for warehousing parks, comprising: Based on the power distribution network topology and spatial distribution of energy-consuming equipment in the warehousing park, target control units are defined and bound to each power distribution link; Synchronously collect multi-source spatiotemporal operational data from the target control unit across the entire chain, and obtain a standardized temporal feature set through data cleaning and multi-scale feature coding; Based on a standardized time-series feature set, the physical safety compliance threshold matrix and the multi-objective benchmark set for carbon reduction benefits of the target control unit are calibrated. An initial collaborative scheduling scheme is generated using the physical security compliance threshold matrix as a hard constraint. After full-time compliance verification and iterative correction, a security-compliant scheduling scheme is obtained. Within the feasible domain of the safety and compliance scheduling scheme, the multi-objective optimization is completed using the multi-objective benchmark set of carbon reduction benefits as the optimization objective, and the optimal carbon reduction scheduling scheme is output and issued for execution.

[0006] Furthermore, target control units are defined and bound one-to-one with each power distribution link, including: Taking the power distribution room of the warehouse park as a node, and combining the physical route of the power distribution circuit, it is divided into several independent power supply unit grids, with each grid corresponding to a unique power distribution circuit. Grid cells that meet the preset grid screening conditions are identified as target control units. The preset grid screening conditions are that they contain at least two types of photovoltaic, energy storage and charging equipment, contain core energy loads for storage, and have carbon emission and power quality monitoring devices deployed.

[0007] Furthermore, multi-source spatiotemporal operational data from the target control unit are collected synchronously across the entire chain. After data cleaning and multi-scale feature encoding, a standardized time-series feature set is obtained, including: With a preset collection cycle, multi-source spatiotemporal operation data from the grid side, photovoltaic side, energy storage side, charging pile side, load side, and carbon emission side of the target control unit are collected simultaneously. All data are marked with a unified timestamp and a unique code of the target control unit. Data cleaning was completed by removing outliers using the improved 3σ criterion and filling in missing data using the sliding window interpolation method. The standardized temporal feature set was then obtained through multi-scale feature embedding encoding.

[0008] Furthermore, the dynamic calibration method for the physical security compliance threshold matrix is ​​as follows: Based on multi-source spatiotemporal operational data, combined with power grid dispatching procedures, equipment operation specifications, and park power supply guarantee requirements, hard constraint thresholds are calibrated for the power grid side, equipment side, and load side respectively, and a two-dimensional physical safety compliance threshold matrix is ​​constructed for each time period.

[0009] Furthermore, the dynamic calibration method for the multi-objective benchmark set of carbon reduction benefits is as follows: Acquire real-time carbon emission factors of target management units, green electricity ratio of power grid in different time periods, peak and valley electricity prices, energy storage life decay curves and carbon emission quotas for different time periods; With the goal of minimizing carbon emission intensity, and taking into account electricity costs, energy storage life loss and grid peak-valley difference smoothing, a multi-objective optimization function is constructed, and a multi-objective benchmark set of carbon reduction benefits is obtained through calibration.

[0010] Furthermore, an initial collaborative scheduling scheme is generated using the physical security compliance threshold matrix as a hard constraint. After full-time compliance verification and iterative correction, a security-compliant scheduling scheme is obtained, including: Based on power system flow calculation, the initial coordinated scheduling scheme is simulated for full-time operation within a future preset period, with each preset duration. Compare the simulation data with the physical safety compliance threshold matrix on a time-by-time basis; If all data is within the threshold range, the verification passes; otherwise, the threshold-exceeding items are located, a safety iterative correction factor is generated, the scheme is corrected in a targeted manner, and then the verification is repeated until a safe and compliant scheduling scheme is obtained.

[0011] Furthermore, the multi-objective optimization adopts an improved non-dominated sorting genetic algorithm. In the optimization process, carbon emission intensity, electricity cost, energy storage life loss and grid peak-valley difference smoothing are taken as optimization objectives, and corresponding dynamic weights are configured for each. The dynamic weights are adjusted in real time based on the energy consumption characteristics of the park during different time periods and the requirements of the power grid, and the optimal carbon reduction scheduling scheme is output from the Pareto optimal solution set.

[0012] Furthermore, during the multi-objective optimization process, a safe boundary projection mechanism is applied to the generated offspring individuals: If the decision variables of a child individual exceed the range defined by the physical security compliance threshold matrix, the variable value is projected to the nearest security boundary point, or the child individual is discarded and regenerated to ensure that all individuals entering the next generation meet the physical security requirements.

[0013] Furthermore, after the optimal carbon reduction scheduling plan is issued and implemented, a closed-loop correction is performed: The actual operating data of the target control unit is monitored in real time. If the actual deviation exceeds the allowable tolerance, a second correction is made within the allowable range of the physical safety compliance threshold matrix, and the corrected state is used as the initial state for the next round of rolling optimization.

[0014] On the other hand, the present invention also provides a smart flexible carbon reduction system for photovoltaic, energy storage, and charging in a warehouse park, comprising: The unit division module is used to divide target management units that are bound one-to-one with the power distribution links, based on the topology of the power distribution network and the spatial distribution of energy-consuming equipment in the warehouse park. The data governance module is used to synchronously collect multi-source spatiotemporal operational data from the target control unit across the entire chain, and obtain a standardized time-series feature set through data cleaning and multi-scale feature encoding. The benchmark calibration module is used to calibrate the physical safety compliance threshold matrix and the multi-objective benchmark set of carbon reduction benefits of the target control unit based on a standardized time-series feature set. The security verification module is used to generate an initial collaborative scheduling scheme with the physical security compliance threshold matrix as a hard constraint. After full-time compliance verification and iterative correction, a safe and compliant scheduling scheme is obtained. The optimization execution module is used to perform multi-objective optimization within the feasible domain of the safety and compliance scheduling scheme, using the multi-objective benchmark set of carbon reduction benefits as the optimization objective, outputting the optimal carbon reduction scheduling scheme and issuing it for execution.

[0015] The technical solution of this invention can achieve the following technical effects: By dividing the target control units into one-to-one bindings with the power distribution links, and combining the accurate collection, cleaning and coding of multi-source data across the entire link, the physical safety compliance threshold matrix and carbon reduction benefit multi-objective benchmark set adapted to the park's operating conditions are dynamically calibrated. Through full-time compliance verification, iterative correction and improvement of the non-dominated sorting genetic algorithm for multi-objective optimization, coupled with the safety boundary projection mechanism and closed-loop correction, the carbon reduction benefits are maximized while ensuring the full-time and full-link physical safety compliance of the photovoltaic-storage-charging system in the warehousing park. At the same time, it synergistically reduces electricity costs, extends the life of energy storage equipment, smooths the peak-valley difference of the power grid, and improves the stability, adaptability and economy of the photovoltaic-storage-charging system. It adapts to the energy consumption characteristics of the warehousing park and achieves the optimal synergy between safety, carbon reduction and equipment operation and maintenance, and power grid stability.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in 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 recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the process of a smart and flexible carbon reduction method for photovoltaic, energy storage and charging in a warehouse park according to the present invention. Figure 2 This is a schematic diagram of the structure of a smart flexible carbon reduction system for photovoltaic storage and charging in a warehouse park according to the present invention. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] like Figure 1As shown, the present invention provides a smart and flexible carbon reduction method for photovoltaic, energy storage, and charging facilities in a warehouse park, which specifically includes the following steps: Step S1: Based on the power distribution network topology and spatial distribution of energy-consuming equipment in the warehouse park, divide the target control units into one-to-one bindings with the power distribution links; Step S2: Synchronously collect multi-source spatiotemporal operation data of the target control unit across the entire link, and obtain a standardized time series feature set through data cleaning and multi-scale feature coding; Step S3: Based on the standardized time-series feature set, calibrate the physical safety compliance threshold matrix and the multi-objective benchmark set of carbon reduction benefits for the target control unit; Step S4: Generate an initial collaborative scheduling scheme using the physical security compliance threshold matrix as a hard constraint. After full-time compliance verification and iterative correction, obtain a security-compliant scheduling scheme. Step S5: Within the feasible domain of the safety and compliance scheduling scheme, use the multi-objective benchmark set of carbon reduction benefits as the optimization objective to complete the multi-objective optimization, output the optimal carbon reduction scheduling scheme and issue it for execution.

[0022] In this embodiment, by using an independent control unit physically bound to the power distribution link as the spatial basis, and combining it with a time scheduling logic that first locks the safe and feasible domain through full-time power flow simulation verification, and then performs dynamic weighted multi-objective optimization within the feasible domain, the algorithm does not need to repeatedly explore the safety boundary or reserve a global conservative margin during the optimization process. Instead, it directly uses a calibrated dynamic benchmark set to accurately mine the carbon reduction potential within the verified physical safety closed loop. This solves the problem of safety overstepping the boundary in local time periods due to source-load fluctuations or excessively conservative compression of carbon reduction space. Under the premise of ensuring full-time and full-link physical safety compliance, it maximizes the carbon reduction benefits of the photovoltaic-storage-charging system in the warehouse park, extends the life of energy storage equipment, and smooths the peak-valley difference of the power grid.

[0023] In a specific implementation, as one example, given that the centralized scheduling mode does not finely decompose the power distribution links in the warehousing park, the safety constraints and carbon reduction goals mutually restrain each other when multiple links are coupled for scheduling, making it impossible to achieve dual protection of safety and carbon reduction benefits at all times. Therefore, it is necessary to achieve one-to-one binding, precise screening, and independent scheduling optimization of target management units and power distribution links. This embodiment achieves distributed fine-grained scheduling by dividing the power supply unit grid based on the power distribution circuit, screening target management units according to preset conditions, and establishing independently optimized scheduling links, as detailed below: Step S11: Obtain the topology of the power distribution network and the spatial distribution data of energy-consuming equipment in the warehousing park. The power distribution network topology data is obtained through a combination of drawing review and on-site survey. First, the original power distribution engineering design drawings and power distribution equipment ledgers of the park are retrieved to clarify the location of the power distribution room, the number of power distribution circuits, the outgoing ports of the circuits, the cable laying paths, and the power supply range of each circuit. Then, a cable path detector is used to trace the cable route to confirm the actual physical boundaries of the power distribution circuits, avoiding division deviations caused by discrepancies between the drawings and the actual site. The spatial distribution data of energy-consuming equipment is obtained through a park equipment location survey. All photovoltaic, energy storage, and charging equipment, core energy loads of the warehousing, and monitoring devices in the park are located one by one, and the installation coordinates and power supply circuits of each device are recorded to form a spatial distribution ledger of energy-consuming equipment. By obtaining the above two types of data, the correspondence between the physical carriers of the power distribution links and the energy-consuming equipment is clarified, providing a physical basis for grid division and realizing one-to-one binding with the power distribution links. Step S12: Divide the grid into independent power supply units; taking all the power distribution rooms in the warehouse park as core nodes, each power distribution room is an independent power supply head, and the grid is divided according to the physical routing of each power distribution circuit; the physical routing of the power distribution circuit is based on the cable laying path, following the principles of independent circuit, clear range, and no overlapping boundaries, dividing the power supply range of each power distribution circuit into an independent power supply unit grid, that is, one grid corresponds to a unique power distribution circuit; during the division process, for different functional areas of the warehouse park, the grid boundaries are reasonably defined in combination with the power supply capacity and load distribution of the power distribution circuits; for example, a power distribution circuit led out from a certain power distribution room This circuit is specifically designed to power refrigeration units, photovoltaic modules, and two charging piles within the cold chain storage area. The power supply range of this circuit constitutes an independent grid, with the grid boundary defined by the boundary of the cold chain storage area covered by the circuit's cable. This ensures that all energy-consuming equipment within the grid is powered independently by its corresponding distribution circuit, with no cross-circuit power supply. The distribution room is the starting point of the power distribution link, and all distribution circuits originate from it. By using the distribution room as a node, the grid division and the source of the power distribution link are ensured to correspond. Each grid corresponds to a unique distribution circuit, enabling precise binding between grid management and power distribution link management. This provides physical support for independent scheduling and optimization, avoiding energy consumption interference between different links. Step S13: Set grid screening conditions and screen target control units; the grid screening conditions are set based on the actual needs of photovoltaic-storage-charging coordinated carbon reduction and safe scheduling, including: a. The grid contains at least two types of photovoltaic, energy storage and charging equipment. Photovoltaic, energy storage and charging equipment is the core carrier of carbon reduction in the park. A single type of equipment cannot achieve synergistic carbon reduction. At least two types of equipment can form a basic synergistic effect. For example, the synergy between photovoltaic modules and energy storage devices can realize the storage and stable release of photovoltaic power. The synergy between photovoltaic modules and charging piles can realize the direct supply of clean power to new energy vehicles, providing hardware support for carbon reduction optimization. b. The grid contains the core energy load of warehousing; the core energy load of warehousing is the main component of the park's energy consumption and the key target of load regulation. Its energy consumption is rigid and its load fluctuation is regular. By managing the grid containing the core load, carbon reduction can be achieved in the main energy consumption links of the park. If the grid does not contain the core energy load and only contains auxiliary energy equipment, the carbon reduction effect of scheduling optimization is limited and the park-level carbon reduction target cannot be achieved. c. Carbon emission and power quality monitoring devices are deployed within the grid. The carbon emission monitoring devices are used to collect carbon emission data within the grid in real time, providing data for carbon reduction benefit assessment and optimization. The power quality monitoring devices are used to monitor electrical parameters such as voltage, current and frequency within the grid in real time, providing a basis for safety compliance verification. Without either monitoring device, threshold calibration, scheduling optimization and safety verification cannot be achieved, and the safety compliance of the scheduling scheme and the accurate assessment of carbon reduction benefits cannot be guaranteed. The screening process employs a grid-by-grid verification and multi-dimensional validation approach, systematically checking each of the divided independent power supply unit grids. First, the type and quantity of photovoltaic, energy storage, and charging equipment within the grid are checked, confirming at least two types through equipment ledgers and on-site inspections. Second, the grid is checked to ensure it contains core energy loads for warehousing, confirming load type and attribution against the core load list. Finally, the deployment of monitoring devices is checked, confirming that carbon emission monitors and power quality monitors are installed correctly and capable of collecting data. If all three conditions are met, the grid meets the screening criteria and is identified as a target control unit. If any condition is not met, the grid is excluded from the target control unit scope. Such grids only undergo basic power supply monitoring and do not participate in the photovoltaic-energy storage-charging collaborative carbon reduction scheduling optimization. Due to the lack of necessary hardware infrastructure or data support, including such grids would reduce scheduling optimization efficiency and even affect the overall safety and compliance of the solution. Step S14: Establish independent scheduling and optimization links for each target management unit. Each target management unit corresponds to a unique power distribution link. The establishment of independent scheduling and optimization links is based on the principles of link independence, data independence, and optimization independence. This is achieved by configuring a dedicated scheduling and control unit for each target management unit and establishing dedicated communication links with the photovoltaic, energy storage, and charging equipment, core energy loads, and monitoring devices within the corresponding grid. Communication adopts a combination of industrial Ethernet and wireless communication to ensure the real-time performance and stability of data transmission. Under this architecture, the scheduling and control unit independently collects the operating data of each target management unit, independently performs threshold calibration, and independently conducts scheduling optimization and safety verification without cross-interference with the scheduling links of other target management units. Each target management unit can fully explore carbon reduction potential within the safety boundary of its own power distribution link, ensuring safety and compliance during local periods in the global optimization process, while fully releasing the carbon reduction potential within the safety boundary.

[0024] In this embodiment, by binding the target management unit to the power distribution link one-to-one, the scheduling optimization of each unit is only for the corresponding power distribution link, which can accurately locate the safety hazards and carbon reduction potential within the link and avoid safety constraint overruns caused by cross-link interference. By screening conditions, it is ensured that the target management unit has the hardware and data foundation for photovoltaic-storage-charging collaborative carbon reduction, avoiding ineffective management and improving the targeting and efficiency of scheduling optimization. By independently scheduling and optimizing the link, each unit can maximize the carbon reduction potential within its own safe and feasible domain without reserving a global conservative safety margin, while also being able to quickly respond to source-load fluctuations and ensure safety and compliance at all times.

[0025] In some embodiments of the present invention, existing methods for collecting multi-source operational data of target control units typically employ synchronous collection. However, this synchronous collection only achieves the synchronous triggering of collection actions and lacks a unified time calibration and data verification mechanism. This results in nominally synchronized data from different sides having discrepancies in actual timestamps and data anomalies. Furthermore, feature encoding usually adopts a single-scale approach, which cannot adapt to the spatiotemporal characteristics and multi-scale fluctuation patterns of multi-source data in storage parks. This leads to deviations in related data processing and optimization, thereby affecting the safety and carbon reduction effectiveness of scheduling schemes. Therefore, it is necessary to achieve accurate synchronous collection, accurate cleaning, and adaptable feature encoding of multi-source spatiotemporal operational data. This embodiment completes the construction of a standardized time-series feature set by constructing a full-link synchronous collection mechanism, optimizing data cleaning strategies, and designing multi-scale feature embedding encoding. Specifically, the following operations are performed: Step S21: Synchronously collect multi-source spatiotemporal operational data of the target control unit across the entire chain, and all data are marked with a unified timestamp and a unique code for the target control unit; the collection scope includes: a. Grid side: The data collection target is the common connection point of the target control unit connected to the grid. The collected data includes the voltage, current, frequency, active power and reactive power of the connection point. The data is collected in real time by the power quality monitor deployed at the connection point. The collected data directly reflects the power supply status of the grid side and provides basic data for threshold calibration. b. Photovoltaic side: The data collection targets are all photovoltaic modules and combiner boxes within the target control unit. The collected data includes the output power of photovoltaic modules, irradiance, and module temperature. The data is collected in conjunction with the monitoring module built into the photovoltaic combiner box and independently deployed irradiance and temperature sensors. This type of data reflects the real-time fluctuations and potential of photovoltaic output, providing clean energy-related data for optimizing carbon reduction benefits. c. Energy storage side: The data collected is collected from the battery management system of the energy storage device. The collected data includes the remaining capacity of the energy storage battery, charging and discharging power, battery temperature, and number of charge and discharge cycles. The data is uploaded to the scheduling and control unit in real time by the energy storage battery management system. This type of data is used to evaluate the operating status and lifespan of the energy storage device and to support the construction of a multi-objective benchmark set for carbon reduction benefits. d. Charging pile side: The data collected is from new energy charging piles. The data collected includes the charging power, charging time, charging voltage and charging current of the charging piles. The data is collected and uploaded by the charging pile controller. This type of data reflects the load fluctuation characteristics of the charging piles and provides a basis for load-side regulation and peak-valley difference smoothing of the power grid. e. Load side: The data collection targets are the core energy loads of the warehouse, including the operating power and start / stop status of the cold chain refrigeration units, the operating power of the shelf lighting system and the cargo conveying equipment. This data is collected in conjunction with the load controller and the electricity metering device. This type of data reflects the rigid demand and fluctuation pattern of energy consumption in the park and is used for physical safety constraint threshold calibration and load adjustment. f. Carbon emission side: The data collected is from carbon emission monitoring instruments. The collected data includes real-time carbon emission concentration and total carbon emissions within the unit. The data is collected and uploaded by the carbon emission monitoring instruments and is directly used for carbon reduction benefit assessment and multi-objective optimization. The data collection cycle is determined based on the energy consumption fluctuation patterns and data usage needs of the warehousing park, and is reasonably set according to the frequency of data changes on each side. For example, photovoltaic output fluctuates rapidly due to sunlight, and charging pile load fluctuates randomly due to vehicle charging demand. The data collection cycle for these two types of data is set to a shorter cycle. The data fluctuations on the grid side and load side are relatively smooth, and the collection cycle can be appropriately extended. The data collection cycle for all sides is uniformly calibrated to ensure that the collection actions are triggered synchronously. The unified timestamp adopts the park's unified standard time, and marks each set of synchronously collected multi-source data with the same timestamp to ensure the spatiotemporal correlation of multi-source data. The target control unit adopts a preset unique code, and each set of collected data is bound to the corresponding code to ensure data traceability and avoid confusion. Step S22: Clean the collected multi-source spatiotemporal operation data by removing outliers using the improved 3σ criterion and filling in missing data using the sliding window interpolation method to complete the data cleaning operation; The improved 3σ criterion is adaptable to the distribution characteristics of multi-source operational data in warehousing parks and can cope with non-normal data fluctuations. By introducing a data volatility coefficient to adjust the value of σ, it avoids misjudgment and missed judgment of outliers, ensuring the accuracy and adaptability of outlier identification. The process is as follows: First, statistical analysis is performed on each type of collected data to calculate the mean and standard deviation σ of the data, and then the volatility coefficient, i.e., the ratio of the standard deviation to the mean, is obtained. Then, the weight of σ is adjusted using the volatility coefficient. If the volatility coefficient is large, the weight of σ is appropriately increased to accommodate large fluctuations and avoid misjudgment; if the volatility coefficient is small, the weight of σ is decreased to improve sensitivity and prevent missed judgment. Based on the adjusted σ, a judgment interval is set, and data falling outside this interval is identified as outliers. For the identified outliers, the mean of the adjacent normal data is used to replace them to preserve the temporal regularity of the data. The sliding window interpolation method is adapted to the temporal continuity of multi-source data in warehousing parks. It can utilize normal data before and after missing data to fill in missing values ​​through linear interpolation, maximizing the preservation of the temporal fluctuation characteristics of the data, reducing the deviation between the filled data and the actual data, and ensuring the integrity and accuracy of the data. The process is as follows: Set the size of the sliding window, which is determined according to the data collection cycle and the possible duration of the missing data. For example, when the collection cycle is short, the window size is set to a smaller value to ensure the accuracy of interpolation; when the collection cycle is long, the window size is appropriately increased to avoid the interpolated data lacking representativeness due to the window being too small. For a missing data point, take that point as the center, select the normal data before and after within the sliding window, calculate the slope of the data change, and calculate the filled value of the missing data through the linear interpolation formula. If there are multiple consecutive missing data points, interpolation calculation is performed on each missing point in turn to ensure that the data sequence after filling in the missing data is continuous, smooth, and consistent with the fluctuation pattern of the original data. Step S23: Perform multi-scale feature embedding encoding on the cleaned multi-source time-series data sequence to obtain a standardized time-series feature set. The multi-source data from the warehousing park exhibits fluctuation characteristics at different time scales. For example, photovoltaic output shows short-term and periodic fluctuations within the day, while load-side data shows periodic and long-term seasonal fluctuations. Therefore, multi-scale feature embedding encoding is used to extract and encode features at different time scales, achieving data standardization and ensuring that the feature set comprehensively reflects the spatiotemporal characteristics and fluctuation patterns of the multi-source data. This step first decomposes the multi-source time-series data sequence into three dimensions: short-scale, medium-scale, and long-scale. The division of each scale is based on the park's... The data is determined based on energy fluctuation patterns and data characteristics. Features are extracted for each scale, including the fluctuation amplitude, mean, peak, trough, and fluctuation period at that scale. The fluctuation amplitude, peak, and trough are used to calibrate the physical safety compliance threshold, while the mean and fluctuation period are used to construct a multi-objective optimization function for carbon reduction benefits. Finally, the above multi-scale features are embedded and encoded using a normalization method, mapping all types of features to a unified interval of [0,1] to eliminate dimensional differences. The encoded features are all bound to the data source, time scale, control unit, and timestamp, ultimately forming a standardized time-series feature set. This feature set integrates multi-source data from various sides with multi-scale fluctuation features.

[0026] In this embodiment, a full-link synchronous acquisition mechanism is constructed to achieve synchronous acquisition and accurate traceability of multi-source data. By improving the 3σ criterion and sliding window interpolation method, the data characteristics of the park are adapted to ensure the accuracy and integrity of data cleaning. Through multi-scale feature embedding and coding, the multi-scale fluctuation characteristics of the data are fully captured, and feature standardization is achieved. A complete data processing link is formed from data acquisition, cleaning to coding, which is directly used for relevant threshold calibration, multi-objective optimization and scheduling scheme generation, ensuring the orderly advancement and implementation effect of the smart flexible carbon reduction method of photovoltaic storage and charging in the entire warehousing park.

[0027] In practical implementation, as one example, existing methods often set fixed values ​​for physical safety constraint threshold calibration. However, the source load fluctuations in warehousing parks are random and diverse, making fixed thresholds prone to deviating from actual operating conditions. Furthermore, the carbon reduction benefit benchmark set is statically set, often focusing only on a single carbon emission target without considering related targets such as electricity costs, energy storage lifespan loss, and grid peak-valley leveling. This makes it unable to respond to real-time changes in operating conditions, resulting in a lack of scientifically sound optimization benchmarks for multi-objective optimization and difficulty in achieving a synergistic balance between carbon reduction benefits and related targets within the safety boundary. This embodiment constructs a dynamic calibration mechanism based on a standardized time-series feature set to calibrate the physical safety compliance threshold matrix and the multi-objective carbon reduction benefit benchmark set. The specific implementation steps are as follows: Step S31: Analyze the grid-side voltage amplitude, frequency deviation, and branch current data from the standardized time-series feature set. Combine the allowable deviation range of rated voltage and rated frequency specified in the grid dispatching regulations to extract the grid-side hard constraint boundary. Simultaneously, read the maximum output current of the photovoltaic inverter, the upper limit of the charge / discharge rate of the energy storage battery, and the maximum power limit of the charging pile on the equipment side. Compare these with the nameplate parameters and thermal stability limits in the equipment operation specifications, eliminate instantaneous derating interference caused by changes in ambient temperature, and determine the equipment-side physical load hard constraint boundary. Then, extract the starting inrush current of the cold chain compressor and the maximum load rate of the lighting circuit on the load side. Based on the power supply guarantee requirements for the core loads in the park, define the minimum power supply reliability hard constraint boundary on the load side. Align the above three types of hard constraint boundaries along the time axis and construct a physical safety compliance threshold matrix with a two-dimensional structure of constraint type and time segment using a preset acquisition period as the granularity. Each row represents the union safety interval of the grid, equipment, and load at a specific time. If the dispatching instruction at any time exceeds the interval defined in that row, it is judged as physically unsafe. Step S32: Call the carbon emission monitoring data from the standardized time-series feature set to obtain the grid average carbon emission factor of the target control unit access point in real time. This factor fluctuates hourly with changes in the grid power structure. Simultaneously analyze the green electricity ratio data issued by the grid dispatch center for each time period to quantify the proportion of zero carbon components in a unit of electricity. Read the peak-valley time-of-use electricity price signal from the electricity market trading interface to identify whether the current time is in the peak, high, flat, or low price range. Retrieve the cumulative cycle count and real-time temperature data recorded by the energy storage battery management system, substitute them into the preset energy storage life decay experience curve, and calculate the equivalent life loss cost corresponding to the current charging and discharging action. Obtain the time period carbon emission quota upper limit issued by the park management department to clarify the remaining emission quota within the current time window. Use the above five types of dynamic parameters as input variables for the multi-objective benchmark set of carbon reduction benefits. Among them, the carbon emission factor and green electricity ratio directly determine the carbon emission reduction potential, the peak-valley electricity price determines the economic adjustment space, the life decay curve determines the cost of energy storage use, and the carbon emission quota determines the emission red line. Step S33: Construct a multi-objective optimization function framework with minimizing carbon emission intensity as the primary guiding principle, while simultaneously considering minimizing electricity costs, minimizing energy storage lifespan loss, and optimizing peak-valley shift smoothing of the power grid. In this framework, instead of pre-setting fixed weight coefficients for each objective, a dynamic weight adjustment mechanism is established based on the energy consumption characteristics of the industrial park during specific time periods and the real-time requirements of the power grid. When the standardized time-series feature set indicates that the current time is during the peak load period of the power grid and the proportion of green electricity is low, the priority weights of peak-valley shift smoothing and carbon emission intensity minimization in the benchmark set are increased to guide subsequent optimization. The system prioritizes energy storage discharge to support load and reduce reliance on thermal power. When the current time indicates a period of low electricity prices and sufficient remaining lifespan for energy storage, the weight for minimizing electricity costs is increased to guide subsequent optimization towards energy storage charging and reserving low-cost electricity. When carbon emission allowances are about to be exhausted, the weight for minimizing carbon emission intensity is forcibly raised to the veto level to restrict the generation of any high-carbon emission dispatch strategies. Through this dynamic weight adjustment mechanism, a multi-objective benchmark set of carbon reduction benefits that evolves over time and adapts to different operating conditions is calibrated. This benchmark set defines the search direction and evaluation criteria for the optimal solution under different operating conditions.

[0028] In this embodiment, the physical safety compliance threshold matrix is ​​constructed based on deterministic specifications such as power grid dispatching procedures, equipment nameplate parameters, and park supply guarantee levels. This eliminates the risk of random boundary violations during the optimization process and ensures safety and compliance across all time periods and links. The multi-objective benchmark set for carbon reduction benefits is constructed based on dynamic operating data such as real-time carbon factors, green electricity ratios, electricity price signals, and lifetime curves. A dynamic weighting mechanism is used to achieve real-time response of optimization objectives to park operating conditions and power grid demands. By treating safety verification as an independent step, multi-objective optimization no longer requires the allocation of computing power to handle safety constraints. Computing power can be concentrated within a defined safe and feasible domain to explore carbon reduction potential. This avoids the waste of carbon reduction space caused by excessive conservative margins for safety and eliminates the risk of sacrificing safe operation for carbon reduction, thereby achieving precise maximization of carbon reduction benefits within the safety boundary.

[0029] In a specific implementation, as one example, existing photovoltaic-storage-charging collaborative scheduling technology often directly solves the objective function when generating the initial scheme, embedding safety constraints as penalty terms into the objective function. This coupling method leads to the algorithm easily testing the safety boundary in pursuit of carbon reduction extremes under scenarios of drastic source-load fluctuations, causing voltage overruns or line overloads in local time periods. However, increasing the penalty coefficient to force compliance with safety can cause the solution space to converge too quickly, getting stuck in local optima and sacrificing a large amount of carbon reduction potential. This embodiment uses the physical safety compliance threshold matrix as an absolute veto criterion, pre-imagines the future state through full-time power flow simulation, and uses a deviation feedback mechanism to generate targeted safety iteration correction factors to perform closed-loop correction of the initial scheme, ensuring that the output scheme strictly meets the full-link physical safety requirements before entering the carbon reduction optimization stage. The specific implementation steps are as follows: Step S41: Based on the physical safety compliance threshold matrix, and combined with the photovoltaic predicted output, energy storage charge status, charging pile reserved load and cold chain basic load data of the target control unit at the current moment, an initial collaborative scheduling scheme is generated using heuristic rules. This initial scheme prioritizes meeting the load power supply demand and preliminarily arranges the energy storage charging and discharging strategy based on the electricity price signal. It does not verify whether it fully complies with the time-by-time hard constraints in the physical safety compliance threshold matrix, but only ensures the correctness of power balance and equipment start-up and shutdown logic, thereby obtaining a benchmark scheduling sequence with carbon reduction potential but potentially safety hazards. Step S42: Using power system power flow calculation methods, a full-time operation simulation model of the initial collaborative scheduling scheme is built. Power system power flow calculation can simulate the real-time operation status of the distribution link and calculate the operating parameters such as voltage, current, and power of each node in the distribution network. This is consistent with the actual operation logic of the distribution system in the warehousing park, avoiding deviations between simulation data and actual operation status, and ensuring the reliability of the verification results. When building the simulation model, the input parameters include multi-source operation data in the standardized time-series feature set, the control parameters of each device in the initial collaborative scheduling scheme, and the time-period constraint parameters in the physical safety compliance threshold matrix. At the same time, the basic parameters such as the distribution network topology and line impedance of the target management unit are also incorporated to ensure that the simulation model can completely replicate the operation scenario of the distribution link. The simulation scope covers the entire distribution link of the target management unit without omitting any power consumption node or device, avoiding verification omissions due to incomplete simulation scope. Step S43: Based on the simulation model, conduct full-time operation simulation of the initial collaborative scheduling scheme within a future preset period for each preset duration. The future preset period is determined based on the periodicity of energy consumption fluctuations in the warehousing park, combined with the park's daily operation rhythm and source-load fluctuation characteristics, ensuring that the period can cover the complete cycle of source-load fluctuations and avoiding incomplete verification due to an excessively short period. The setting of each preset duration is consistent with the time-period division of the threshold matrix, ensuring that the simulation data of each time period can be accurately compared with the threshold of the corresponding time period. During the simulation, the operation simulation data of each time period, each power distribution link, and each piece of equipment are recorded in real time, including grid-side node voltage, current, and frequency, equipment-side operating power and temperature, and actual operating load on the load side. Step S44: Compare the full-time operation simulation data with the physical safety compliance threshold matrix time by time to complete the full-time compliance verification. The comparison process adopts a precise comparison method of time by time and parameter by parameter. The simulation data of each time period is compared with the corresponding constraint parameter in the threshold matrix of that time period one by one. For example, the grid-side simulation frequency of a certain time period is compared with the grid-side frequency threshold of that time period, and the equipment-side simulation power is compared with the equipment-side power threshold of that time period to ensure that every operating parameter can be verified and no constraint item is missed. If all simulation parameters of a certain time period are within the corresponding threshold range, the verification of that time period is passed. If any simulation parameter of a certain time period exceeds the corresponding threshold range, the verification of that time period is determined to be unqualified. At the same time, the specific information of the exceeding threshold item is recorded, including the exceeding threshold time period, the exceeding threshold parameter type, the exceeding threshold value, the corresponding power distribution link and equipment. Step S45: For the threshold items that fail the verification, locate the root cause of the problem and generate a safety iterative correction factor. For example, the threshold may be caused by excessive photovoltaic output leading to excessive grid voltage, excessive energy storage charging and discharging rate leading to excessive equipment power, or excessive load fluctuation leading to excessive load constraints. The location process needs to be accurate to specific equipment and specific parameters to avoid ineffective correction due to ambiguous location. The generation of the safety iterative correction factor is based on the specific information of the threshold item, combined with the constraint requirements of the threshold matrix, to determine the correction direction and correction magnitude. The correction factor corresponds one-to-one with the threshold item. For example, for voltage exceeding the threshold, a photovoltaic output adjustment correction factor is generated, and for power exceeding the threshold, an energy storage charging and discharging rate adjustment correction factor is generated. The correction magnitude is based on the principle of just returning the threshold parameter to the threshold range, without over-correction to avoid compressing the carbon reduction optimization space, and at the same time ensuring that the correction does not cause other parameters to exceed the threshold. Step S46: Based on the safety iterative correction factor, perform targeted correction on the initial collaborative scheduling scheme. After the correction is completed, repeat steps S42 to S45 until the full-time verification is passed, and obtain a safe and compliant scheduling scheme. The targeted correction process focuses on the control parameters corresponding to the threshold items, and only adjusts the parameters with problems. For example, according to the photovoltaic output adjustment correction factor, adjust the photovoltaic output allocation strategy for the corresponding time period, and according to the energy storage charge and discharge rate correction factor, adjust the energy storage charge and discharge parameters without changing other parameters that meet the constraints. This ensures the correction efficiency while retaining the reasonable carbon reduction optimization space in the initial scheme. After each correction, conduct full-time operation simulation, compliance verification and correction again until all simulation parameters for all time periods are within the constraint range of the threshold matrix. The scheduling scheme obtained at this time is the safe and compliant scheduling scheme, which can ensure safety and compliance throughout the entire time period and the entire link.

[0030] In this embodiment, by replacing static constraint judgment with full-time power flow simulation, dynamic voltage fluctuations and heat accumulation effects caused by source load fluctuations can be captured, solving the problem that static models cannot reflect the time-series coupling security issue. By introducing a security iterative correction factor mechanism, the security boundary violation problem is transformed into a directional variable adjustment problem, which can avoid the convergence difficulties or solution quality degradation caused by the difficulty in tuning the weight coefficients in the traditional penalty function method. By placing security verification in advance and independently of the carbon reduction optimization objective, it is ensured that the solution set entering the final optimization stage itself has full-time and full-link security attributes, eliminating the risk of security accidents caused by search randomness during the optimization process. At the same time, since the correction process follows the principle of minimum action, the carbon reduction characteristics of the initial solution are preserved to the greatest extent, providing a compliant initial solution space for maximizing carbon reduction benefits within the safe and feasible domain.

[0031] In practical implementation, as one example, existing photovoltaic-storage-charging coordinated scheduling technologies often suffer from limited solution space search during the multi-objective optimization stage due to the coupling between security constraints and optimization objectives. Furthermore, the use of fixed weights fails to respond to real-time fluctuations in source load within storage parks and the dynamic scheduling requirements of the power grid. Consequently, the generated scheduling schemes struggle to achieve a multi-dimensional balance of carbon reduction, economy, lifespan, and grid friendliness while ensuring end-to-end physical security. This embodiment, within the defined feasible domain of a safe and compliant scheduling scheme, utilizes an improved non-dominated sorting genetic algorithm, combined with real-time dynamic adjustment of optimization objective weights, to complete multi-objective optimization and output the optimal carbon reduction scheduling scheme. The specific implementation steps are as follows: Step S51: Read the safety and compliance scheduling plan and use the photovoltaic output sequence, energy storage charging and discharging power sequence, charging pile load sequence, and controllable load adjustment sequence contained therein as the baseline individual. Centered on this baseline individual, within the feasible region defined by the calibrated physical safety and compliance threshold matrix, introduce a Gaussian distributed random perturbation operator to generate several mutated individuals, which together form the initial population. Using a real number matrix encoding method, the matrix row vectors correspond to time slices within a future preset period, and the column vectors respectively map to four decision variables: active and reactive power setpoints of photovoltaic inverters, charging and discharging power commands of energy storage converters, start and stop status and power limits of charging piles, and flexible adjustment rate of cold chain load, to ensure that all individuals in the population meet the voltage, current, and equipment thermal stability limits at the physical level. Step S52: For each individual in the initial population, simultaneously and in parallel calculate four objective function values: carbon emission intensity, electricity cost, energy storage lifetime loss, and grid peak-valley difference smoothing degree. Based on the real-time grid carbon emission factor, combined with the proportion of photovoltaic green electricity self-consumption and grid-purchased electricity in the individual, integrate to obtain the full-cycle carbon footprint. Combined with the real-time peak-valley time-of-use electricity price signal, calculate the difference between electricity purchase expenditure and surplus electricity grid connection revenue under the individual scheduling strategy to obtain the electricity cost. Call the energy storage lifetime decay curve, map the charge and discharge depth and rate of each time period in the individual to the equivalent number of cycles, quantify the battery health cost to obtain the energy storage lifetime loss. Characterize the grid peak-valley difference smoothing degree by calculating the variance of the total load curve of the park after individual scheduling. Construct the fitness vector of the individual with the above four objective values ​​for non-dominated ranking. Step S53: Before each generation of evolutionary iteration, read the energy consumption characteristics data of the current period and the real-time operation data of the power grid from the standardized time series feature set, and calculate the dynamic weight coefficients of the four optimization objectives according to the preset rule logic. The rule logic is set as follows: when it is detected that the load rate of the current period is higher than the preset peak threshold and the green electricity ratio of the power grid is lower than the preset low carbon threshold, increase the weight coefficients of the two indicators of power grid peak-valley difference smoothing and carbon emission intensity, and decrease the weight coefficient of electricity cost; when it is detected that the current period is in the low-valley electricity price range and the remaining life of energy storage is higher than the preset health threshold, increase the weight coefficient of electricity cost; when it is detected that the cumulative carbon emissions are close to the upper limit of the period quota, adjust the carbon emission intensity weight coefficient to the maximum value and forcibly restrict the high carbon emission strategy; use the calculated dynamic weight coefficients to perform a corrected non-dominated ranking of the population individuals, divide the non-dominated individuals into different frontiers, and determine the optimal solution set of the current generation. Step S54: Introduce a density-adaptive crowding distance operator, calculate the Euclidean distance between adjacent individuals in the target space and multiply it by a density correction factor, increase the crowding distance value in sparsely distributed areas of the target space and decrease the value in dense areas; perform a selection operation, prioritize retaining individuals with higher rank, and if the ranks are the same, select individuals with larger crowding distances to form a parent population, ensuring that the final output solution set covers the entire spectrum of scheduling strategies from extreme carbon reduction to extreme economy, and avoids the solution set from clustering at a single target extreme point; Step S55: Perform a simulated binary crossover operator on the selected parent individuals to generate offspring by linearly combining the decision variables of the two parent individuals. The crossover probability decreases linearly with the number of generations. Then, execute a polynomial mutation operator to apply a small perturbation to the decision variables of the offspring individuals. During the mutation process, implement a safety boundary projection mechanism. If the decision variables of the new individuals generated by the mutation exceed the range of the safety compliance threshold matrix, immediately project the variable value to the nearest safety boundary point, or directly discard the individual and regenerate it to ensure that all offspring individuals entering the next generation strictly meet the physical safety requirements. Step S56: Repeat steps S52 to S55 until the preset maximum number of iterations is reached or the Pareto front hypervolume change rate is below the convergence threshold for several consecutive generations; output the final Pareto optimal solution set; select the optimal individual from the solution set according to the current priority instructions of the park: if the priority instruction is low carbon priority, select the individual with the lowest carbon emission intensity; if the priority instruction is economic priority, select the individual with the lowest electricity cost; if the priority instruction is grid friendly priority, select the individual with the best peak-valley difference smoothing; if there is no specific priority instruction, calculate the comprehensive score using the dot product of the dynamic weight vector and the normalized value of each target, and select the individual with the highest score as the optimal carbon reduction scheduling scheme. This scheme includes the detailed control instruction sequence of each device in the future preset period; Step S57: The optimal carbon reduction scheduling scheme is parsed into specific control commands and sent to the scheduling control units of each target management unit via industrial Ethernet to drive the photovoltaic inverter, energy storage converter, charging pile and load controller to operate. During the execution, the scheduling control unit monitors the actual operating data in real time. If the actual deviation is found to exceed the allowable tolerance, a fast response mechanism is triggered to make a secondary correction within the allowable range of the safety compliance threshold matrix. The corrected state is then uploaded to the cloud optimization engine as the initial state for the next round of rolling optimization, forming a closed-loop control.

[0032] In this embodiment, by removing security constraints from the penalty term of the optimization objective function and transforming them into hard boundaries of the search space, a dynamic weight adjustment mechanism based on real-time data and rule logic is used to solve the adaptive decision-making problem under multi-objective conflicts. This achieves multi-dimensional and dynamic optimization of carbon reduction benefits for the photovoltaic-storage-charging system in the warehouse park while strictly ensuring physical security throughout all time periods and links. It avoids the solution space compression and local optimum traps caused by the coupling of security constraints, eliminates the waste of carbon reduction potential caused by reserving excessive conservative margins for safety, and can respond in real time to the grid dispatching needs and changes in the energy consumption characteristics of the park, outputting the optimal dispatching strategy that takes into account low carbon, economy, equipment lifespan and grid stability, ensuring that the photovoltaic-storage-charging system in the warehouse park always operates at the optimal state point within the safe and feasible domain under complex operating conditions.

[0033] Based on the same inventive concept as the aforementioned embodiment of the intelligent flexible carbon reduction method for photovoltaic, energy storage, and charging in a warehouse park, this invention also provides an intelligent flexible carbon reduction system for photovoltaic, energy storage, and charging in a warehouse park, such as... Figure 2 As shown, the system includes: The unit division module is used to divide target management units that are bound one-to-one with the power distribution links, based on the topology of the power distribution network and the spatial distribution of energy-consuming equipment in the warehouse park. The data governance module is used to synchronously collect multi-source spatiotemporal operational data from the target control unit across the entire chain, and obtain a standardized time-series feature set through data cleaning and multi-scale feature encoding. The benchmark calibration module is used to calibrate the physical safety compliance threshold matrix and the multi-objective benchmark set of carbon reduction benefits of the target control unit based on a standardized time-series feature set. The security verification module is used to generate an initial collaborative scheduling scheme with the physical security compliance threshold matrix as a hard constraint. After full-time compliance verification and iterative correction, a safe and compliant scheduling scheme is obtained. The optimization execution module is used to perform multi-objective optimization within the feasible domain of the safety and compliance scheduling scheme, using the multi-objective benchmark set of carbon reduction benefits as the optimization objective, outputting the optimal carbon reduction scheduling scheme and issuing it for execution.

[0034] The system described above in this invention can effectively realize a smart and flexible carbon reduction method for photovoltaic, energy storage and charging in warehousing parks. The technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0035] Although this application has been described in conjunction with specific features and embodiments, it is apparent that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and accompanying drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application is intended to include such modifications and modifications.

Claims

1. A smart, flexible carbon reduction method for photovoltaic, energy storage, and charging systems in warehousing parks, characterized in that... include: Based on the topology of the power distribution network in the warehousing park and the spatial distribution of energy-consuming equipment, target control units are divided and bound to the power distribution links one by one. Synchronously collect multi-source spatiotemporal operation data of the target control unit across the entire link, and obtain a standardized time series feature set through data cleaning and multi-scale feature encoding; Based on the standardized time-series feature set, the physical safety compliance threshold matrix and the multi-objective benchmark set of carbon reduction benefits of the target control unit are calibrated; An initial collaborative scheduling scheme is generated using the physical safety compliance threshold matrix as a hard constraint. After full-time compliance verification and iterative correction, a safe and compliant scheduling scheme is obtained. This includes: based on power system power flow calculation, performing full-time operation simulation of the initial collaborative scheduling scheme for a preset period of time; comparing the simulation data with the physical safety compliance threshold matrix for each time period; if all data for the entire time period are within the threshold range, the verification is passed; otherwise, the out-of-threshold items are located, a safety iterative correction factor is generated, the scheme is corrected in a targeted manner, and then re-verified until a safe and compliant scheduling scheme is obtained. Within the feasible domain of the safety and compliance scheduling scheme, the multi-objective optimization is completed using the multi-objective benchmark set of carbon reduction benefits as the optimization objective, and the optimal carbon reduction scheduling scheme is output and issued for execution.

2. The intelligent flexible carbon reduction method for photovoltaic, energy storage, and charging in warehousing parks according to claim 1, characterized in that, The target control unit, which is bound to each distribution link, includes: Taking the power distribution room of the warehouse park as a node, and combining the physical route of the power distribution circuit, it is divided into several independent power supply unit grids, with each grid corresponding to a unique power distribution circuit. Grid cells that meet the preset grid screening conditions are identified as target control units; the preset grid screening conditions are that they simultaneously contain at least two types of photovoltaic, energy storage and charging equipment, contain core energy loads for storage, and are equipped with carbon emission and power quality monitoring devices.

3. The intelligent flexible carbon reduction method for photovoltaic, energy storage, and charging in warehousing parks according to claim 2, characterized in that, The synchronous acquisition of multi-source spatiotemporal operational data from the target control unit across the entire chain, followed by data cleaning and multi-scale feature encoding, yields a standardized time-series feature set, including: At a preset collection cycle, multi-source spatiotemporal operation data of the target control unit from the grid side, photovoltaic side, energy storage side, charging pile side, load side and carbon emission side are collected synchronously. All data are marked with a unified timestamp and a unique code of the target control unit. Data cleaning was completed by removing outliers using the improved 3σ criterion and filling in missing data using the sliding window interpolation method. The standardized temporal feature set was then obtained through multi-scale feature embedding encoding.

4. The intelligent flexible carbon reduction method for photovoltaic, energy storage, and charging in warehousing parks according to claim 1, characterized in that, The dynamic calibration method for the physical security compliance threshold matrix is ​​as follows: Based on multi-source spatiotemporal operational data, combined with power grid dispatching procedures, equipment operation specifications, and park power supply guarantee requirements, hard constraint thresholds are calibrated for the power grid side, equipment side, and load side respectively, and a two-dimensional physical safety compliance threshold matrix is ​​constructed for each time period.

5. The intelligent flexible carbon reduction method for photovoltaic-storage-charging systems in warehousing parks according to claim 1, characterized in that, The dynamic calibration method for the multi-objective benchmark set of carbon reduction benefits is as follows: Acquire real-time carbon emission factors of target management units, green electricity ratio of power grid in different time periods, peak and valley electricity prices, energy storage life decay curves and carbon emission quotas for different time periods; With the goal of minimizing carbon emission intensity, and taking into account electricity costs, energy storage life loss and grid peak-valley difference smoothing, a multi-objective optimization function is constructed, and a multi-objective benchmark set of carbon reduction benefits is obtained through calibration.

6. The intelligent flexible carbon reduction method for photovoltaic, energy storage, and charging in warehousing parks according to claim 1, characterized in that, The multi-objective optimization adopts an improved non-dominated sorting genetic algorithm. In the optimization process, carbon emission intensity, electricity cost, energy storage life loss and grid peak-valley difference smoothing are taken as optimization objectives, and corresponding dynamic weights are configured for each. The dynamic weights are adjusted in real time based on the energy consumption characteristics of the park during different time periods and the requirements of the power grid, and the optimal carbon reduction scheduling scheme is output from the Pareto optimal solution set.

7. The intelligent flexible carbon reduction method for photovoltaic, energy storage, and charging in warehousing parks according to claim 1, characterized in that, In the multi-objective optimization process, a safe boundary projection mechanism is performed on the generated offspring individuals: If the decision variables of a child individual exceed the range defined by the physical security compliance threshold matrix, the variable value is projected to the nearest security boundary point, or the child individual is discarded and regenerated to ensure that all individuals entering the next generation meet the physical security requirements.

8. The intelligent flexible carbon reduction method for photovoltaic, energy storage, and charging in warehousing parks according to claim 1, characterized in that, After the optimal carbon reduction scheduling scheme is issued and executed, a closed-loop correction is performed: The actual operating data of the target control unit is monitored in real time. If the actual deviation exceeds the allowable tolerance, a second correction is made within the allowable range of the physical safety compliance threshold matrix, and the corrected state is used as the initial state for the next round of rolling optimization.

9. A smart flexible carbon reduction system for photovoltaic, energy storage, and charging in a warehouse park, wherein the system is applied to the smart flexible carbon reduction method for photovoltaic, energy storage, and charging in a warehouse park as described in claim 1, characterized in that... include: The unit division module is used to divide target management units that are bound one-to-one with the power distribution links, based on the topology of the power distribution network and the spatial distribution of energy-consuming equipment in the warehouse park. The data governance module is used to synchronously collect multi-source spatiotemporal operational data from the target control unit across the entire chain, and obtain a standardized time-series feature set through data cleaning and multi-scale feature encoding. The benchmark calibration module is used to calibrate the physical safety compliance threshold matrix and the multi-objective benchmark set of carbon reduction benefits of the target control unit based on the standardized time-series feature set. The security verification module is used to generate an initial collaborative scheduling scheme with the physical security compliance threshold matrix as a hard constraint, and obtain a security compliant scheduling scheme through full-time compliance verification and iterative correction. The optimization execution module is used to perform multi-objective optimization within the feasible domain of the safety and compliance scheduling scheme, using the multi-objective benchmark set of carbon reduction benefits as the optimization objective, outputting the optimal carbon reduction scheduling scheme and issuing it for execution.

Images