An adaptive scheduling method and system for an autonomous power supply system based on data feedback

Abstract

The application discloses a kind of based on data feedback's self-powered system adaptive scheduling method and system, belong to self-powered scheduling technical field;The method is for the specific scene that offshore photovoltaic power station and energy storage cooperate for underwater robot power supply, collects the operation data of light storage and the data of grid interconnection state;Power supply scene data set is constructed, and mapping is to operating state description space;Underwater robot power supply stability feature and energy demand elasticity feature are constructed;Establish supply-demand matching state evaluation model, the comprehensive index of power supply stability feature and the comprehensive index of energy demand elasticity feature are two-dimensional coordinate mapping;Adaptive determination scheduling strategy;After acquisition, power supply quality parameter, energy utilization parameter and load operation parameter are executed;Generation scheduling log, update historical data sample library, improve the power supply stability of self-powered system, energy utilization efficiency and load satisfaction rate, solve the problem that scheduling adaptability is insufficient, lack of self-optimization capability.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, specifically to an adaptive scheduling method and system for an autonomous power supply system based on data feedback. Background Technology

[0002] With the rapid development of new energy technologies and distributed power supply networks, autonomous power supply systems, with their advantages of flexible energy supply, strong environmental adaptability, and low dependence on the main power grid, have been widely applied in scenarios such as power supply in remote areas, emergency power supply, and distributed energy networking. In underwater robot operation scenarios, to ensure their continuous and stable operation, a power supply scheme combining shore-based photovoltaic power stations and energy storage systems is often adopted. This enables local energy production, storage, and dispatch, effectively supporting the long-term operation needs of underwater robots. However, this underwater robot operation scenario faces multiple complex scheduling problems: photovoltaic power generation is significantly affected by weather and sunlight, resulting in intermittent and fluctuating output; the underwater robot's workload has temporal differences, with concentrated charging demand at night; photovoltaic power stations require regular maintenance, during which they need to rely on energy storage systems for independent power supply; and there are time-based differences in electricity prices when connected to the grid. Energy storage charging and discharging can ensure uninterrupted energy supply, meeting the charging needs of underwater robots for multiple days of operation through off-grid energy storage and photovoltaic systems when photovoltaic power stations are shut down for maintenance. Existing scheduling methods are mostly based on fixed rules or static models, which are difficult to adapt to multi-dimensional dynamic scenarios such as photovoltaic fluctuations, load time-varying, maintenance off-grid, and electricity price changes, resulting in problems such as insufficient power supply continuity, insufficient utilization of energy storage, and lack of interruption mechanisms.

[0003] Existing research primarily employs fixed-threshold scheduling methods, rule-based static allocation strategies, or traditional model predictive control methods. While these methods are effective in handling single steady-state scenarios, they suffer from several shortcomings when dealing with continuous power supply tasks for special objects such as underwater robots. Firstly, most scheduling strategies do not adequately consider the energy elasticity and operational priority of the load, making it difficult to ensure power supply to critical loads during periods of insufficient photovoltaic output or system maintenance. Secondly, they lack refined design of grid interaction strategies, failing to effectively utilize market mechanisms such as time-of-use pricing to achieve "energy continuity" for energy storage, thus undermining the system's economic potential. Thirdly, existing methods often rely on fixed models and parameters, limiting their adaptability to dynamic factors such as photovoltaic fluctuations, energy storage degradation, and changes in load demand. Especially during planned power outages and maintenance of photovoltaic power plants, it is difficult to maintain the multi-day continuous charging needs of underwater robots through off-grid operation, resulting in insufficient self-optimization and fault tolerance. Therefore, achieving coordinated optimization of energy supply stability, load power supply continuity, and system economy under complex operating environments has become a critical issue urgently needing resolution in this field. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive scheduling method and system for autonomous power supply systems based on data feedback, so as to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: An adaptive scheduling method for an autonomous power supply system based on data feedback is proposed. This method includes the following steps: Step S1: Synchronously collect operational data and grid interconnection status data of offshore photovoltaic (PV) power generation, energy storage, and underwater robots according to a preset sampling period; construct a continuously updatable power supply scenario dataset for offshore PV and energy storage supplying power to underwater robots, and map it to a unified operational status description space; Step S2: Based on the power supply scenario dataset for offshore PV and energy storage supplying power to underwater robots, analyze the PV power generation fluctuation characteristic parameters, adjustable margin, and grid-side available support capacity coefficient to construct the stability characteristics of underwater robot power supply; Step S3: Based on the power supply scenario dataset for offshore PV and energy storage supplying power to underwater robots... The process involves: extracting the priority attributes of underwater robot power supply; obtaining parameters of underwater robot charging power fluctuation characteristics, upper limit of delay time, and upper limit of power reduction ratio to construct energy demand elasticity characteristics; Step S4: establishing a supply and demand matching state evaluation model, mapping the comprehensive index of underwater robot power supply stability characteristics with the comprehensive index of energy demand elasticity characteristics using two-dimensional coordinates; adaptively determining scheduling strategies for different matching states; Step S5: tracking the execution effect of scheduling decisions in real time through monitoring modules deployed in each unit, collecting power quality parameters, energy utilization parameters, and load operation parameters after execution; generating scheduling logs and feeding them back to the power supply scenario dataset along with abnormal early warning information, and updating the historical data sample library.

[0006] As a preferred embodiment of the adaptive scheduling method for an autonomous power supply system based on data feedback described in this invention, the operating data of photovoltaic-storage-load and grid interconnection status data are synchronously collected according to a preset sampling period. The photovoltaic-storage-load includes offshore photovoltaic units, energy storage units, and underwater robot units. The operating data of the offshore photovoltaic units includes output power, photoelectric conversion efficiency, and illuminance response coefficient; the operating data of the energy storage units includes current state of charge, upper limit of charging and discharging power, charging and discharging efficiency, and remaining cycle life; the operating data of the underwater robot units includes real-time power demand, voltage stability threshold, and continuous operating time requirement; and the grid interconnection status data includes grid-side voltage frequency, access point power limit, and interconnection line transmission loss parameters.

[0007] As a preferred embodiment of the adaptive scheduling method for an autonomous power supply system based on data feedback described in this invention, a continuously updatable power supply scenario dataset for supplying power to underwater robots from offshore photovoltaic units, energy storage units, and underwater robot units is constructed based on the operational data of these units and the grid interconnection status data. The details are as follows: A three-segment unified coding rule of unit type identification, data dimension encoding, and timestamp sequence is adopted to identify multi-source data and establish a mapping relationship between data and system operation status. System operation status includes energy supply status, load operation status, and interconnection status; energy supply status includes surplus status, balanced status, and shortage status; load operation status includes full load status, partial load status, and standby status; interconnection status includes grid-connected status, off-grid status, and transition status. The power supply scenario dataset is mapped to a unified operation status description space. The operation status description space adopts a standardized data matrix form, with the matrix row dimension corresponding to the optical storage load type and the column dimension corresponding to the data acquisition dimension.

[0008] As a preferred embodiment of the adaptive scheduling method for an autonomous power supply system based on data feedback described in this invention, based on a power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots, historical photovoltaic power generation time series data is extracted, and the power fluctuation variance is analyzed using the sliding window method to obtain photovoltaic power generation fluctuation characteristic parameters. The adjustable margin of energy storage is calculated by the difference between the current state of charge of the energy storage unit and the upper limit of the charging and discharging power. The actual usable value of the adjustable margin is then corrected by combining the energy storage charging and discharging efficiency decay coefficient. Collect voltage and frequency fluctuation data and access point power limitation parameters from the power grid side, and calculate the available support capacity coefficient of the power grid side using a weighted summation method; The stability characteristics of the underwater robot's power supply are constructed using a three-dimensional vector form. ,in, The stability coefficient of photovoltaic power generation, This is the energy storage regulation capacity coefficient. The power grid support coefficient is used as the weight, and the weighted average is used to obtain the comprehensive supply stability index.

[0009] As a preferred embodiment of the adaptive scheduling method for an autonomous power supply system based on data feedback described in this invention, the priority attributes of the underwater robot power supply are extracted based on a power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots, including critical loads, important loads, and ordinary loads; the power change behavior of the underwater robot power supply is analyzed, and the charging power fluctuation characteristic parameters of the underwater robot are obtained by calculating the standard deviation of the power change rate. Based on the operational constraints of the underwater robot, identify the upper limit of the delay time of the delayable load and the upper limit of the power reduction ratio of the load that can be reduced; Construct energy demand elasticity characteristics, wherein the energy demand elasticity characteristics are in three-dimensional vector form. ,in, This is a load priority weighting factor. For power fluctuation sensitivity coefficient, As an adjustable potential coefficient, the weights are determined by the entropy weight method, and the weighted average is used to obtain the comprehensive demand elasticity index.

[0010] As a preferred embodiment of the adaptive scheduling method for an autonomous power supply system based on data feedback described in this invention, a supply-demand matching state evaluation model is established. This model maps the comprehensive index of the underwater robot's power supply stability characteristics with the comprehensive index of its energy demand elasticity characteristics using two-dimensional coordinates, classifying supply-demand matching states as: sufficient supply – high demand elasticity, sufficient supply – low demand elasticity, insufficient supply – high demand elasticity, and insufficient supply – low demand elasticity. A scheduling strategy is then adaptively determined for each matching state. Ample supply and high demand elasticity: The power distribution strategy adopts the approach of maximizing local consumption, shallow charging of energy storage, and grid surplus power. Power balance aims to supply the load to the full capacity. The grid-connected mode is maintained when the power is not connected to the grid, and the grid-connected power is dynamically adjusted. Ample supply and low demand elasticity: The power distribution strategy adopts full load supply + full charging of energy storage + limiting grid power. Power balancing aims to stabilize the load voltage and frequency. It maintains grid-connected mode when not in grid-connected state and solidifies core interface parameters. Supply shortage - high demand elasticity: The power distribution strategy adopts the key load priority supply + ordinary load peak shaving / delay + energy storage discharge to supplement energy. The power balance aims to minimize the supply and demand gap. The tendency to be connected to or off the grid is dynamically adjusted according to the grid support capacity. If the grid support is sufficient, the grid connection is maintained; otherwise, the off-grid mode is switched. Insufficient supply—low demand elasticity: Adopt a power distribution strategy of full supply to critical loads + power limitation for important loads + deep discharge of energy storage + emergency grid connection application. Power balance aims to ensure the operation of core loads, activate the off-grid and on-grid rapid switching mechanism, and optimize interface parameters to achieve a smooth transition.

[0011] As a preferred embodiment of the adaptive scheduling method for an autonomous power supply system based on data feedback described in this invention, the monitoring modules deployed in each unit track the execution effect of scheduling decisions in real time and collect power quality parameters, energy utilization parameters and load operation parameters after execution. Generate a scheduling log, which includes scheduling time, matching status determination result, scheduling strategy details, and execution effect indicators; set an abnormal warning threshold. When the power supply quality parameter exceeds the preset quality threshold and the energy utilization parameter is lower than the preset utilization parameter threshold, an abnormal warning is triggered and the warning type and triggering reason are recorded. The scheduling logs and anomaly warning information are fed back to the power supply scenario dataset to update the historical data sample library; based on the new samples, the incremental learning algorithm is used to correct the calculation model parameters of the underwater robot's power supply stability characteristics and energy demand elasticity characteristics; at the same time, the regional division threshold of the supply and demand matching status assessment model is adjusted according to the anomaly warning information.

[0012] An adaptive scheduling system for autonomous power supply based on data feedback, comprising: a data acquisition module, a feature extraction module, a scheduling decision module, a feedback update module, and a storage module; The data acquisition module: synchronously collects the operation data of photovoltaic storage and load and the grid interconnection status data according to a preset sampling period; constructs a continuously updatable power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots, and maps it to a unified operation status description space; The feature extraction module: Based on the power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, it analyzes the fluctuation characteristic parameters of photovoltaic power generation, adjustable margin and grid-side available support capacity coefficient, and constructs the power supply stability characteristics of underwater robots. The scheduling decision module extracts the priority attributes of underwater robot power supply based on the power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots; obtains underwater robot charging power fluctuation characteristic parameters, delay time upper limit and power reduction ratio upper limit, and constructs energy demand elasticity characteristics; The feedback update module establishes a supply and demand matching status assessment model, maps the comprehensive index of the underwater robot's power supply stability characteristics with the comprehensive index of its energy demand elasticity characteristics using two-dimensional coordinates, and adaptively determines the scheduling strategy for different matching states. The storage module: tracks the execution effect of scheduling decisions in real time through the monitoring modules deployed in each unit, collects power quality parameters, energy utilization parameters and load operation parameters after execution; generates scheduling logs and feeds them back to the power supply scenario dataset along with abnormal early warning information, and updates the historical data sample library.

[0013] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: This invention provides an adaptive scheduling method and system for autonomous power supply systems based on data feedback. By synchronously collecting multi-source data from photovoltaic, energy storage, load, and grid interconnection at preset cycles, and constructing a power supply scenario dataset using unified coding rules and standardized processing, it achieves a normalized expression of multi-dimensional and heterogeneous data, providing a high-quality data foundation for subsequent feature extraction and scheduling decisions. Subsequently, by extracting the stability characteristics of underwater robot power supply and the elasticity characteristics of energy demand in a hierarchical manner, it characterizes the system's supply capacity and adjustable demand space, respectively, achieving accurate quantification of the core characteristics on both the supply and demand sides, providing a key basis for supply and demand matching analysis. Furthermore, based on two-dimensional coordinate mapping, it divides four supply and demand matching states, and adaptively formulates differentiated scheduling strategies for different states, realizing a smooth transition between on-grid and off-grid states and dynamic adaptation of interface parameters, effectively improving the scheduling flexibility and adaptability in complex scenarios. Finally, by continuously tracking the scheduling execution effect, generating logs and early warning information and feeding them back to the dataset, and using incremental learning algorithms to correct model parameters, a closed-loop self-optimization mechanism is constructed, enabling the system to continuously optimize the scheduling strategy according to changes in operating status. Overall, this invention not only improves the power supply stability, energy utilization efficiency, and load fulfillment rate of the autonomous power supply system, but also has good environmental adaptability and scalability. It effectively solves the problems of insufficient scheduling adaptability and lack of self-optimization capability in existing technologies under dynamic and complex scenarios, and has important engineering application value. Attached Figure Description

[0014] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0015] Figure 1 This is a schematic diagram illustrating the steps of an adaptive scheduling method for an autonomous power supply system based on data feedback, according to the present invention. Figure 2 This is a schematic diagram of the structure of an adaptive scheduling system for an autonomous power supply system based on data feedback, according to the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 In this first embodiment: an adaptive scheduling method for an autonomous power supply system based on data feedback is provided, which includes the following steps: Step S1: Synchronously collect operational data and grid interconnection status data of marine photovoltaic, energy storage and underwater robots according to the preset sampling period; construct a continuously updatable power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, and map it to a unified operational status description space.

[0018] Specifically, the operation data of photovoltaic storage and energy storage units and grid interconnection status data are collected synchronously according to a preset sampling period. The photovoltaic storage and energy storage units include offshore photovoltaic units, energy storage units, and underwater robot units. The operation data of offshore photovoltaic units include output power, photoelectric conversion efficiency, and light intensity response coefficient; the operation data of energy storage units include current state of charge, upper limit of charging and discharging power, charging and discharging efficiency, and remaining cycle life; the operation data of underwater robot units includes real-time power demand, voltage stability threshold, and continuous operation time requirements; and the grid interconnection status data includes grid-side voltage frequency, access point power limit, and interconnection line transmission loss parameters.

[0019] Furthermore, based on the operational data of offshore photovoltaic units, energy storage units, and underwater robot units, as well as the grid interconnection status data, a continuously updatable dataset of power supply scenarios for offshore photovoltaic and energy storage to power underwater robots is constructed, as follows: A three-segment unified coding rule of unit type identification, data dimension encoding, and timestamp sequence is adopted to identify multi-source data and establish a mapping relationship between data and system operation status. System operation status includes energy supply status, load operation status, and interconnection status; energy supply status includes surplus status, balanced status, and shortage status; load operation status includes full load status, partial load status, and standby status; interconnection status includes grid-connected status, off-grid status, and transition status. The power supply scenario dataset is mapped to a unified operation status description space. The operation status description space adopts a standardized data matrix form, with the matrix row dimension corresponding to the optical storage load type and the column dimension corresponding to the data acquisition dimension.

[0020] In this invention, the operating status of the autonomous power supply system is jointly determined by multiple components, including photovoltaics, energy storage, load, and grid interconnection. The data of each component has temporal correlation (e.g., fluctuations in photovoltaic output directly affect the charging and discharging status of energy storage). Therefore, it is necessary to collect data synchronously according to a preset sampling period (e.g., 1 minute / time) to avoid misjudgment of status due to data time sequence misalignment. The multi-source data (photovoltaic output power, energy storage state of charge, etc.) is heterogeneous (different dimensions and units). A one-to-one mapping between data and system operating status (supply / load / interconnection status) is established through a three-segment encoding of "unit type identifier (e.g., photovoltaic = 01, energy storage = 02) + data dimension encoding (e.g., output power = 001, state of charge = 002) + timestamp sequence (e.g., 20240520100000)", realizing the normalization identification of heterogeneous data. A unified description space is constructed in matrix form (row = unit type, column = data collection dimension). Essentially, this is to eliminate the dimensional differences of different unit data through data standardization processing (e.g., normalization, unit unification), providing a unified computational basis for subsequent cross-unit feature extraction.

[0021] The collected core data, such as photovoltaic output power, energy storage charging and discharging efficiency, and real-time load demand, are the original basis for characterizing the system's "supply capacity" and "demand scale." For example, photovoltaic photoelectric conversion efficiency data directly reflects the energy supply potential, and the load voltage stability threshold determines the bottom line of power supply quality. By classifying energy supply (surplus / balance / shortage), load operation (full load / partial load / standby), and interconnection status (grid-connected / off-grid / transition), the system can accurately classify operating scenarios, such as differentiated scenarios like "grid-connected status + supply surplus" and "off-grid status + supply shortage," providing scenario labels for subsequent targeted scheduling.

[0022] Step S2: Based on the power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, analyze the photovoltaic power generation fluctuation characteristic parameters, adjustable margin and grid-side available support capacity coefficient, and construct the power supply stability characteristics of underwater robots.

[0023] Specifically, based on the power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots, historical photovoltaic power generation time series data is extracted, and the power fluctuation variance is analyzed using the sliding window method to obtain photovoltaic power generation fluctuation characteristic parameters; The adjustable margin of energy storage is calculated by the difference between the current state of charge of the energy storage unit and the upper limit of the charging and discharging power. The actual usable value of the adjustable margin is then corrected by combining the energy storage charging and discharging efficiency decay coefficient. Collect voltage and frequency fluctuation data and access point power limitation parameters from the power grid side, and calculate the available support capacity coefficient of the power grid side using a weighted summation method; The stability characteristics of the underwater robot's power supply are constructed using a three-dimensional vector form. ,in, The stability coefficient of photovoltaic power generation, This is the energy storage regulation capacity coefficient. The power grid support coefficient is used as the weight, and the weighted average is used to obtain the comprehensive supply stability index.

[0024] In this invention, a sliding window method (e.g., window size = 10 minutes) is used to calculate the power fluctuation variance of historical time-series data. A larger variance indicates more unstable photovoltaic output (e.g., sudden power changes due to cloud cover on cloudy days). This parameter directly reflects the "natural fluctuation risk" on the supply side. The theoretical value of the adjustable margin for energy storage is "current state of charge - upper limit of charge / discharge power." However, in reality, the energy storage charge / discharge efficiency will decrease with cycle life (e.g., efficiency drops from 90% to 80% after 5 years of use). Therefore, a decay coefficient correction needs to be introduced to ensure that the calculation results closely match the actual performance of energy storage. Grid-side voltage... Frequency stability (smaller fluctuations indicate stronger support capability) and access point power limitation (larger limitations allow for greater power transmission) are core influencing factors of grid support capability. A weighted summation (e.g., voltage and frequency weighted at 0.6, power limitation weighted at 0.4) is used to comprehensively quantify the grid's support for the system. The photovoltaic power generation stability coefficient, energy storage regulation capability coefficient, and grid support coefficient have different degrees of influence on supply stability (e.g., energy storage regulation capability has a higher weight in off-grid scenarios). The Analytic Hierarchy Process (AHP) method is used to invite field experts to score the data, determine reasonable weights, and then weight the results to obtain a comprehensive index, ensuring the scientific validity of the index. Three-dimensional vector features (photovoltaic stability coefficient, energy storage regulation coefficient, and grid support coefficient) comprehensively characterize the system's supply capacity from three dimensions: "natural output stability," "artificial regulation capability," and "external support strength"—for example... Low values ​​indicate large fluctuations in photovoltaic power, requiring reliance on... (Energy storage) or (Power grid) to compensate; The higher the overall supply stability index, the stronger the system's supply capacity and the better its stability, allowing for more flexible scheduling strategies (such as grid connection of surplus power); the lower the index, the higher the supply risk, requiring priority to ensure power supply to critical loads. For example, when the overall index is below 0.3, emergency discharge of energy storage needs to be initiated.

[0025] Step S3: Based on the power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, extract the priority attribute of power supply to underwater robots; obtain the charging power fluctuation characteristic parameters, delay time upper limit and power reduction ratio upper limit of underwater robots, and construct the energy demand elasticity characteristics.

[0026] Specifically, based on the power supply scenario dataset of underwater robots powered by marine photovoltaic and energy storage, priority attributes of underwater robot power supply are extracted, including critical load, important load, and ordinary load; the power change behavior of underwater robot power supply is analyzed, and the charging power fluctuation characteristic parameters of underwater robot are obtained by calculating the standard deviation of the power change rate. Based on the operational constraints of the underwater robot, identify the upper limit of the delay time of the delayable load and the upper limit of the power reduction ratio of the load that can be reduced; Energy demand elasticity characteristics are constructed using a three-dimensional vector form. ,in, This is a load priority weighting factor. For power fluctuation sensitivity coefficient, As an adjustable potential coefficient, the weights are determined by the entropy weight method, and the weighted average is used to obtain the comprehensive demand elasticity index.

[0027] In this invention, based on the importance of loads to system operation (e.g., hospital emergency equipment is a critical load, residential lighting is a common load), priorities are assigned and weighted coefficients are given (e.g., critical loads have a weight of 1.0, important loads 0.7, and common loads 0.3) to ensure that "important loads are given priority" in scheduling. The standard deviation of the load power change rate is calculated; a larger standard deviation indicates a more sensitive load to power fluctuations (e.g., precision instruments cannot withstand sudden power changes), and this coefficient determines the smoothness of power adjustments in scheduling. Combined with load operating constraints (e.g., washing machines can delay starting by 1 hour, air conditioners can reduce power by 10%), the "delayable space" and "reduction ratio" of the load are quantified; a higher coefficient indicates greater load adjustment flexibility and a larger space for participating in supply and demand balancing. The importance of load priority, fluctuation sensitivity, and adjustability potential is determined by the information content of the data itself (e.g., in a certain scenario, most loads are adjustable, so the adjustability potential has a higher weight). The entropy weighting method determines the weights by calculating the data entropy value, avoiding subjective assumptions and ensuring the objectivity of the indicators. Three-dimensional vector features (load priority coefficient, fluctuation sensitivity coefficient, and adjustability potential coefficient) accurately characterize the "rigid demand scale" and "flexible adjustment space" on the demand side—for example... High and A low load indicates that the load is mainly critical rigid load, which needs to be fully guaranteed during scheduling; Based on characteristics, scheduling strategies for different loads can be clearly defined, such as fluctuation sensitivity coefficients. High-precision equipment requires avoiding sudden power fluctuations during scheduling; adjustable potential coefficients are necessary. High normal loads can be prioritized for peak shaving or delayed operation when supply is insufficient.

[0028] Step S4: Establish a supply and demand matching status assessment model, and perform two-dimensional coordinate mapping between the comprehensive index of the stability characteristics of the underwater robot's power supply and the comprehensive index of the elasticity characteristics of energy demand; adaptively determine the scheduling strategy for different matching states.

[0029] Specifically, a supply-demand matching status assessment model is established, which maps the comprehensive index of the underwater robot's power supply stability characteristics with the comprehensive index of its energy demand elasticity characteristics using two-dimensional coordinates to classify the supply-demand matching status, including: sufficient supply - high demand elasticity, sufficient supply - low demand elasticity, insufficient supply - high demand elasticity, and insufficient supply - low demand elasticity. An adaptive scheduling strategy is then determined for each matching status. Ample supply and high demand elasticity: The power distribution strategy adopts the approach of maximizing local consumption, shallow charging of energy storage, and grid surplus power. Power balance aims to supply the load to the full capacity. The grid-connected mode is maintained when the power is not connected to the grid, and the grid-connected power is dynamically adjusted. Ample supply and low demand elasticity: The power distribution strategy adopts full load supply + full charging of energy storage + limiting grid power. Power balancing aims to stabilize the load voltage and frequency. It maintains grid-connected mode when not in grid-connected state and solidifies core interface parameters. Supply shortage - high demand elasticity: The power distribution strategy adopts the key load priority supply + ordinary load peak shaving / delay + energy storage discharge to supplement energy. The power balance aims to minimize the supply and demand gap. The tendency to be connected to or off the grid is dynamically adjusted according to the grid support capacity. If the grid support is sufficient, the grid connection is maintained; otherwise, the off-grid mode is switched. Insufficient supply—low demand elasticity: Adopt a power distribution strategy of full supply to critical loads + power limitation for important loads + deep discharge of energy storage + emergency grid connection application. Power balance aims to ensure the operation of core loads, activate the off-grid and on-grid rapid switching mechanism, and optimize interface parameters to achieve a smooth transition.

[0030] In this invention, the core of the supply-demand contradiction differs under different matching states, requiring targeted strategy design: Ample supply and high demand elasticity: The core issue is "excess energy consumption". Therefore, the approach of "local consumption + shallow charging of energy storage + grid connection of surplus electricity" is adopted to avoid energy waste and generate revenue through grid connection. Ample supply – low demand elasticity: The core challenge is “ensuring load stability”, therefore, the approach is to adopt “full supply + full charging of energy storage + limiting grid connection” to avoid the impact of grid power fluctuations on load voltage frequency. Supply shortage - high demand elasticity: The core contradiction is to "minimize the supply-demand gap". Therefore, the supply shortage is made up by "prioritizing critical loads and adjusting ordinary loads" and using demand-side flexibility. Insufficient supply and low demand elasticity: The core contradiction is "ensuring core loads", so "deep discharge of energy storage + emergency grid connection" is initiated to ensure rigid demand to the greatest extent.

[0031] To address the "one-size-fits-all" problem of traditional fixed strategies, for example, the same system can adopt a surplus power grid connection strategy at noon (when supply is sufficient and residential demand is highly elastic) and a core load guarantee strategy at night (when supply is insufficient and industrial critical load demand is low elastic), thereby improving adaptability to different scenarios. The strategy clearly defines the rules for adjusting the grid connection and off-grid status (e.g., maintaining grid connection when supply is insufficient and grid support is sufficient, otherwise switching to off-grid), and optimizes interface parameters (e.g., voltage and frequency adjustment range) to avoid power surges or power outages during the switching process. For example, when operating off-grid in remote areas, if supply is insufficient, the system can quickly switch to grid connection mode to obtain grid support.

[0032] Step S5: Track the execution effect of scheduling decisions in real time through monitoring modules deployed in each unit, collect power quality parameters, energy utilization parameters and load operation parameters after execution; generate scheduling logs and feed them back to the power supply scenario dataset along with abnormal early warning information, and update the historical data sample library.

[0033] Specifically, the monitoring modules deployed in each unit track the execution effect of scheduling decisions in real time and collect power quality parameters, energy utilization parameters and load operation parameters after execution. Generate a scheduling log, which includes scheduling time, matching status determination result, scheduling strategy details, and execution effect indicators; set an abnormal warning threshold. When the power supply quality parameter exceeds the preset quality threshold and the energy utilization parameter is lower than the preset utilization parameter threshold, an abnormal warning is triggered and the warning type and triggering reason are recorded. The scheduling logs and anomaly warning information are fed back to the power supply scenario dataset to update the historical data sample library; based on the new samples, the incremental learning algorithm is used to correct the calculation model parameters of the underwater robot's power supply stability characteristics and energy demand elasticity characteristics; at the same time, the regional division threshold of the supply and demand matching status assessment model is adjusted according to the anomaly warning information.

[0034] In this invention, monitoring modules deployed in each unit (such as photovoltaic output monitoring and load voltage monitoring) collect real-time data on power supply quality (voltage frequency, harmonic content), energy utilization (energy storage charging and discharging efficiency, photovoltaic absorption rate), and load operation (continuous running time, power compliance rate) to comprehensively evaluate the actual effect of the scheduling strategy. A dual threshold (power supply quality threshold + energy utilization threshold) is set, triggering an early warning only when "power supply quality is substandard and energy utilization is inefficient" (e.g., voltage fluctuations exceeding ±5% and photovoltaic absorption rate below 60%), avoiding false warnings caused by a single abnormal indicator and ensuring the accuracy of the warnings. New scheduling logs and early warning information are added as samples, and incremental learning algorithms (such as incremental SVM and online gradient descent) are used to correct the calculation model parameters of supply stability characteristics and demand elasticity characteristics, avoiding feature calculation deviations caused by system aging (e.g., energy storage degradation) and environmental changes (e.g., changes in sunlight patterns). The regional division threshold for supply and demand matching status is adjusted based on abnormal early warning information (e.g., if "insufficient supply but no early warning is triggered" occurs multiple times, the threshold for sufficient supply is lowered), making the matching status division more consistent with the actual system operation.

[0035] By monitoring data to quantify scheduling effectiveness—for example, power quality parameters reflecting whether load power requirements are met, and energy utilization parameters reflecting energy waste—data is provided for strategy optimization. For instance, if a strategy is found to have a low photovoltaic absorption rate, the proportion of surplus power fed into the grid in subsequent scheduling needs adjustment. This addresses the "lack of adaptability" in traditional scheduling methods by incrementally learning to correct model parameters and adjust matching thresholds, enabling the system to "learn from operation." For example, after the energy storage cycle life decays, the model automatically corrects the calculation method of the energy storage regulation capacity coefficient, ensuring the strategy remains adapted to the actual system performance. Anomaly warnings are promptly fed back to the dataset. For example, when grid voltage and frequency fluctuations are abnormal, subsequent scheduling will reduce the weight of the grid support coefficient and increase the priority of energy storage regulation to avoid power outages due to grid faults, thus improving system reliability.

[0036] Please see Figure 2 In this second embodiment: an adaptive scheduling system for autonomous power supply system based on data feedback is provided. The system includes: a data acquisition module, a feature extraction module, a scheduling decision module, a feedback update module, and a storage module. The data acquisition module: synchronously collects the operation data of photovoltaic storage and load and the grid interconnection status data according to a preset sampling period; constructs a continuously updatable power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots, and maps it to a unified operation status description space; The feature extraction module: Based on the power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, it analyzes the fluctuation characteristic parameters of photovoltaic power generation, adjustable margin and grid-side available support capacity coefficient, and constructs the power supply stability characteristics of underwater robots. The scheduling decision module extracts the priority attributes of underwater robot power supply based on the power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots; obtains underwater robot charging power fluctuation characteristic parameters, delay time upper limit and power reduction ratio upper limit, and constructs energy demand elasticity characteristics; The feedback update module establishes a supply and demand matching status assessment model, maps the comprehensive index of the underwater robot's power supply stability characteristics with the comprehensive index of its energy demand elasticity characteristics using two-dimensional coordinates, and adaptively determines the scheduling strategy for different matching states. The storage module: tracks the execution effect of scheduling decisions in real time through the monitoring modules deployed in each unit, collects power quality parameters, energy utilization parameters and load operation parameters after execution; generates scheduling logs and feeds them back to the power supply scenario dataset along with abnormal early warning information, and updates the historical data sample library.

[0037] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0038] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive scheduling method for an autonomous power supply system based on data feedback, characterized in that, The method includes the following steps: Step S1: Synchronously collect operational data and grid interconnection status data of marine photovoltaic, energy storage and underwater robots according to the preset sampling period; construct a continuously updatable power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, and map it to a unified operational status description space; Step S2: Based on the power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, analyze the photovoltaic power generation fluctuation characteristic parameters, adjustable margin and grid-side available support capacity coefficient, and construct the power supply stability characteristics of underwater robots; Step S3: Based on the power supply scenario dataset of underwater robots powered by marine photovoltaic and energy storage, extract the priority attribute of underwater robot power supply; obtain the underwater robot charging power fluctuation characteristic parameters, delay time upper limit and power reduction ratio upper limit, and construct energy demand elasticity characteristics; Step S4: Establish a supply and demand matching status assessment model, and perform two-dimensional coordinate mapping between the comprehensive index of the stability characteristics of the underwater robot's power supply and the comprehensive index of the elasticity characteristics of energy demand; adaptively determine the scheduling strategy for different matching states. Step S5: Track the execution effect of scheduling decisions in real time through monitoring modules deployed in each unit, collect power quality parameters, energy utilization parameters and load operation parameters after execution; generate scheduling logs and feed them back to the power supply scenario dataset along with abnormal early warning information, and update the historical data sample library.

2. The adaptive scheduling method for an autonomous power supply system based on data feedback according to claim 1, characterized in that, The specific implementation process of step S1 includes: The operation data of photovoltaic storage and energy storage units and grid interconnection status data are collected synchronously according to a preset sampling period. The photovoltaic storage and energy storage units include offshore photovoltaic units, energy storage units, and underwater robot units. The operation data of offshore photovoltaic units include output power, photoelectric conversion efficiency, and light intensity response coefficient; the operation data of energy storage units include current state of charge, upper limit of charging and discharging power, charging and discharging efficiency, and remaining cycle life; the operation data of underwater robot units includes real-time power demand, voltage stability threshold, and continuous operation time requirement; the grid interconnection status data includes grid-side voltage frequency, access point power limit, and interconnection line transmission loss parameters.

3. The adaptive scheduling method for an autonomous power supply system based on data feedback according to claim 2, characterized in that, The specific implementation process of step S1 also includes: Based on the operational data of offshore photovoltaic units, energy storage units, and underwater robot units, as well as grid interconnection status data, a continuously updatable power supply scenario dataset for offshore photovoltaic and energy storage to supply power to underwater robots is constructed, as follows: A three-segment unified coding rule of unit type identification, data dimension encoding, and timestamp sequence is adopted to identify multi-source data and establish a mapping relationship between data and system operation status. System operation status includes energy supply status, load operation status, and interconnection status; energy supply status includes surplus status, balanced status, and shortage status; load operation status includes full load status, partial load status, and standby status; interconnection status includes grid-connected status, off-grid status, and transition status. The power supply scenario dataset is mapped to a unified operation status description space. The operation status description space adopts a standardized data matrix form, with the matrix row dimension corresponding to the optical storage load type and the column dimension corresponding to the data acquisition dimension.

4. The adaptive scheduling method for an autonomous power supply system based on data feedback according to claim 3, characterized in that, The specific implementation process of step S2 includes: Based on the power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, historical photovoltaic power generation time series data are extracted, and the power fluctuation variance is analyzed using the sliding window method to obtain photovoltaic power generation fluctuation characteristic parameters. The adjustable margin of energy storage is calculated by the difference between the current state of charge of the energy storage unit and the upper limit of the charging and discharging power. The actual usable value of the adjustable margin is then corrected by combining the energy storage charging and discharging efficiency decay coefficient. Collect voltage and frequency fluctuation data and access point power limitation parameters from the power grid side, and calculate the available support capacity coefficient of the power grid side using a weighted summation method; The stability characteristics of the underwater robot's power supply are constructed using a three-dimensional vector form. ,in, The stability coefficient of photovoltaic power generation, This is the energy storage regulation capacity coefficient. The power grid support coefficient is used as the weight, and the weighted average is used to obtain the comprehensive supply stability index.

5. The adaptive scheduling method for an autonomous power supply system based on data feedback according to claim 4, characterized in that, The specific implementation process of step S3 includes: Based on a power supply scenario dataset of underwater robots powered by offshore photovoltaic and energy storage, priority attributes of underwater robot power supply are extracted, including critical loads, important loads, and ordinary loads; the power change behavior of underwater robot power supply is analyzed, and the charging power fluctuation characteristics of underwater robots are obtained by calculating the standard deviation of the power change rate. Based on the operational constraints of the underwater robot, identify the upper limit of the delay time of the delayable load and the upper limit of the power reduction ratio of the load that can be reduced; Energy demand elasticity characteristics are constructed using a three-dimensional vector form. ,in, This is a load priority weighting factor. For power fluctuation sensitivity coefficient, As an adjustable potential coefficient, the weights are determined by the entropy weight method, and the weighted average is used to obtain the comprehensive demand elasticity index.

6. The adaptive scheduling method for an autonomous power supply system based on data feedback according to claim 5, characterized in that, The specific implementation process of step S4 includes: A supply-demand matching status assessment model is established, which maps the comprehensive index of the stability of the underwater robot's power supply with the comprehensive index of the elasticity of energy demand in two dimensions to classify the supply-demand matching status, including: sufficient supply - high demand elasticity, sufficient supply - low demand elasticity, insufficient supply - high demand elasticity, and insufficient supply - low demand elasticity. The scheduling strategy is adaptively determined for different matching statuses.

7. The adaptive scheduling method for an autonomous power supply system based on data feedback according to claim 6, characterized in that, The specific implementation process of step S5 includes: The monitoring modules deployed in each unit track the execution effect of scheduling decisions in real time and collect power quality parameters, energy utilization parameters and load operation parameters after execution. Generate a scheduling log, which includes scheduling time, matching status determination result, scheduling strategy details, and execution effect indicators; set an abnormal warning threshold. When the power supply quality parameter exceeds the preset quality threshold and the energy utilization parameter is lower than the preset utilization parameter threshold, an abnormal warning is triggered and the warning type and triggering reason are recorded. The scheduling logs and anomaly warning information are fed back to the power supply scenario dataset to update the historical data sample library; based on the new samples, the incremental learning algorithm is used to correct the calculation model parameters of the underwater robot's power supply stability characteristics and energy demand elasticity characteristics; at the same time, the regional division threshold of the supply and demand matching status assessment model is adjusted according to the anomaly warning information.

8. An adaptive scheduling system for an autonomous power supply system based on data feedback, executing the adaptive scheduling method for an autonomous power supply system based on data feedback as described in any one of claims 1-7, characterized in that, The system includes: a data acquisition module, a feature extraction module, a scheduling decision module, a feedback update module, and a storage module; The data acquisition module: synchronously collects the operation data of photovoltaic storage and load and the grid interconnection status data according to a preset sampling period; constructs a continuously updatable power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots, and maps it to a unified operation status description space; The feature extraction module: Based on the power supply scenario dataset of marine photovoltaic and energy storage supplying power to underwater robots, it analyzes the fluctuation characteristic parameters of photovoltaic power generation, adjustable margin and grid-side available support capacity coefficient, and constructs the power supply stability characteristics of underwater robots. The scheduling decision module extracts the priority attributes of underwater robot power supply based on the power supply scenario dataset of offshore photovoltaic and energy storage supplying power to underwater robots; obtains underwater robot charging power fluctuation characteristic parameters, delay time upper limit and power reduction ratio upper limit, and constructs energy demand elasticity characteristics; The feedback update module establishes a supply and demand matching status assessment model, maps the comprehensive index of the underwater robot's power supply stability characteristics with the comprehensive index of its energy demand elasticity characteristics using two-dimensional coordinates, and adaptively determines the scheduling strategy for different matching states. The storage module: tracks the execution effect of scheduling decisions in real time through the monitoring modules deployed in each unit, collects power quality parameters, energy utilization parameters and load operation parameters after execution; generates scheduling logs and feeds them back to the power supply scenario dataset along with abnormal early warning information, and updates the historical data sample library.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the adaptive scheduling method for an autonomous power supply system based on data feedback as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the adaptive scheduling method for an autonomous power supply system based on data feedback as described in any one of claims 1 to 7.

Images