Energy management and adaptive power consumption control method and device for unmanned system

Abstract

This invention relates to the field of energy management technology, and in particular to a method and apparatus for energy management and adaptive power consumption control of unmanned systems. The method acquires multi-source operating data and real-time load location, extracts target energy operating data, collects dynamic power consumption fluctuation data for the next cycle based on the initial power consumption allocation scheme, and optimizes and generates an updated power consumption scheme based on the target data to achieve dynamic adaptation. Based on the current energy storage status, target data, and load priority, it calculates the extreme power supply range and determines the target power-protected load set to ensure priority power supply to critical loads. When it is determined that the system is about to run out of energy, it intelligently generates a target load-cutting sequence by comprehensively considering real-time operating conditions, priorities, and energy storage status, and executes orderly graded power outages, improving energy utilization efficiency. Under extreme operating conditions, it can scientifically plan power protection and load-cutting strategies, effectively avoiding disorderly power outages, and significantly enhancing the power supply reliability, safety, and intelligence level of unmanned energy systems.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to a method and apparatus for energy management and adaptive power consumption control of unmanned systems. Background Technology

[0002] With the rapid development of applications such as the Internet of Things, distributed microgrids, and unattended base stations, unmanned energy systems (such as independent photovoltaic energy storage systems, wind-solar hybrid power supply systems, and remote monitoring power supply systems) are playing an increasingly important role in ensuring the continuous operation of critical loads. These systems are typically deployed in areas with complex environments and difficult maintenance, and their core task is to achieve stable power supply and efficient management of multiple types of loads with limited energy input and storage capacity.

[0003] However, existing unmanned energy systems still have many shortcomings in management and power consumption control. First, traditional energy management strategies are often based on static historical data or fixed scheduling rules, lacking the ability to dynamically perceive real-time operating conditions and multi-source operational data (such as instantaneous weather changes, load surges, etc.). This makes it difficult for the system to accurately extract key target data affecting the current power quality, causing the initial power allocation scheme to lag behind in the face of sudden dynamic power fluctuations, failing to generate optimized updated schemes in a timely manner, and easily leading to energy waste or unstable power supply. In addition, the emergency handling mechanism for when the system is about to enter a state of energy depletion is not perfect. Existing technologies rarely integrate multi-dimensional factors such as real-time location, load priority, and remaining power to intelligently plan the optimal target load switching sequence. This deficiency makes the power adjustment in the final stage of the system lack foresight and orderliness, which may not only accelerate battery damage but also cause greater social value loss or equipment safety hazards due to disordered power outages.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a method and apparatus for energy management and adaptive power consumption control of unmanned systems. This invention aims to solve the technical problems of existing unmanned energy systems, which lack the ability to dynamically perceive and finely evaluate real-time operating conditions and multi-source data. As a result, they suffer from scheduling lag when facing dynamic power consumption fluctuations, and are unable to scientifically plan power supply sets and load shedding sequences based on load priorities in emergency situations where energy is depleted. This leads to unstable power supply, interruption of critical services, and value loss caused by disorderly power outages.

[0006] To achieve the above objectives, the present invention provides an energy management and adaptive power consumption control method for unmanned systems, the method comprising: Acquire multi-source operation data of the unmanned energy system and the real-time operating condition location of the load end, and extract data from the multi-source operation data based on the real-time operating condition location to obtain the target energy operation data; The initial power consumption allocation scheme of the unmanned energy system is obtained. Based on the initial power consumption allocation scheme and the real-time operating condition location, dynamic power consumption fluctuation data in the next control cycle is collected. Based on the dynamic power consumption fluctuation data and the target energy operation data, the initial power consumption allocation scheme is re-optimized to obtain an updated power consumption scheme. Based on the updated power consumption scheme, the current energy storage status and all load priority data of the unmanned energy system are obtained. Based on the current energy storage status and the target energy operation data, the limit power supply range of the unmanned energy system is obtained. Based on the limit power supply range and the load priority data, the target power supply load set is determined. Based on the current energy storage status, it is determined whether the unmanned energy system is about to enter an energy depletion state. If it is determined that the unmanned energy system is about to enter an energy depletion state, the target load switching sequence is determined based on the real-time operating location, the load priority data, and the current energy storage status.

[0007] Optionally, the step of acquiring multi-source operating data of the unmanned energy system and the real-time operating condition location of the load end, and extracting data from the multi-source operating data based on the real-time operating condition location to obtain the target energy operating data, specifically includes: Acquire multi-source operation data of the unmanned energy system and the real-time operating condition location of the load end, and divide the multi-source operation data into time periods according to the real-time operating condition location to obtain time-segmented energy operation data; Based on the operating scenario of the unmanned energy system, the influencing factors that affect the power consumption of the unmanned energy system are extracted. Based on the aforementioned influencing factors, the time-segmented energy operation data is filtered to obtain the target energy operation data.

[0008] Optionally, the step of collecting dynamic power consumption fluctuation data within the next control cycle based on the initial power consumption allocation scheme and the real-time operating condition location specifically includes: Based on the initial power consumption allocation scheme and the real-time operating condition location, the power position of the unmanned energy system in the current control cycle is determined; Based on the power location, the subsequent power change trend of the unmanned energy system is determined; Based on the subsequent power change trend, collect the base load power consumption data, dynamic fluctuation power consumption data, renewable energy input data and energy storage loss data in the next control cycle; Based on the basic load power consumption data and the dynamic fluctuation power consumption data, the power consumption demand distribution in the next control cycle is obtained, and the initial power consumption fluctuation data is generated based on the power consumption demand distribution. Based on the renewable energy input data and the energy storage loss data, an auxiliary power supply influencing factor table is generated, and the auxiliary power supply influencing factor table is added to the initial power consumption fluctuation data to generate dynamic power consumption fluctuation data.

[0009] Optionally, the step of re-optimizing the initial power allocation scheme based on the dynamic power consumption fluctuation data and the target energy operation data to obtain an updated power consumption scheme specifically includes: Based on the dynamic power consumption fluctuation data, the power distribution data in the next control cycle is obtained, and a power consumption demand prediction model is constructed based on the power distribution data. The energy storage parameters and power supply constraint data of the unmanned energy system are obtained, and the limit discharge power, limit voltage regulation accuracy, limit power supply duration and limit load carrying capacity of the unmanned energy system are obtained based on the energy storage parameters and the power supply constraint data. The constraint conditions of the unmanned energy system are generated based on the ultimate discharge power, the ultimate voltage regulation accuracy, the ultimate power supply duration, and the ultimate load carrying capacity. Based on the aforementioned constrained operating conditions, power consumption data is matched in the power consumption demand prediction model to generate a safe power supply range. Extract the power distribution characteristics of the safe power supply range, generate an optimized power consumption allocation scheme based on the power distribution characteristics, and re-optimize the initial power consumption allocation scheme based on the optimized power consumption allocation scheme to obtain an updated power consumption scheme.

[0010] Optionally, the step of obtaining the limit power supply range of the unmanned energy system based on the current energy storage state and the target energy operation data, and determining the target power supply load set based on the limit power supply range and the load priority data, specifically includes: Based on the current energy storage state, the power supply safety value of the unmanned energy system is obtained, and it is determined whether the power supply safety value is lower than a preset safety threshold. If it is determined that the power supply safety value is lower than the safety threshold, then it is determined that the unmanned energy system needs to perform load reduction and power preservation operations. Based on the current energy storage status, the limit power supply range of the unmanned energy system is extracted, and data is extracted from all load data based on the limit power supply range to obtain the candidate power supply range; The system iterates through the candidate power supply range to determine if there are any power supply loads that meet the priority requirements. If it is determined that there are power supply loads that meet the requirements, the target power supply set of the power supply loads is extracted. Extract the power demand data of the target power supply set, generate a load reduction control scheme for the unmanned energy system based on the power demand data and the current energy storage status, and control the unmanned energy system to continuously supply power to the target power supply load according to the load reduction control scheme.

[0011] Optionally, after determining that there are qualified power-protected loads, the step of extracting the target power-protected load set further includes: If it is determined that there is no power supply load that meets the priority requirements, then the power supply requirements, minimum power supply power, and minimum power supply reliability requirements of the unmanned energy system are obtained, and the power supply requirements, minimum power supply power, and minimum power supply reliability requirements are integrated to generate minimum power supply conditions. Based on the minimum power supply guarantee conditions, all load data, and the candidate power supply guarantee range, load queries are performed to obtain multiple divisible non-core load sets; power supply guarantee condition data and total power demand data for each divisible non-core load set are extracted; based on the current energy storage status, the urgency value of the power supply guarantee required by the unmanned energy system is obtained. Based on the power supply condition data, the total power demand data, and the urgency level value, multiple sets of divisible non-core loads are filtered to obtain a target divisible power supply set; Based on the target segmentable power supply set and the current energy storage state of the unmanned energy system, a backup load reduction control scheme for the unmanned energy system is generated, and the unmanned energy system is controlled to provide tiered power supply to the target segmentable power supply set according to the backup load reduction control scheme.

[0012] Optionally, the step of determining the target load-switching sequence based on the real-time operating location, the load priority data, and the current energy storage state specifically includes: Based on the current energy storage status, the energy consumption trend of the unmanned energy system is obtained, and the initial load shedding sequence is obtained based on the energy consumption trend; The power supply architecture data of the unmanned energy system is obtained based on the operating parameters of the unmanned energy system. Based on the power supply architecture data and the energy storage parameters, the load shedding constraints of the unmanned energy system before power failure are obtained. By combining the real-time operating location, the target energy operation data, and the load shedding constraints, the initial load shedding sequence is corrected to obtain the final load shedding direction; Based on the final load shedding direction and the load shedding constraints, the load shedding execution route of the unmanned energy system is generated, and the target load shedding sequence is determined by combining the load shedding execution route and the load priority data.

[0013] Optionally, after the step of determining the target load-switching sequence by combining the load-switching execution route and the load priority data, the method further includes: Based on the current energy storage status, extract the last remaining energy data that the unmanned energy system can provide before a complete power outage; Based on the load shedding constraints and the final remaining energy data, the final load shedding adjustment amount that the unmanned energy system can make before the power outage is obtained. Based on the final load shedding adjustment amount and the target load shedding sequence, an optional load shedding range is generated, and the total power requirement of the target load shedding sequence is extracted; The optional load shedding range is divided according to the total power demand, multiple alternative load shedding schemes are generated, and the social value loss data of each alternative load shedding scheme is obtained. The alternative load shedding scheme with the lowest social value loss data is selected as the final load shedding scheme, and the execution parameter data of the final load shedding scheme is extracted. Based on the execution parameter data, the current energy storage state, and the load shedding constraints, a final load shedding control scheme for the unmanned energy system is generated; the unmanned energy system adjusts its power consumption according to the final load shedding control scheme.

[0014] Furthermore, to achieve the above objectives, the present invention also provides an unmanned system energy management and adaptive power consumption control device, the device comprising: The operating condition extraction module is used to acquire multi-source operating data of the unmanned energy system and the real-time operating condition location of the load end, and to extract data from the multi-source operating data based on the real-time operating condition location to obtain the target energy operating data. The scheme optimization module is used to obtain the initial power consumption allocation scheme of the unmanned energy system, collect dynamic power consumption fluctuation data in the next control cycle based on the initial power consumption allocation scheme and the real-time operating position, and re-optimize the initial power consumption allocation scheme based on the dynamic power consumption fluctuation data and the target energy operation data to obtain an updated power consumption scheme. The power supply determination module is used to obtain the current energy storage status and all load priority data of the unmanned energy system based on the updated power consumption scheme, obtain the limit power supply range of the unmanned energy system based on the current energy storage status and the target energy operation data, and determine the target power supply load set based on the limit power supply range and the load priority data. The load shedding decision module is used to determine whether the unmanned energy system is about to enter an energy depletion state based on the current energy storage state. If it is determined that the unmanned energy system is about to enter an energy depletion state, the target load shedding sequence is determined based on the real-time operating location, the load priority data, and the current energy storage state.

[0015] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing an unmanned system energy management and adaptive power consumption control program, wherein when the unmanned system energy management and adaptive power consumption control program is executed by a processor, it implements the steps of the unmanned system energy management and adaptive power consumption control method as described above.

[0016] This invention provides an energy management and adaptive power consumption control method for unmanned systems. This method combines multi-source operational data with real-time load location to accurately extract target energy operation data, enhancing the system's dynamic perception of the operating environment. Based on dynamic power consumption fluctuation data and target energy operation data, the initial power allocation scheme is optimized in real time, improving the response speed and adaptability of power consumption control and effectively increasing energy utilization efficiency. By integrating current energy storage status, target energy operation data, and load priorities, the extreme power supply range is scientifically defined, and a target power-protection load set is determined, ensuring that critical loads receive priority protection when energy is limited. When the system is about to enter an energy depletion state, a target load-cutting sequence is intelligently generated by comprehensively considering real-time operating conditions, load priorities, and energy storage status, achieving orderly and hierarchical emergency power outage control. This significantly reduces the risk of unnecessary power outages and social value loss, and overall improves the power supply reliability, operational safety, and intelligent management level of unmanned energy systems. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an embodiment of the unmanned system energy management and adaptive power consumption control method of the present invention; Figure 2 This is a structural block diagram of an embodiment of the unmanned system energy management and adaptive power consumption control device of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] Reference Figure 1 , Figure 1This is a flowchart illustrating an embodiment of the unmanned system energy management and adaptive power consumption control method of the present invention, which presents an embodiment of the unmanned system energy management and adaptive power consumption control method of the present invention.

[0021] In one embodiment, the unmanned system energy management and adaptive power consumption control method includes: Step S100: Obtain multi-source operation data of the unmanned energy system and the real-time operating location of the load end. Based on the real-time operating location, extract data from the multi-source operation data to obtain the target energy operation data.

[0022] The unmanned energy system can be a distributed energy supply system deployed in an unattended environment with autonomous operation capabilities. It can provide a continuous and stable power supply to critical loads and is suitable for areas with difficult maintenance or complex environments. In this embodiment, the unmanned energy system may include, but is not limited to, independent photovoltaic energy storage systems, wind-solar hybrid power supply systems, and remotely monitored power supply systems. Multi-source operational data can be a set of real-time operating parameters from multiple heterogeneous data sources, reflecting the dynamic state of the system's internal and external systems. It can be used to provide input for energy management decisions, supporting dynamic perception and optimized control. Furthermore, multi-source operational data can collect environmental and equipment operation information through sensor networks, communication interfaces, and meteorological service interfaces. For example, multi-source operational data may include instantaneous meteorological data, load current and voltage data, and energy storage SOC / SOH status data.

[0023] The real-time operating location of the load can be its current deployment location in the physical space or logical topology, along with its operating status identifier. This can be used as spatiotemporal context information for targeted filtering and correlation analysis of multi-source operating data. In an exemplary embodiment, the real-time operating location of the load can be obtained through a geolocation module, a device ID mapping table, or a network topology discovery protocol. Furthermore, the real-time operating location of the load can collaborate with multi-source operating data to achieve spatial alignment and semantic focus. Target energy operating data can be a subset of energy states directly related to the current control task, extracted from multi-source operating data guided by the real-time operating location. This can be used to improve data relevance, reduce noise interference, and support accurate power consumption decisions. In a specific embodiment, target energy operating data can be obtained by filtering, aggregating, or weighted fusion of multi-source data using location tags.

[0024] Step S200: Obtain the initial power consumption allocation scheme of the unmanned energy system; collect dynamic power consumption fluctuation data in the next control cycle based on the initial power consumption allocation scheme and the real-time operating condition location; re-optimize the initial power consumption allocation scheme based on the dynamic power consumption fluctuation data and the target energy operation data to obtain an updated power consumption scheme.

[0025] The initial power allocation scheme can be a power allocation plan for each load set by the system before the start of the current control cycle. It can be used as a benchmark reference for dynamic optimization, for comparison and iterative updates. In a specific embodiment, the initial power allocation scheme can be generated through historical scheduling strategies, preset rules, or optimization results from the previous cycle. The next control cycle can be a preset control time interval immediately following the current time window. It can be used to define the time boundary for collecting dynamic power fluctuation data, ensuring synchronization between prediction and control. The dynamic power fluctuation data can be the actual observed load power change sequence within the next control cycle. It can be used to reflect deviations in actual load behavior and to correct the initial allocation scheme. In an exemplary embodiment, the dynamic power fluctuation data can be obtained through continuous sampling by a power monitoring unit. Furthermore, this data can be based on the initial power allocation scheme and real-time operating conditions, with smart meters sampling the actual power consumption of each load at a frequency of seconds; or edge computing nodes can be used to perform real-time FFT analysis on the load current waveform to identify abrupt events. Updating the power consumption scheme can be achieved by constructing a future power demand prediction model using dynamic power consumption fluctuation data, and iteratively solving the model in conjunction with system physical constraints to directly match the optimal power distribution characteristics that meet safety requirements, thereby generating an updated scheme. Alternatively, it can involve extracting the peak values ​​and trends in dynamic power consumption fluctuations, performing boundary checks on the initial scheme based on the system's limiting capabilities (such as maximum discharge power and voltage regulation accuracy) determined by the target energy operation data, correcting allocation items that exceed the safety range, and obtaining an updated scheme. Another approach is to comprehensively calculate the net value of dynamic load demand fluctuations and renewable energy auxiliary supply, adjust the energy storage charging and discharging strategy based on the energy time-shift characteristics in the target energy operation data, and optimize the initial allocation scheme by balancing the real-time supply and demand gap, thereby achieving rapid response to sudden load changes and improving energy utilization efficiency and power supply stability. Obtaining the initial power consumption allocation scheme of the unmanned energy system can be achieved by reading the preset power allocation plan for the current period from the scheduling controller's memory or configuration database, thus providing an optimization starting point and ensuring the continuity of regulation.

[0026] Step S300: Based on the updated power consumption scheme, obtain the current energy storage status of the unmanned energy system and all load priority data. Based on the current energy storage status and target energy operation data, obtain the limit power supply range of the unmanned energy system. Based on the limit power supply range and load priority data, determine the target power supply load set.

[0027] The current energy storage status can be the remaining energy level and related health indicators of the energy storage unit at the current moment after the power consumption allocation is updated. This can be used as an energy constraint condition to participate in the delineation of the extreme power supply range and emergency load shedding decisions. In one specific embodiment, the current energy storage status can be obtained through real-time reporting of parameters such as SOC, voltage, and temperature by the battery management system (BMS). All load priority data can be numerical priority identifiers for each load in the system after the power consumption allocation is updated, based on its business importance or security level. This can be used to guide power supply and load shedding decisions, ensuring that high-value loads are given priority power. Furthermore, all load priority data can be preset by the system configuration file or dynamically distributed through a remote management platform. The extreme power supply range can be the maximum time or power boundary that the system can maintain power supply under the constraints of the current energy storage status and target energy operation data. This can be used to quantify the upper limit of the system's power supply capacity and provide a threshold basis for selecting power supply sets. In one exemplary embodiment, the extreme power supply range can be calculated jointly by an energy conservation model and a load prediction algorithm. Furthermore, this range can be estimated based on the linear energy balance equation for the maximum power supply time; or a Monte Carlo simulation can be used to consider the power supply probability boundary under meteorological uncertainties. The target power supply load set can be a subset of loads selected within the extreme power supply range that must be continuously powered. This subset can be used to ensure the continuity of critical business operations and prevent high-value loads from being interrupted due to insufficient energy. In one specific embodiment, the target power supply load set can be sorted and filtered based on load priority data and the extreme power supply range. Furthermore, this set can employ a greedy algorithm to select the load combination with the highest energy consumption per unit priority; or an integer programming model can be introduced to solve for the optimal power supply load set, thereby maximizing the coverage of critical load protection under resource-constrained conditions.

[0028] Obtaining the current energy storage status and all load priority data of the unmanned energy system can be achieved by reading energy storage parameters from the BMS and loading the load priority list from the configuration library. This allows for the construction of dual constraints on resources and demand, supporting power supply decisions. The ultimate power supply range of the unmanned energy system can be determined based on the current energy storage status and target energy operation data. This can be done by combining remaining power capacity with future energy input forecasts to calculate the sustainable power supply duration or power ceiling. Furthermore, this operation can be achieved by integrating short-term weather forecasts and load trend models, enabling the quantification of the system's power supply capacity boundary and avoiding over-commitment of power supply. Determining the target power supply load set based on the ultimate power supply range and load priority data can be achieved by accumulating load power consumption from high to low priority until approaching the ultimate power supply range threshold. This maximizes the coverage of critical loads under resource-constrained conditions.

[0029] Step S400: Based on the current energy storage status, determine whether the unmanned energy system is about to enter an energy depletion state. If it is determined that the unmanned energy system is about to enter an energy depletion state, then determine the target load switching sequence based on the real-time operating location, load priority data, and the current energy storage status.

[0030] The energy depletion state can be defined as a state where the remaining energy of the energy storage unit is below the critical threshold required to maintain basic system operation. This state can be used to trigger an emergency load shedding mechanism to prevent disorderly power outages. In one specific embodiment, the energy depletion state can be determined by comparing the current energy storage state with a preset threshold. Furthermore, this determination can be based on a dynamic threshold, automatically adjusting the critical point according to changes in the total load, or combined with trend prediction, providing an early warning if the SOC decline slope exceeds the threshold. The target load shedding sequence can be a list of load power-off sequences determined by priority and operating conditions when the system is about to enter the energy depletion state. This can be used to achieve orderly, graded power outages, minimizing social value loss and equipment risk. In an exemplary embodiment, the target load shedding sequence can be obtained by multi-dimensional scoring and sorting by integrating real-time operating location, load priority data, and the current energy storage state. Furthermore, the sequence can be designed with a weighted scoring function: Score = α × Priority + β × LocationRisk + γ × EnergyDrainRate, where α, β, and γ are weighting factors, Priority is the load priority, LocationRisk is the real-time operating location, and EnergyDrainRate is the current energy storage status. Alternatively, a decision tree model can be constructed to output the load shedding order based on rule branches.

[0031] Based on the current energy storage status, determining whether the unmanned energy system is about to enter an energy depletion state can be achieved by comparing the current energy storage status with a preset critical threshold; if it falls below, it is considered about to be depleted. Furthermore, this determination can be aided by using a sliding window to statistically analyze the SOC change rate, thereby triggering emergency mechanisms in advance to avoid sudden power outages. If the unmanned energy system is determined to be about to enter an energy depletion state, a target load-cutting sequence is determined based on real-time operating location, load priority data, and the current energy storage status. This can be achieved by comprehensively evaluating and ranking the loads using a multi-dimensional approach, forming a power-off sequence. For example, this operation can be weighted by combining the environmental risks of the loads (e.g., outdoor equipment is more susceptible to low temperatures) with the current power consumption rate, enabling orderly and controllable tiered power outages and reducing unnecessary losses.

[0032] Taking the power supply management of unmanned base stations on plateaus as an example, the energy management and adaptive power consumption control method of the unmanned system in this embodiment can be as follows: A communication base station deployed at an altitude of 4,500 meters adopts an independent photovoltaic energy storage system. The system acquires local irradiance, wind speed, battery SOC, and location information of each communication device in real time; based on the GPS coordinates of the equipment cabinet, it extracts cloud cover prediction data of the corresponding point from the regional meteorological grid to generate target energy operation data; when a sudden sandstorm causes a sharp drop in photovoltaic output, the system collects dynamic fluctuation data of abnormal increase in RRU (radio frequency unit) power consumption, immediately optimizes the initial allocation scheme, and temporarily reduces the power of non-core monitoring cameras; at the same time, based on the fact that the remaining power can only support 6 hours, combined with the fact that the core gateway has the highest priority, the power supply range is defined and the power supply set is determined to be the gateway and the main control board; when the SOC drops to the critical value of 15%, the system generates a load-cutting sequence by comprehensively considering the location (indoor / outdoor), priority, and current power consumption of each device: first, disconnect the outdoor lighting, then disconnect the environmental sensors, and finally retain the core communication link until the maintenance personnel arrive.

[0033] In one embodiment, the steps of acquiring multi-source operational data of the unmanned energy system and the real-time operating location of the load end, and extracting data from the multi-source operational data based on the real-time operating location to obtain the target energy operation data are as follows: Acquire multi-source operation data of the unmanned energy system and the real-time operating condition location of the load end. Divide the multi-source operation data into time periods based on the real-time operating condition location to obtain time-segmented energy operation data. Based on the operating scenarios of unmanned energy systems, the factors affecting the power consumption of unmanned energy systems are extracted. Data on energy operation in different time periods is filtered based on influencing factors to obtain target energy operation data.

[0034] Time segmentation can be an operation process that divides continuous multi-source operating data into multiple time segments with semantic consistency based on the real-time operating status change points corresponding to the load end's operating condition location. This can be used to align energy data with the actual operating stage and improve the temporal relevance of subsequent analysis. In this embodiment, time segmentation can trigger the division of time window boundaries by detecting operating condition location-related events (such as equipment start / stop, geographical location switching, or communication link status changes). Furthermore, time segmentation can form an input-output relationship with time-segmented energy operating data, enabling the original data to be organized in a structured manner according to the operating status. For example, time segmentation can include, but is not limited to, starting a new time segment record and marking the start timestamp when the operating condition location identifier switches from "standby" to "running," or reconstructing time segment boundaries by reversing historical data based on location-related event logs. Time-segmented energy operating data can be a subset of multi-source energy parameters obtained after time segmentation and organized in segments according to operating status. This can be used to provide structured input for identifying influencing factors and avoid cross-state data mixing interference. In an exemplary embodiment, time-segmented energy operating data can be obtained by slicing the original multi-source operating data according to the segmentation points and labeling it with time segments.

[0035] The operational scenario can be a combined description of the deployment environment and typical operating modes of an unmanned energy system, reflecting the system's external constraints and internal behavioral characteristics. It can serve as a contextual knowledge base to guide the extraction logic of influencing factors. In one specific embodiment, the operational scenario can be preset by system configuration metadata or automatically identified through long-term operational clustering. Further, operational scenarios can include, but are not limited to, independent photovoltaic stations on plateaus, wind-solar-storage microgrids on islands, and communication relay base stations in deserts. Influencing factors can be environmental or load variables that significantly affect the current power consumption under a specific operational scenario. They can be used as criteria for data filtering, focusing on the key variables that truly drive power consumption changes. In this embodiment, influencing factors can be matched to a preset influencing factor rule base based on the operational scenario, or dynamically identified through an online feature importance assessment algorithm. For example, influencing factors can include, but are not limited to, a sudden drop in irradiance caused by cloud cover, a surge in RRU power consumption caused by sudden communication traffic, and an increase in battery internal resistance under low-temperature conditions. Data filtering can be a process of retaining highly correlated variables and removing redundant or irrelevant data from time-segmented energy operation data based on influencing factors. This can be used to generate target energy operation data with a high signal-to-noise ratio, improving the quality of subsequent decision-making. In one specific embodiment, data filtering can be achieved through rule matching, correlation coefficient threshold filtering, or machine learning feature selection.

[0036] Multi-source operational data is segmented into time periods based on real-time operating location to obtain time-segmented energy operation data. This can be achieved by using changes in the real-time operating location of the load as a trigger signal to slice the continuously collected multi-source operational data stream. Furthermore, this operation can be implemented by starting a new time period recording and marking the start timestamp when the operating location identifier switches from "standby" to "operation," or by reconstructing time period boundaries based on location-related event logs through reverse alignment of historical data. This achieves spatiotemporal alignment between data and actual operating status, avoiding feature distortion caused by time misalignment. Based on the operating scenario of the unmanned energy system, factors affecting the power consumption of the unmanned energy system are extracted. This can be done by querying a rule base of influencing factors matching the current operating scenario, or by calling an online feature analysis module to identify key variables. Further, this operation can be achieved by using a pre-set scenario-influencing factor mapping table (e.g., "plateau photovoltaic station" → {direct irradiance, module temperature, inverter efficiency}), or by using mutual information calculation within a sliding window to dynamically evaluate the correlation between each variable and total power consumption and select Top-K as influencing factors. This transforms general multi-source data into scenario-specific key variables, enhancing the targeting of perception. By filtering time-segmented energy operation data based on influencing factors, target energy operation data can be obtained. This can be achieved by retaining only the variable dimensions belonging to the set of influencing factors in the time-segmented data, forming a simplified dataset. Furthermore, this operation can directly project the required variable columns through field name matching, or synthesize and supplement highly correlated derived variables (such as power = voltage × current) that are not explicitly listed. This allows for the output of highly correlated, low-redundancy target energy operation data, providing accurate input for optimization and emergency decision-making.

[0037] For example, in the scenario of emergency power supply for island microgrids, the unmanned system energy management and adaptive power consumption control method of this embodiment can be as follows: A certain island microgrid system is deployed in a typhoon-prone area, and the operating scenario is marked as "island wind-solar-storage microgrid". The system acquires multi-source operating data including wind speed, surge current, diesel generator fuel consumption, and base station load location. When the communication tower load switches from "low service" to "emergency broadcast mode" (real-time operating condition location change), time period segmentation is triggered, and data segments corresponding to the high load period are generated; according to the "island" scenario rule base, the system extracts the influencing factors as "sudden gusts", "contact resistance increase caused by high humidity and high salinity environment" and "peak power consumption of broadcast equipment"; then only these three types of variables are retained in the time period data to form target energy operation data, which is used to quickly adjust the energy storage discharge strategy and predict whether the backup generator needs to be started.

[0038] In one embodiment, the step of collecting dynamic power consumption fluctuation data within the next control cycle based on the initial power consumption allocation scheme and the real-time operating condition location specifically includes: Based on the initial power consumption allocation scheme and real-time operating conditions, determine the power position of the unmanned energy system in the current control cycle; Based on power location, determine the subsequent power change trend of the unmanned energy system; Based on subsequent power change trends, collect base load power consumption data, dynamic fluctuation power consumption data, renewable energy input data, and energy storage loss data for the next control cycle; Based on the basic load power consumption data and dynamic fluctuation power consumption data, the power consumption demand distribution in the next control cycle is obtained, and the initial power consumption fluctuation data is generated based on the power consumption demand distribution. Based on renewable energy input data and energy storage loss data, an auxiliary power supply influencing factor table is generated, and the auxiliary power supply influencing factor table is added to the initial power consumption fluctuation data to generate dynamic power consumption fluctuation data.

[0039] The power location can be a composite state identifier representing the current power operating point of the system, jointly determined by the initial power allocation scheme and the real-time operating condition location of the load. It can serve as a precise anchor point for the current operating state of the system and for predicting subsequent power evolution paths. In this embodiment, the power location can be generated by weighted mapping of the initially allocated load power values ​​to their corresponding operating conditions (such as geographical coordinates or operating modes) to create a spatial-power joint vector. Furthermore, the power location can be the result of coupling power state with physical or logical location. The subsequent power change trend can be the evolution direction and rate of the total system power or key load power in the next control cycle, derived from the current power location. This can be used to guide the forward-looking configuration of data acquisition, ensuring the capture of key dynamic variables. In an exemplary embodiment, the subsequent power change trend can be obtained by extrapolating historical and current power locations using time series models, state transition diagrams, or physical simulation models. The basic load power consumption data can be a stable and predictable load power consumption sequence necessary for maintaining the basic functions of the system in the next control cycle. It can be used to form a rigid base for power demand distribution, reflecting the unreducible energy demand. In one specific embodiment, the basic load power consumption data can be obtained through device nameplate parameters, historical steady-state operation records, or low-frequency sampling.

[0040] Dynamic fluctuation power consumption data can be instantaneous or short-term power consumption changes deviating from the base load, caused by sudden business operations, environmental disturbances, or equipment anomalies. It can be used to capture load-side uncertainties and support emergency response and resource reservation. Furthermore, dynamic fluctuation power consumption data can be obtained by high-frequency power monitoring units through targeted sampling of high-variance areas guided by trends. Renewable energy input data can be the predicted or measured power generation sequence of renewable energy sources such as photovoltaics and wind power in the next control cycle. It can be used to characterize external energy supply capacity and participate in supply and demand balance modeling. In an exemplary embodiment, renewable energy input data can be generated by fusing meteorological forecasts, irradiance / wind speed sensors, and inverter output data. Energy storage loss data can be a quantitative indicator of energy loss caused by internal resistance, aging, or temperature effects during the charging and discharging process of energy storage units. It can be used to reflect internal system efficiency bottlenecks and correct ideal energy supply assumptions. For example, energy storage loss data can be obtained in real time by the battery management system (BMS) based on parameters such as state of charge (SOC), current, and temperature through a loss model.

[0041] The power demand distribution can be a structured expression of total power demand organized by time granularity or load category within the next control cycle, which can be used to provide a complete demand-side view for generating initial power fluctuation data. In one specific embodiment, the power demand distribution can be formed by superimposing and classifying basic load power consumption data and dynamic fluctuation power consumption data after time alignment. The initial power fluctuation data can be a raw fluctuation description constructed solely based on the load-side power demand distribution, without considering the influence of the energy supply side, and can be used as a preliminary version of the dynamic power fluctuation data, pending enhancement of supply-side information. Furthermore, the initial power fluctuation data can be obtained by extracting fluctuation features (such as peak-to-valley difference, rate of change) from the power demand distribution to form a time-series vector. The auxiliary power supply influencing factor table can be a structured set of influencing factors integrating the uncertainty of renewable energy input and the characteristics of energy storage loss, which can be used to quantify the impact of non-ideal characteristics of the supply side on actual available power and to correct the load-side fluctuation model. In an exemplary embodiment, the auxiliary power supply influencing factor table can be generated by normalizing renewable energy input data and energy storage loss data, extracting features, and organizing them into key-value pairs or weight tables. For example, the auxiliary power supply influencing factor table may include, but is not limited to, a photovoltaic cloud shading fluctuation coefficient table, a discharge efficiency decay table caused by battery aging, and a temperature-internal resistance coupling loss mapping table.

[0042] Based on the initial power allocation scheme and real-time operating conditions, the power position of the unmanned energy system in the current control cycle is determined. This can be achieved by associating the initial power value allocated to each load with its corresponding real-time operating condition position (such as device ID or geographic tag) to form a system-level power state vector. Furthermore, this operation can be implemented using a location-weighted averaging method or by constructing a power-location two-dimensional heatmap, thereby achieving a coupled representation of power state and physical / logical location, improving state identification accuracy. Based on the power position, the subsequent power change trend of the unmanned energy system is determined. This can be achieved by inputting the current power position into a trend prediction model, which outputs the power evolution path for the next cycle. Further, this operation can be achieved by using a Long Short-Term Memory (LSTM) network to learn the mapping relationship between historical power position sequences and future trends, or by using a rule engine to determine the trend type under specific operating conditions, thereby enabling forward-looking prediction of the system's dynamic behavior and guiding efficient data acquisition.

[0043] Based on subsequent power change trends, the system collects baseline load power consumption data, dynamic fluctuation power consumption data, renewable energy input data, and energy storage loss data for the next control cycle. This can be achieved by adjusting sensor sampling strategies based on predicted trends, increasing the sampling frequency during critical periods or high-variability areas. Furthermore, this operation can be implemented by enabling millisecond-level meter sampling for periods predicted to see sharp power increases, minute-level sampling for stable periods, or prioritizing dynamic fluctuation data acquisition from nodes closer to high-priority loads. This allows for on-demand focus in data acquisition, improving data quality and system resource utilization efficiency. Based on the baseline load power consumption data and dynamic fluctuation power consumption data, the power demand distribution for the next control cycle is obtained. This can be achieved by aligning and overlaying the two types of data on the time axis, and organizing them in a structured manner according to load type or time window. Further, this operation can be achieved by generating a histogram with buckets every 5 minutes to statistically analyze the total power consumption and fluctuation percentage for each period, or by hierarchically summarizing by load priority to form a multi-level demand distribution matrix. This allows for the construction of a complete load-side demand profile, supporting refined power consumption modeling.

[0044] Generating initial power consumption fluctuation data based on the power consumption demand distribution can be achieved by extracting fluctuation features (such as standard deviation, maximum slope, and number of abrupt change points) from the power consumption demand distribution to form a compact representation. Furthermore, this operation can be implemented by using wavelet transform to extract multi-scale fluctuation components, or by using the difference sequence obtained from comparing the demand distribution with a steady-state benchmark as the initial fluctuation data, thereby transforming the original demand into a fluctuation signal that can be used for optimization algorithms. Generating an auxiliary power supply influencing factor table based on renewable energy input data and energy storage loss data can be achieved by performing feature engineering on both types of data to extract key factors affecting actual available power and storing them in a structured manner. Furthermore, this operation can be achieved by calculating the renewable energy prediction error bandwidth as a fluctuation coefficient and storing it in the influencing factor table, or by fitting energy storage loss data to a state-of-charge-efficiency curve and then discretizing it to form lookup table entries, thereby explicitly modeling supply-side uncertainties and providing a basis for correcting load fluctuation data.

[0045] Adding the auxiliary power supply influencing factor table to the initial power consumption fluctuation data to generate dynamic power consumption fluctuation data can be achieved by fusing the correction coefficients or offsets in the influencing factor table with the initial power consumption fluctuation data. Furthermore, this operation can be achieved by multiplying the initial fluctuation data by the corresponding renewable energy availability factor at each time point, or by superimposing the equivalent power gap caused by energy storage losses at the fluctuation peak. This generates high-fidelity dynamic power consumption fluctuation data that simultaneously incorporates load-side dynamics and power supply-side uncertainties, improving the robustness of subsequent optimizations.

[0046] For example, in the scenario of a communication relay station on the edge of a desert, the unmanned system energy management and adaptive power consumption control method of this embodiment can be as follows: An unmanned base station deployed in the Gobi Desert is currently operating in a high-temperature, low-load state. Based on the initial power consumption allocation (10W for the main control board and 5W for monitoring) and the device's GPS location (east rack), the system determines the current power position as "low-power east rack steady state." Based on this position and historical data, it predicts that the next cycle will trigger a data transmission task due to satellite transit, and the power will show an upward trend. Accordingly, the system will collect dynamic fluctuation power consumption data of the radio frequency remote unit module at high frequency in the next 15 minutes, and simultaneously obtain the measured output of the photovoltaic panel (reduced by 30% due to sandstorm) and the discharge loss of the lithium battery at 55°C (efficiency reduced to 88%). The power demand distribution is obtained by superimposing the base load (15W) and dynamic fluctuation (peak + 25W). After generating the initial power fluctuation data, the auxiliary power supply influencing factor table consisting of "dust shading coefficient 0.7" and "high temperature loss factor 0.88" is introduced. The final generated dynamic power fluctuation data shows that although the load demand reaches 40W, the actual available power is only about 31W, triggering the subsequent power protection strategy to start in advance.

[0047] In one embodiment, the step of re-optimizing the initial power allocation scheme based on dynamic power consumption fluctuation data and target energy operation data to obtain an updated power consumption scheme is as follows: Based on dynamic power consumption fluctuation data, power distribution data for the next control cycle is obtained, and a power consumption demand prediction model is constructed based on the power distribution data. Obtain the energy storage parameters and power supply constraint data of the unmanned energy system, and based on the energy storage parameters and power supply constraint data, obtain the ultimate discharge power, ultimate voltage regulation accuracy, ultimate power supply duration and ultimate load carrying capacity of the unmanned energy system. The constraint conditions of the unmanned energy system are generated based on the ultimate discharge power, ultimate voltage regulation accuracy, ultimate power supply duration, and ultimate load carrying capacity. Based on the constrained operating conditions, power consumption data is matched in the power consumption demand prediction model to generate a safe power supply range. Extract the power distribution characteristics of the safe power supply range, generate an optimized power allocation scheme based on the power distribution characteristics, and re-optimize the initial power allocation scheme based on the optimized power allocation scheme to obtain an updated power allocation scheme.

[0048] The power distribution data can be a probabilistic or deterministic distribution of the total system power demand organized by time or load dimension in the next control cycle. It can be used as the basic input for constructing a power demand prediction model, reflecting the structured characteristics of load demand. In this embodiment, the power distribution data can be obtained by statistical modeling, clustering, or time-series decomposition of dynamic power fluctuation data. For example, the power distribution data can be generated by dividing the fluctuation data into multiple time buckets and calculating the power mean and variance within each bucket to form a distribution histogram, or by using kernel density estimation to fit a continuous power probability density function. The power demand prediction model can employ a deep neural network structure, comprising a sequentially connected data input layer, multiple hidden layers for feature nonlinear mapping, and an output layer that outputs the prediction results. Its input parameters primarily cover dynamic power fluctuation data such as base load power consumption, dynamic fluctuation power consumption, renewable energy input, and energy storage losses within the next regulation cycle. The output is the power distribution data for that cycle. The model undergoes supervised training using historical operating data and actual operating results, iteratively adjusting network weight parameters through a backpropagation algorithm to minimize prediction errors. The core evaluation logic involves inputting the power distribution data output by the model into a constraint condition generated by the ultimate discharge power, ultimate voltage regulation accuracy, ultimate power supply duration, and ultimate load carrying capacity for matching calculations. Based on this, power distribution features that meet the requirements of a safe power supply range are extracted, serving as a key basis for evaluating the accuracy of the prediction model and the feasibility of generating optimization schemes.

[0049] Based on dynamic power consumption fluctuation data, power distribution data for the next control cycle can be obtained. This can be achieved by statistically summarizing, probabilistically modeling, or segmenting the dynamic power consumption fluctuation data. Furthermore, this operation can be implemented by dividing the fluctuation data into multiple time buckets and calculating the power mean and variance within each bucket to form a distribution histogram, or by using kernel density estimation to fit a continuous power probability density function. This transforms the raw fluctuation signal into a structured demand expression, facilitating modeling and prediction. A power demand prediction model can be constructed based on the power distribution data. This can be done by using the power distribution data as a training set to fit the mapping relationship from input (e.g., time, operating condition) to output (power). Further, this operation can be achieved by training a lightweight LSTM network (taking the distribution of the past 30 minutes as input and predicting the power sequence of the next 15 minutes) or by constructing a state transition model based on a Markov chain (describing the transition probabilities of different power ranges), thereby achieving high-precision, low-latency prediction of future load demand.

[0050] Energy storage parameters can be a set of technical indicators describing the current state and capacity of an energy storage unit, and can be used to calculate the system's power supply capacity boundary. In this embodiment, energy storage parameters can be reported in real time by the battery management system (BMS). For example, energy storage parameters can include, but are not limited to, remaining charge (SOC), maximum discharge current, state of battery health (SOH), and temperature. Power supply constraint data can be the operating restrictions that the system must meet in terms of electrical safety and equipment protection, and can be used to define the legal range of electrical quantities such as voltage, current, and power. Further, power supply constraint data can be preset by system design specifications or obtained through grid interface protocols. In a specific embodiment, power supply constraint data can include, but is not limited to, the allowable fluctuation range of bus voltage, the maximum current carrying capacity of the line, and the inverter output harmonic limit. Obtaining the energy storage parameters and power supply constraint data of the unmanned energy system can be achieved by reading the real-time energy storage status and preset electrical constraints from the BMS and system configuration database, respectively, thereby providing basic parameters for calculating the system's capacity boundary.

[0051] The limiting discharge power can be the maximum instantaneous power that the energy storage system can safely output under current energy storage parameters and power supply constraints. It can be used to prevent battery damage or protection tripping caused by overcurrent discharge. In this embodiment, the limiting discharge power can be calculated based on SOC, temperature, internal resistance, and BMS protection threshold. The limiting voltage regulation accuracy can be the minimum deviation range of the output voltage stability that the system can guarantee during power supply maintenance. It can be used to ensure that sensitive loads receive power quality that meets standards. Furthermore, the limiting voltage regulation accuracy can be jointly determined by the inverter control bandwidth, line impedance, and load change response capability. The limiting power supply duration can be the longest continuous time that the system can maintain basic power supply under the current energy storage state and predicted load. It can be used to provide a time window for power supply strategies and emergency load shedding. In an exemplary embodiment, the limiting power supply duration can be calculated based on the energy conservation equation and power prediction integral. The limiting load carrying capacity can be the maximum total load or combination that the system can simultaneously support without triggering protection mechanisms. It can be used to avoid equipment shutdown or fire risks due to overload. Furthermore, the limiting load carrying capacity can be evaluated by comprehensively considering line capacity, inverter rated power, and heat dissipation capacity.

[0052] Based on energy storage parameters and power supply constraint data, the ultimate discharge power, ultimate voltage regulation accuracy, ultimate power supply duration, and ultimate load carrying capacity of the unmanned energy system are obtained. This can be achieved by calling a pre-set physical model or using a lookup table method to map the parameters into four types of limit indicators. Furthermore, this operation calculates the maximum discharge power under different SOCs using the Thevenin equivalent circuit model of the battery, or derives the ultimate voltage regulation accuracy based on the line impedance matrix and voltage stability criteria, thereby quantifying the multi-dimensional operating boundaries of the system in the current state. The constraint conditions can be a set of system operating boundaries composed of ultimate discharge power, ultimate voltage regulation accuracy, ultimate power supply duration, and ultimate load carrying capacity. These can be used as hard constraints embedded in the optimization process to ensure the physical feasibility of the solution. In this embodiment, the constraint conditions can structurally integrate the four types of limit parameters into a multi-dimensional constraint vector or feasible domain description. For example, constraint conditions may include constructing a four-dimensional constraint hypercube (each dimension corresponding to a type of limit) or generating a list of constraint rules. The constraint conditions for the unmanned energy system are generated based on the limiting discharge power, limiting voltage regulation accuracy, limiting power supply duration, and limiting load carrying capacity. This can be achieved by encapsulating the four types of limiting indices into a unified constraint description structure (such as a system of inequalities or a feasible region). Furthermore, this operation can be implemented by constructing a four-dimensional constraint hypercube or generating a list of constraint rules, thereby forming standardized constraint inputs that can be used for optimization solutions.

[0053] A safe power supply interval can be a set of feasible power allocation solutions that satisfy the constraints in the power demand prediction model output. It can be used to define a safe operating area that meets load requirements without exceeding system capacity. In this embodiment, the safe power supply interval can be extracted from the prediction space through constraint satisfaction, feasible region projection, or sampling screening. Based on the constraints, power data matching is performed in the power demand prediction model to generate the safe power supply interval. This can be achieved by screening a subset of power allocation schemes that satisfy all constraints in the prediction model output space. Furthermore, this operation can be achieved by performing Monte Carlo sampling on the prediction output and retaining the sample points that satisfy the constraints to form the safe interval, or by using a projection operator to map the prediction results to the constrained feasible region. This ensures that all candidate schemes are within the physically safe range, avoiding ineffective optimization. Power distribution characteristics can be key statistical or structural characteristics of power changes over time or load within the safe power supply interval. These can be used to guide the generation of optimized power allocation schemes, focusing on key time periods and coupling relationships. In this embodiment, power distribution characteristics can be extracted using feature engineering methods (such as peak detection, spectrum analysis, and correlation calculation). For example, power distribution characteristics may include, but are not limited to, daytime power peak periods, nighttime base load stability, and the negative correlation between sudden loads and renewable energy output.

[0054] Extracting power distribution characteristics within a safe power supply zone can be achieved by performing feature engineering on the power sequence within the safe zone to identify key patterns. Further, this operation can be implemented by using wavelet transform to extract multi-scale fluctuation features or by calculating the cross-correlation matrix between the power of each load, thereby extracting guiding principles from feasible solutions and improving optimization efficiency. The optimized power allocation scheme can be a new power allocation strategy generated within the safe power supply zone based on power distribution characteristics, balancing efficiency and reliability. It can be used as an improved version of the initial scheme to generate the final updated power allocation scheme. In this embodiment, the optimized power allocation scheme can employ a multi-objective optimization algorithm to minimize energy consumption deviation or maximize the critical load guarantee rate while satisfying constraints. For example, the optimized power allocation scheme can be generated by prioritizing the reduction of low-priority loads during peak periods to smooth the total power curve, or by utilizing the negative correlation between loads for peak-shifting scheduling to reduce the peak total demand.

[0055] Generating an optimized power allocation scheme based on power distribution characteristics can be achieved by using these characteristics as the optimization objective or weighting basis, and solving for the optimal allocation within a safe power supply range. Furthermore, this operation can be implemented by prioritizing the reduction of low-priority loads during peak periods to smooth the total power curve, or by utilizing the negative correlation between loads for peak-shaving scheduling to reduce the peak total demand, thereby generating an efficient allocation strategy that balances system capacity and load requirements. The initial power allocation scheme is then re-optimized based on the optimized scheme to obtain an updated power allocation scheme. This can be achieved by replacing corresponding items in the initial scheme with power quotas in the optimized scheme, forming the final execution scheme. Further, this operation can be achieved through weighted fusion or by directly overriding conflicting items in the initial scheme while retaining non-conflicting items to reduce control disturbances, thus enabling a closed-loop iterative update from a static initial scheme to a dynamic optimized scheme.

[0056] For example, in the scenario of an independent photovoltaic energy storage base station on a plateau, the unmanned system energy management and adaptive power consumption control method of this embodiment can be as follows: A communication base station at an altitude of 5000 meters currently has a SOC of 40% and an ambient temperature of -15℃. The system generates power distribution data for the next 15 minutes based on dynamic power consumption fluctuation data (including RRU burst traffic) and constructs an LSTM power demand prediction model. Simultaneously, it obtains energy storage parameters from the BMS (maximum discharge current is limited at low temperatures) and reads power supply constraints (DC bus voltage must be maintained at 48±2V). Based on this, the following calculations are made: the maximum discharge power is 3.2kW (lower than 4.5kW at normal temperature), the maximum voltage regulation accuracy is ±1.5V, the maximum power supply duration is 2.1 hours, and the maximum load carrying capacity is 3.8kW. After these four factors constitute the constrained operating conditions, the system selects a safe power supply range from the prediction model output—a distribution combination with a total power ≤3.2kW and voltage fluctuation ≤1.5V. Feature extraction revealed that the nighttime period from 02:00 to 04:00 is a low-power period, allowing for reserved energy storage; while the RRU power consumption peaks at 09:00 when the satellite passes overhead. Based on this, an optimized power allocation scheme was generated: charging in advance during low-power periods and temporarily reducing the frame rate of the surveillance cameras during peak periods. This scheme replaces the initial allocation, forming an updated power allocation plan that is then implemented.

[0057] In one embodiment, the step of determining the limit power supply range of the unmanned energy system based on the current energy storage status and target energy operation data, and determining the target power supply load set based on the limit power supply range and load priority data, specifically includes: Based on the current energy storage status, the power supply safety value of the unmanned energy system is obtained, and it is determined whether the power supply safety value is lower than the preset safety threshold. The power supply safety value is a quantitative indicator that characterizes the system's remaining power supply capacity and safety margin, calculated based on the current energy storage state. It can be used to assess whether the system is in a power supply risk zone and serve as a basis for triggering load shedding and power preservation operations. In this embodiment, the power supply safety value can be generated by combining the current energy storage state (such as State of Charge (SOC), voltage, and temperature) with a battery health model or energy decay curve mapping. For example, the power supply safety value can be calculated using a linear weighted method, i.e., by weighting the SOC and normalized voltage; or it can be obtained using a lookup table method, indexing a pre-stored safety level table based on a combination of SOC and temperature. The preset safety threshold can be a pre-set critical point for power supply safety, used to determine whether an energy shortage state has been entered. It can provide a standardized comparison benchmark, achieving a binary judgment of the power supply safety state. In an exemplary embodiment, the preset safety threshold is statically set by the system configuration file or operation and maintenance strategy, and can also be dynamically adjusted according to the load type.

[0058] Based on the current energy storage state, the power supply safety value of the unmanned energy system is obtained. This can be achieved by inputting the current energy storage state parameters into a preset mapping function or evaluation model, and outputting a normalized safety score. Furthermore, this operation can be implemented using a linear weighting method or a lookup table method, thereby transforming the multi-dimensional energy storage state into a single comparable indicator, facilitating threshold judgment. Determining whether the power supply safety value is lower than a preset safety threshold can be achieved by comparing the power supply safety value with the safety threshold value, and outputting a Boolean judgment result. This operation enables the system to achieve early identification of power supply risks, providing triggering conditions for load reduction decisions.

[0059] If it is determined that the power supply safety value is lower than the safety threshold, then it is determined that the unmanned energy system needs to perform load reduction and power preservation operations. Load shedding and power preservation operations can be a control behavior that proactively reduces non-critical loads to ensure continuous power supply to high-priority loads under energy-constrained conditions. This can be used to extend the operating time of critical loads, prevent deep discharge of energy storage units, and maintain basic system functions. In one specific embodiment, if it is determined that the power supply safety value is below the safety threshold, then it is determined that the unmanned energy system needs to perform load shedding and power preservation operations. This can be achieved by activating a load shedding and power preservation process flag or state machine when the determination is valid, thereby initiating an emergency energy management sub-process to avoid passive power outages.

[0060] Based on the current energy storage status, the limit power supply range of the unmanned energy system is extracted, and data is extracted from all load data based on the limit power supply range to obtain the candidate power supply range; The limit power supply range reflects the system's maximum sustainable power supply capacity without triggering over-discharge protection. Its extraction is based on the current energy storage status and combined with target energy operation data (such as future photovoltaic input forecasts) to quantify the system's maximum power supply boundary. The candidate power supply range can be a set of load candidates selected from the limit power supply range that meet physical power supply capacity constraints. This can be used to narrow the power supply decision space and ensure that subsequent priority selection is conducted within the feasible domain. In an exemplary embodiment, the candidate power supply range can be obtained by filtering all load data based on total power or time dimension using the limit power supply range. Further, this process can be arranged in ascending order of load power, accumulating sequentially until approaching the limit value to form a candidate set; alternatively, it can be based on the load topology, retaining only devices directly connected to the main line and with controllable power consumption. Exemplarily, the candidate power supply range works in conjunction with the limit power supply range as its logical extension; simultaneously, it works in conjunction with priority-compliant power supply loads to constitute the screening input.

[0061] Based on the current energy storage status, the limit of power supply range of the unmanned energy system can be extracted. This can be achieved by combining the remaining power capacity with the target energy operation data to calculate the sustainable power supply boundary. This operation allows the system to quantify its maximum power supply capacity without triggering over-discharge protection.

[0062] Data is extracted from all load data based on the limit power supply range to obtain the candidate power supply range. This can be a selection of load combinations or individual loads whose total power consumption does not exceed the limit power supply range. Furthermore, this operation can be implemented through the two methods described above, thereby ensuring that subsequent power supply decisions are made within the physically feasible domain and avoiding exceeding the commitment limits.

[0063] The system iterates through the candidate power supply range to determine if there are any power supply loads that meet the priority requirements. If it is determined that there are power supply loads that meet the requirements, the target power supply set of the power supply loads is extracted. Priority-compliant loads for power protection can be individual loads within the candidate power protection range whose priority data reaches or exceeds a preset power protection threshold. These loads serve as components of the target power protection set, ensuring resources are allocated to high-value loads. The target power protection set can be a subset of loads requiring continuous power supply, determined after filtering by both priority and power supply capacity constraints. This subset clarifies the specific boundaries of load reduction control, guiding subsequent power allocation and power outage execution. A traversal query is performed within the candidate power protection range to determine if any priority-compliant loads exist. This can involve checking each candidate load's priority data against the preset power protection threshold. Furthermore, this operation can set an absolute priority threshold (e.g., ≥3 levels) as a filtering condition, or use relative sorting to select the top N% of high-priority loads as eligible loads, thus achieving value-oriented load filtering that balances feasibility and importance. If eligible loads are identified, the target power protection set for these loads is extracted. This can involve aggregating all eligible load IDs or identifiers into a set object. This operation allows the system to clearly define power protection objects, providing a target list for power calculation and control execution.

[0064] Extract the power demand data of the target power supply set, generate a load reduction control scheme for the unmanned energy system based on the power demand data and the current energy storage status, and control the unmanned energy system to continuously supply power to the target power supply load according to the load reduction control scheme.

[0065] Power demand data can be the sum of real-time or predicted power consumption of each load in the target power supply set under current operating conditions. It can be used as a key input for generating load shedding control schemes and for verifying energy storage support capabilities. In one specific embodiment, power demand data is collected by a load monitoring unit or estimated based on a historical load model. The load shedding control scheme can be a specific set of instructions generated based on matching power demand data with the current energy storage state, used to execute load shedding and power supply scheduling. It can be used to realize the transformation from decision-making to execution, ensuring that the target power supply loads receive stable power. In an exemplary embodiment, the load shedding control scheme generates switching control signals or power limits through a power balance verification and control logic engine. Furthermore, if the power demand is less than or equal to the available power, a full power supply instruction is generated; otherwise, it is truncated to a supportable level in descending order of priority; a soft start-stop sequence can also be generated to implement gradual power reduction for non-power supply loads to reduce impact.

[0066] Extracting power demand data from the target power supply set can involve summarizing the current or predicted power values ​​of each load within the set. This operation allows the system to quantify the resources required for power supply protection, facilitating matching and verification with energy storage capacity. Based on the power demand data and the current energy storage status, a load shedding control scheme for the unmanned energy system is generated. This can involve verifying whether the power demand is within the energy storage support range and generating corresponding load switching or power adjustment commands. Furthermore, this operation, implemented as described above, forms an executable and verifiable control strategy, ensuring the safe and effective operation of power supply protection. Controlling the unmanned energy system to continuously supply power to the target power supply load according to the load shedding control scheme can involve issuing a power supply protection command to the power management unit, locking the power supply path to the target load, and cutting off the remaining loads. This operation enables the system to implement power supply decisions and maintain the continuous operation of critical business processes.

[0067] Taking emergency power supply for an island microgrid as an example, the unmanned system energy management and adaptive power consumption control method in this embodiment can be as follows: After continuous rainy days, the battery SOC of a certain island wind-solar-storage microgrid drops to 25%. The system calculates the power supply safety value as 0.28, which is lower than the preset safety threshold of 0.3, triggering a load reduction power supply operation. Based on the current energy storage status and the forecast of no sunlight for the next 24 hours, the extreme power supply range is set at 8 kWh. The system selects combinations with a total power consumption ≤ 8 kWh from all 12 loads to form a candidate power supply range (including seawater desalination pumps, communication base stations, and weather stations). It is found that the communication base station (priority 5) and the weather station (priority 4) meet the power supply threshold (≥4), forming the target power supply set. Its total power demand is 1.2 kW, and the current energy storage can support it for 6.7 hours. The system generates a load reduction control scheme: cut off low-priority loads such as lighting and air conditioning, lock the power supply circuits of communication and weather equipment, and ensure the continuous operation of critical facilities until the weather clears up.

[0068] In one embodiment, after determining that there are qualified power-protected loads, the step of extracting the target power-protected load set further includes: If it is determined that there is no power supply load that meets the priority requirements, then the power supply requirements, minimum power supply power, and minimum power supply reliability requirements of the unmanned energy system are obtained, and the power supply requirements, minimum power supply power, and minimum power supply reliability requirements are integrated to generate the minimum power supply conditions. Based on the minimum power supply guarantee conditions, all load data, and the candidate power supply guarantee range, load queries are performed to obtain multiple sets of divisible non-core loads; power supply guarantee condition data and total power demand data are extracted for each set of divisible non-core loads; based on the current energy storage status, the urgency value of the unmanned energy system requiring power supply guarantee is obtained. Based on power supply condition data, total power demand data, and urgency level values, multiple sets of divisible non-core loads are filtered to obtain the target divisible power supply set; Based on the target segmentable power supply set and the current energy storage status of the unmanned energy system, a backup load reduction control scheme for the unmanned energy system is generated. According to the backup load reduction control scheme, the unmanned energy system is controlled to supply power to the target segmentable power supply set in stages.

[0069] The power supply guarantee requirement can be a pre-defined normative description of the basic functions or service types that must be maintained, serving as a semantic constraint for the minimum power supply guarantee condition and defining the core capabilities of the system that cannot be completely interrupted. In this embodiment, the power supply guarantee requirement can be imported from an operation and maintenance strategy configuration file or industry standards. The minimum power supply guarantee power can be the minimum total power threshold required to maintain basic system operation, used to quantify the energy consumption boundary of the power supply guarantee bottom line and prevent system paralysis due to excessive reduction. For example, the minimum power supply guarantee power can be derived based on the minimum operating voltage / current characteristics of the equipment or statistical analysis of historical operating data. The minimum power supply reliability requirement can be the minimum acceptable power supply continuity or availability index of the system in an emergency state, used to constrain the power supply quality of the backup plan and avoid damage to equipment from frequent start-stop cycles. In an exemplary embodiment, the minimum power supply reliability requirement can be set in the form of time percentage or maximum interruption duration.

[0070] The minimum power supply guarantee condition can be a system power supply backup standard formed by integrating power supply requirements, minimum power supply power, and minimum power supply reliability requirements. It can be used to provide an alternative power supply decision-making basis when no high-priority loads are available, preventing complete functional loss. Furthermore, the minimum power supply guarantee condition can integrate the three types of inputs into a unified judgment condition through a rule engine or logical combination. The set of shardable non-core loads can be a combination of non-critical loads identified within the candidate power supply range that has partial power supply feasibility. It can be used to overcome the binary load-switching limitation and achieve refined utilization of low-value but retainable functions. In a specific embodiment, the set of shardable non-core loads can include, but is not limited to, one or more of the following: environmental monitoring units supporting intermittent operation, communication relay modules that can operate at reduced power, and data acquisition terminals with functional tailoring capabilities. Power supply guarantee condition data can be a set of capability attributes describing whether the set of shardable non-core loads meets the minimum power supply guarantee condition. It can be used to evaluate whether the set meets the system backup operation standard. In this embodiment, the power supply guarantee condition data can be extracted from the load technical specifications and operation logs to determine its flexible power supply compatibility parameters.

[0071] Total power demand data can be the total energy consumption of a set of shardable non-core loads under a specific power supply mode, and can be used as a key resource accounting basis for matching the current energy storage status. The urgency value can be a quantitative score reflecting the urgency of the system's energy crisis, calculated based on the current energy storage status, and can be used to dynamically adjust screening strategies, allowing for lower protection levels to extend survival time in extreme cases. For example, the urgency value can be generated by weighting multiple factors such as the SOC decline rate, remaining power supply time, and temperature anomaly. The target shardable power supply set can be the optimal subset of retainable non-core loads determined after multi-dimensional screening under the current urgency level, and can be used to clarify the specific objects of standby load shedding control, achieving flexible power supply protection. In an exemplary embodiment, the target shardable power supply set can be obtained by sorting or optimizing the comprehensive power supply condition data, total power demand data, and urgency value. The standby load shedding control scheme can be an emergency dispatch instruction set supporting tiered power supply generated for the target shardable power supply set, and can be used to maintain the minimum system functionality when no high-priority loads are available, improving operational resilience under extreme conditions. Furthermore, the standby load shedding control scheme can generate differentiated power supply strategies based on the load resilience and energy storage status.

[0072] If it is determined that no loads meet the priority requirements for power supply, the power supply requirements, minimum power supply power, and minimum power supply reliability requirements of the unmanned energy system are obtained. These three fallback parameters can be read from the system configuration library. Furthermore, this operation can be triggered when the target power supply set is empty, thereby activating a flexible power supply mechanism to prevent a complete system power outage due to priority selection failure. Integrating the power supply requirements, minimum power supply power, and minimum power supply reliability requirements to generate minimum power supply conditions can be achieved by merging these three heterogeneous requirements into a unified judgment standard through logical rules. Further, this operation can simultaneously satisfy power, reliability, and functional requirements using AND logic, or by constructing a fuzzy rule base to support weakened implementation of some conditions, thus forming a structured fallback constraint to support subsequent load querying and filtering. Based on the minimum power supply conditions, all load data, and the candidate power supply range, load queries are performed to obtain multiple divisible non-core load sets. This can identify load groups within the candidate power supply range that support flexible power supply and, when combined, meet the minimum power supply conditions. Furthermore, this operation can be achieved by clustering the load into independently separable units based on power supply dependencies using a graph segmentation algorithm, or by enumerating all subsets that meet the minimum power threshold and labeling their power supply mode compatibility. This can help discover intermediate power supply possibilities that are ignored by traditional binary load shedding and expand emergency options.

[0073] Extracting power supply condition data and total power demand data for each separable non-core load set can be done by analyzing whether each candidate set meets the minimum power supply conditions and calculating the corresponding energy consumption. Furthermore, this operation can be achieved through structured analysis of load specifications and operating records, providing structured input for multi-dimensional screening and supporting quantitative comparison. Based on the current energy storage state, the urgency level of power supply required for the unmanned energy system can be obtained, which can be achieved by mapping parameters such as SOC, discharge rate, and temperature to an emergency score within the 0-1 range. Further, this operation can be calculated by combining the ratio of state of charge to its nominal value with an exponential decay term for the discharge rate, or by using a pre-trained lightweight neural network to input multi-dimensional states and output an emergency level. This allows for dynamic adjustment of the power supply tolerance, relaxing conditions to extend system survival during periods of heightened crisis.

[0074] Based on power supply condition data, total power demand data, and urgency levels, multiple sets of divisible non-core loads are screened to obtain a target divisible power supply set. This can be achieved by constructing a multi-objective scoring function, ranking the candidate sets based on the three types of data, and selecting the optimal one. Furthermore, this operation can be implemented by selecting the solution with the lowest power and best matching conditions from the non-dominated solution set through linear weighted scoring or Pareto front analysis, thereby achieving coordinated optimization of resources, constraints, and crisis levels to ensure the maximization of feasibility and value of the solution. Based on the target divisible power supply set and the current energy storage state of the unmanned energy system, a backup load shedding control scheme for the unmanned energy system is generated, which can be achieved by assigning specific power supply modes to each load within the set. Further, this operation can be achieved by issuing duty cycle commands to loads supporting PWM modulation, or by enabling low-power mode and limiting the sampling frequency for digital loads, thereby transforming abstract power supply decisions into executable tiered power supply commands. The unmanned energy system is then controlled to provide tiered power supply to the target divisible power supply set according to the backup load shedding control scheme. This can be achieved by implementing differentiated power supply strategies through the power management unit while simultaneously cutting off other loads. Furthermore, this operation can be performed by adjusting the power supply parameters of each load in real time, thereby maintaining the minimum functional operation of the system in scenarios without high-priority loads and improving survivability under extreme conditions.

[0075] Taking the emergency power supply of a polar research station in winter as an example, the energy management and adaptive power consumption control method of the unmanned system in this embodiment can be as follows: A certain Antarctic research station encounters continuous blizzards, photovoltaic failure occurs, and the battery SOC drops to 8%. The system determines that there are no loads with a priority ≥4 that can be guaranteed (only the meteorological sensor with a priority of 3 remains). The backup mechanism is then activated: the power guarantee requirements (basic data transmission must be maintained), minimum power guarantee power (300W), and minimum power supply reliability (at least 6 hours of continuous power supply per day) are read. Within the candidate power guarantee range (total power consumption ≤1.2kWh), two sets of non-core loads that can be divided are identified: set A (meteorological station + satellite terminal, supporting intermittent power supply) and set B (environmental recorder only, power can be reduced to 200W). The power guarantee condition data are extracted and both meet the requirements, with total power of 280W and 200W respectively. The current urgency level value reaches 0.92 (extremely high risk). After comprehensive evaluation, the system selects set B (lower power) and generates a backup load reduction control scheme: the environmental recorder runs continuously at 200W, and all others are powered off. Ultimately, basic data collection was maintained at -40℃ for 72 hours until the weather cleared up.

[0076] In one embodiment, the step of determining the target load-switching sequence based on real-time operating location, evening data of the load priority day, and the current energy storage status is as follows: Based on the current energy storage status, the energy consumption trend of the unmanned energy system is obtained, and the initial load shedding sequence is obtained based on the energy consumption trend; Based on the operating parameters of the unmanned energy system, the power supply architecture data of the unmanned energy system is obtained. Based on the power supply architecture data and energy storage parameters, the load shedding constraints of the unmanned energy system before power failure are obtained. By combining real-time operating conditions, target energy operation data, and load shedding constraints, the initial load shedding sequence is corrected to obtain the final load shedding direction; Based on the final load shedding direction and load shedding constraints, the load shedding execution route of the unmanned energy system is generated, and the target load shedding sequence is determined by combining the load shedding execution route and load priority data.

[0077] The energy consumption trend can be derived from the current energy storage state, showing the rate and pattern of system remaining energy decay over time. This reflects the urgency of the system before energy depletion, providing a basis for load shedding timing. In this embodiment, the energy consumption trend can be obtained by jointly fitting the historical SOC change slope, total load power consumption, and renewable energy input predictions. For example, the energy consumption trend can be fitted with nonlinear discharge characteristics using linear extrapolation or an exponential decay model, particularly suitable for scenarios such as low temperatures or aging batteries. The initial load shedding sequence can be a preliminary load power-off order generated solely based on the energy consumption trend, without considering system topology and operating conditions. This can be used as an initial benchmark for load shedding decisions and will be subsequently corrected by multi-dimensional information. In an exemplary embodiment, the initial load shedding sequence can arrange loads in reverse order of energy consumption rate, or set power-off priorities within a fixed time window. Furthermore, the initial load shedding sequence can sort loads by energy consumption per unit time, prioritizing high-energy-consuming loads in the shedding queue, or setting a time-segmentation strategy, such as shedding low-priority loads less than 1 hour before depletion, and medium-priority loads 1 to 2 hours before depletion.

[0078] Based on the current energy storage status, the energy consumption trend of the unmanned energy system can be obtained by analyzing the current State of Charge (SOC) and its rate of change, and combining this with the total load power to predict the future energy decay curve. Furthermore, this operation can be achieved by using linear extrapolation (e.g., if the SOC decreases by 5% per hour, it is expected that 20% of the SOC will be depleted in 4 hours) or by using an exponential decay model to fit the nonlinear discharge characteristics. This allows for quantifying the remaining operating time window of the system and providing a dynamic threshold for load shedding. The initial load shedding sequence is obtained based on the energy consumption trend. This can be achieved by arranging loads in reverse order of energy depletion time, or by prioritizing loads closer to their depletion time. In a specific embodiment, this operation can be achieved by sorting loads by energy consumption per unit time, prioritizing high-energy-consuming loads in the shedding queue, or by setting a time-segmentation strategy. This establishes a preliminary load shedding logic based on time urgency, avoiding a disconnect between static priorities and dynamic timing.

[0079] The operating parameters of an unmanned energy system can be a set of configuration data describing the system's electrical structure and control logic. These parameters can be used to reconstruct the power supply physical architecture and support constraint derivation. In this embodiment, the operating parameters of the unmanned energy system can be read from the system's digital twin model or device configuration files. For example, the operating parameters of the unmanned energy system may include, but are not limited to, one or more of the following: bus topology, load grouping relationships, and switch control logic tables. The power supply architecture data can be the system power supply network and hierarchical structure obtained from the operating parameters. This data can be used to reveal the electrical coupling relationships between loads and avoid cascading failures caused by load shedding. In an exemplary embodiment, the power supply architecture data can be generated by performing graph theory modeling or rule parsing on the operating parameters to produce a load-power supply correlation matrix. Furthermore, the power supply architecture data can be obtained by deriving a three-level structure tree of bus-feeder-load from a SCADA system, or by reverse deducing the power supply branch to which the load belongs based on a Modbus register mapping table.

[0080] Load shedding constraints can be electrical safety and timing logic limitations that must be met before a power outage in an unmanned energy system. They can be used to ensure that load shedding does not lead to overvoltage, circulating current, protection malfunction, or topology conflicts. In this embodiment, load shedding constraints can be derived by combining power supply architecture data and energy storage parameters (such as minimum sustaining voltage and discharge cutoff point). For example, load shedding constraints may include, but are not limited to, bus voltage stability constraints, load group synchronous disconnection constraints, and switch operation timing constraints. The power supply architecture data of the unmanned energy system is obtained based on the operating parameters of the unmanned energy system. This can be achieved by parsing the electrical data in the system configuration file and constructing a load-power supply topology diagram. Furthermore, this operation can be achieved by deriving a three-level structure tree of bus-feeder-load from the SCADA system, or by reverse deducing the power supply branch to which the load belongs based on the Modbus register mapping table, thereby restoring the physical power supply structure of the system and providing a topology basis for constraint generation.

[0081] Based on power supply architecture data and energy storage parameters, load shedding constraints before power outages in unmanned energy systems are obtained. This can be achieved by identifying key electrical limitations (such as a busbar that cannot be disconnected alone) and combining them with energy storage cutoff conditions to form a rule set. In one embodiment, this operation can be implemented by prohibiting the independent disconnection of a load if it shares an inverter output port with core equipment, or by prohibiting the sudden unloading of high-power loads when the battery voltage is close to the discharge cutoff value to prevent voltage surges. This ensures that the load shedding action is electrically feasible and safe, preventing secondary failures. The final load shedding direction can be a load shedding optimization guide obtained by integrating real-time operating conditions, target energy operation data, and load shedding constraints. This guide can be used to reflect the impact of environmental context and energy supply potential on the load shedding strategy. In this embodiment, the final load shedding direction can be obtained by reordering the initial load shedding sequence after applying spatial, energy, and topology correction factors. For example, the final load shedding direction can incorporate spatial distribution (such as outdoor loads being more susceptible to environmental influences) and energy forecasting (such as the possibility of increased sunlight in the next 10 minutes) to adjust the load shedding priority.

[0082] By combining real-time operating conditions, target energy operation data, and load shedding constraints, the initial load shedding sequence is modified to obtain the final load shedding direction. This can be achieved by incorporating spatial distribution (e.g., outdoor loads are more susceptible to environmental influences) and energy forecasting (e.g., a rebound in sunlight within the next 10 minutes) to adjust load shedding priorities. Furthermore, this operation can be implemented by prioritizing shedding loads located on a sandstorm path if the target energy operation data indicates an impending photovoltaic outage, or prioritizing the outdoor load if two loads belong to the same group but one is indoors (lower risk). This allows the load shedding strategy to possess environmental awareness and energy supply prediction capabilities, improving decision-making intelligence. The load shedding execution route can be a specific power-off action sequence and timing arrangement planned according to the final load shedding direction under load shedding constraints, which can be used to ensure that the load shedding process complies with electrical safety regulations and equipment operation logic. In this embodiment, the load shedding execution route can generate a time-sequential instruction chain by mapping the final load shedding direction to operable switch paths in the power supply architecture. For example, the load shedding execution route can generate a relay control sequence with a delay (first disconnect branch A, wait 5 seconds and then disconnect branch B), or be planned from bottom to top according to the power supply level (first disconnect the end load, then disconnect the intermediate feeder).

[0083] Based on the final load shedding direction and constraints, a load shedding execution route for the unmanned energy system is generated. This can be achieved by converting the modified load shedding direction into a specific instruction sequence that conforms to the switching operation logic and timing requirements. In one specific embodiment, this operation can be implemented by generating a relay control sequence with delay or by planning from the bottom up according to the power supply level, thereby transforming the strategy into executable actions and ensuring that the process is controllable and orderly. The target load shedding sequence is determined by combining the load shedding execution route and load priority data. This can be achieved by fine-tuning the power-off order based on the load priority data within the framework of the load shedding execution route, ensuring that high-value loads are de-energized last. Furthermore, this operation can be implemented by prioritizing the higher-priority load if multiple loads can be de-energized simultaneously in the same load shedding step, or by directly moving a low-priority load to the front of the queue when the execution route allows skipping it. This maximizes critical business continuity while ensuring electrical safety and execution feasibility.

[0084] Taking emergency load shedding of an island microgrid as an example, the unmanned system energy management and adaptive power consumption control method in this embodiment can be: the SOC of a certain island microgrid drops to 18%, and the BMS predicts that it will be exhausted within 2 hours. The system first generates an initial load shedding sequence based on the energy consumption trend: lighting → environmental sensor → video monitoring → communication gateway. Then, the operating parameters are analyzed to find that: lighting and video monitoring share AC bus A, and the communication gateway occupies DC bus B exclusively; the energy storage parameters show that the battery cutoff voltage is 44V, and the sudden unloading of a load of >1kW will cause the voltage rebound to exceed the limit. Thus, load shedding constraints are generated: (1) lighting must not be cut off without video; (2) the single load shedding power is ≤800W. Combining the real-time operating location (the video monitoring is an outdoor camera, which is affected by the typhoon) and the target energy operation data (no sunshine in the next 30 minutes), the system corrects the initial sequence and advances the outdoor video monitoring before the lighting to form the final load shedding direction. According to this, the load shedding execution route is planned: first, the environmental sensor (300W) is cut off, and after 5 seconds, the lighting and video (total 750W) are cut off simultaneously, and finally the communication gateway is retained. The load priority is superimposed (gateway is the highest), and the final target load switching sequence is determined as [environmental sensor, lighting + video, communication gateway (power supply)].

[0085] In one embodiment, after determining the target load-cutting sequence by combining the load-cutting execution route and load priority data, the method further includes: Based on the current energy storage status, extract the last remaining energy data that the unmanned energy system can provide before a complete power outage; The final remaining energy data can be the total marginal energy that the system can safely release based on the current energy storage state before a complete power outage. It can be used as the upper limit of resources for load shedding and to define the dispatchable energy boundary during emergency phases. For example, the final remaining energy data can be obtained through the ampere-hour integration method, that is, integrating the product of current and voltage over time from the current moment to the time point corresponding to the discharge cutoff voltage; or by looking up the SOC-available energy mapping table pre-stored in the BMS and correcting it with a temperature compensation factor.

[0086] Based on load shedding constraints and the final remaining energy data, the final load shedding adjustment that the unmanned energy system can make before the power outage is obtained. The final load shedding adjustment amount can be the total power or energy adjustment limit of the load shedding actions that the system can perform before power failure under load shedding constraints. It can be used to limit the adjustment range of the selectable load shedding range to ensure that all schemes are physically feasible. In an exemplary embodiment, the final load shedding adjustment amount can be converted into a specific operating limit (e.g., 1kW@72 seconds) by combining the shedding capacity supported by the remaining energy (e.g., 1.2kW·min) with the single maximum shedding power (e.g., 1kW) constraint; or it can be expressed as the upper limit of the number of loads that can be shedding (e.g., a maximum of 3 more non-critical loads can be shedding).

[0087] Based on the final load shedding adjustment and the target load shedding sequence, a selectable load shedding range is generated, and the total power requirement of the target load shedding sequence is extracted. The selectable load shedding range can be a subset of loads that can be flexibly trimmed or recombined based on the final load shedding adjustment, based on the target load shedding sequence. This can provide multiple technically feasible load shedding candidate spaces, supporting multi-scheme comparison. For example, the selectable load shedding range can fix the high-priority power-saving portion at the front end, only combining and enumerating the low-priority portions at the back end; or it can allow skipping certain items in the sequence, but the total power shedding cannot exceed the adjustment limit. The total power requirement can be the total power consumption value of all loads to be shedding in the target load shedding sequence, which can be used as a criterion for dividing the selectable load shedding range and for structurally generating alternative schemes. In an exemplary embodiment, the total power requirement can be conservatively estimated by summing peak power; or a weighted average power can be used to reflect typical operating conditions.

[0088] The available load shedding range is divided according to the total power demand, multiple alternative load shedding schemes are generated, and the social value loss data of each alternative load shedding scheme is obtained. The alternative load shedding schemes can be candidate strategies with different load shedding combinations and social impacts, divided from the available load shedding range according to total power demand. These can be used to form a decision option set for multi-objective optimization. Furthermore, alternative load shedding schemes can be generated as discretized schemes based on power tiers (e.g., 0.5kW increments) or load combination granularity. Social value loss data can be a comprehensive indicator quantifying the public service, economic, or safety losses caused by load interruption for each alternative load shedding scheme. This can be used as a core evaluation criterion for scheme selection, achieving optimization in both technical and social dimensions. In an exemplary embodiment, social value loss data may include a communication interruption loss index, public safety risk score, and estimated operation and maintenance recovery costs. Furthermore, social value loss data can be used to construct a weighted loss function and calculate a score based on load type, service object level, and business continuity requirements. For example, social value loss data can be calculated by multiplying the loss by the priority weight of each load, the interruption duration, and the service impact coefficient; or by using an expert rule engine, where if the core gateway is interrupted, the loss value is directly set to extremely high.

[0089] The alternative load shedding scheme with the lowest social value loss data is selected as the final load shedding scheme, and the execution parameter data of the final load shedding scheme is extracted. The final load shedding scheme can be the selected strategy with the lowest social value loss among all alternative load shedding schemes. It can be used as the decision-making basis for generating control instructions to ensure that the impact of power outages is minimized. In a specific embodiment, if the losses of multiple schemes are similar, the one with fewer load shedding steps can be selected first to reduce execution risk; or Pareto front screening can be introduced to balance loss and execution complexity. Execution parameter data can be a set of specific operation instruction parameters included in the final load shedding scheme. It can be used to bridge the decision-making layer and the execution layer to ensure that the control scheme can be implemented. In an exemplary embodiment, execution parameter data can include switching action timestamps, load group identifiers, operation confirmation timeout thresholds, etc. For example, execution parameter data can generate JSON format instruction packages containing a list of load IDs and corresponding delays; or output Modbus write register sequences corresponding to the control addresses of each relay.

[0090] Based on the execution parameter data, the current energy storage status, and the load shedding constraints, the final load shedding control scheme of the unmanned energy system is generated; the unmanned energy system adjusts its power consumption according to the final load shedding control scheme.

[0091] The final load shedding control scheme can be a complete set of control instructions that can be directly issued and executed, generated by integrating execution parameter data, current energy storage status, and load shedding constraints. This can be used to drive the unmanned energy system to implement orderly power outages according to the optimal strategy. In one specific embodiment, the final load shedding control scheme can add a SOC and voltage snapshot to the instruction packet header for secondary verification by the execution unit; or embed load shedding constraint check logic, automatically terminating if the SOC rises before execution. The unmanned energy system adjusts its power consumption according to the final load shedding control scheme by having the control unit drive relays or solid-state switches according to the instruction timing sequence to gradually disconnect specified loads. Furthermore, the unmanned energy system can adjust its power consumption according to the final load shedding control scheme by adopting a soft start-stop strategy (such as gradually reducing the PWM duty cycle instead of hard shutting down) or by sending back status confirmation after execution to form an execution-feedback closed loop, thereby achieving the technical effect of orderly, controllable, and traceable emergency power consumption adjustment.

[0092] Taking the extreme emergency response of a border unmanned communication station as an example, the unmanned system energy management and adaptive power consumption control method in this embodiment can be as follows: The SOC of a communication station in a high-altitude cold region drops to 12%, and the BMS predicts that it will reach the discharge cutoff voltage in 18 minutes. The system has generated a target load shedding sequence of [outdoor lighting, environmental sensors, video surveillance, and backup RRU]. Further calculation shows that the final remaining energy data is 0.9kWh. Combined with the load shedding constraints (single shedding ≤600W, minimum interval 3 seconds), the final load shedding adjustment is to shear off a total of 800W of load. The total power requirement of the target sequence is 1.1kW. Based on this, three alternative load shedding schemes are generated: Scheme A shedding all four items (high loss, including backup RRU interruption); Scheme B shedding the first three items (retaining backup RRU, medium loss); Scheme C shedding only lighting and sensors (low loss, but insufficient energy saving). Social value loss model evaluation: Scheme C loss = 30 (only lighting interruption), Scheme B = 65 (video surveillance interruption affects security), Scheme A = 120 (loss of communication redundancy). The system selects Option C as the final load shedding scheme, extracting the execution parameters: first cut off lighting (300W), then cut off sensors (200W) 3 seconds later. Combining the current SOC of 12% and the load shedding constraint (voltage rebound must not exceed 54V), the final load shedding control scheme is generated and issued. After system execution, the remaining energy extends power supply to 25 minutes, the core communication link remains intact, and social value loss is minimized.

[0093] In addition, refer to Figure 2 To achieve the above objectives, the present invention also provides an energy management and adaptive power consumption control device for unmanned systems, the device comprising: The operating condition extraction module 10 is used to acquire multi-source operating data of the unmanned energy system and the real-time operating condition location of the load end, and extract data from the multi-source operating data based on the real-time operating condition location to obtain target energy operating data. The scheme optimization module 20 is used to obtain the initial power consumption allocation scheme of the unmanned energy system, collect dynamic power consumption fluctuation data in the next control cycle based on the initial power consumption allocation scheme and the real-time operating position, and re-optimize the initial power consumption allocation scheme based on the dynamic power consumption fluctuation data and the target energy operation data to obtain an updated power consumption scheme. The power supply determination module 30 is used to obtain the current energy storage status and all load priority data of the unmanned energy system based on the updated power consumption scheme, obtain the limit power supply range of the unmanned energy system according to the current energy storage status and the target energy operation data, and determine the target power supply load set according to the limit power supply range and the load priority data. The load shedding decision module 40 is used to determine whether the unmanned energy system is about to enter an energy depletion state based on the current energy storage state. If it is determined that the unmanned energy system is about to enter an energy depletion state, the target load shedding sequence is determined based on the real-time operating location, the load priority data, and the current energy storage state.

[0094] Other embodiments or specific implementations of the unmanned system energy management and adaptive power consumption control device of the present invention can be referred to the above-described method embodiments, and will not be repeated here.

[0095] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing an unmanned system energy management and adaptive power consumption control program, wherein when the unmanned system energy management and adaptive power consumption control program is executed by a processor, it implements the steps of the unmanned system energy management and adaptive power consumption control method as described above.

[0096] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for energy management and adaptive power consumption control of an unmanned system, characterized in that, The method includes: Acquire multi-source operation data of the unmanned energy system and the real-time operating condition location of the load end, and extract data from the multi-source operation data based on the real-time operating condition location to obtain the target energy operation data; The initial power consumption allocation scheme of the unmanned energy system is obtained. Based on the initial power consumption allocation scheme and the real-time operating condition location, dynamic power consumption fluctuation data in the next control cycle is collected. Based on the dynamic power consumption fluctuation data and the target energy operation data, the initial power consumption allocation scheme is re-optimized to obtain an updated power consumption scheme. Based on the updated power consumption scheme, the current energy storage status and all load priority data of the unmanned energy system are obtained. Based on the current energy storage status and the target energy operation data, the limit power supply range of the unmanned energy system is obtained. Based on the limit power supply range and the load priority data, the target power supply load set is determined. Based on the current energy storage status, it is determined whether the unmanned energy system is about to enter an energy depletion state. If it is determined that the unmanned energy system is about to enter an energy depletion state, the target load switching sequence is determined based on the real-time operating location, the load priority data, and the current energy storage status.

2. The unmanned system energy management and adaptive power consumption control method as described in claim 1, characterized in that, The steps of acquiring multi-source operating data of the unmanned energy system and the real-time operating condition location of the load end, and extracting data from the multi-source operating data based on the real-time operating condition location to obtain the target energy operating data are as follows: Acquire multi-source operation data of the unmanned energy system and the real-time operating condition location of the load end, and divide the multi-source operation data into time periods according to the real-time operating condition location to obtain time-segmented energy operation data; Based on the operating scenario of the unmanned energy system, the influencing factors that affect the power consumption of the unmanned energy system are extracted. Based on the aforementioned influencing factors, the time-segmented energy operation data is filtered to obtain the target energy operation data.

3. The unmanned system energy management and adaptive power consumption control method as described in claim 2, characterized in that, The step of collecting dynamic power consumption fluctuation data in the next control cycle based on the initial power consumption allocation scheme and the real-time operating condition location is as follows: Based on the initial power consumption allocation scheme and the real-time operating condition location, the power position of the unmanned energy system in the current control cycle is determined; Based on the power location, the subsequent power change trend of the unmanned energy system is determined; Based on the subsequent power change trend, collect the base load power consumption data, dynamic fluctuation power consumption data, renewable energy input data and energy storage loss data in the next control cycle; Based on the basic load power consumption data and the dynamic fluctuation power consumption data, the power consumption demand distribution in the next control cycle is obtained, and the initial power consumption fluctuation data is generated based on the power consumption demand distribution. Based on the renewable energy input data and the energy storage loss data, an auxiliary power supply influencing factor table is generated, and the auxiliary power supply influencing factor table is added to the initial power consumption fluctuation data to generate dynamic power consumption fluctuation data.

4. The unmanned system energy management and adaptive power consumption control method as described in claim 3, characterized in that, The step of re-optimizing the initial power allocation scheme based on the dynamic power consumption fluctuation data and the target energy operation data to obtain an updated power consumption scheme specifically includes: Based on the dynamic power consumption fluctuation data, the power distribution data in the next control cycle is obtained, and a power consumption demand prediction model is constructed based on the power distribution data. The energy storage parameters and power supply constraint data of the unmanned energy system are obtained, and the limit discharge power, limit voltage regulation accuracy, limit power supply duration and limit load carrying capacity of the unmanned energy system are obtained based on the energy storage parameters and the power supply constraint data. The constraint conditions of the unmanned energy system are generated based on the ultimate discharge power, the ultimate voltage regulation accuracy, the ultimate power supply duration, and the ultimate load carrying capacity. Based on the aforementioned constrained operating conditions, power consumption data is matched in the power consumption demand prediction model to generate a safe power supply range. Extract the power distribution characteristics of the safe power supply range, generate an optimized power consumption allocation scheme based on the power distribution characteristics, and re-optimize the initial power consumption allocation scheme based on the optimized power consumption allocation scheme to obtain an updated power consumption scheme.

5. The unmanned system energy management and adaptive power consumption control method as described in claim 1, characterized in that, The step of obtaining the limit power supply range of the unmanned energy system based on the current energy storage status and the target energy operation data, and determining the target power supply load set based on the limit power supply range and the load priority data, specifically includes: Based on the current energy storage state, the power supply safety value of the unmanned energy system is obtained, and it is determined whether the power supply safety value is lower than a preset safety threshold. If it is determined that the power supply safety value is lower than the safety threshold, then it is determined that the unmanned energy system needs to perform load reduction and power preservation operations. Based on the current energy storage status, the limit power supply range of the unmanned energy system is extracted, and data is extracted from all load data based on the limit power supply range to obtain the candidate power supply range; The system iterates through the candidate power supply range to determine if there are any power supply loads that meet the priority requirements. If it is determined that there are power supply loads that meet the requirements, the target power supply set of the power supply loads is extracted. Extract the power demand data of the target power supply set, generate a load reduction control scheme for the unmanned energy system based on the power demand data and the current energy storage status, and control the unmanned energy system to continuously supply power to the target power supply load according to the load reduction control scheme.

6. The unmanned system energy management and adaptive power consumption control method as described in claim 5, characterized in that, After determining that a qualified power supply load exists, the step of extracting the target power supply set of the qualified power supply load further includes: If it is determined that there is no power supply load that meets the priority requirements, then the power supply requirements, minimum power supply power, and minimum power supply reliability requirements of the unmanned energy system are obtained, and the power supply requirements, minimum power supply power, and minimum power supply reliability requirements are integrated to generate minimum power supply conditions. Based on the minimum power supply guarantee conditions, all load data, and the candidate power supply guarantee range, load queries are performed to obtain multiple divisible non-core load sets; power supply guarantee condition data and total power demand data for each divisible non-core load set are extracted; based on the current energy storage status, the urgency value of the power supply guarantee required by the unmanned energy system is obtained. Based on the power supply condition data, the total power demand data, and the urgency level value, multiple sets of divisible non-core loads are filtered to obtain a target divisible power supply set; Based on the target segmentable power supply set and the current energy storage state of the unmanned energy system, a backup load reduction control scheme for the unmanned energy system is generated, and the unmanned energy system is controlled to provide tiered power supply to the target segmentable power supply set according to the backup load reduction control scheme.

7. The unmanned system energy management and adaptive power consumption control method as described in claim 4, characterized in that, The step of determining the target load switching sequence based on the real-time operating location, the load priority data, and the current energy storage state is as follows: Based on the current energy storage status, the energy consumption trend of the unmanned energy system is obtained, and the initial load shedding sequence is obtained based on the energy consumption trend; The power supply architecture data of the unmanned energy system is obtained based on the operating parameters of the unmanned energy system. Based on the power supply architecture data and the energy storage parameters, the load shedding constraints of the unmanned energy system before power failure are obtained. By combining the real-time operating location, the target energy operation data, and the load shedding constraints, the initial load shedding sequence is corrected to obtain the final load shedding direction; Based on the final load shedding direction and the load shedding constraints, the load shedding execution route of the unmanned energy system is generated, and the target load shedding sequence is determined by combining the load shedding execution route and the load priority data.

8. The unmanned system energy management and adaptive power consumption control method as described in claim 7, characterized in that, After the step of determining the target load-switching sequence by combining the load-switching execution route and the load priority data, the method further includes: Based on the current energy storage status, extract the last remaining energy data that the unmanned energy system can provide before a complete power outage; Based on the load shedding constraints and the final remaining energy data, the final load shedding adjustment amount that the unmanned energy system can make before the power outage is obtained. Based on the final load shedding adjustment amount and the target load shedding sequence, an optional load shedding range is generated, and the total power requirement of the target load shedding sequence is extracted; The optional load shedding range is divided according to the total power demand, multiple alternative load shedding schemes are generated, and the social value loss data of each alternative load shedding scheme is obtained. The alternative load shedding scheme with the lowest social value loss data is selected as the final load shedding scheme, and the execution parameter data of the final load shedding scheme is extracted. Based on the execution parameter data, the current energy storage state, and the load shedding constraints, a final load shedding control scheme for the unmanned energy system is generated; the unmanned energy system adjusts its power consumption according to the final load shedding control scheme.

9. An energy management and adaptive power consumption control device for an unmanned system, characterized in that, The device includes: The operating condition extraction module is used to acquire multi-source operating data of the unmanned energy system and the real-time operating condition location of the load end, and to extract data from the multi-source operating data based on the real-time operating condition location to obtain the target energy operating data. The scheme optimization module is used to obtain the initial power consumption allocation scheme of the unmanned energy system, collect dynamic power consumption fluctuation data in the next control cycle based on the initial power consumption allocation scheme and the real-time operating position, and re-optimize the initial power consumption allocation scheme based on the dynamic power consumption fluctuation data and the target energy operation data to obtain an updated power consumption scheme. The power supply determination module is used to obtain the current energy storage status and all load priority data of the unmanned energy system based on the updated power consumption scheme, obtain the limit power supply range of the unmanned energy system based on the current energy storage status and the target energy operation data, and determine the target power supply load set based on the limit power supply range and the load priority data. The load shedding decision module is used to determine whether the unmanned energy system is about to enter an energy depletion state based on the current energy storage state. If it is determined that the unmanned energy system is about to enter an energy depletion state, the target load shedding sequence is determined based on the real-time operating location, the load priority data, and the current energy storage state.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an unmanned system energy management and adaptive power consumption control program, which, when executed by a processor, implements the steps of the unmanned system energy management and adaptive power consumption control method as described in any one of claims 1 to 8.

Images