Intelligent management system of distributed energy storage equipment

Abstract

The invention discloses an intelligent management system for distributed energy storage equipment, belongs to the technical field of photovoltaic energy storage equipment management, and is designed for solving the problems of poor sampling adaptation, undifferentiated operation of functions and rigid resource scheduling in a remote power-grid-free scene. The system comprises a perceptual adaptation module, a function grading and scene judging module and a time sequence resource scheduling module. The perceptual adaptation module predicts the illumination and load change rate through a fitting model and optimizes the sampling frequency in combination with a wave combining variance and a dynamic power consumption threshold value. The function grading and scene judging module divides function priorities based on multi-dimensional indexes, and judges an operation scene by linking illumination-load fluctuation and energy storage electric quantity; the time sequence resource scheduling module dynamically adjusts an operation period and a time slot ratio according to a priority and a scene, and realizes accurate resource allocation through redundancy correction; according to the invention, collaborative adaptation of sampling, functions and scheduling is realized, energy supply-load pre-judgment precision is improved, resource waste is avoided, stable operation of core functions is guaranteed, and complex scene requirements are met.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic energy storage equipment management technology, specifically relating to an intelligent management system for distributed energy storage equipment. Background Technology

[0002] The application of photovoltaic energy storage in remote, grid-free scenarios is becoming increasingly widespread, but existing technologies lack adaptability: the sampling frequency is difficult to adjust with fluctuations in environmental load, functions operate indiscriminately, and resource scheduling timing is fixed; therefore, this invention urgently needs to solve the following technical problems: In remote, grid-free photovoltaic energy storage scenarios, the sampling frequency of the sensing end cannot be adaptively adjusted according to the dynamic changes in light and load, and the power consumption control threshold is statically set, resulting in low accuracy in predicting the power supply and load status. The various operational functions of edge nodes lack a scientific priority classification standard, and all of them operate without distinction. Furthermore, they do not differentiate between scenarios based on the actual operating status of the devices, which can easily lead to a contradiction between insufficient core functional resources and waste of non-core functional resources. The resource scheduling sequence of energy storage devices is fixed and cannot be flexibly adapted to the dynamic changes in functional importance and equipment operation scenarios, resulting in low resource allocation efficiency and difficulty in meeting the stable operation requirements of core functions in different scenarios. To address this, we propose an intelligent management system for distributed energy storage devices. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent management system for distributed energy storage devices to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent management system for distributed energy storage devices, comprising: The sensing and adaptation module collects the core operating parameters of the energy storage device at the basic acquisition frequency, builds a fitting model to predict the rate of change of light intensity and the rate of change of load power in the period to be predicted, analyzes the variance of the combined wave, matches the sampling frequency adjustment range, determines the optimal sampling frequency, and adjusts the optimal sampling frequency in combination with the remaining energy storage capacity and the temperature correction power consumption threshold. Functional grading and scenario determination module: Analyzes the functional priority evaluation values ​​of all running functions of the edge sensing terminal according to the evaluation cycle, matches the corresponding functional priority level, analyzes the comprehensive fluctuation value of illumination-load in the evaluation cycle, and determines the scheduling scenario type of the corresponding cycle. The timing resource scheduling module obtains the optimal sampling frequency and scheduling scenario type for the current evaluation period, matches the basic cycle adjustment coefficient corresponding to the function priority level and adjusts the running cycle; calculates the total number of time slots, dynamically adjusts the proportion of time slots of each priority according to the scenario, splits and arranges the running time slots of the orchestration function, and generates a distribution plan to be sent to the edge perception end scheduling execution unit.

[0005] Preferably, the specific process for predicting the rate of change of light intensity and the rate of change of load power in the period to be predicted is as follows: The core operating parameters of the energy storage device are collected at the basic acquisition frequency. These parameters include the light intensity on the surface of the photovoltaic module, the real-time power on the load side, the remaining power of the energy storage unit, and the ambient temperature inside the energy storage chamber. Set the sampling frequency update period, and for each sampling moment, obtain the current light intensity value, load power value and time interval between the current and previous sampling moments. Subtract the corresponding parameters at the two moments and divide by the time interval to obtain the corresponding light intensity change rate and load power change rate. At each acquisition moment, the optimal operating temperature of the energy storage unit cell and the ambient temperature inside the energy storage chamber are simultaneously obtained, and the temperature correction coefficient is calculated. At the end of the update cycle, extract all light intensity change rates and load power change rates within the cycle, and construct linear fitting models using the least squares method respectively; Set the next update cycle as the prediction cycle, substitute the data collection time to be predicted into the model, and output the corresponding predicted values ​​of light intensity change rate and load power change rate.

[0006] Preferably, the specific process of analyzing the variance of the multiplexed signal and matching the sampling frequency adjustment range is as follows: Calculate the fluctuation difference between the predicted values ​​of the rate of change of light intensity and the rate of change of load power within the predicted period, normalize and remove dimensions, and then sum them up by combining them with preset weighting coefficients to obtain the predicted value of the summation variance. Divide the waveform into several summation variance prediction value intervals, and pre-set each summation variance prediction value interval to correspond to a sampling frequency adjustment range; Match the predicted summation variance value within the period to be predicted with all summation variance prediction value intervals, and output the corresponding sampling frequency adjustment range. Extract all candidate sampling frequencies within the range and construct a candidate sampling frequency set; obtain all temperature correction coefficients within the current sampling frequency update period and calculate the mean to obtain the average temperature correction coefficient; A comprehensive objective function model is constructed to consider both sensing accuracy loss and sampling power consumption. By finding the candidate sampling frequency that minimizes the model result, the optimal sampling frequency for the period to be predicted is output.

[0007] Preferably, the specific process for adjusting the optimal sampling frequency is as follows: After entering the next update cycle, the corresponding remaining energy storage capacity and real-time temperature correction coefficient are obtained at each acquisition moment. After normalization and dimensionless processing, the power consumption control threshold at that acquisition moment is calculated. Obtain the optimal sampling frequency corresponding to the update cycle. If the optimal sampling frequency meets the power consumption control threshold requirement, keep it unchanged; otherwise, adjust the optimal sampling frequency to the maximum feasible value that meets the power consumption control threshold.

[0008] Preferably, the specific process of analyzing the function priority evaluation value of the running function and matching the corresponding function priority level is as follows: Acquire all running functions of the edge sensing terminal, set the sampling frequency update cycle as the priority evaluation cycle, and trigger the priority update when the cumulative runtime reaches the target. For each function, retrieve historical operation data from the most recent consecutive preset collection times, and calculate the contribution ratio of function data calls, the correlation ratio of function missing alarms, and the core resource occupancy coefficient of the function. The three indicators are weighted and summed using preset weighting coefficients to obtain the functional priority evaluation value. The system is divided into three priority levels: core support level, important collaboration level, and non-core auxiliary level, with each level corresponding to a unique evaluation value range. Match the evaluation values ​​of each operating function with the interval, and output the function priority level corresponding to each operating function within the current evaluation period.

[0009] Preferably, the specific process for analyzing the combined light and load fluctuation value of the evaluation period and determining the scheduling scenario type for the corresponding period is as follows: When priority update is triggered, the scheduling scenario is determined synchronously, a historical statistical time window is set, the cumulative fluctuation of light intensity and load power within the statistical window is calculated, and the weighted sum is used to obtain the comprehensive fluctuation value of light intensity and load. At the same time, the remaining energy storage power at the last collection time of the window is obtained. Preset relevant fluctuation thresholds and power thresholds. Based on the correspondence between the comprehensive fluctuation value of solar radiation and load and the remaining power of energy storage, the scheduling scenarios in this evaluation period are marked as stable scenarios, fluctuating scenarios, or tense scenarios, respectively.

[0010] Preferably, the specific process of adjusting the running cycle according to the basic cycle adjustment coefficient corresponding to the function priority level is as follows: Obtain the optimal sampling frequency for the current evaluation period, and set corresponding basic period adjustment coefficients for the core support level, important collaborative level, and non-core auxiliary level respectively; Each operating function is matched with a basic cycle adjustment coefficient according to its own priority level, and the basic operating cycle is calculated. Execution coefficients are adjusted based on the scheduling scenario type: For stable scenarios, the baseline coefficients are used. Correcting important collaborative level coefficients and non-core auxiliary level coefficients in fluctuating scenarios; In stressful scenarios, critical coordination-level coefficients are adjusted, while non-core auxiliary-level operations are suspended. When the optimal sampling frequency triggers an update, the basic operating cycle of each function is recalculated synchronously.

[0011] Preferably, the specific process of calculating the total number of time slots and dynamically adjusting the proportion of each priority time slot according to the scenario is as follows: Obtain the total duration of the evaluation cycle, and calculate the total number of time slots for the edge sensing end based on the optimal sampling frequency of the current evaluation cycle; Construct a functional priority time slot ratio configuration library, and configure initial time slot ratio coefficients for core protection level, important collaboration level, non-core auxiliary level and redundancy respectively. Each coefficient meets the total ratio requirement, and calculate the initial allocated time slots for each priority level accordingly. Dynamically adjust according to scheduling scenario: for stable scenarios, directly use the initial allocation; In fluctuating scenarios, the proportion of non-core auxiliary time slots is reduced; the proportion of core support time slots is increased or all remaining time slots are allocated based on the remaining allocable time slot space; and the initial proportion of important coordination time slots is retained. In high-pressure scenarios, the proportion of important collaborative time slots will be reduced, the proportion of non-core auxiliary time slots will be fully reduced, and the proportion of core support time slots will be increased to the preset maximum threshold.

[0012] Preferably, the specific process of splitting the orchestration function runtime slots and generating a distribution plan to be sent to the edge-aware scheduling and execution unit is as follows: Verify whether the sum of the proportions of each priority time slot after the adjustment meets the total time slot allocation requirements. If there is a slight deviation, it is corrected by preset redundant time slots to ensure that time slot resources are not wasted. All operational functions are categorized into core support level, important collaboration level, and non-core auxiliary level. Based on the preset basic runtime slot quotas for each function, the allocated time slots are split and adjusted according to the quota ratio of the same priority category. Arrange the runtime slots of each function in chronological order, generate a time slot allocation scheme for the function in the cost evaluation cycle, and directly send it to the edge sensing end scheduling and execution unit for execution.

[0013] Compared with the prior art, the beneficial effects of the present invention are: (1) The intelligent management system of the distributed energy storage device first predicts the rate of change of light and load by linear fitting through the sensing and adaptation module, then locks the sampling interval by summing variance, and performs secondary trimming on the optimal sampling frequency by the remaining power and temperature correction threshold. The system can automatically increase sampling to capture peak when the light changes drastically, and actively reduce frequency to extend the battery life when the power is stable or low. It achieves high-precision prediction and low power consumption in parallel, which significantly improves the power supply reliability in remote gridless scenarios, while avoiding battery anxiety caused by excessive power consumption.

[0014] (2) The intelligent management system of this distributed energy storage device calculates the three-dimensional indicators of "data contribution, missing alarm and resource occupation" through the function classification and scenario judgment module. It assigns three levels of priority to each operation task: core guarantee, important collaboration and non-core auxiliary. It also crosses three types of scenarios: stable, fluctuating and tense. When fluctuating, the non-core frequency is reduced and when tense, the non-core is directly suspended. The computing power and energy consumption are concentrated to key functions such as charging and discharging control, which significantly eliminates the resource mismatch caused by the traditional "one-size-fits-all" operation and ensures that the core tasks always get the minimum resource guarantee.

[0015] (3) The intelligent management system of this distributed energy storage device divides the evaluation cycle into time slot pools through the time-series resource scheduling module. It first gives the initial quota according to priority, then dynamically reclaims non-core time slots according to the scenario and adds them to the core guarantee level. Redundant fragments are corrected at once. Finally, it arranges exclusive running time slots according to the time axis, so that core functions can obtain continuous hardware windows first when needed and low priority tasks give way in an orderly manner. It realizes real-time scheduling of three-dimensional linkage of "priority-sampling frequency-scenario", effectively shortens the response time and improves the utilization rate of time slots. Attached Figure Description

[0016] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1; Please see Figure 1 The present invention provides an intelligent management system for distributed energy storage devices, including: a perception and adaptation module, a function classification and scenario determination module, and a time-series resource scheduling module; The sensing and adaptation module collects core operating parameters of the energy storage device at a basic acquisition frequency, constructs a fitting model to predict the rate of change of light intensity and the rate of change of load power in the period to be predicted, analyzes the variance of the combined wave, matches the sampling frequency adjustment range, determines the optimal sampling frequency, and adjusts the optimal sampling frequency in combination with the remaining energy storage capacity and the temperature correction power consumption threshold. The specific process is as follows: In this distributed energy storage device, multiple types of monitoring sensors are deployed at the photovoltaic modules, load-side incoming line, energy storage unit cell area, and energy storage compartment through edge sensing terminals. According to the basic acquisition frequency, the core operating parameters of the energy storage device are collected, including: the light intensity on the surface of the photovoltaic modules, the real-time power on the load side, the remaining power of the energy storage unit, and the ambient temperature inside the energy storage compartment. Set the sampling frequency update period. For each sampling moment in the current update period, obtain the light intensity value, load power value, and time interval between the current sampling moment and the previous sampling moment. Subtract the parameter values ​​corresponding to the previous acquisition time from the light intensity value and load power value corresponding to the current acquisition time, and divide by the time interval to obtain the light intensity change rate G and load power change rate P corresponding to the current acquisition time. At each data acquisition moment in the current update cycle, the optimal operating temperature of the energy storage unit cell, as specified by the factory calibration, is obtained. Simultaneously, the ambient temperature inside the energy storage compartment at the current acquisition moment is obtained, using the formula: The temperature correction coefficient is obtained. ;in, This represents the ambient temperature inside the energy storage chamber at the current data collection moment. For optimal operating temperature, The preset temperature influence coefficient; At the end of the current update cycle, extract the rate of change of light intensity G and the rate of change of load power P corresponding to all collection times within this cycle, and construct linear fitting models using the least squares method. The fitting model expression is: , where y is the rate of change of light intensity G or the rate of change of load power P at the corresponding moment; denoted as the fitting slope, b as the fitting intercept, and x as the time sequence identifier for the acquisition time. Mark the next update cycle as the prediction cycle, substitute the prediction acquisition time of the prediction cycle into the fitting model, and output the predicted values ​​of the light intensity change rate and the load power change rate corresponding to each prediction acquisition time of the next prediction cycle. The fluctuation difference (mean square error) between the predicted rate of change of illuminance and the predicted rate of change of load power within the forecast period is calculated separately to obtain the predicted value YG of the fluctuation difference of the rate of change of illuminance and the predicted value YP of the fluctuation difference of the rate of change of load power. After normalization and dimensionless processing, the summation variance prediction value HP is obtained using the formula: HP=YG×b1+YP×b2, where b1 and b2 are preset weighting coefficients. The summation variance prediction value is used to characterize the overall fluctuation of illuminance and load within the future forecast period. Divide the waveform into several summation variance prediction value intervals, and pre-set each summation variance prediction value interval to correspond to a sampling frequency adjustment range; Match the predicted summation variance value within the period to be predicted with all summation variance prediction value intervals, and output the corresponding sampling frequency adjustment range. Extract all candidate sampling frequencies within the sampling frequency adjustment range and construct a candidate sampling frequency set F; Obtain all temperature correction coefficients within the current sampling frequency update period, and calculate the average temperature correction coefficient. ; A comprehensive objective function model considering both sensing accuracy loss and sampling power consumption is constructed, outputting the optimal sampling frequency corresponding to the period to be predicted. The formula is as follows: ; in, The optimal sampling frequency; Let i be the i-th candidate sampling frequency in the candidate sampling frequency set F, where i is the label of the candidate sampling frequency; To optimize the operator, we are looking for the value of the independent variable that minimizes the objective function value within the curly braces; The power consumption per unit frequency sampling (a fixed value determined by actual measurements based on the sensor hardware specifications, representing the power consumption increase for every 1Hz increase in sampling frequency); It should be noted that all parameters used in the objective function model have been normalized and dimensionless. Once the system enters the next update cycle, it acquires the remaining energy storage capacity at each acquisition time within that cycle. With real-time temperature correction factor After normalization and dimensionless processing, the formula is used: The power consumption control threshold corresponding to the acquisition time is obtained. Where k is the preset power consumption control benchmark coefficient for the edge sensing end; Obtain the optimal sampling frequency corresponding to this update period. ,like If so, the optimal sampling frequency remains unchanged; like Then Adjusted to meet The maximum feasible value (to avoid excessive consumption of energy storage capacity).

[0019] It should be noted that by deploying multiple types of monitoring sensors in key areas of the energy storage equipment, core operating parameters are accurately collected at the basic acquisition frequency, providing comprehensive data support for subsequent analysis. Then, by calculating the difference between parameters at two acquisition times and combining the time interval to obtain the rate of change, a linear fitting model is constructed using the least squares method to achieve accurate prediction of future periodic rates of change, providing a scientific basis for sampling frequency adaptation.

[0020] Simultaneously, a temperature correction coefficient is introduced to optimize model accuracy, and the sampling frequency range is matched with the summing variance to break the limitations of the traditional fixed sampling mode. Finally, based on the remaining energy storage capacity and the dynamically calculated power consumption control threshold, the optimal sampling frequency is adjusted a second time. This ensures the accuracy of energy supply-load status prediction while avoiding excessive power consumption, achieving a dynamic balance between sensing accuracy and energy consumption control. This provides accurate and efficient basic data support for subsequent function classification and resource scheduling, ensuring the stable operation of energy storage equipment in complex scenarios.

[0021] Functional grading and scenario determination module: Analyzes the functional priority evaluation values ​​of all operating functions at the edge sensing end according to the evaluation cycle, matches the corresponding functional priority level, analyzes the combined light-load fluctuation value of the evaluation cycle, and determines the scheduling scenario type for the corresponding cycle. The specific process is as follows: Acquire all operational functions of the edge sensing terminal, including: sensing and acquisition functions, charging and discharging control functions, communication and transmission functions, status prediction functions, data storage functions, and log recording functions; The sampling frequency update cycle and the priority evaluation cycle are set to a unified cycle. Whenever the cumulative runtime reaches the evaluation cycle, the priority level of each running function is triggered to update, specifically as follows: For each operating function, retrieve the historical operating data of the most recent M consecutive acquisition times (M is a preset positive integer), count the total number of times the data output by the operating function is called by the charging and discharging control core function, divide the number of calls by the charging and discharging control core function by the total number of times all data is called to obtain the function data call contribution ratio I1; Based on the historical operation data of the most recent M consecutive collection times, the timing scheduling unit counts the total number of time periods when the operation function is not running (including natural absence or temporary non-startup), and at the same time counts the number of times the edge sensing terminal triggers core indicator alarms (such as voltage over-limit, charging and discharging abnormality) during these inactive periods; divide the above alarm count by the total number of inactive periods to obtain the function absence alarm correlation ratio I2; The processor computing power ratio (i.e., the actual computing power used by the running function ÷ the total available computing power of the edge sensing end) and the energy consumption ratio (i.e., the actual energy consumption of the running function ÷ the total energy consumption of the edge sensing end during the same period) of the most recent M consecutive collection times are statistically analyzed. After assigning different preset weights to the two ratios, they are weighted and summed to obtain the core resource occupancy coefficient I3 of the function. By substituting the contribution ratio of function data calls I1, the correlation ratio of function missing alarms I2, and the core resource occupancy coefficient of the function into the formula: GYP=I1×w1+I2×w2+I3×w3, the function priority evaluation value GYP is obtained; where w1, w2, and w3 are preset weight coefficients. The system is divided into three functional priority levels: core support level, important collaboration level, and non-core auxiliary level. Each priority level is pre-defined to correspond to a unique functional priority evaluation value range. The priority evaluation value corresponding to each running function is matched with the evaluation value range corresponding to the three function priority levels, and the function priority level corresponding to each running function in the current evaluation period is output (as the core basis for subsequent resource allocation and timing adjustment). The system acquires real-time dynamic parameters collected by the sensing and adaptation module at each acquisition moment, including: real-time values ​​of light intensity, load power, remaining energy storage capacity, power consumption control threshold, light intensity change rate, load power change rate, and optimal sampling frequency, and performs preprocessing and normalization to remove dimensions. Each time a priority update is triggered, the current scheduling scenario is determined synchronously, specifically as follows: A historical statistical time window is set, and the cumulative fluctuation of light intensity at the edge sensing end (the sum of the absolute values ​​of the differences between light intensity at each collection time and the previous time) and the cumulative fluctuation of load power (the sum of the absolute values ​​of the differences between load power at each collection time and the previous time) are calculated within the historical statistical time window. After assigning preset weights to the two types of fluctuations, the values ​​are weighted and summed to obtain the comprehensive fluctuation value of light intensity and load. At the same time, the real-time value of the remaining energy storage capacity at the last collection time of the historical statistical time window is taken. Preset low fluctuation threshold σL, high fluctuation threshold σH, low power threshold SL, and high power threshold SH; If the combined fluctuation value of illumination and load in the current evaluation period is ≤σL, and the remaining energy storage capacity at the last collection time of the historical statistical period window is ≥SH, then the scheduling scenario of the edge sensing terminal in this evaluation period will be marked as a stable scenario. If σL < the combined fluctuation value of illumination and load in the current evaluation period ≤ σH, and SL ≤ the remaining energy storage capacity at the last collection moment of the historical statistical time window < SH, then the scheduling scenario of the edge sensing terminal in this evaluation period will be marked as a fluctuation scenario. If the combined light-load fluctuation value of the current evaluation period is greater than σH, or the remaining energy storage capacity at the last collection moment of the historical statistical time window is less than SL, then the scheduling scenario of the edge sensing terminal in this evaluation period will be marked as a tense scenario.

[0022] It should be noted that by constructing a multi-dimensional priority evaluation system, the system comprehensively considers the contribution ratio of function data calls, the correlation ratio of function missing alarms, and the core resource occupancy coefficient, rather than determining priority based on a single dimension. This makes the division of core protection level, important collaboration level, and non-core auxiliary level more in line with actual operational needs, ensuring that key functions such as charging and discharging control can receive priority resource support, and avoiding non-core functions from consuming too much computing power and energy. At the same time, priority updates and scheduling scenario determination are triggered synchronously. By statistically analyzing the comprehensive fluctuation value of light and load within the historical time window, and combining the dual-dimensional indicators of remaining energy storage power, three scenarios are divided into stable, fluctuating, and tense. This breaks the limitations of single indicator determination and can accurately reflect the dynamic environment and energy status of equipment operation. In addition, the real-time acquisition of accurate dynamic data from the perception and adaptation module, after preprocessing, is used for scene determination, achieving seamless connection of the data link. This ensures that the determination of priority level and scene type is based on real and real-time operating data, providing accurate and reliable decision-making basis for the time-series resource scheduling module. This optimizes the resource allocation logic from the root and improves the operating efficiency and stability of energy storage equipment.

[0023] The time-series resource scheduling module obtains the optimal sampling frequency and scheduling scenario type for the current evaluation period, matches the basic cycle adjustment coefficient corresponding to the function priority level, and adjusts the running cycle; calculates the total number of time slots, dynamically adjusts the proportion of time slots for each priority according to the scenario, splits and arranges the function running slots, and generates a distribution plan to be sent to the edge perception end scheduling execution unit. The specific process is as follows: Obtain the optimal sampling frequency of the edge sensing terminal within the current evaluation period, and set a basic period adjustment coefficient corresponding to each functional priority level, namely: the core guarantee level corresponds to... Important collaborative level correspondence Non-core auxiliary level correspondence ,and ; For each running function at the edge sensing end, its corresponding function priority level is obtained, and its corresponding basic cycle adjustment coefficient is matched accordingly; using the formula: , to obtain the basic operating cycle T of the running function within the current evaluation cycle; m is the label of the basic cycle adjustment coefficient, m=1, 2, 3; Obtain the scheduling scenario type of the edge perception terminal within the current evaluation period, and execute the corresponding coefficient adjustment logic, specifically as follows: In a stable scenario, all running functions will use the base adjustment factor of their respective function priority level (i.e.: , , ); In the case of fluctuating scenarios, the corresponding adjustment coefficients for each important collaborative level function will be corrected to... ,and >1, The preset adjustment factor is used; for each non-core auxiliary function, the corresponding adjustment coefficient is adjusted to... ,and , This is a preset adjustment factor; In high-pressure scenarios, the adjustment coefficients for each critical collaborative function will be corrected to... , , The preset adjustment factor will be used to pause the operation of each non-core auxiliary function during this evaluation period. Whenever the optimal sampling frequency of the edge sensing terminal is updated, the basic operating cycle of each function is recalculated synchronously. Obtain the total duration of the evaluation period and use the formula. The total number of time slots of the edge sensing end in the current evaluation period is obtained. ;in, The duration of the evaluation period; Build a time slot ratio configuration library for edge sensing functions. In the library, an initial time slot ratio coefficient is configured for each function priority level. The initial time slot ratio coefficient for core support functions is denoted as: The initial time slot ratio coefficient for important collaborative functions is denoted as: The initial time slot ratio coefficient for non-core auxiliary functions is denoted as: The redundancy time slot ratio coefficient is denoted as And satisfy: ; The initial allocated time slots for each function priority level are obtained by multiplying the total number of time slots in the current evaluation period by the initial time slot ratio coefficient corresponding to each function priority level. These include: the initial allocated time slots for core support functions, the initial allocated time slots for important collaborative functions, the initial allocated time slots for non-core auxiliary functions, and the initial allocated time slots for redundancy. Obtain the scheduling scenario type for the current edge sensing terminal in this evaluation cycle, and dynamically adjust the initial time slot ratio coefficient and initial allocated time slot number for each function priority level based on the scheduling scenario type. Specifically: If the current scheduling scenario is a stable scenario, the number of time slots allocated to each function priority level will be directly allocated according to the initial time slot ratio coefficient; If the current scheduling scenario is a fluctuating scenario, then the following time slot adjustment operation will be performed: For non-core auxiliary functions, the preset non-core auxiliary time slot compression ratio is applied. ( ), its initial time slot proportion coefficient After compression, the proportion of non-core auxiliary time slots is: The corresponding number of allocated time slots is ; For core protection-level functions, the total available time slot space for the edge sensing end within this evaluation period is first obtained. (That is, the remaining time slots after deducting the initial allocated time slots for important coordination level, the compressed allocated time slots for non-core auxiliary level, and the initial allocated time slots for core support level from the total number of time slots). If the total available time slot space ,in, The increase in the proportion of core support-level time slots is preset; then according to Initial time slot ratio coefficient for core protection level The upgrade will be implemented, and the core support-level time slot ratio coefficient will be [value missing]. The corresponding number of time slots allocated is ; If the remaining time slots Then the remaining time slots All allocated to core support functions; the corresponding number of time slots allocated is ; Important collaborative functions will continue to allocate time slots using the initial time slot ratio coefficient. For the time slot ratio of important collaborative functions, the initial time slot ratio coefficient is compressed according to the pre-set important collaborative time slot compression ratio. After compression, the adjusted ratio coefficient is multiplied by the total number of time slots in this evaluation cycle to calculate the number of allocated time slots corresponding to the important collaborative functions. For the time slot ratio of non-core auxiliary functions, the initial time slot ratio coefficient is compressed according to the pre-set full compression ratio of non-core auxiliary time slots. After compression, the adjusted ratio coefficient is multiplied by the total number of time slots in this evaluation period to calculate the allocated time slots for non-core auxiliary functions. The full compression ratio of non-core auxiliary time slots is not only greater than the compression ratio of important collaborative time slots, but also greater than the compression ratio of non-core auxiliary time slots set in fluctuating scenarios, ensuring that non-core auxiliary functions release resources to the maximum extent in stressful scenarios. For the time slot ratio of core protection-level functions, the corresponding time slot ratio coefficient is directly increased to the maximum threshold of core protection-level time slots preset by the edge perception system. This maximum threshold is the highest time slot ratio that the system supports core protection-level functions to occupy, which can meet the extreme resource requirements of core protection-level functions in stressful scenarios. After the time slot ratios of core protection, important collaboration, and non-core auxiliary functions are all adjusted, verify whether the sum of the adjusted time slot ratios of each priority level meets the system's total time slot allocation requirements. If there is a slight deviation, it will be corrected by using preset redundant time slots to ensure that there is no waste or idleness of total time slot resources. All operational functions at the edge sensing end are categorized and grouped according to their established priority levels: core protection level, important collaboration level, and non-core auxiliary level. Then, referring to the pre-defined basic runtime slot quota for each function (a fixed value determined based on the function's hardware operating specifications and data processing volume), the number of time slots is allocated to each priority level after adjustment. The time slots are equally divided according to the proportion of the basic runtime slot quota for each function under the same priority level. Core protection level functions are allocated time slots in full according to their basic quota proportion. Important collaboration level functions are allocated time slots after being compressed according to their basic quota proportion. Non-core auxiliary level functions are not allocated any time slots. After the split is completed, the runtime slots of each function are arranged in chronological order (ensuring that the slots of different functions do not overlap), generating the function slot allocation scheme for the current evaluation cycle, and directly sending it to the scheduling and execution unit of the edge sensing end for execution.

[0024] It should be noted that the timing resource scheduling module, as the core execution unit of the system, undertakes the optimal sampling frequency and function classification of the perception and adaptation module and the priority and scenario decision of the scenario judgment module. It specifically solves the inefficiency and preemption problems caused by fixed timing and rigid resource allocation in traditional scheduling, and realizes dynamic optimization and configuration of resources. Based on the optimal sampling frequency, a dedicated basic cycle adjustment coefficient is matched for functions of different priorities, and the coefficient is dynamically corrected according to three scenarios: stable, fluctuating, and tense. In fluctuating scenarios, the efficiency of important collaborative functions is enhanced, while in tense scenarios, the core support functions are focused on and non-core auxiliary functions are suspended. This ensures that the rhythm of function operation is precisely matched with the actual operating status of the equipment, avoiding the waste of resources caused by a "one-size-fits-all" timing arrangement. At the time slot allocation level, the total number of time slots is calculated by the evaluation period and the optimal sampling frequency, and a multi-priority initial time slot ratio configuration library is constructed. Then, it is dynamically adjusted according to the scenario: in fluctuating scenarios, non-core auxiliary time slots are compressed and core support time slots are increased as needed; in tense scenarios, non-core auxiliary time slots are fully compressed and core support time slots are maximized to the maximum threshold to ensure that key functions receive sufficient resource support in extreme scenarios. At the same time, the deviation is corrected by redundant time slots to achieve no waste and no idle time slot resources. Furthermore, by classifying and grouping time slots according to functional priority and splitting them with reference to the basic runtime slot quota, and arranging non-overlapping runtime slot segments in chronological order, a distribution plan is generated and directly issued for execution. This ensures the fairness of resource allocation among functions of the same priority and avoids functional operation conflicts, transforming resource scheduling from "passive allocation" to "active adaptation." This significantly improves the operating efficiency and response speed of the edge sensing end, providing a solid guarantee for the stable and efficient operation of energy storage devices in complex and ever-changing remote off-grid scenarios.

[0025] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent management system for distributed energy storage devices, characterized in that, include: The sensing and adaptation module collects the core operating parameters of the energy storage device at the basic acquisition frequency, builds a fitting model to predict the rate of change of light intensity and the rate of change of load power in the period to be predicted, analyzes the variance of the combined wave, matches the sampling frequency adjustment range, determines the optimal sampling frequency, and adjusts the optimal sampling frequency in combination with the remaining energy storage capacity and the temperature correction power consumption threshold. Functional grading and scenario determination module: Analyzes the functional priority evaluation values ​​of all running functions of the edge sensing terminal according to the evaluation cycle, matches the corresponding functional priority level, analyzes the comprehensive fluctuation value of illumination-load in the evaluation cycle, and determines the scheduling scenario type of the corresponding cycle. The timing resource scheduling module obtains the optimal sampling frequency and scheduling scenario type for the current evaluation period, matches the basic cycle adjustment coefficient corresponding to the function priority level and adjusts the running cycle; calculates the total number of time slots, dynamically adjusts the proportion of time slots of each priority according to the scenario, splits and arranges the running time slots of the orchestration function, and generates a distribution plan to be sent to the edge perception end scheduling execution unit.

2. The intelligent management system for a distributed energy storage device according to claim 1, characterized in that: The specific process for predicting the rate of change of light intensity and the rate of change of load power in the period to be predicted is as follows: The core operating parameters of the energy storage device are collected at the basic acquisition frequency. These parameters include the light intensity on the surface of the photovoltaic module, the real-time power on the load side, the remaining power of the energy storage unit, and the ambient temperature inside the energy storage chamber. Set the sampling frequency update period, and for each sampling moment, obtain the current light intensity value, load power value and time interval between the current and previous sampling moments. Subtract the corresponding parameters at the two moments and divide by the time interval to obtain the corresponding light intensity change rate and load power change rate. At each acquisition moment, the optimal operating temperature of the energy storage unit cell and the ambient temperature inside the energy storage chamber are simultaneously obtained, and the temperature correction coefficient is calculated. At the end of the update cycle, extract all light intensity change rates and load power change rates within the cycle, and construct linear fitting models using the least squares method respectively; Set the next update cycle as the prediction cycle, substitute the data collection time to be predicted into the model, and output the corresponding predicted values ​​of light intensity change rate and load power change rate.

3. The intelligent management system for a distributed energy storage device according to claim 2, characterized in that: The specific process of analyzing the variance of the combined waveform and matching the sampling frequency adjustment range is as follows: Calculate the fluctuation difference between the predicted values ​​of the rate of change of light intensity and the rate of change of load power within the predicted period, normalize and remove dimensions, and then sum them up by combining them with preset weighting coefficients to obtain the predicted value of the summation variance. Divide the waveform into several summation variance prediction value intervals, and pre-set each summation variance prediction value interval to correspond to a sampling frequency adjustment range; Match the predicted summation variance value within the period to be predicted with all summation variance prediction value intervals, and output the corresponding sampling frequency adjustment range. Extract all candidate sampling frequencies within the range and construct a candidate sampling frequency set; obtain all temperature correction coefficients within the current sampling frequency update period and calculate the mean to obtain the average temperature correction coefficient; A comprehensive objective function model is constructed to consider both sensing accuracy loss and sampling power consumption. By finding the candidate sampling frequency that minimizes the model result, the optimal sampling frequency for the period to be predicted is output.

4. The intelligent management system for a distributed energy storage device according to claim 3, characterized in that: The specific process for adjusting the optimal sampling frequency is as follows: After entering the next update cycle, the corresponding remaining energy storage capacity and real-time temperature correction coefficient are obtained at each acquisition moment. After normalization and dimensionless processing, the power consumption control threshold at that acquisition moment is calculated. Obtain the optimal sampling frequency corresponding to the update cycle. If the optimal sampling frequency meets the power consumption control threshold requirement, keep it unchanged; otherwise, adjust the optimal sampling frequency to the maximum feasible value that meets the power consumption control threshold.

5. The intelligent management system for a distributed energy storage device according to claim 4, characterized in that: The specific process of analyzing the function priority evaluation values ​​of the running functions and matching them with the corresponding function priority levels is as follows: Acquire all running functions of the edge sensing terminal, set the sampling frequency update cycle as the priority evaluation cycle, and trigger the priority update when the cumulative runtime reaches the target. For each function, retrieve historical operation data from the most recent preset number of consecutive collection times, and calculate the contribution ratio of function data calls, the correlation ratio of function missing alarms, and the core resource occupancy coefficient of the function. The three indicators are weighted and summed using preset weighting coefficients to obtain the functional priority evaluation value. The system is divided into three priority levels: core support level, important collaboration level, and non-core auxiliary level, with each level corresponding to a unique evaluation value range. Match the evaluation values ​​of each operating function with the interval, and output the function priority level corresponding to each operating function within the current evaluation period.

6. The intelligent management system for a distributed energy storage device according to claim 5, characterized in that: The specific process for analyzing the combined light and load fluctuation values ​​during the evaluation period and determining the scheduling scenario type for the corresponding period is as follows: When priority update is triggered, the scheduling scenario is determined synchronously, a historical statistical time window is set, the cumulative fluctuation of light intensity and load power within the statistical window is calculated, and the weighted sum is used to obtain the comprehensive fluctuation value of light intensity and load. At the same time, the remaining energy storage power at the last collection time of the window is obtained. Preset relevant fluctuation thresholds and power thresholds. Based on the correspondence between the comprehensive fluctuation value of solar radiation and load and the remaining power of energy storage, the scheduling scenarios in this evaluation period are marked as stable scenarios, fluctuating scenarios, or tense scenarios, respectively.

7. The intelligent management system for a distributed energy storage device according to claim 6, characterized in that: The specific process of adjusting the running cycle according to the basic cycle adjustment coefficient corresponding to the function priority level is as follows: Obtain the optimal sampling frequency for the current evaluation period, and set corresponding basic period adjustment coefficients for the core support level, important collaborative level, and non-core auxiliary level respectively; Each operating function is matched with a basic cycle adjustment coefficient according to its own priority level, and the basic operating cycle is calculated. Execution coefficients are adjusted based on the scheduling scenario type: For stable scenarios, the baseline coefficients are used. Correcting important collaborative level coefficients and non-core auxiliary level coefficients in fluctuating scenarios; In stressful scenarios, critical coordination-level coefficients are adjusted, while non-core auxiliary-level operations are suspended. When the optimal sampling frequency triggers an update, the basic operating cycle of each function is recalculated synchronously.

8. The intelligent management system for a distributed energy storage device according to claim 7, characterized in that: The specific process for calculating the total number of time slots and dynamically adjusting the proportion of each priority time slot according to the scenario is as follows: Obtain the total duration of the evaluation cycle, and calculate the total number of time slots for the edge sensing end based on the optimal sampling frequency of the current evaluation cycle; Construct a functional priority time slot ratio configuration library, and configure initial time slot ratio coefficients for core protection level, important collaboration level, non-core auxiliary level and redundancy respectively. Each coefficient meets the total ratio requirement, and calculate the initial allocated time slots for each priority level accordingly. Dynamically adjust according to scheduling scenario: for stable scenarios, directly use the initial allocation; In fluctuating scenarios, the proportion of non-core auxiliary time slots is reduced; the proportion of core support time slots is increased or all remaining time slots are allocated based on the remaining allocable time slot space; and the initial proportion of important coordination time slots is retained. In high-pressure scenarios, the proportion of important collaborative time slots will be reduced, the proportion of non-core auxiliary time slots will be fully reduced, and the proportion of core support time slots will be increased to the preset maximum threshold.

9. The intelligent management system for a distributed energy storage device according to claim 8, characterized in that: The specific process of splitting the orchestration function runtime slots, generating a distribution plan, and sending it to the edge-aware scheduling and execution unit is as follows: Verify whether the sum of the proportions of each priority time slot after the adjustment meets the total time slot allocation requirements. If there is a slight deviation, it is corrected by preset redundant time slots to ensure that time slot resources are not wasted. All operational functions are categorized into core support level, important collaboration level, and non-core auxiliary level. Based on the preset basic runtime slot quotas for each function, the allocated time slots are split and adjusted according to the quota ratio of the same priority category. Arrange the runtime slots of each function in chronological order, generate a time slot allocation scheme for the function in the cost evaluation cycle, and directly send it to the edge sensing end scheduling and execution unit for execution.

Images