Intelligent monitoring and early warning method and device for photovoltaic energy storage equipment

Abstract

The invention provides an intelligent monitoring and early warning method and device for photovoltaic energy storage equipment, and the method comprises the steps: carrying out the multi-time scale sliding window feature extraction of the operation data of a photovoltaic energy storage system, calculating a power coupling degree quantitative index and a system response feature index, and generating a real-time feature data set; based on the real-time feature data set, performing deviation propagation path identification under thermal dynamic constraint to obtain a real-time deviation propagation path diagram marking deviation sensitive nodes; performing multi-level dynamic deduction in combination with the real-time feature data set and the deviation propagation path map to obtain a dynamic deviation accumulation situation map displaying a deviation accumulation risk level and a development trend; and performing adaptive threshold early warning judgment according to the dynamic deviation accumulation situation map, and executing response control according to a hierarchical early warning mechanism to obtain a hierarchical early warning response control instruction. According to the method, the cumulative effect of the power prediction deviation can be effectively identified, and early warning of coordination imbalance of the photovoltaic energy storage system is realized.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an intelligent monitoring and early warning method and device for photovoltaic energy storage equipment. Background Art

[0002] With the widespread adoption of photovoltaic energy storage systems in distributed energy systems, intelligent monitoring and early warning technologies are becoming crucial for ensuring the safe and stable operation of these systems. Existing monitoring and early warning methods for photovoltaic energy storage devices primarily rely on single-point threshold detection and statistical analysis of historical data. These methods determine the device's operating status by setting fixed thresholds for parameters such as voltage, current, and temperature. Warning signals are triggered when monitored parameters exceed preset ranges.

[0003] However, a major problem with existing technologies is their inability to effectively identify the cumulative effects of photovoltaic power forecast deviations on energy storage systems. Due to the random and intermittent nature of photovoltaic power generation, there are often deviations between actual generated power and predicted power. These deviations propagate and accumulate within the energy storage system through the charging and discharging regulation process, ultimately leading to energy shortages or overcharge risks at critical moments. Traditional fixed-threshold monitoring methods struggle to identify these cumulative coordination imbalances in advance. Summary of the Invention

[0004] The main purpose of the present invention is to solve the technical problem that photovoltaic energy storage equipment monitoring and early warning methods cannot effectively identify the cumulative effect of power prediction deviations, resulting in the inability to provide early warning of system coordination imbalance risks; A first aspect of the present invention provides an intelligent monitoring and early warning method for photovoltaic energy storage equipment, the intelligent monitoring and early warning method for photovoltaic energy storage equipment comprising: Performing feature extraction processing on the operating data of the photovoltaic energy storage system based on a multi-time scale sliding window, calculating a power coupling quantification index and a system response characteristic index, and generating a real-time feature data set based on the power coupling quantification index and the system response characteristic index; Based on the real-time feature data set, a deviation propagation path identification process is performed on the photovoltaic energy storage system based on thermodynamic constraints to obtain a real-time deviation propagation path map with deviation-sensitive nodes marked; Based on the real-time feature data set and the deviation propagation path diagram, a multi-level dynamic deduction process is performed on the prediction deviation accumulation process to obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend; According to the dynamic deviation accumulation situation diagram, an adaptive threshold warning judgment process is performed on the coordinated imbalance state of the photovoltaic energy storage system, and corresponding response control is executed according to the hierarchical warning mechanism to obtain a hierarchical warning response control instruction.

[0005] Optionally, in a first implementation of the first aspect of the present invention, performing feature extraction processing on the operating data of the photovoltaic energy storage system based on a multi-time-scale sliding window, calculating a power coupling quantification index and a system response characteristic index, and generating a real-time feature data set based on the power coupling quantification index and the system response characteristic index includes: Perform multi-time-scale sliding window processing on the power generation data, charge and discharge power data, and battery SOC status data of each group of photovoltaic energy storage systems, and calculate the power change rate within each time window, where the power change rate includes the photovoltaic power generation power change rate and the energy storage response power change rate; The Pearson correlation coefficient of the photovoltaic power generation rate of change and the energy storage response power change rate is calculated to obtain a quantitative index of power coupling; The time difference between the time the PV energy storage system issues a command and the time the power output responds is calculated to obtain the energy storage response delay factor. A variance analysis is then performed on the power generated by strings in different orientations to calculate the array imbalance index. The energy storage response delay factor and the array imbalance index are combined to obtain a system response characteristic index, and the power coupling quantization index and the system response characteristic index are combined in time series to obtain a real-time characteristic data set.

[0006] Optionally, in a second implementation of the first aspect of the present invention, performing thermodynamic constraint-based deviation propagation path identification processing on the photovoltaic energy storage system based on the real-time feature dataset to obtain a real-time deviation propagation path graph marking deviation-sensitive nodes includes: The difference between the actual power and the predicted power of the photovoltaic power generation system is calculated to obtain the power prediction deviation; Calculating the charging and discharging power adjustment requirements of the photovoltaic energy storage system based on the power prediction deviation, and quantitatively analyzing the battery internal resistance change rate, charge and discharge rate, and current SOC state response degree in combination with the power coupling quantification index and the system response characteristic index to obtain a deviation amplification factor; According to the battery thermal time constant and the current temperature state, the temperature change lag time is calculated by fitting the exponential decay function to obtain the thermal inertia delay compensation parameter; Performing matrix operations on the deviation amplification factor and the thermal inertia delay compensation parameter to construct a two-dimensional vector field of the deviation propagation speed and direction; Nodes whose gradient values ​​in the two-dimensional vector field exceed a critical threshold are marked to obtain a real-time deviation propagation path diagram of the marked deviation-sensitive nodes.

[0007] Optionally, in a third implementation of the first aspect of the present invention, performing matrix operation on the deviation amplification factor and the thermal inertia delay compensation parameter to construct a two-dimensional vector field of the deviation propagation speed and direction includes: Performing a two-dimensional matrix arrangement processing on the deviation amplification coefficient according to the spatial distribution position of the photovoltaic energy storage system to obtain a coefficient spatial distribution matrix; Calculating the temperature response time constant of each layout position of the battery module according to the thermal inertia delay compensation parameter, and performing spatial difference operation processing on the temperature response time constant between adjacent layout positions to obtain a temperature propagation gradient field; Performing a Hadamard product operation on the coefficient space distribution matrix and the temperature propagation gradient field to obtain a deviation propagation intensity distribution considering thermodynamic constraints; The deviation propagation intensity distribution is spatially calculated and processed between battery module layout positions to obtain a deviation propagation direction field, and the deviation propagation intensity distribution and the deviation propagation direction field are vector-constructed to obtain a two-dimensional vector field.

[0008] Optionally, in a fourth implementation of the first aspect of the present invention, performing multi-level dynamic deduction processing on the predicted deviation accumulation process based on the real-time feature data set and the deviation propagation path diagram to obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend includes: According to the real-time feature data set and the deviation propagation path diagram, the prediction deviation accumulation process is decomposed into three levels: the first level, the second level, and the third level; At the first level, the deviation-sensitive node information in the deviation propagation path diagram is combined to perform a minimum calculation under the constraints of the power ramp rate, available capacity margin, and real-time temperature status of the photovoltaic energy storage system to obtain the instantaneous deviation absorption capacity. At the second level, based on the system response characteristic index, a linear regression fitting process is performed on the deviation accumulation trend within the past preset time, and a rolling time domain integral calculation process is performed on the load forecast value within the future preset time to obtain a deviation accumulation forecast value; At the third level, the power coupling quantification index in the real-time feature data set is combined to calculate the time series matching between the all-day illumination change curve and the energy storage charging and discharging plan to obtain the intraday energy balance deviation index; The instantaneous deviation absorption capacity, deviation cumulative prediction value and intraday energy balance deviation index of the three levels are superimposed in time series and graded and colored according to the risk level classification algorithm to obtain a dynamic deviation cumulative situation diagram showing the deviation cumulative risk level and development trend.

[0009] Optionally, in a fifth implementation of the first aspect of the present invention, at the second level, linear regression fitting processing is performed on the deviation accumulation trend within a past preset time based on the system response characteristic indicator, and rolling time domain integral calculation processing is performed in combination with the load forecast value within a future preset time, to obtain the deviation accumulation forecast value, including: At the second level, a time series rearrangement process is performed on the power forecast deviation data within a preset time period in the past according to the energy storage response delay factor in the system response characteristic index; The rearranged deviation data were subjected to least squares linear regression fitting, the slope coefficient and intercept parameter of the deviation accumulation were calculated, and the deviation accumulation trend function was obtained; According to the array imbalance index in the system response characteristic index, a weighted correction process is performed on the load prediction value within a preset time in the future to obtain a corrected load prediction sequence; Performing numerical integration calculation on the deviation cumulative trend function within a future preset time window, and performing rolling time domain superposition calculation on the modified load forecast sequence to obtain a superposition calculation result; The superposition operation result is dynamically updated according to the sliding window to obtain a deviation cumulative prediction value.

[0010] Optionally, in a sixth implementation of the first aspect of the present invention, performing adaptive threshold warning judgment processing on the coordinated imbalance state of the photovoltaic energy storage system based on the dynamic deviation accumulation situation diagram, executing corresponding response control according to the hierarchical warning mechanism, and obtaining the hierarchical warning response control instructions include: According to the risk level value in the dynamic deviation accumulation situation diagram, a multi-factor adaptive threshold calculation process is performed on the operating status parameters of the photovoltaic energy storage system to obtain a dynamic warning threshold; Compare and judge the risk level value with the dynamic warning threshold, and classify the warning levels according to the hierarchical warning mechanism to obtain the corresponding warning level; According to different warning levels, corresponding hierarchical response control processing is executed, and the warning judgment results are subjected to feedback learning optimization processing to obtain hierarchical warning response control instructions including warning levels and control actions.

[0011] A second aspect of the present invention provides an intelligent monitoring and early warning device for photovoltaic energy storage equipment, the intelligent monitoring and early warning device for photovoltaic energy storage equipment comprising: a feature extraction module for performing feature extraction processing on the operating data of the photovoltaic energy storage system based on a multi-time-scale sliding window, calculating a power coupling quantification index and a system response characteristic index, and generating a real-time feature data set based on the power coupling quantification index and the system response characteristic index; a path identification module, configured to perform a deviation propagation path identification process based on thermodynamic constraints on the photovoltaic energy storage system according to the real-time feature data set, and obtain a real-time deviation propagation path map with deviation-sensitive nodes marked; A situation deduction module is used to perform multi-level dynamic deduction processing on the prediction deviation accumulation process based on the real-time feature data set and the deviation propagation path diagram, and obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend; The early warning control module is used to perform adaptive threshold early warning judgment processing on the coordinated imbalance state of the photovoltaic energy storage system according to the dynamic deviation accumulation situation diagram, execute corresponding response control according to the hierarchical early warning mechanism, and obtain hierarchical early warning response control instructions.

[0012] The above-mentioned intelligent monitoring and early warning method and device for photovoltaic energy storage equipment performs multi-time-scale sliding window feature extraction on the photovoltaic energy storage system operation data, calculates power coupling quantification indicators and system response characteristic indicators, and generates a real-time feature data set; based on the real-time feature data set, it identifies the deviation propagation path under thermodynamic constraints to obtain a real-time deviation propagation path diagram that marks deviation-sensitive nodes; combines the real-time feature data set and the deviation propagation path diagram to perform multi-level dynamic deduction to obtain a dynamic deviation accumulation situation diagram that displays the deviation accumulation risk level and development trend; performs adaptive threshold warning judgment based on the dynamic deviation accumulation situation diagram, executes response control according to the hierarchical warning mechanism, and obtains a hierarchical warning response control instruction. The present invention can effectively identify the cumulative effect of power forecast deviations and achieve early warning of coordinated imbalances in photovoltaic energy storage systems.

[0013] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the present invention. The purposes and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0014] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a schematic diagram of a first embodiment of an intelligent monitoring and early warning method for photovoltaic energy storage equipment according to an embodiment of the present invention; Figure 2 Schematic diagram of an embodiment of an intelligent monitoring and early warning device for photovoltaic energy storage equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0017] The terms "including," "having," and any variations thereof, as used in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to the process, method, product, or device.

[0018] To facilitate understanding of this embodiment, firstly, a method for intelligent monitoring and early warning of photovoltaic energy storage equipment disclosed in an embodiment of the present invention is described in detail. Figure 1 As shown, this method includes the following steps: 101. Perform feature extraction processing on the operating data of the photovoltaic energy storage system based on a multi-time scale sliding window, calculate the power coupling quantification index and the system response characteristic index, and generate a real-time feature data set based on the power coupling quantification index and the system response characteristic index; In one embodiment of the present invention, performing feature extraction processing based on a multi-time-scale sliding window on the operating data of the photovoltaic energy storage system, calculating a power coupling quantification index and a system response characteristic index, and generating a real-time feature data set based on the power coupling quantification index and the system response characteristic index includes: performing multi-time-scale sliding window division processing on the power generation data, charge and discharge power data, and battery SOC status data of each string of the photovoltaic energy storage system, calculating the power change rate within each time window, wherein the power change rate includes the photovoltaic power generation power change rate and the energy storage response power change rate; calculating the Pearson correlation coefficient of the photovoltaic power generation power change rate and the energy storage response power change rate to obtain the power coupling quantification index; calculating the time difference between the command issuance time and the power output response time of the photovoltaic energy storage system to obtain the energy storage response delay factor, performing variance analysis processing on the power generation power of strings in different orientations, and calculating the array imbalance index; combining the energy storage response delay factor and the array imbalance index to obtain the system response characteristic index, and combining the power coupling quantification index and the system response characteristic index in time series to obtain the real-time feature data set.

[0019] Specifically, the system first acquires multi-dimensional operational data from the PV energy storage system's data acquisition system. This data includes the real-time power generation of each PV array string, the energy storage system's charge and discharge power, and the battery pack's state-of-charge (SOC) status. The system preprocesses this raw data, eliminating significant outliers by setting a reasonable value range and applying a moving average filter to eliminate high-frequency noise. After data preprocessing, the system establishes three sliding window mechanisms with different time scales: 1-minute, 15-minute, and 1-hour windows. This sliding window implementation utilizes a circular buffer data structure. Each buffer stores historical data for a specific time period, automatically overwriting the oldest data as new data arrives to maintain a constant window size. Within each sliding window, the system calculates the power rate of change by iterating over all data points within the window. Specifically, the system subtracts the power value of the previous data point from the power value of the current data point, then divides the result by the time interval between the two data points. The PV power generation rate of change and the energy storage response power rate of change are calculated using the same differential method. The system performs time differential operations on the PV string power generation sequence and the energy storage system charge and discharge power sequence to obtain the corresponding rate of change data series.

[0020] Specifically, the power coupling metric is quantified through the Pearson correlation coefficient calculation process. This process first calculates the arithmetic mean of the PV power generation rate of change series and the energy storage response power change rate series. The system iterates through all values ​​in each series, adds them up, and divides them by the total number of values ​​to obtain the mean. The system then calculates the deviation of each value in each series from its corresponding mean value, that is, subtracting the mean value of the series from each value. Next, the system calculates the covariance between the two series by multiplying the deviation of each value in the PV power generation rate of change series with the deviation of the corresponding value in the energy storage response power change rate series, then summing all the products and dividing them by the total number of data points minus one. The system also calculates the standard deviation of the two series by summing the squares of the deviations of each value in each series, dividing them by the total number of data points minus one, and then taking the square root. The Pearson correlation coefficient is calculated by dividing the covariance by the product of the standard deviations of the two series. This value reflects the strength of the linear correlation between the PV power generation and energy storage response power changes.

[0021] Specifically, the calculation of the system response characteristic index includes two components: the energy storage response delay factor and the array imbalance index. The energy storage response delay factor is calculated using the monitoring system's timestamp recording function. The system assigns a precise timestamp to each control command and each power output change event. When the energy management system issues a charge or discharge command to the energy storage device, the system records the time the command was issued and continuously monitors the actual power output of the energy storage device. The system determines the occurrence of a power change event by comparing the power output values ​​at adjacent time points. A valid power response is determined when the power change exceeds a preset threshold. The energy storage response delay is calculated by subtracting the timestamp of the corresponding control command from the timestamp of the power response. The array imbalance index is calculated using variance analysis. The system first collects the generated power values ​​of all PV strings at the same time and calculates the arithmetic mean of these values ​​as the overall average power. The system then calculates the difference between the generated power of each string and the overall average power. The squares of all these differences are summed and divided by the total number of strings minus one. This yields the variance of the generated power, which serves as the array imbalance index. The system normalizes the energy storage response delay factor and array imbalance index by subtracting the historical minimum value of the index from the original value, and then dividing it by the difference between the historical maximum and minimum values, so that the numerical range of both indicators is unified between zero and one. After normalization, the two indicators are weighted averaged by setting weight coefficients to obtain a comprehensive system response characteristic index.

[0022] Specifically, the real-time feature dataset is constructed using the storage structure of a time series database. The system creates a data record for each time point, which contains the power coupling quantification indicator value and system response characteristic indicator value calculated at the current moment. The data record also contains metadata such as timestamp information, data source identification, and calculation parameter configuration. The system uses a first-in-first-out queue mechanism to manage the dataset. When the dataset reaches the preset maximum capacity, the oldest data record is automatically deleted to free up storage space. The dataset update frequency is synchronized with the data collection frequency. Whenever new raw data arrives, the system immediately triggers the feature calculation process and updates the dataset content. The system also establishes a data integrity check mechanism, verifying the correctness of each data record through a checksum algorithm to ensure that the data is not corrupted during storage and transmission. The dataset uses an index structure to facilitate rapid retrieval of data records within a specific time range. The index is sorted by timestamp and supports range query operations.

[0023] 102. Based on the real-time feature data set, perform deviation propagation path identification processing on the photovoltaic energy storage system based on thermodynamic constraints to obtain a real-time deviation propagation path map that marks deviation-sensitive nodes; In one embodiment of the present invention, the step of performing thermodynamic constraint-based deviation propagation path identification processing on the photovoltaic energy storage system based on the real-time feature data set to obtain a real-time deviation propagation path diagram for marking deviation-sensitive nodes includes: performing difference calculation processing on the actual power and predicted power of the photovoltaic power generation system to obtain a power prediction deviation; calculating the charging and discharging power adjustment requirement of the photovoltaic energy storage system based on the power prediction deviation, and quantitatively analyzing the battery internal resistance change rate, charge and discharge rate, and response degree of the current SOC state in combination with the power coupling quantification index and the system response characteristic index to obtain a deviation amplification coefficient; performing exponential decay function fitting calculation processing on the temperature change lag time based on the battery thermal time constant and the current temperature state to obtain a thermal inertia delay compensation parameter; performing matrix operation processing on the deviation amplification coefficient and the thermal inertia delay compensation parameter to construct a two-dimensional vector field of the deviation propagation speed and direction; and marking nodes in the two-dimensional vector field whose gradient values ​​exceed a critical threshold to obtain a real-time deviation propagation path diagram for marking deviation-sensitive nodes.

[0024] Specifically, the system obtains theoretical power forecasts from the photovoltaic power generation prediction system and actual power measurements from the photovoltaic array monitoring system. The power forecast deviation is calculated by subtracting the predicted power from the actual power. The system performs this difference calculation at each sampling moment. A positive deviation is generated when the actual power exceeds the predicted power, and a negative deviation is generated when the actual power falls below the predicted power. The system establishes a deviation data caching mechanism, storing deviation values ​​over consecutive time periods in a time series array. This deviation data reflects the changes in the accuracy of photovoltaic power generation forecasts. Next, the system calculates the required charging and discharging power adjustments for the energy storage system based on the power forecast deviation. This calculation is performed by multiplying the power forecast deviation by the system's power balance coefficient. A positive deviation indicates that the energy storage system needs to increase charging power or reduce discharging power, while a negative deviation indicates that the system needs to reduce charging power or increase discharging power. The system extracts power coupling quantification indicators and system response characteristic indicators from the real-time feature data set. These two indicators reflect the coordinated operation status and response characteristics of the photovoltaic energy storage system. The power coupling quantification index is used to correct the calculation results of the power adjustment demand. When the coupling degree is high, the system coordination is good and the calculation accuracy of the power adjustment demand is high. When the coupling degree is low, there are problems with the system coordination and the power adjustment demand needs to be corrected accordingly.

[0025] Specifically, the calculation of the deviation amplification factor involves quantitative analysis of three key parameters: the battery's internal resistance change rate, the charge / discharge rate, and the current state of charge (SOC). The battery's internal resistance change rate is calculated by monitoring the real-time changes in the battery terminal voltage and current. The system uses Ohm's law to calculate the instantaneous internal resistance value at each moment, and then calculates the internal resistance change rate by taking the difference between the internal resistance values ​​at adjacent moments. The charge / discharge rate represents the ratio of the current charge / discharge current to the battery's rated capacity. The system calculates this by dividing the real-time current value by the battery's rated capacity. The current state of charge (SOC) is obtained using the battery management system's coulomb counting algorithm, which calculates the battery's state of charge based on the integration of the battery's charge and discharge current. The system performs cross-correlation analysis on these three parameters with the power coupling quantification index and the system response characteristic index. Cross-correlation analysis is used to quantify the degree of mutual influence between different parameters. The specific calculation process involves performing a sliding window correlation calculation on the numerical sequence of each battery parameter with the power coupling quantification index sequence to obtain a correlation coefficient matrix. The system also calculates the correlation coefficients between each parameter and the system response characteristic index to form another correlation coefficient matrix. The deviation amplification factor is obtained by performing element-wise multiplication of two correlation coefficient matrices and summing them up. This coefficient quantifies the degree to which the power prediction deviation is amplified in the battery system.

[0026] Specifically, the calculation of thermal inertia delay compensation parameters is based on the battery's thermodynamic characteristics. The system first obtains the battery's thermal time constant and current temperature data. The thermal time constant is a physical parameter that describes the battery's temperature response speed, representing the time required for the battery's internal temperature to reach 63 percent of the external temperature change. The system obtains the standard thermal time constant value from the battery manufacturer's technical specifications and then adjusts it based on current environmental conditions and battery aging. The current temperature state is measured by temperature sensors distributed throughout the battery module. The system calculates the weighted average of all temperature measurement points as the overall temperature state. The temperature change lag time is calculated using an exponential decay function fitting method, which assumes that the battery's internal temperature changes follow an exponential decay law. The system collects historical temperature change data over a period of time and uses the least squares method to fit an exponential function to this data. The fitting process continuously adjusts the exponential function parameters through an iterative algorithm until the fitting error is minimized. The fitted exponential decay function parameters include the decay time constant and the initial amplitude. The thermal inertia delay compensation parameters are calculated by multiplying the decay time constant by the deviation from the current temperature state.

[0027] Specifically, the deviation propagation path map is constructed through matrix processing. The system arranges the deviation amplification factors into a two-dimensional matrix based on the physical layout of the battery modules. The physical layout of the battery modules includes row and column position information. The system fills the corresponding deviation amplification factors into the corresponding positions in the matrix based on the coordinate position of each battery module. The thermal inertia delay compensation parameters are also arranged into a two-dimensional matrix based on the battery module positions, with the dimensions and position correspondence of the two matrices consistent. The system performs a Hadamard product on the two matrices, multiplying the elements at corresponding positions to obtain a propagation intensity matrix that comprehensively accounts for the deviation amplification and thermal inertia effects. The system then calculates the spatial gradient of the propagation intensity matrix using the finite difference method. The gradient vector is calculated by taking the difference between the values ​​of each position in the matrix and its adjacent positions. The direction of the gradient vector indicates the main direction of the deviation propagation, and the magnitude of the gradient vector indicates the speed and intensity of the deviation propagation. The system combines the gradient vectors at all positions to form a two-dimensional vector field, which describes the propagation path and intensity distribution of the deviation in the energy storage system.

[0028] Specifically, the marking process for deviation-sensitive nodes is achieved by analyzing the gradient characteristics of a two-dimensional vector field. The system calculates the gradient magnitude, or modulus, of each location in the vector field. The gradient magnitude reflects the level of deviation propagation activity at that location; larger values ​​indicate greater sensitivity to deviation propagation. The system sets a critical threshold as a judgment criterion, determined by analyzing the statistical characteristics of deviation propagation in historical data. The specific method involves collecting the distribution of gradient magnitudes from long-term operational data, calculating the mean and standard deviation, and using the mean plus twice the standard deviation as the critical threshold. The system traverses all locations in the vector field and marks those with gradient magnitudes exceeding the critical threshold as deviation-sensitive nodes. The marking process uses a binarization method, setting locations that meet the criteria to 1 and those that do not to 0, thus forming a distribution map of deviation-sensitive nodes. Finally, the system overlays the deviation-sensitive node distribution map with the two-dimensional vector field to generate a real-time deviation propagation path map, which includes propagation paths and sensitive node labels. This path map uses color coding to intuitively display the direction and intensity of deviation propagation, while also marking the locations of deviation-sensitive nodes with special symbols, creating a comprehensive display of the deviation propagation situation.

[0029] Furthermore, the deviation amplification coefficient and the thermal inertia delay compensation parameter are subjected to matrix operation processing to construct a two-dimensional vector field of the deviation propagation speed and direction, including: performing two-dimensional matrix arrangement processing on the deviation amplification coefficient according to the spatial distribution position of the photovoltaic energy storage system to obtain a coefficient spatial distribution matrix; calculating the temperature response time constant of each layout position of the battery module according to the thermal inertia delay compensation parameter, and performing spatial difference operation processing on the temperature response time constant between adjacent layout positions to obtain a temperature propagation gradient field; performing Hadamard product operation processing on the coefficient spatial distribution matrix and the temperature propagation gradient field to obtain a deviation propagation intensity distribution considering thermodynamic constraints; performing spatial gradient calculation processing on the deviation propagation intensity distribution between the battery module layout positions to obtain a deviation propagation direction field, and performing vector field construction processing on the deviation propagation intensity distribution and the deviation propagation direction field to obtain a two-dimensional vector field.

[0030] Specifically, the system spatially distributes the deviation amplification factors. A two-dimensional coordinate system is created based on the actual physical layout of the PV energy storage system. This coordinate system is centered at the geometric center of the energy storage system. The horizontal axis represents the east-west position, and the vertical axis represents the north-south position. Coordinate units are in meters. The system obtains the actual installation coordinates of each battery module, including the module number, horizontal and vertical coordinates. The dimensions of the spatial distribution matrix are determined by the maximum span of the energy storage system. The system calculates the bounding range of all battery module locations and then divides the spatial area into equally spaced grid cells. The grid cell size is set to half the battery module spacing to ensure that each battery module is assigned a unique grid location. The system iterates through all battery modules, determines their corresponding grid locations based on their physical coordinates, and populates the matrix with the deviation amplification factor value for each module. For grid locations without a battery module, the system uses nearest neighbor interpolation to calculate the value at that location. This interpolation process weights the distance between that location and the four nearest battery module locations and then takes the weighted average of the deviation amplification factors for these four locations. After the system spatial distribution matrix is ​​completed, each matrix element represents the degree of deviation amplification of the corresponding spatial position, and the row and column indexes of the matrix directly correspond to the spatial coordinates of the energy storage system.

[0031] Specifically, the temperature propagation gradient field is calculated based on thermal inertia delay compensation parameters and battery module thermal characteristics analysis. The system first calculates the temperature response time constant at each battery module location. The temperature response time constant represents the battery module's response speed to external temperature changes and is affected by factors such as the battery's thermal capacity, thermal resistance, and heat dissipation conditions. The system extracts a base time constant from the thermal inertia delay compensation parameters and then modifies it based on the specific heat dissipation conditions of each module. Heat dissipation assessment includes factors such as air circulation around the module, distance from adjacent modules, and external ambient temperature. The system establishes a heat dissipation scoring mechanism to quantitatively assess each module's heat dissipation environment. Modules with good heat dissipation conditions have smaller temperature response time constants, while modules with poor heat dissipation conditions have larger time constants. The modified temperature response time constants are arranged into a two-dimensional matrix based on the spatial location of the battery modules. This matrix has the same dimensions and coordinates as the coefficient spatial distribution matrix. A spatial differencing operation uses the central difference method to calculate the spatial gradient of the temperature response time constant. Specifically, the horizontal and vertical difference values ​​are calculated for each location in the matrix. The horizontal difference is calculated by subtracting the left-hand element from the right-hand element at that location, divided by twice the grid spacing. The vertical difference is calculated by subtracting the right-hand element from the left-hand element at that location, divided by twice the grid spacing. For matrix boundary locations, the system uses forward or backward differencing. The temperature propagation gradient field is composed of the horizontal and vertical difference values ​​at all locations, forming two independent gradient component matrices.

[0032] Specifically, the deviation propagation intensity distribution is calculated using the Hadamard product, an element-wise multiplication operation in matrix operations that multiplies the elements at corresponding positions in two matrices of the same dimension. The system performs Hadamard products on the coefficient spatial distribution matrix and the two component matrices of the temperature propagation gradient field. Because the temperature propagation gradient field consists of horizontal and vertical components, the system first calculates the gradient field's magnitude matrix by adding the squares of the horizontal and vertical components at each position and taking the square root. After obtaining the gradient magnitude matrix, the system performs a Hadamard product on it with the coefficient spatial distribution matrix, resulting in the deviation propagation intensity distribution matrix. Each element in this matrix represents the deviation propagation intensity at the corresponding spatial location, and the magnitude reflects the combined impact of the deviation amplification effect and thermal propagation effect at that location. The deviation propagation intensity distribution takes into account the spatial heterogeneity of the energy storage system. Different locations exhibit different deviation propagation intensities due to differences in battery characteristics, heat dissipation conditions, and spatial layout. The system performs range checking and outlier processing on the deviation propagation intensity distribution to ensure that all elements in the matrix are within a reasonable range.

[0033] Specifically, the deviation propagation direction field is calculated by performing a spatial gradient operation on the deviation propagation intensity distribution. This gradient operation determines the direction of the most dramatic intensity change, which is the primary direction of deviation propagation. The system uses the same central difference method used for temperature gradient calculation to calculate the horizontal and vertical gradient components of the deviation propagation intensity distribution matrix. The horizontal gradient component reflects the east-west trend of deviation propagation, while the vertical gradient component reflects the north-south trend of deviation propagation. The system combines these two gradient components to form a direction vector. Each spatial location corresponds to a direction vector. The direction of the vector indicates the primary direction of deviation propagation at that location, and the length of the vector indicates the strength of the directionality. Once the deviation propagation direction field is constructed, the system uses the deviation propagation intensity distribution as the magnitude information of the vector field and the deviation propagation direction field as the direction information of the vector field. These two components are combined to form a complete two-dimensional vector field. The vector field construction process uses a vector synthesis method, multiplying the intensity value at each spatial location by the corresponding direction vector to obtain the propagation vector at that location. The vector at each location in the two-dimensional vector field contains horizontal and vertical components. The vector's modulus indicates the velocity strength of the deviation propagation, and the vector's angle indicates the direction of the deviation propagation. The system performs a continuity check on the constructed two-dimensional vector field to ensure smooth and continuous changes between adjacent vectors, avoiding undesirable mutations. Vector field data is stored in a three-dimensional array structure, where the first and second dimensions correspond to the spatial position coordinates, and the third dimension stores the horizontal and vertical components of the position vector.

[0034] 103. Based on the real-time feature data set and the deviation propagation path diagram, the prediction deviation accumulation process is dynamically deduced at multiple levels to obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend; In one embodiment of the present invention, the multi-level dynamic deduction processing of the predicted deviation accumulation process is performed according to the real-time feature data set and the deviation propagation path diagram to obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend, including: performing a three-level decomposition processing of the predicted deviation accumulation process at the first level, the second level and the third level according to the real-time feature data set and the deviation propagation path diagram; at the first level, combining the deviation sensitive node information in the deviation propagation path diagram, performing a minimum value calculation processing under the constraint conditions on the power climbing rate, the available capacity margin and the real-time temperature state of the photovoltaic energy storage system, and obtaining the instantaneous deviation absorption capacity; at the second level, based on the system In response to the characteristic indicators, a linear regression fitting process is performed on the deviation accumulation trend within the past preset time, and a rolling time domain integral calculation process is performed in combination with the load forecast value within the future preset time to obtain the deviation accumulation forecast value; at the third level, the power coupling quantification indicator in the real-time characteristic data set is combined to perform a time series matching calculation process on the all-day illumination change curve and the energy storage charging and discharging plan to obtain the intraday energy balance deviation index; the instantaneous deviation absorption capacity, deviation accumulation forecast value and intraday energy balance deviation index of the three levels are superimposed in time series, and graded and colored according to the risk level classification algorithm to obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend.

[0035] Specifically, the system divides the processing process into different timescales based on the temporal characteristics of deviation accumulation. The first level focuses on transient response analysis from seconds to minutes, the second level handles medium-term trend forecasting from minutes to hours, and the third level is responsible for long-term balance assessment from hours to days. The principle of this three-level decomposition is that deviation accumulation exhibits different physical mechanisms and influencing factors at different timescales. In the short term, it is primarily constrained by the response capability of the energy storage system, in the medium term by the system's dynamic characteristics, and in the long term by the energy balance strategy. The system establishes independent data processing channels for each level and establishes an information exchange mechanism between levels to ensure consistent processing results. The data processing channels adopt a pipeline structure. The real-time feature dataset and the deviation propagation path map are simultaneously fed into three processing channels. Each channel extracts a corresponding data subset for analysis based on its timescale characteristics. Information exchange between levels is achieved through a shared memory area. The processing results of the upper level serve as constraints or correction parameters for the lower level, ensuring the coordination of multi-level analysis.

[0036] Specifically, the first-level instantaneous deviation absorption capacity calculation process incorporates information about deviation-sensitive nodes in the deviation propagation path graph. The system first extracts the locations of all nodes marked as deviation-sensitive and their corresponding sensitivity values ​​from the deviation propagation path graph. The sensitivity of a deviation-sensitive node is represented by its gradient magnitude in the path graph; a larger value indicates a node's greater sensitivity to deviation changes. The system also calculates the power ramp rate of the PV energy storage system, which represents the maximum change in the energy storage system's power output per unit time. This parameter is calculated by analyzing the maximum rate of change in historical power change data. The available capacity margin is calculated by the difference between the current SOC and the safe operating boundary. The system sets upper and lower safety boundaries at 90% and 10%, respectively. The current available charging capacity is equal to the upper boundary minus the current SOC value, multiplied by the battery's rated capacity. The available discharging capacity is equal to the current SOC value minus the lower boundary, multiplied by the battery's rated capacity. The real-time temperature status is calculated by taking a weighted average of the temperature measurements at all deviation-sensitive nodes, with the weight coefficient proportional to each node's sensitivity. The minimum value calculation under the constraints utilizes a multi-constraint optimization method. The system considers the power ramp rate, available capacity margin, and temperature constraint coefficient as three independent constraints. The system then calculates the maximum deviation the system can absorb if all three constraints are met simultaneously. The temperature constraint coefficient is determined by the relationship between the current temperature and the safe operating temperature range. The constraint coefficient is smaller when the temperature approaches the upper limit and larger when the temperature is within the normal range. The instantaneous deviation absorption capacity is equal to the minimum of the calculated results of the three constraints. This value represents the deviation magnitude that the energy storage system can immediately respond to and absorb under the current state.

[0037] Specifically, the second-level cumulative deviation forecast calculation is based on system response characteristic indicators and time series analysis methods. The system extracts historical time series data for system response characteristic indicators from real-time characteristic datasets, with a time span set to a preset time period (typically 15 minutes to one hour). Linear regression fitting utilizes a least-squares algorithm, using time as the independent variable and cumulative deviation as the dependent variable. The system determines the optimal linear function parameters by minimizing the squared error between the predicted and actual values. During the fitting process, the system uses the system response characteristic indicator as a weighting factor, assigning higher weights to time periods with good response characteristics and lower weights to time periods with poor response characteristics. The fitted linear function contains two parameters: the slope and the intercept. The slope represents the trend of cumulative deviation, while the intercept represents the baseline deviation level. A rolling time-domain integral calculation combines future load forecast information to obtain a load power forecast series for a preset future time period, typically spanning one to four hours. The integral calculation utilizes a trapezoidal numerical integration algorithm, dividing the forecast period into several small time intervals. The cumulative deviation within each time interval is calculated using a linear function, and the cumulative deviations for all time intervals are then summed to obtain the total cumulative forecast value. The rolling horizon is implemented through a sliding window mechanism. Whenever a new time point arrives, the system automatically updates the forecast start time and recalculates the cumulative deviation forecast for the future time period. The cumulative deviation forecast reflects the cumulative development trend of deviations in the system over the medium-term time scale.

[0038] Specifically, the third-level calculation of the intraday energy balance deviation index combines a power coupling metric with a time-series matching analysis. The system first obtains a 24-hour daily sunlight variation forecast curve and the energy storage system's charge and discharge schedule curve. The sunlight variation curve is calculated using meteorological forecast data and a photovoltaic power generation model, reflecting the daily variations in solar radiation intensity. The energy storage charge and discharge schedule curve is developed based on load demand forecasts and grid dispatch requirements, describing the expected charge and discharge power of the energy storage system in each time period. The time-series matching calculation utilizes a cross-correlation function. The system time-aligns the sunlight variation curve and the charge and discharge schedule curve, then calculates the correlation coefficient between the two curves at different time offsets. The matching calculation uses the power coupling metric as a correction factor. A high coupling indicates good coordination between the photovoltaic energy storage system and a high reliability of the matching calculation. A low coupling indicates system coordination issues, requiring appropriate corrections to the matching results. The intraday energy balance deviation index is calculated by calculating the area difference between the sunlight variation curve and the charge and discharge schedule curve. This area difference indicates the degree of imbalance in energy supply and demand throughout the day. A positive area difference indicates that PV power generation exceeds energy storage charging demand, while a negative area difference indicates that energy storage discharge demand exceeds PV power generation. The system normalizes the area difference by the total daily energy demand to obtain a dimensionless daily energy balance deviation index.

[0039] Specifically, the dynamic deviation accumulation situation map is generated through time series overlay processing and a risk grading algorithm. The system aligns and overlays the instantaneous deviation absorption capacity, cumulative deviation forecast, and daily energy balance deviation index calculated at the three levels along a unified time axis. The time axis resolution is set to minute levels, and the system interpolates the calculated results at each level onto a unified time grid. The overlay process uses a weighted summation method, with the weight coefficients for the three levels determined by the distance between the current time and the forecast time range of each level, with closer levels receiving greater weights. The risk grading algorithm classifies the cumulative deviation values ​​after overlay into four risk levels: low, medium, high, and extremely high. The risk grading thresholds are determined through statistical analysis of historical data. The system analyzes the distribution characteristics of past cumulative deviation events and categorizes the cumulative deviation values ​​into different levels based on percentiles. The grading process uses a color mapping method, with low risk corresponding to green, medium risk corresponding to yellow, high risk corresponding to orange, and extremely high risk corresponding to red. The situation map is displayed as a time-series heat map, with time on the horizontal axis and spatial location or system component on the vertical axis. The color of each pixel in the map indicates the risk level at that time and location. The system also overlays trend information on the situation map, using arrows or contour lines to indicate the direction and intensity of deviation accumulation, creating a complete dynamic deviation accumulation situation map.

[0040] Furthermore, at the second level, based on the system response characteristic index, a linear regression fitting process is performed on the deviation accumulation trend within the past preset time, and a rolling time domain integral calculation process is performed in combination with the load forecast value within the future preset time to obtain the deviation accumulation prediction value, including: at the second level, according to the energy storage response delay factor in the system response characteristic index, the power forecast deviation data within the past preset time is time series rearranged; the rearranged deviation data is subjected to least squares linear regression fitting process, and the slope coefficient and intercept parameter of the deviation accumulation are calculated to obtain the deviation accumulation trend function; according to the array imbalance index in the system response characteristic index, the load forecast value within the future preset time is subjected to imbalance weighted correction process to obtain a corrected load forecast sequence; the deviation accumulation trend function is numerically integrated and calculated within the future preset time window, and a rolling time domain superposition operation is performed in combination with the corrected load forecast sequence to obtain a superposition operation result; the superposition operation result is dynamically updated according to the sliding window to obtain the deviation accumulation prediction value.

[0041] Specifically, the system extracts historical data series of the energy storage response delay factor from the real-time feature data set. This factor reflects the time delay between the energy storage system receiving control commands and its actual response. The value of the energy storage response delay factor typically varies from a few seconds to a few minutes, depending on the technical characteristics and current operating status of the energy storage device. Based on the current value of the energy storage response delay factor, the system adjusts the time axis of the power forecast deviation data for a preset period of time. The principle of time series reordering is based on the physical process of deviation propagation. When photovoltaic power generation deviations occur, the energy storage system has an inherent delay in responding, which affects the accumulation of deviations in the system. The reordering process is achieved by time-shifting the historical deviation data according to the delay time. Specifically, the deviation data at each time point is shifted forward by the corresponding delay time. The system establishes a time mapping table to record the correspondence between the original time points and the adjusted time points, ensuring the accuracy and traceability of the data reordering process. The reordered deviation data series reflects the actual time course of deviation accumulation after accounting for system response delays. This series eliminates the time deviation caused by response delays, making subsequent trend analysis more accurate.

[0042] Specifically, the cumulative deviation trend function is constructed using the least squares linear regression fitting algorithm, a classic statistical method for determining linear relationships between variables. The system uses the rearranged deviation data as the input for regression analysis, with time as the independent variable and the cumulative deviation as the dependent variable to construct a regression model. The cumulative deviation is calculated by integrating the deviation data over time. This integration process uses the trapezoidal method, dividing the time axis into equally spaced segments. The deviation values ​​within each time segment are estimated using linear interpolation, and then the cumulative sum of the deviation values ​​across all time segments is calculated. The least squares regression process involves constructing a system of normal equations and solving for the parameter vector. The system first calculates the cumulative values ​​of the linear and quadratic terms of the time variable, and then calculates the cumulative value of the product of the time variable and the cumulative deviation. The slope coefficient is calculated by dividing the covariance of the time variable and the dependent variable by the variance of the time variable. The intercept parameter is calculated by subtracting the slope coefficient from the mean of the dependent variable and multiplying it by the mean of the time variable. The quality of the regression fit is assessed by calculating the coefficient of determination, which indicates the degree to which the regression model explains the data variation. A value closer to 1 indicates a better fit. The deviation cumulative trend function adopts the form of a linear function. The slope coefficient of the function represents the rate of change of the deviation accumulation. A positive value indicates that the deviation has a cumulative growth trend, and a negative value indicates that the deviation has a decaying trend. The intercept parameter represents the baseline deviation level of the trend function.

[0043] Specifically, the load forecast sequence is corrected based on the array imbalance indicator, a system response characteristic indicator. This indicator quantifies the degree of power disparity between different strings in a photovoltaic array. Array imbalance can affect the overall stability of photovoltaic power generation, and thus the accuracy of the energy storage system's load demand forecast. The system obtains an original load forecast sequence for a preset future timeframe. This sequence is typically generated by a load forecasting algorithm based on historical load patterns and external factors. Weighted imbalance correction is based on the correlation between array imbalance and forecast uncertainty. A higher array imbalance indicates greater variability in photovoltaic power generation, and correspondingly, higher load forecast uncertainty. The correction utilizes a dynamic weight adjustment method. The system calculates a corrected weight coefficient based on the current array imbalance indicator value. Higher imbalances result in smaller weight coefficients, indicating lower confidence in the original forecast value. The weight coefficients are calculated using an inverse proportional function. The system normalizes the imbalance indicator and takes the inverse as the base weight. The system then adjusts the weights using a smoothing function to avoid drastic weight changes. The revised load forecast sequence is obtained by multiplying the original forecast value by the corresponding weight coefficient. The correction process also takes into account the time distance factor. The farther the forecast value is from the current time, the greater the correction effect. The revised load forecast sequence has better robustness and can reflect the impact of PV array imbalance on system operation.

[0044] Specifically, the cumulative deviation forecast value is calculated through numerical integration and rolling time-domain superposition. The system first integrates the cumulative deviation trend function over a preset future time window. The length of the time window is determined by the second-level forecast range and is typically one to four hours. Numerical integration employs Simpson's rule, which offers higher accuracy than the trapezoidal method and is particularly suitable for integrating smooth functions. The integration process divides the time window into an even number of equally spaced subintervals. Within each subinterval, a quadratic polynomial is used to approximate the trend function, which is then analytically integrated. The result represents the cumulative deviation expected over the future time window based on the current trend. The rolling time-domain superposition combines the trend function integration result with the revised load forecast sequence. This superposition process accounts for the impact of load changes on cumulative deviation. An increase in load increases the energy storage system's discharge demand, thereby affecting the cumulative deviation. A decrease in load reduces the energy storage system's discharge pressure, facilitating the dissipation of the deviation. The superposition operation employs a convolution integral method, convolving the difference between the trend function and the load forecast sequence to produce a comprehensive cumulative deviation forecast that accounts for load influence. The rolling time domain is implemented through a sliding window mechanism. The system sets a fixed-length forecast window, automatically updating the start and end times of the window with each time step. Dynamic update processing ensures that forecast results always reflect the latest system status and trend information. The system uses exponential smoothing to smooth continuous forecast results, reducing jumps in forecast results caused by data fluctuations. The resulting cumulative deviation forecast value represents the expected development of system deviation accumulation within the second-level time scale. This value comprehensively considers the impact of system response characteristics, historical trends, and future load changes.

[0045] 104. Based on the dynamic deviation accumulation situation diagram, an adaptive threshold warning judgment process is performed on the coordinated imbalance state of the photovoltaic energy storage system, and corresponding response control is executed according to the hierarchical warning mechanism to obtain a hierarchical warning response control instruction.

[0046] In one embodiment of the present invention, the coordinated imbalance state of the photovoltaic energy storage system is adaptively warned and judged based on the dynamic deviation accumulation situation diagram, and corresponding response control is executed according to the hierarchical warning mechanism to obtain a hierarchical warning response control instruction, including: performing multi-factor adaptive threshold calculation processing on the operating state parameters of the photovoltaic energy storage system according to the risk level value in the dynamic deviation accumulation situation diagram to obtain a dynamic warning threshold; comparing and judging the risk level value with the dynamic warning threshold, dividing the warning level according to the hierarchical warning mechanism to obtain the corresponding warning level; executing corresponding hierarchical response control processing according to different warning levels, and performing feedback learning optimization processing on the warning judgment results to obtain a hierarchical warning response control instruction including warning level and control action.

[0047] Specifically, the system uses an image processing algorithm to traverse all pixels in the situation map and identify the risk level information corresponding to different colored areas. Risk levels are coded using a four-level system: green areas correspond to a value of 1, indicating low risk; yellow areas correspond to a value of 2, indicating medium risk; orange areas correspond to a value of 3, indicating high risk; and red areas correspond to a value of 4, indicating extremely high risk. The system establishes a risk level statistical mechanism to calculate the area share and distribution pattern of each risk level area at the current moment. The area share reflects the scope of influence of different risk levels, and the distribution pattern reflects the spatial clustering characteristics of risk. The multi-factor adaptive threshold calculation process is based on multiple operating parameters of the PV energy storage system, including the system's state of charge (SOC), battery temperature distribution, PV power fluctuation, load variation, and environmental factors. SOC parameters are acquired in real time by the battery management system. The system calculates the mean and standard deviation of the SOC values ​​for all battery modules. The mean reflects the overall charge level, and the standard deviation reflects the degree of balance between modules. Battery temperature distribution parameters are acquired through a distributed temperature sensor network. The system calculates the spatial gradient and temporal rate of change of the temperature field. The spatial gradient reflects the uniformity of the temperature distribution, and the temporal rate of change reflects the activity of thermodynamic processes. The degree of PV power fluctuation is calculated by analyzing the statistical characteristics of recent power output. The system uses a sliding window method to calculate the variance and coefficient of variation of the power series. The variance reflects the absolute intensity of the fluctuation, while the coefficient of variation reflects the relative intensity of the fluctuation. The magnitude of load fluctuation is calculated using the time derivative of the load power. The system analyzes the rate and direction of change in load demand. Rapidly changing loads place higher demands on the responsiveness of the energy storage system.

[0048] Specifically, the dynamic warning threshold is calculated using a multivariate linear regression method. The system uses various operating parameters as independent variables and historical warning trigger events as dependent variables to establish a regression model. The regression model is trained using historical operating data. The system collects operating parameter values ​​and corresponding warning trigger records over a period of time. Through statistical analysis, the system determines the weight of each parameter on the warning threshold. The weight is calculated using partial correlation analysis, which eliminates interference from other variables and accurately assesses the independent impact of a single variable on the warning threshold. The system implements a dynamic weight adjustment mechanism that adjusts the weight coefficients of each parameter based on the current operating conditions. The weight of the temperature parameter is increased in high-temperature environments, while the weight of the power fluctuation parameter is increased when the light intensity fluctuates significantly. The dynamic warning threshold is calculated through a weighted summation. The calculation also includes a time correction factor, which accounts for differences in warning sensitivity at different times of the day. Typically, the warning threshold is lowered during peak hours to enhance sensitivity, while it is appropriately raised during low hours to avoid false alarms. This modified dynamic warning threshold adapts to changes in system operating conditions and offers greater adaptability and accuracy than a fixed threshold.

[0049] Specifically, the warning level classification process is implemented through a comparative judgment algorithm. The system compares the risk level value in the dynamic deviation accumulation situation map with the dynamic warning threshold. This comparison and judgment utilizes a hierarchical threshold mechanism. The system sets four different threshold levels corresponding to four warning levels: the first threshold corresponds to normal status, the second threshold corresponds to caution status, the third threshold corresponds to warning status, and the fourth threshold corresponds to emergency status. The threshold levels are set based on the baseline value of the dynamic warning threshold. The system scales the baseline value according to different scaling factors to obtain the threshold values. The scaling factors are determined based on statistical analysis of historical warning events. The comparative judgment process utilizes a hysteresis comparator principle, which can avoid frequent state switching near threshold boundaries. The hysteresis comparator sets two values: a rising threshold and a falling threshold. When the risk level value crosses the rising threshold from low to high, the state is upgraded; when the value crosses the falling threshold from high to low, the state is downgraded. The difference between the two thresholds forms the hysteresis interval. The system also implements a time confirmation mechanism, requiring the risk level value to remain within the new threshold interval for a certain period of time before confirming a state change, to avoid misjudgments caused by transient fluctuations. The results of the warning level classification are managed using a state machine model, which records information such as the current warning level, state change history, and state duration.

[0050] Specifically, the hierarchical response control process implements corresponding control strategies based on different warning levels. The system establishes a mapping table between warning levels and control actions. Under normal conditions, the system implements conventional optimization control strategies, primarily aiming to maintain efficient operation and energy balance of the PV energy storage system. Under caution conditions, the system initiates preventive control measures, including adjusting the energy storage system's charging and discharging strategies, optimizing the PV array's operating point settings, and increasing system monitoring frequency. Under warning conditions, the system implements proactive intervention control, including limiting the energy storage system's power output range, activating backup energy storage units, and adjusting load priority. Under emergency conditions, the system implements protective control measures, including disconnecting some PV strings, initiating emergency discharge procedures, and shutting off power to non-critical loads. Control actions are executed using a priority queue mechanism, with the system assigning priorities based on their importance and urgency, with higher-priority actions being executed first. Feedback learning optimization optimizes system parameters by analyzing the accuracy of warning judgments and the effectiveness of control effects. The system establishes a warning effectiveness evaluation mechanism to track the development and final outcome of each warning event and assess the timeliness, accuracy, and necessity of the warning. Timeliness is assessed by comparing the difference between the warning trigger time and the actual problem occurrence time. Accuracy is assessed by comparing the warning prediction results with the actual situation. Necessity is assessed by evaluating the effectiveness of the control measures implemented after the warning. The system uses a reinforcement learning algorithm to optimize and adjust warning parameters. The algorithm adjusts parameters such as threshold settings, weight distribution, and control strategy selection based on the evaluation results of the warning effect. The generation of hierarchical warning response control instructions uses a structured data format. The instructions contain information such as the warning level, trigger time, risk assessment, recommended control action, execution priority, and expected effect. These instructions are sent to the various subsystems of the photovoltaic energy storage system via the communication interface to execute the corresponding control operations.

[0051] In this embodiment, by performing multi-time-scale sliding window feature extraction on the photovoltaic energy storage system operating data, power coupling quantification indicators and system response characteristic indicators are calculated to generate a real-time feature data set; based on the real-time feature data set, deviation propagation paths are identified under thermodynamic constraints to obtain a real-time deviation propagation path diagram that marks deviation-sensitive nodes; multi-level dynamic deduction is performed in combination with the real-time feature data set and the deviation propagation path diagram to obtain a dynamic deviation accumulation situation diagram that displays the deviation accumulation risk level and development trend; adaptive threshold warning judgment is performed based on the dynamic deviation accumulation situation diagram, and response control is executed according to the hierarchical warning mechanism to obtain a hierarchical warning response control instruction. The present invention can effectively identify the cumulative effect of power forecast deviations and achieve early warning of coordinated imbalances in photovoltaic energy storage systems.

[0052] The above describes the intelligent monitoring and early warning method of photovoltaic energy storage equipment in the embodiment of the present invention. The following describes the intelligent monitoring and early warning device of photovoltaic energy storage equipment in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, an intelligent monitoring and early warning device for photovoltaic energy storage equipment includes: A feature extraction module 201 is configured to perform feature extraction processing on the operating data of the photovoltaic energy storage system based on a multi-time-scale sliding window, calculate a power coupling quantification index and a system response characteristic index, and generate a real-time feature data set based on the power coupling quantification index and the system response characteristic index; A path identification module 202 is configured to perform a deviation propagation path identification process based on thermodynamic constraints on the photovoltaic energy storage system according to the real-time feature data set, and obtain a real-time deviation propagation path map with deviation-sensitive nodes marked; A situation deduction module 203 is configured to perform multi-level dynamic deduction processing on the prediction deviation accumulation process based on the real-time feature data set and the deviation propagation path diagram, and obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend; The early warning control module 204 is used to perform adaptive threshold early warning judgment processing on the coordinated imbalance state of the photovoltaic energy storage system according to the dynamic deviation accumulation situation diagram, execute corresponding response control according to the hierarchical early warning mechanism, and obtain hierarchical early warning response control instructions.

[0053] In an embodiment of the present invention, the intelligent monitoring and early warning device of the photovoltaic energy storage equipment runs the intelligent monitoring and early warning method of the photovoltaic energy storage equipment described above. The intelligent monitoring and early warning device of the photovoltaic energy storage equipment generates a real-time feature data set by performing multi-time-scale sliding window feature extraction on the photovoltaic energy storage system operation data, calculating power coupling quantification indicators and system response characteristic indicators; based on the real-time feature data set, the deviation propagation path is identified under thermodynamic constraints to obtain a real-time deviation propagation path diagram that marks deviation-sensitive nodes; multi-level dynamic deduction is performed in combination with the real-time feature data set and the deviation propagation path diagram to obtain a dynamic deviation accumulation situation diagram that displays the deviation accumulation risk level and development trend; adaptive threshold warning judgment is performed based on the dynamic deviation accumulation situation diagram, and response control is executed according to the hierarchical warning mechanism to obtain a hierarchical warning response control instruction. The present invention can effectively identify the cumulative effect of power forecast deviations and achieve early warning of coordinated imbalances in photovoltaic energy storage systems.

[0054] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0055] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0056] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent monitoring and early warning method for photovoltaic energy storage equipment, characterized in that: The intelligent monitoring and early warning method for photovoltaic energy storage equipment includes: Performing feature extraction processing on the operating data of the photovoltaic energy storage system based on a multi-time scale sliding window, calculating a power coupling quantification index and a system response characteristic index, and generating a real-time feature data set based on the power coupling quantification index and the system response characteristic index; Based on the real-time feature data set, a deviation propagation path identification process is performed on the photovoltaic energy storage system based on thermodynamic constraints to obtain a real-time deviation propagation path map with deviation-sensitive nodes marked; Based on the real-time feature data set and the deviation propagation path diagram, a multi-level dynamic deduction process is performed on the prediction deviation accumulation process to obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend; According to the dynamic deviation accumulation situation diagram, an adaptive threshold warning judgment process is performed on the coordinated imbalance state of the photovoltaic energy storage system, and corresponding response control is executed according to the hierarchical warning mechanism to obtain a hierarchical warning response control instruction.

2. The intelligent monitoring and early warning method for photovoltaic energy storage equipment according to claim 1 is characterized in that: The performing of feature extraction processing on the operation data of the photovoltaic energy storage system based on a multi-time scale sliding window, calculating a power coupling quantification index and a system response characteristic index, and generating a real-time feature data set based on the power coupling quantification index and the system response characteristic index comprises: Perform multi-time-scale sliding window processing on the power generation data, charge and discharge power data, and battery SOC status data of each group of photovoltaic energy storage systems, and calculate the power change rate within each time window, where the power change rate includes the photovoltaic power generation power change rate and the energy storage response power change rate; The Pearson correlation coefficient of the photovoltaic power generation rate of change and the energy storage response power change rate is calculated to obtain a quantitative index of power coupling; The time difference between the time the PV energy storage system issues a command and the time the power output responds is calculated to obtain the energy storage response delay factor. A variance analysis is then performed on the power generated by strings in different orientations to calculate the array imbalance index. The energy storage response delay factor and the array imbalance index are combined to obtain a system response characteristic index, and the power coupling quantization index and the system response characteristic index are combined in time series to obtain a real-time characteristic data set.

3. The intelligent monitoring and early warning method for photovoltaic energy storage equipment according to claim 1 is characterized in that: The step of performing a deviation propagation path identification process based on thermodynamic constraints on the photovoltaic energy storage system according to the real-time feature data set to obtain a real-time deviation propagation path graph with deviation-sensitive nodes marked includes: The difference between the actual power and the predicted power of the photovoltaic power generation system is calculated to obtain the power prediction deviation; Calculating the charging and discharging power adjustment requirements of the photovoltaic energy storage system based on the power prediction deviation, and quantitatively analyzing the battery internal resistance change rate, charge and discharge rate, and current SOC state response degree in combination with the power coupling quantification index and the system response characteristic index to obtain a deviation amplification factor; According to the battery thermal time constant and the current temperature state, the temperature change lag time is calculated by fitting the exponential decay function to obtain the thermal inertia delay compensation parameter; Performing matrix operations on the deviation amplification factor and the thermal inertia delay compensation parameter to construct a two-dimensional vector field of the deviation propagation speed and direction; Nodes whose gradient values ​​in the two-dimensional vector field exceed a critical threshold are marked to obtain a real-time deviation propagation path diagram of the marked deviation-sensitive nodes.

4. The intelligent monitoring and early warning method for photovoltaic energy storage equipment according to claim 3 is characterized in that: The performing matrix operation on the deviation amplification factor and the thermal inertia delay compensation parameter to construct a two-dimensional vector field of the deviation propagation speed and direction includes: Performing a two-dimensional matrix arrangement processing on the deviation amplification coefficient according to the spatial distribution position of the photovoltaic energy storage system to obtain a coefficient spatial distribution matrix; Calculating the temperature response time constant of each layout position of the battery module according to the thermal inertia delay compensation parameter, and performing spatial difference operation processing on the temperature response time constant between adjacent layout positions to obtain a temperature propagation gradient field; Performing a Hadamard product operation on the coefficient space distribution matrix and the temperature propagation gradient field to obtain a deviation propagation intensity distribution considering thermodynamic constraints; The deviation propagation intensity distribution is spatially calculated and processed between battery module layout positions to obtain a deviation propagation direction field, and the deviation propagation intensity distribution and the deviation propagation direction field are vector-constructed to obtain a two-dimensional vector field.

5. The intelligent monitoring and early warning method for photovoltaic energy storage equipment according to claim 1 is characterized in that: The multi-level dynamic deduction process of the predicted deviation accumulation process is performed based on the real-time feature data set and the deviation propagation path diagram to obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend, including: According to the real-time feature data set and the deviation propagation path diagram, the prediction deviation accumulation process is decomposed into three levels: the first level, the second level, and the third level; At the first level, the deviation-sensitive node information in the deviation propagation path diagram is combined to perform a minimum calculation under the constraints of the power ramp rate, available capacity margin, and real-time temperature status of the photovoltaic energy storage system to obtain the instantaneous deviation absorption capacity. At the second level, based on the system response characteristic index, a linear regression fitting process is performed on the deviation accumulation trend within the past preset time, and a rolling time domain integral calculation process is performed on the load forecast value within the future preset time to obtain a deviation accumulation forecast value; At the third level, the power coupling quantification index in the real-time feature data set is combined to calculate the time series matching between the all-day illumination change curve and the energy storage charging and discharging plan to obtain the intraday energy balance deviation index; The instantaneous deviation absorption capacity, deviation cumulative prediction value and intraday energy balance deviation index of the three levels are superimposed in time series and graded and colored according to the risk level classification algorithm to obtain a dynamic deviation cumulative situation diagram showing the deviation cumulative risk level and development trend.

6. The intelligent monitoring and early warning method for photovoltaic energy storage equipment according to claim 5 is characterized in that: At the second level, based on the system response characteristic index, a linear regression fitting process is performed on the deviation accumulation trend within the past preset time, and a rolling time domain integral calculation process is performed on the load forecast value within the future preset time to obtain the deviation accumulation forecast value, which includes: At the second level, a time series rearrangement process is performed on the power forecast deviation data within a preset time period in the past according to the energy storage response delay factor in the system response characteristic index; The rearranged deviation data were subjected to least squares linear regression fitting, the slope coefficient and intercept parameter of the deviation accumulation were calculated, and the deviation accumulation trend function was obtained; According to the array imbalance index in the system response characteristic index, a weighted correction process is performed on the load prediction value within a preset time in the future to obtain a corrected load prediction sequence; Performing numerical integration calculation on the deviation cumulative trend function within a future preset time window, and performing rolling time domain superposition calculation on the modified load forecast sequence to obtain a superposition calculation result; The superposition operation result is dynamically updated according to the sliding window to obtain a deviation cumulative prediction value.

7. The intelligent monitoring and early warning method for photovoltaic energy storage equipment according to claim 1, characterized in that: According to the dynamic deviation accumulation situation diagram, the photovoltaic energy storage system coordinated imbalance state is subjected to adaptive threshold warning judgment processing, and corresponding response control is executed according to the hierarchical warning mechanism to obtain the hierarchical warning response control instructions, including: According to the risk level value in the dynamic deviation accumulation situation diagram, a multi-factor adaptive threshold calculation process is performed on the operating status parameters of the photovoltaic energy storage system to obtain a dynamic warning threshold; Compare and judge the risk level value with the dynamic warning threshold, and classify the warning levels according to the hierarchical warning mechanism to obtain the corresponding warning level; According to different warning levels, corresponding hierarchical response control processing is executed, and the warning judgment results are subjected to feedback learning optimization processing to obtain hierarchical warning response control instructions including warning levels and control actions.

8. An intelligent monitoring and early warning device for photovoltaic energy storage equipment, characterized in that: The intelligent monitoring and early warning device of the photovoltaic energy storage equipment includes: a feature extraction module for performing feature extraction processing on the operating data of the photovoltaic energy storage system based on a multi-time-scale sliding window, calculating a power coupling quantification index and a system response characteristic index, and generating a real-time feature data set based on the power coupling quantification index and the system response characteristic index; a path identification module, configured to perform a deviation propagation path identification process based on thermodynamic constraints on the photovoltaic energy storage system according to the real-time feature data set, and obtain a real-time deviation propagation path diagram that marks deviation-sensitive nodes; A situation deduction module is used to perform multi-level dynamic deduction processing on the prediction deviation accumulation process based on the real-time feature data set and the deviation propagation path diagram, and obtain a dynamic deviation accumulation situation diagram showing the deviation accumulation risk level and development trend; The early warning control module is used to perform adaptive threshold early warning judgment processing on the coordinated imbalance state of the photovoltaic energy storage system according to the dynamic deviation accumulation situation diagram, execute corresponding response control according to the hierarchical early warning mechanism, and obtain hierarchical early warning response control instructions.

Images