A load monitoring method and system for photovoltaic energy storage batteries

Abstract

The present invention relates to the field of intelligent monitoring of photovoltaic energy storage, and specifically to a load monitoring method and system for photovoltaic energy storage batteries. The method collects multi-dimensional operating data such as current, voltage, temperature and power of photovoltaic energy storage batteries, and fuses and normalizes them into unified "electric-thermal parameters". Next, a dynamic variance model is constructed. The model evaluates a baseline variance reflecting the macro environment based on the real-time output power of the photovoltaic array, and calculates the dynamic fluctuation term in combination with recent data fluctuations. The two are fused through dynamic weights to obtain a dynamic variance that can adapt to the current working conditions. Finally, within the framework of the random outlier selection SOS algorithm, this dynamic variance is applied to calculate the similarity of data points, and ultimately the abnormal probability of the load is obtained, thereby achieving accurate and adaptive monitoring of the load status. The present invention can effectively improve the accuracy and reliability of monitoring.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring of photovoltaic energy storage, and in particular to a load monitoring method and system for photovoltaic energy storage batteries. Background Art

[0002] Photovoltaic energy storage technology is a key component of renewable energy development, and the safe and stable operation of its core component, photovoltaic energy storage batteries, is crucial. Photovoltaic energy storage battery systems typically operate in complex and volatile environments, such as outdoors. Their operating status is influenced not only by their own charging and discharging behavior but also by external factors such as light intensity, ambient temperature, and load demand. Therefore, real-time and accurate monitoring of the load status of photovoltaic energy storage batteries is a core technical requirement for early warning of potential failures, ensuring system safety, and extending battery life. However, establishing an accurate load monitoring model for photovoltaic energy storage batteries presents several challenges. These include: battery operating status is determined by multiple physical quantities, which are multi-dimensional and have varying dimensions, making them difficult to assess uniformly; and the highly dynamic nature of system operating conditions. For example, clear and cloudless weather can lead to significant differences in photovoltaic output power; and load switching can also cause instantaneous fluctuations in system parameters. Traditional load monitoring methods typically use fixed thresholds or static models, which are unsuitable for complex and variable real-world operating conditions. Under conditions where the light and load are stable, its monitoring sensitivity may be insufficient to detect early minor anomalies in a timely manner; however, when the light fluctuates violently or the load switches frequently, it is easy to misjudge normal system fluctuations as faults, resulting in a large number of false alarms and reducing the reliability of the monitoring system.

[0003] To address these issues, some technical solutions have proposed using anomaly detection algorithms (such as the random outlier selection (SOS) algorithm) for load monitoring. These algorithms determine whether a data point is an outlier by calculating its similarity with neighboring data points. However, traditional SOS algorithms typically rely on a fixed variance parameter to define the "closeness" relationship between data points when calculating similarity. This static variance setting is a core flaw, as it fails to reflect the "normal fluctuation range" of the dynamic PV system. For example, a strict variance set for stable sunny conditions will be overly sensitive and generate false alarms under fluctuating cloudy conditions. Conversely, a loose variance set for cloudy conditions will be insufficiently sensitive and miss true fault signs on sunny days. Therefore, this monitoring method, which lacks adaptive operating conditions, struggles to maintain high sensitivity while effectively suppressing false alarms, and cannot meet the high-precision and high-reliability requirements of PV energy storage battery load monitoring. Summary of the Invention

[0004] In response to the above-mentioned problem of difficulty in effectively suppressing false alarms while ensuring high sensitivity, in the first aspect, the present invention proposes a load monitoring method for photovoltaic energy storage batteries, including: collecting multi-dimensional operating data including real-time current, voltage, temperature and instantaneous power of the photovoltaic energy storage battery, and fusing and normalizing them to construct an electro-thermal parameter that can comprehensively characterize the comprehensive operating state of the battery at any time; calculating the dynamic variance of the electro-thermal parameter, and the calculation method includes: evaluating the baseline variance associated with the actual output power of the current photovoltaic array, and the baseline variance is used to quantify the system basic fluctuation level determined by the macroscopic lighting conditions. ; Calculate the dynamic fluctuation term of the electric-thermal parameter within a preset time window, and the dynamic fluctuation term is used to quantify the intensity of the short-term coordinated fluctuation of the system caused by factors such as load changes; Use a dynamic weight coefficient determined by the recent power fluctuation degree to perform a weighted combination of the baseline variance and the dynamic fluctuation term to obtain the dynamic variance; Apply the dynamic variance to the Gaussian kernel function as a criterion for judging dynamic changes, and calculate the similarity between the current electric-thermal parameter and other electric-thermal parameters within the time window; Based on the average value of the similarity, obtain the abnormal probability of the current photovoltaic energy storage battery load state to achieve load monitoring.

[0005] Furthermore, the calculation method of the probability of load anomaly is specifically as follows:

[0006] ;

[0007] in Indicates the Abnormal probability of electric-thermal parameters at each moment; Indicates the The electro-thermal parameters at the moment Similarity of electrical-thermal parameters at each moment; Indicates the length of the preset time window.

[0008] Compared to simple methods that directly set thresholds based on distance or similarity, this method provides a more intuitive and standardized anomaly score by calculating the average similarity between a data point and all points in a window and converting it into a probability value. The degree of anomaly of a point no longer depends on its relationship to a specific point, but rather on its degree of sociality within its local neighborhood. This method is more sensitive to isolated outliers, and the output probability value facilitates the setting of unified and standardized alarm thresholds.

[0009] Furthermore, the similarity is calculated as follows:

[0010] ;

[0011] in Indicates the The electro-thermal parameters at the moment Similarity of electrical-thermal parameters at each moment; represents the square of the Euclidean distance between two electro-thermal parameter vectors in the feature space; Indicates the The dynamic variance at the moment; represents the natural exponential function.

[0012] The present invention uses dynamic variance As the bandwidth parameter of the Gaussian kernel function, it is a significant improvement to the SOS algorithm that uses a fixed bandwidth. In the traditional method, the scale of the two points is fixed. In the present invention, this scale (i.e., dynamic variance) changes in real time. When the system is stable ( Small), a small data difference will also be judged as "far", with low similarity; when the system fluctuates ( The same data difference may be judged as “close” and have high similarity.

[0013] Furthermore, the dynamic variance is calculated as follows:

[0014] ;

[0015] in Indicates the Dynamic variance corresponding to the electro-thermal parameters at each moment; Represents the dynamic weight coefficient, which is used to balance the contribution of baseline variance and dynamic fluctuation term; represents the baseline variance; represents the dynamic fluctuation term, which is used to quantify the recent and short-term coordinated fluctuation intensity of various parameters within the time window; Indicates the length of the preset time window; and Respectively Time electro-thermal parameters The variance of and its corresponding Euclidean norm.

[0016] Compared with the existing technology that uses fixed variance parameters, the dynamic variance calculation formula proposed in this invention uses dynamic weight coefficients It intelligently balances the baseline variance reflecting the macro-environment with the dynamic fluctuation term reflecting recent operational fluctuations. This weighted fusion mechanism allows the variance to be determined by taking into account both the long-term system context (such as weather conditions) and recent transient dynamics (such as load changes). The resulting evaluation is more reasonable and closer to actual operating conditions than any single fixed variance or variance that only considers a single factor, laying the core computational foundation for accurate adaptive monitoring.

[0017] Furthermore, the baseline variance is calculated as follows:

[0018] ;

[0019] in represents the baseline variance; Indicates a preset constant; Indicates the actual output power of the current photovoltaic array, which directly reflects the current light intensity level; Indicates the rated power of the photovoltaic energy storage battery as a normalized benchmark; Expressed as a natural constant An exponential function with base .

[0020] Furthermore, the dynamic weight coefficient is calculated as follows:

[0021] ;

[0022] in Represents the dynamic weight coefficient; Indicates the The variance of instantaneous power in all electrical-thermal parameters within the preset time window corresponding to the moment directly reflects the degree of recent load fluctuation; Indicates the rated power of the photovoltaic energy storage battery, which serves as the normalized benchmark; Expressed as a natural constant An exponential function with base .

[0023] Compared with the existing technology that uses fixed weights or relies on manual experience to set, the present invention uses dynamic weight coefficients Directly linked to the severity of recent power fluctuations, automatic and intelligent adjustment of weights is achieved. The value tends to the lower limit, making the monitoring standard more inclined to the baseline variance determined by the macro environment; when the power fluctuates violently, As the value approaches the upper limit, the weight of the dynamic fluctuation term is increased to adapt to the changes.

[0024] Furthermore, the fusion method of the electric-thermal parameters is specifically as follows:

[0025] ;

[0026] in Indicates the The electro-thermal parameters at the time; and Respectively Real-time current at the moment and rated current of photovoltaic energy storage battery; and Respectively represent Real-time voltage at the moment and open-circuit voltage of photovoltaic energy storage battery; 、 and Respectively Real-time battery temperature, ambient temperature, and the maximum operating temperature allowed by the battery at all times; and Respectively The instantaneous power at the moment and the rated power of the photovoltaic energy storage battery; Represents the transpose symbol of a matrix.

[0027] Furthermore, the multi-dimensional operation data is obtained by installing current sensors, voltage sensors, temperature sensors and power sensors at key nodes of the photovoltaic energy storage battery and synchronously collecting them at the same collection frequency.

[0028] Furthermore, the method further includes comparing the abnormal probability with a preset abnormal alarm threshold, and determining that the load state is abnormal when the abnormal probability is higher than the abnormal alarm threshold.

[0029] In a second aspect, the present invention provides a load monitoring system for a photovoltaic energy storage battery, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a load monitoring method for a photovoltaic energy storage battery of the present invention is implemented.

[0030] The technical effects of the present invention are:

[0031] This invention constructs a dual-mode driven dynamic variance model. First, a baseline variance associated with the actual output power of the photovoltaic array is used to automatically adjust the monitoring system's basic tolerance based on long-term, macro-environmental factors such as light intensity. Monitoring is more stringent when light intensity is high, and tolerance is higher when light intensity is low. Then, a dynamic fluctuation term is used to quantify the intensity of recent data-coordinated fluctuations, with its weight determined by the variance of recent power fluctuations. This allows the monitoring system to flexibly respond to short-term, severe system disturbances such as load switching, avoiding misinterpretation of normal fluctuations as anomalies. This achieves dynamic integration of monitoring algorithms.

[0032] This invention cleverly embeds this dynamic variance into the Gaussian kernel function of the SOS algorithm, serving as the core adjustment factor for calculating data point similarity. This eliminates the need for a fixed similarity criterion, allowing it to be tightened or relaxed dynamically in real time based on the system's current operating conditions. This fundamentally addresses the false alarm vulnerability of traditional algorithms under variable operating conditions, significantly improving the sensitivity and accuracy of anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart schematically illustrating a load monitoring method for a photovoltaic energy storage battery according to an embodiment of the present invention;

[0034] Figure 2 The figure schematically shows a structural block diagram of a load monitoring system for photovoltaic energy storage batteries according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0036] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0037] An embodiment of a load monitoring method for photovoltaic energy storage batteries:

[0038] like Figure 1 As shown, a load monitoring method for a photovoltaic energy storage battery of the present invention includes:

[0039] S1. Collect and integrate the operating data of photovoltaic energy storage batteries.

[0040] First, in order to accurately monitor the load status of photovoltaic energy storage batteries, their operating status must be fully perceived. However, the operating status of the battery is determined by multiple different physical quantities such as current, voltage, and temperature. These physical quantities have different units and dimensions, making it difficult to directly compare and calculate them uniformly. Therefore, this step aims to collect multi-dimensional data through sensors and fuse and normalize them to construct a characteristic parameter that can comprehensively and standardizedly characterize the comprehensive operating status of the battery at any time, namely the "electro-thermal parameter."

[0041] In this embodiment, specifically, current sensors, voltage sensors, temperature sensors, and power sensors are installed at key nodes of the photovoltaic energy storage battery, and all sensors are set to collect data at the same frequency (for example, Hz) synchronously collects data to ensure that key operating parameters such as current, voltage, battery temperature and instantaneous power of photovoltaic energy storage batteries are obtained at the same time. In order to capture the dynamic characteristics of the data, a predetermined length (for example, length Moving step length (for example, the sliding step length is ) time window.

[0042] In order to construct the above-mentioned electrical-thermal parameters, this embodiment uses the following formula to fuse the collected multi-dimensional data:

[0043] ;

[0044] in Indicates the The electrical-thermal parameter at the moment, which is a multidimensional vector that comprehensively reflects the electrical and thermal characteristics of the battery at that moment; and Respectively Real-time current at the moment and rated current of photovoltaic energy storage battery; and Respectively represent Real-time voltage at the moment and open-circuit voltage of photovoltaic energy storage battery; 、 and Respectively The real-time battery temperature, ambient temperature and the maximum operating temperature allowed by the battery at all times; and Respectively The instantaneous power at the moment and the rated power of the photovoltaic energy storage battery; Represents the transpose symbol of a matrix.

[0045] The above formula normalizes the data by dividing each real-time parameter by its corresponding reference value (such as rated value or limit value), eliminating the dimension difference between different physical quantities. When a certain operating parameter of the battery changes, the electro-thermal parameter vector The position in the feature space will also change accordingly. For example, when the real-time current Increase and approach the rated current When the first component of the vector Close to , which intuitively reflects in the feature space that the battery is working close to its current limit. Similarly, when the battery temperature When the value of increases, the third component will also increase, which represents the increase of battery thermal stress. For a stable system, the electro-thermal parameter vector A tight cluster will be formed in the feature space; once an anomaly occurs, the vector will deviate significantly from this normal cluster.

[0046] Exemplary explanation: At the end of a normal charge, the voltage Close to open circuit voltage , then the second term Approaching If an internal short circuit occurs at this time, the current May surge sharply, far exceeding the rated current , resulting in the first term Much greater than , while the temperature Rapidly increases. At this point, the entire electro-thermal parameter vector The position in the feature space will instantly move away from the cluster formed under normal working conditions, thus providing a clear quantitative basis for subsequent anomaly detection.

[0047] S2. Calculate the dynamic fluctuation term used to quantify the coordinated fluctuation intensity of various parameters within the time window; and evaluate the baseline variance based on the actual output power of the photovoltaic array and the rated power of the photovoltaic energy storage battery; finally, obtain the dynamic variance of the electro-thermal parameters based on the baseline variance and the dynamic fluctuation term.

[0048] In step S1, the electrical and thermal parameters that can characterize the comprehensive operating state of the photovoltaic energy storage battery at each moment are obtained. The core purpose of this step is to establish a dynamic and flexible "tolerance" standard for subsequent anomaly detection. In photovoltaic energy storage systems, there are two main types of disturbances: normal, environmentally driven system fluctuations (such as power fluctuations caused by cloud cover), for which monitoring criteria should be relaxed to avoid false alarms; and true equipment failures (such as battery aging and short circuits), for which monitoring criteria should be tightened to improve monitoring sensitivity. To achieve this goal, this example constructs a dynamic variance calculation model that combines the system's long-term steady-state characteristics with its short-term mutation characteristics.

[0049] In one embodiment, Dynamic variance corresponding to the electro-thermal parameters at each moment The specific calculation formula is as follows:

[0050] ;

[0051] in Indicates the The dynamic variance corresponding to the electro-thermal parameters at each moment is used as the evaluation criterion for subsequent similarity calculation; Represents the dynamic weight coefficient, which is used to balance the contribution of baseline variance and dynamic fluctuation term; It represents the baseline variance, which mainly reflects the basic fluctuation level of the system determined by macro and long-term factors such as current light intensity; Represents the dynamic fluctuation term, which is used to quantify the recent and short-term coordinated fluctuation intensity of various parameters within the time window; Indicates the length of the time window; and Respectively Time electro-thermal parameters The variance of and its corresponding Euclidean norm.

[0052] The above dynamic variance calculation formula is essentially a weighted average. Controlling dynamic variance When the system is running smoothly, A higher value makes More inclined to the baseline variance determined by macro lighting conditions When the system experiences severe fluctuations, The value decreases, making Taking recent dynamic fluctuations into account more, the variance is dynamically increased, that is, the judgment scale is relaxed. In order to achieve the above dynamic weighting, an indicator that can quantify the current "volatility" of the system is needed and map it to a weight coefficient. This embodiment considers that the fluctuation of system power is a good indicator of its stability. Therefore, the following dynamic weight coefficient is constructed: The calculation formula is:

[0053] ;

[0054] in Represents the dynamic weight coefficient; Indicates the The variance of instantaneous power in all electrical-thermal parameters in the time window corresponding to the moment directly reflects the degree of recent load fluctuation; Indicates the rated power of the photovoltaic energy storage battery, which serves as the normalized benchmark; Expressed as a natural constant An exponential function with base .

[0055] This formula uses the S-type function to smoothly map the variance of power fluctuations to When the system load fluctuation is small, Approaching When , the power of the exponential term approaches , ,at this time Approaching When the power fluctuation is smaller, the system is more stable. Should be close to , so that the baseline variance Together with the dynamic fluctuation term, the dynamic variance is determined; when the power fluctuation is larger, such as frequent load switching, Increase, Should be close to , increase the weight of the dynamic fluctuation term to adapt to this change.

[0056] Example: When a cloud passes by the photovoltaic array quickly, the photovoltaic output power will fluctuate violently, resulting in According to the logic of the present invention, the dynamic weight coefficient will approach , which will increase the dynamic fluctuation term in the dynamic variance The proportion in the calculation is equivalent to temporarily "relaxing" the abnormality judgment standard, thus avoiding the system from misjudging normal power fluctuations caused by weather as battery load failure. On the contrary, on a clear and cloudless afternoon, the power output is stable. Very small, will approach Make the baseline variance By evaluating the dynamic variance together with the dynamic fluctuation term, the system is maintained in a more "strict" monitoring state.

[0057] Next, we need to determine the baseline variance in the above model The baseline variance should reflect the macro-environment, especially the impact of light intensity on the normal fluctuation range of the system. Generally, the stronger the light, the more stable the system power generation, and the smaller its normal fluctuation range should be. This embodiment uses the actual output power of the photovoltaic array as a proxy indicator of light intensity and constructs the following baseline variance The calculation formula is:

[0058] ;

[0059] in represents the baseline variance; represents a preset constant, which can be set to an empirical value in this embodiment; Indicates the actual output power of the current photovoltaic array, which directly reflects the current light intensity level; Indicates the rated power of the photovoltaic energy storage battery as a normalized benchmark; Expressed as a natural constant An exponential function with base .

[0060] The above formula converts the baseline variance The actual output power of the photovoltaic array When the light intensity is high, Large value, ratio is also large, resulting in the exponential term Approaching 0, Approaching the lower limit This means that under conditions of sufficient sunlight and stable power generation, the system should have a small normal fluctuation range and higher monitoring sensitivity. On the contrary, on cloudy days or at dawn and dusk, when the light is weak, The value is small, the exponential term approaches 1, Approaching the upper limit This means that under low-light conditions, the system's inherent instability increases, so its normal fluctuation range should be moderately relaxed.

[0061] Exemplary explanation: Assuming the rated power of the photovoltaic energy storage battery The actual output power of the photovoltaic array under strong sunlight at noon is 5kW. It may reach 4.5kW, at which point the exponential term The value of is small, and the baseline variance will be very close to its lower limit , the system is in a high-sensitivity monitoring state. On a cloudy afternoon, the actual output power Maybe only kW, the exponential term The value of is large (approximately ), baseline variance will approach its upper limit , the system's monitoring tolerance is higher to adapt to unstable power generation under weak light conditions.

[0062] S3. According to the dynamic variance corresponding to the electric-thermal parameters at any moment, the probability of abnormal load of the photovoltaic energy storage battery at that moment is obtained to realize the monitoring of the photovoltaic energy storage battery load.

[0063] This step is to apply the dynamic variance calculated above Within the framework of the SOS algorithm, the probability of a load anomaly is ultimately determined. The core idea is that whether a data point is an anomaly depends on how well it fits in with its neighboring data points. To quantify this degree of "fitness," we first need to define how to calculate the similarity between two points.

[0064] In one embodiment, this step can be completed in a load monitoring module with a built-in SOS algorithm. The module first calculates the current electrical-thermal parameters based on the Gaussian kernel function. The electric-thermal parameters at other times in its time window Similarity , and the key point is that the similarity calculation here uses dynamic variance :

[0065] ;

[0066] in Indicates the The electro-thermal parameters at the moment The similarity of the electric-thermal parameters at the moment is in the range between; It represents the square of the Euclidean distance between two electro-thermal parameter vectors in the feature space, which directly measures the degree of difference between them; Indicates the value calculated in step S2, corresponding to Dynamic variance at each moment; represents the natural exponential function.

[0067] The above formula converts the distance between two points into similarity. The key adjustment is the dynamic variance in the denominator. .when Very small (system stable, strict monitoring), even a small distance It will also cause the exponential part to become a large negative number, making the similarity Declined sharply, approaching On the contrary, when When the distance is large (system fluctuations, loose monitoring), the similarity calculated by the same distance will be significantly higher. Essentially, the dynamic variance Dynamically defines the relative concepts of "far" and "near".

[0068] Exemplary explanation: Assuming the distance between two electro-thermal parameter points Calculated as At a stable moment, the dynamic variance is calculated It may be very small, such as 0.1. In this case, the exponential part of the similarity formula is ,but is a very small value, and the similarity is almost However, if the load fluctuates violently, the dynamic variance Increased to 0.8, the exponential part becomes ,but Approximately equal to , which is a high similarity. It can be seen that the dynamic variance Significantly changes the system's evaluation criteria for differences in the same data.

[0069] After obtaining the pairwise similarity, we need to comprehensively evaluate the overall similarity of a particular point with all points in the entire time window to determine its anomaly probability. If a data point is "out of tune" with most other points in the window (i.e., its similarity is generally low), then it is highly likely to be an anomaly. Based on this, this embodiment uses the following formula to calculate the final anomaly probability:

[0070] ;

[0071] in Indicates the The final abnormal probability of the electric-thermal parameters at the moment, the value range is between [0,1); Indicates the Electrothermal parameters at the time The average similarity to all electro-thermal parameters (including itself) within the time window; Indicates the length of the time window.

[0072] The average similarity in the above formula represents the data point The degree of “sociability”. If is a normal point, it will be very similar to the adjacent normal points, resulting in a high average similarity (approaching ), then the result of 1-(high average similarity) is a very low probability of anomaly (Close to ). On the contrary, if is an outlier point, which is very different from the surrounding normal points, resulting in a low average similarity (approaching ), then the result of 1−(low average similarity) is a high probability of anomaly (Close to ).

[0073] Example description:

[0074] Assuming the time window length For a normal electro-thermal parameter point , which is different from other The similarity of each point may be very high, for example, the average similarity is Then the abnormal probability , far below the alarm threshold. When a fault occurs, the abnormal data points generated The similarity with other normal points in the window may be very low, for example, the average similarity is only Then the abnormal probability , which will most likely exceed the preset alarm threshold (such as ), thereby triggering an alarm.

[0075] Finally, set an early warning threshold for the calculated abnormal probability. For example, the threshold can be set to When monitoring When the system detects that the load status of the photovoltaic energy storage battery is abnormal, it will trigger the corresponding early warning device to remind the operation and maintenance personnel to conduct timely inspection and processing.

[0076] An embodiment of a load monitoring system for photovoltaic energy storage batteries:

[0077] On the other hand, the present invention also provides a load monitoring system for photovoltaic energy storage batteries. Figure 2 As shown, a load monitoring system for a photovoltaic energy storage battery includes a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, a load monitoring method for a photovoltaic energy storage battery according to the first aspect of the present invention is implemented.

[0078] A load monitoring system for a photovoltaic energy storage battery also includes other components well known to those skilled in the art, such as a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0079] In the present invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, the computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store the required information and can be accessed by an application, module, or both. Any such computer storage medium can be part of, accessible to, or connectable to a device. Any application or module described in the present invention can be implemented using computer-readable / executable instructions stored or otherwise maintained by such a computer-readable medium.

Claims

1. A load monitoring method for a photovoltaic energy storage battery, characterized in that: The method includes: collecting multi-dimensional operating data including real-time current, voltage, temperature and instantaneous power of a photovoltaic energy storage battery, and fusing and normalizing the data to construct an electro-thermal parameter that can fully characterize the comprehensive operating state of the battery at any time; Calculating the dynamic variance of the electro-thermal parameter, wherein the calculation method includes: evaluating the baseline variance associated with the current actual output power of the photovoltaic array, wherein the baseline variance is used to quantify the basic fluctuation level of the system determined by the macroscopic lighting conditions; calculating the dynamic fluctuation term of the electro-thermal parameter within a preset time window, wherein the dynamic fluctuation term is used to quantify the intensity of the short-term coordinated fluctuation of the system caused by factors such as load changes; The baseline variance and the dynamic fluctuation term are weighted together using a dynamic weight coefficient determined by the degree of recent power fluctuation to obtain the dynamic variance; the dynamic variance is applied to a Gaussian kernel function as a criterion for evaluating dynamic changes to calculate the similarity between the current electro-thermal parameter and other electro-thermal parameters within the time window; Based on the average value of the similarities, the abnormal probability of the current photovoltaic energy storage battery load state is obtained to achieve load monitoring.

2. A load monitoring method for photovoltaic energy storage batteries according to claim 1, characterized in that: The calculation method of the probability of load anomaly is specifically as follows: ； in Indicates the Abnormal probability of electric-thermal parameters at each moment; Indicates the The electro-thermal parameters at the moment Similarity of electrical-thermal parameters at each moment; Indicates the length of the preset time window.

3. A load monitoring method for a photovoltaic energy storage battery according to claim 2, characterized in that: The similarity calculation method is specifically as follows: ； in Indicates the The electro-thermal parameters at the moment Similarity of electrical-thermal parameters at each moment; represents the square of the Euclidean distance between two electro-thermal parameter vectors in the feature space; Indicates the The dynamic variance at the moment; represents the natural exponential function.

4. A load monitoring method for photovoltaic energy storage batteries according to claim 3, characterized in that: The calculation method of the dynamic variance is specifically as follows: ； in Indicates the Dynamic variance corresponding to the electro-thermal parameters at each moment; Represents the dynamic weight coefficient, which is used to balance the contribution of baseline variance and dynamic fluctuation term; represents the baseline variance; represents the dynamic fluctuation term, which is used to quantify the recent and short-term coordinated fluctuation intensity of various parameters within the time window; Indicates the length of the preset time window; and Respectively Time electro-thermal parameters The variance of and its corresponding Euclidean norm.

5. A load monitoring method for photovoltaic energy storage batteries according to claim 4, characterized in that: The calculation method of the baseline variance is specifically as follows: ； in represents the baseline variance; Indicates a preset constant; Indicates the actual output power of the current photovoltaic array, which directly reflects the current light intensity level; Indicates the rated power of the photovoltaic energy storage battery as a normalized benchmark; Expressed as a natural constant An exponential function with base .

6. A load monitoring method for photovoltaic energy storage batteries according to claim 4, characterized in that: The calculation method of the dynamic weight coefficient is specifically as follows: ； in Represents the dynamic weight coefficient; Indicates the The variance of instantaneous power in all electrical-thermal parameters within the preset time window corresponding to the moment directly reflects the degree of recent load fluctuation; Indicates the rated power of the photovoltaic energy storage battery, which serves as the normalized benchmark; Expressed as a natural constant An exponential function with base .

7. A load monitoring method for photovoltaic energy storage batteries according to claim 1, characterized in that: The fusion method of the electric-thermal parameters is specifically as follows: ； in Indicates the The electro-thermal parameters at the time; and Respectively Real-time current at the moment and rated current of photovoltaic energy storage battery; and Respectively represent Real-time voltage at the moment and open-circuit voltage of photovoltaic energy storage battery; 、 and Respectively The real-time battery temperature, ambient temperature and the maximum operating temperature allowed by the battery at all times; and Respectively The instantaneous power at the moment and the rated power of the photovoltaic energy storage battery; Represents the transpose symbol of a matrix.

8. A load monitoring method for photovoltaic energy storage batteries according to claim 1, characterized in that: The multi-dimensional operation data is obtained by installing current sensors, voltage sensors, temperature sensors and power sensors at key nodes of photovoltaic energy storage batteries and synchronously collecting them at the same collection frequency.

9. A load monitoring method for photovoltaic energy storage batteries according to claim 1, characterized in that: The method further includes comparing the abnormality probability with a preset abnormality alarm threshold, and determining that the load state is abnormal in response to the abnormality probability being higher than the abnormality alarm threshold.

10. A load monitoring system for photovoltaic energy storage batteries, characterized in that: The invention comprises a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a load monitoring method for a photovoltaic energy storage battery according to any one of claims 1 to 9 is implemented.

Images