A method and system for detecting abnormal state of photovoltaic power station equipment

Abstract

The present invention discloses a method and system for detecting abnormal states of photovoltaic power station equipment, which relates to the technical field of photovoltaic power station detection. The method includes: obtaining photovoltaic component operation data, ambient temperature data and historical operation data, and generating a monitoring data set. A correction factor is generated based on real-time light intensity data and historical data, and the environment is classified in combination with the ambient temperature and real-time light intensity to match the corresponding basic threshold. The degree of temperature deviation between the real-time temperature and the historical temperature mean is calculated, and the basic threshold is dynamically adjusted in combination with the correction factor to screen potential abnormal data points. The component corresponding to the potential abnormal data point is used as the target component, and the data collection neighborhood range is adjusted according to the change in real-time light intensity, and the temperature slope and power change rate of the target component are calculated. The components adjacent to the target component are determined, and the mean and standard deviation of the hot spot effect are calculated. When the mean exceeds the abnormal data screening threshold and the standard deviation is lower than the preset threshold, it is determined to be a systematic abnormality.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station detection, and in particular to a method and system for detecting abnormal states of photovoltaic power station equipment. Background Art

[0002] In the actual operation of photovoltaic power plants, detecting abnormal conditions in photovoltaic modules is a key step in ensuring power generation efficiency and equipment safety. The operating environment of photovoltaic power plants is characterized by significant dynamic changes, including seasonal temperature fluctuations, sudden changes in light intensity caused by short-term weather events, and efficiency degradation and aging of modules over time.

[0003] Currently, traditional methods for detecting abnormalities in photovoltaic power plant equipment often rely on fixed thresholds or simple statistical analysis. These methods often fail to fully consider the impact of environmental factors (such as light intensity and ambient temperature) on equipment operating parameters, leading to false or missed detections in diverse environments, such as high temperatures and strong light, or low temperatures and weak light. Summary of the Invention

[0004] In order to solve the technical problem of insufficient environmental adaptability of existing photovoltaic power station equipment abnormality detection methods, the present invention provides a photovoltaic power station equipment abnormality detection method and system.

[0005] The technical solution adopted in the present invention is:

[0006] A first aspect of the present application provides a method for detecting abnormal conditions of photovoltaic power station equipment, comprising the following steps:

[0007] S1: Obtain the operating data, ambient temperature data, and historical operating data of the photovoltaic module to generate a monitoring data set; the operating data includes the real-time temperature data, real-time power data, and real-time light intensity data of the photovoltaic module; the historical operating data includes the historical temperature data, historical power output data, and corresponding historical light intensity data of the photovoltaic module since it was put into use.

[0008] S2: Generate a correction factor based on real-time light intensity data and historical data; classify the current environment based on ambient temperature data and real-time light intensity data to obtain multiple environmental categories, and match the basic threshold corresponding to each environmental category from the pre-stored environmental category threshold library.

[0009] S3: Calculate the temperature deviation between the real-time temperature data and the average historical temperature data of the PV module; dynamically adjust the basic threshold based on the temperature deviation and the correction factor to generate an abnormal data screening threshold; based on the abnormal data screening threshold, filter out potential abnormal data points from the monitoring data set.

[0010] S4: Take the component corresponding to the potential abnormal data point as the target component, adjust the data collection neighborhood range according to the change of real-time light intensity data, and calculate the temperature slope and power change rate of the target component within the adjusted data collection neighborhood range.

[0011] S5: Determine the neighboring components of the target component within the adjusted data acquisition neighborhood, calculate the mean and standard deviation of the hot spot effect degree of the neighboring components, and determine it as a systematic abnormality when the mean exceeds the abnormal data screening threshold and the standard deviation is lower than the preset fluctuation dynamic threshold.

[0012] Preferably, generating the correction factor based on the real-time light intensity data and the historical data includes the following:

[0013] The real-time light intensity data is normalized based on the historical minimum light intensity data and the historical maximum light intensity data of the current season to obtain the normalized value of the real-time light intensity data; based on the historical power output data and the corresponding historical light intensity data, a photovoltaic module efficiency attenuation model is constructed through polynomial fitting to determine the module aging coefficient that reflects the aging degree of the photovoltaic module; the normalized value of the real-time light intensity data and the module aging coefficient are combined to generate a correction factor.

[0014] Preferably, the current environment is classified based on the ambient temperature data and the real-time light intensity data to obtain multiple environmental categories including the following:

[0015] When the ambient temperature data is higher than the preset high temperature threshold and the real-time light intensity data is higher than the preset strong light threshold, it is classified as a high temperature and strong light scene; when the ambient temperature data is lower than the preset low temperature threshold and the real-time light intensity data is lower than the preset weak light threshold, it is classified as a low temperature and weak light scene; when the change rate of the real-time light intensity data within the preset time period is greater than the preset fluctuation threshold and the ambient temperature data is in the preset temperature range, it is classified as a cloudy and fluctuating scene.

[0016] Preferably, the calculation of the temperature deviation degree between the real-time temperature data and the mean value of the historical temperature data of the photovoltaic module includes the following contents: calculating the absolute value of the difference between the mean values ​​of the real-time temperature data and the historical temperature data in the same light intensity range as the temperature deviation degree.

[0017] Preferably, the dynamic adjustment of the basic threshold based on the temperature deviation degree and the correction factor to generate the abnormal data screening threshold includes the following:

[0018] For the basic threshold values ​​corresponding to the matched environmental categories, set a first weight coefficient corresponding to the normalized value of the real-time light intensity data, a second weight coefficient corresponding to the component aging coefficient, and a third weight coefficient corresponding to the temperature deviation degree. Multiply the normalized value of the real-time light intensity data by the first weight coefficient, the component aging coefficient by the second weight coefficient, and the temperature deviation degree by the third weight coefficient to obtain their respective weighted values; sum the weighted values ​​to obtain the adjustment coefficient;

[0019] If the adjustment coefficient is greater than zero, the product of the adjustment coefficient and the preset positive adjustment ratio is added to the basic threshold to obtain the adjusted threshold; if the adjustment coefficient is less than zero, the product of the absolute value of the adjustment coefficient and the preset negative adjustment ratio is subtracted from the basic threshold to obtain the adjusted threshold;

[0020] The adjusted threshold is the abnormal data screening threshold; wherein the first weight coefficient, the second weight coefficient, the third weight coefficient, the positive adjustment ratio and the negative adjustment ratio are all preset constants.

[0021] Preferably, the method of taking the component corresponding to the potential abnormal data point as the target component, adjusting the data acquisition neighborhood range according to the change of the real-time light intensity data, and calculating the temperature slope and power change rate of the target component within the adjusted data acquisition neighborhood range includes the following:

[0022] The PV module corresponding to the potential abnormal data point is used as the target module, and the corresponding light intensity change rate is calculated in real time. When the light intensity change rate is greater than or equal to the preset threshold, the data collection neighborhood range centered on the target module is narrowed. When the light intensity change rate is less than the preset threshold, the default data collection neighborhood range is restored.

[0023] Based on the adjusted data acquisition neighborhood range, temperature data and power data of the target component within the adjusted data acquisition neighborhood range are extracted, and based on the temperature data and power data, the temperature slope and power change rate of the target component are respectively calculated.

[0024] Preferably, the calculation of the mean and standard deviation of the hot spot effect degree of adjacent components includes the following:

[0025] Within the adjusted data acquisition neighborhood, obtain the temperature data and power data of the neighboring components of each target component; calculate the temperature slope and power change rate of each neighboring component respectively; calculate the hot spot effect degree of each neighboring component based on the temperature slope, power change rate of each neighboring component and the preset hot spot effect degree calculation model; count the hot spot effect degrees of all neighboring components and calculate their mean and standard deviation.

[0026] A second aspect of the present application provides a photovoltaic power station equipment abnormal state detection system, which is applied to the above-mentioned photovoltaic power station equipment abnormal state detection method, comprising:

[0027] A data acquisition module is used to acquire the operating data, ambient temperature data and historical operating data of the photovoltaic module to generate a monitoring data set; wherein the operating data includes the real-time temperature data, real-time power data and real-time light intensity data of the photovoltaic module; the historical operating data includes the historical temperature data, historical power output data and corresponding historical light intensity data of the photovoltaic module since it was put into use.

[0028] The correction factor and threshold matching module is used to generate a correction factor based on real-time light intensity data and historical data; classify the current environment based on ambient temperature data and real-time light intensity data to obtain multiple environmental categories, and match the basic threshold corresponding to each environmental category from a pre-stored environmental category threshold library.

[0029] The threshold adjustment and anomaly screening module is used to calculate the degree of temperature deviation between the real-time temperature data and the average of the historical temperature data of the photovoltaic module; based on the temperature deviation degree and the correction factor, the basic threshold is dynamically adjusted to generate an abnormal data screening threshold; based on the abnormal data screening threshold, potential abnormal data points are screened out from the monitoring data set.

[0030] The target component parameter calculation module is used to take the component corresponding to the potential abnormal data point as the target component, adjust the data acquisition neighborhood range according to the change of real-time light intensity data, and calculate the temperature slope and power change rate of the target component within the adjusted data acquisition neighborhood range.

[0031] A systemic anomaly determination module is used to determine the neighboring components of the target component within the adjusted data acquisition neighborhood, calculate the mean and standard deviation of the hot spot effect degree of the neighboring components, and determine it as a systemic anomaly when the mean exceeds the abnormal data screening threshold and the standard deviation is lower than the preset fluctuation dynamic threshold.

[0032] The beneficial effects of the present invention are at least one of the following:

[0033] By classifying the environment by ambient temperature and light intensity, and combining the dynamic threshold adjustment of the correction factor and the degree of temperature deviation, the detection threshold can adaptively change according to the real-time environmental characteristics and component aging status, effectively reducing the false positive rate and missed detection rate in different environmental scenarios.

[0034] Adjusting the data collection neighborhood based on real-time light intensity data allows for more accurate data collection of target components under varying lighting conditions. When light intensity changes, the data collection neighborhood is promptly narrowed or restored, ensuring that the collected data better reflects the target component's true operating status and helps detect potential anomalies.

[0035] By calculating the mean and standard deviation of the hot spot effect of adjacent modules and combining it with anomaly data screening thresholds and fluctuation dynamic thresholds, we can determine systemic anomalies. This approach, based on neighboring module correlation analysis, can identify anomalies from a cluster perspective, effectively distinguishing between occasional anomalies of individual modules and potential problems with the entire system. This helps to proactively detect and address systemic anomalies that could affect the operation of the entire PV plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a schematic diagram of the method flow of embodiment 1 of the present invention;

[0037] Figure 2 This is a system block diagram of embodiment 2 of the present invention. DETAILED DESCRIPTION

[0038] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0039] Currently, traditional methods for detecting abnormalities in photovoltaic power plant equipment often rely on fixed thresholds or simple statistical analysis. These methods often fail to fully consider the impact of environmental factors (such as light intensity and ambient temperature) on equipment operating parameters, resulting in insufficient environmental adaptability.

[0040] In order to solve the above technical problems, this embodiment provides a method for detecting abnormal status of photovoltaic power station equipment, such as Figure 1 As shown, the following steps are included:

[0041] S1: Obtain the operating data, ambient temperature data, and historical operating data of the photovoltaic module to generate a monitoring data set; the operating data includes the real-time temperature data, real-time power data, and real-time light intensity data of the photovoltaic module; the historical operating data includes the historical temperature data, historical power output data, and corresponding historical light intensity data of the photovoltaic module since it was put into use.

[0042] It should be noted that the abnormal status of photovoltaic modules needs to be comprehensively judged through multi-dimensional data (real-time operating parameters, historical performance data, and environmental variables). A single data set cannot fully reflect the status of the equipment, so it is necessary to build a monitoring data set that includes real-time and historical data.

[0043] For reference, the surface temperature of the photovoltaic module is collected in real time through the integrated thermocouple sensor or infrared temperature sensor, with a sampling frequency of once per minute; the power data is obtained through the built-in power module of the inverter, including the DC side and AC side power, and the timestamp is recorded synchronously; the light intensity is measured by the global radiation sensor installed in the power station; the ambient temperature data comes from the temperature sensor of the weather station, which is no more than 5 meters away from the module temperature sensor to avoid interference from the local heat island effect.

[0044] For example, historical operating data is stored in a distributed database, indexed by "module number + timestamp," and includes all temperature, power, and corresponding light intensity data since the module was commissioned. This data is stored for at least 10 years. In specific implementations, the historical operating data can be preprocessed to remove outliers (e.g., data points where the power suddenly drops to zero or exceeds 120% of the rated value) and noise can be smoothed using a sliding average (with a 10-minute window size).

[0045] Establishing a multi-dimensional monitoring data set containing real-time and historical data can provide multi-dimensional data support for environmental classification, threshold adjustment, and abnormal feature extraction, avoiding misjudgments due to missing data or noise.

[0046] S2: Generate a correction factor based on real-time light intensity data and historical data; classify the current environment based on ambient temperature data and real-time light intensity data to obtain multiple environmental categories, and match the basic threshold corresponding to each environmental category from the pre-stored environmental category threshold library.

[0047] It should be noted that the performance of photovoltaic modules is significantly affected by light intensity and aging. Fixed thresholds cannot adapt to dynamic environments and individual differences, and adaptive adjustment of thresholds is required through correction factors and environmental classification.

[0048] In one possible implementation, generating a correction factor based on real-time light intensity data and historical data includes the following:

[0049] The real-time light intensity data is normalized based on the historical minimum light intensity data and the historical maximum light intensity data of the current season to obtain the normalized value of the real-time light intensity data; based on the historical power output data and the corresponding historical light intensity data, a photovoltaic module efficiency attenuation model is constructed through polynomial fitting to determine the module aging coefficient that reflects the aging degree of the photovoltaic module; the normalized value of the real-time light intensity data and the module aging coefficient are combined to generate a correction factor.

[0050] For example, historical light intensity data is divided by the current season (spring / summer / autumn / winter), and the minimum historical light intensity data in the season is calculated. and historical maximum light intensity data .

[0051] Normalization formula: ,in, is real-time light intensity data, Represents the normalized value of real-time light intensity data, ranging from [0,1].

[0052] For example, historical power data Corresponding light intensity data Normalization is performed (power is normalized to the percentage of rated power) and quadratic polynomial fitting is used ,in is the fitting coefficient.

[0053] For example, the aging factor It is defined as the slope change rate of the fitting curve in year t, reflecting the degree of attenuation of component efficiency over time. The correction factor is a vector , which is used for subsequent threshold adjustment. The correction factor integrates the normalized light value and the aging coefficient to form a comprehensive representation of the environment and individual characteristics.

[0054] In one possible implementation, the current environment is classified based on the ambient temperature data and the real-time light intensity data to obtain multiple environment categories including the following:

[0055] When the ambient temperature data is higher than the preset high temperature threshold and the real-time light intensity data is higher than the preset strong light threshold, it is classified as a high temperature and strong light scene; when the ambient temperature data is lower than the preset low temperature threshold and the real-time light intensity data is lower than the preset weak light threshold, it is classified as a low temperature and weak light scene; when the change rate of the real-time light intensity data within the preset time period is greater than the preset fluctuation threshold and the ambient temperature data is in the preset temperature range, it is classified as a cloudy and fluctuating scene.

[0056] For example, in a high temperature and strong light scene: the ambient temperature data T>40℃ and the real-time light intensity data >1000W / m 2 ;Wherein, T represents the ambient temperature data.

[0057] Low temperature and low light scene: ambient temperature data T<-10℃ and real-time light intensity data <200W / m 2 ;

[0058] Cloudy and fluctuating scene: real-time light intensity data Rate of change within 5 minutes >100W / m / min and the ambient temperature is 10℃≤T≤30℃. T represents the ambient temperature.

[0059] Exemplarily, basic thresholds for different environmental categories are pre-stored in the basic threshold library, for example:

[0060] The basic temperature anomaly threshold in high-temperature and strong-light scenarios is 75°C (the upper limit of component surface temperature safety);

[0061] The basic threshold for power anomalies in low-temperature, low-light scenarios is 15% of the rated power. The threshold library is stored in the system configuration file and supports offline updates. The environment is divided into three typical scenarios: high temperature and strong light; low temperature and low light; and cloudy and fluctuating conditions. Corresponding basic thresholds are pre-stored (for example, focusing on the upper temperature limit in high-temperature scenarios and the lower power limit in low-light scenarios). This solves the "one-size-fits-all" problem of traditional fixed thresholds in extreme environments. For example, the sensitivity to temperature anomalies is dynamically increased in high-temperature, strong light conditions, while false positives in low-temperature, low-light conditions are reduced, ensuring that the detection thresholds are more closely aligned with environmental characteristics.

[0062] S3: Calculate the temperature deviation between the real-time temperature data and the average historical temperature data of the PV module; dynamically adjust the basic threshold based on the temperature deviation and the correction factor to generate an abnormal data screening threshold; based on the abnormal data screening threshold, filter out potential abnormal data points from the monitoring data set.

[0063] It should be noted that the degree of temperature deviation between the real-time temperature data and the average value of the historical temperature data of the photovoltaic module reflects the abnormal trend of the current operating status. Dynamic adjustment of the threshold value combined with the correction factor can reduce the misjudgment of the fixed threshold value under complex working conditions.

[0064] In a possible embodiment, the calculation of the temperature deviation degree between the real-time temperature data and the average value of the historical temperature data of the photovoltaic component includes the following: calculating the absolute value of the difference between the average values ​​of the real-time temperature data and the historical temperature data in the same light intensity range as the temperature deviation degree.

[0065] For example, the temperature deviation degree is calculated as follows: the light intensity is divided into 5 intervals: [0, 200], (200, 500], (500, 800], (800, 1000], (1000, +∞) W / m², and the real-time temperature data T is calculated. real , find the historical temperature mean μ in the corresponding light intensity range T , temperature deviation degree: .

[0066] In a possible implementation, dynamically adjusting the basic threshold based on the temperature deviation degree and the correction factor to generate the abnormal data screening threshold includes the following:

[0067] For the basic threshold values ​​corresponding to the matched environmental categories, set a first weight coefficient corresponding to the normalized value of the real-time light intensity data, a second weight coefficient corresponding to the component aging coefficient, and a third weight coefficient corresponding to the temperature deviation degree. Multiply the normalized value of the real-time light intensity data by the first weight coefficient, the component aging coefficient by the second weight coefficient, and the temperature deviation degree by the third weight coefficient to obtain their respective weighted values; sum the weighted values ​​to obtain the adjustment coefficient;

[0068] If the adjustment coefficient is greater than zero, the product of the adjustment coefficient and the preset positive adjustment ratio is added to the basic threshold to obtain the adjusted threshold; if the adjustment coefficient is less than zero, the product of the absolute value of the adjustment coefficient and the preset negative adjustment ratio is subtracted from the basic threshold to obtain the adjusted threshold; the adjusted threshold is the abnormal data screening threshold; among them, the first weight coefficient, the second weight coefficient, the third weight coefficient, the positive adjustment ratio and the negative adjustment ratio are all pre-set constants.

[0069] For example, the preset weight coefficients are: first weight coefficient k1 = 0.5 (normalized value weight of real-time light intensity data), second weight coefficient k2 = 0.3 (component aging coefficient weight), and third weight coefficient k3 = 0.2 (temperature deviation degree weight);

[0070] Adjustment factor Assuming the positive adjustment ratio is 10% and the negative adjustment ratio is 5%, the adjustment formula is:

[0071] ;

[0072] Represents the basic threshold, the generated As the abnormal data screening threshold, Represents the normalized value of real-time light intensity data, Indicates the degree of temperature deviation.

[0073] Correction factors are generated by normalizing light intensity (eliminating seasonal variations in sunlight) and the component aging coefficient (quantifying efficiency degradation). The threshold is then dynamically adjusted based on the degree of temperature deviation. For example, the sensitivity of detecting temperature fluctuations in aging components is automatically reduced to avoid false alarms caused by natural degradation. This allows the detection system to identify sudden anomalies in new components while also adapting to the gradual degradation of older components.

[0074] S4: Take the component corresponding to the potential abnormal data point as the target component, adjust the data collection neighborhood range according to the change of real-time light intensity data, and calculate the temperature slope and power change rate of the target component within the adjusted data collection neighborhood range.

[0075] It should be noted that sudden changes in real-time light intensity data will cause fluctuations in component temperature and power within the data collection neighborhood. Fixed collection ranges are prone to introduce noise, and the data collection neighborhood range needs to be dynamically adjusted according to light stability.

[0076] In one possible implementation, taking the component corresponding to the potential abnormal data point as the target component, adjusting the data acquisition neighborhood range according to the change in real-time light intensity data, and calculating the temperature slope and power change rate of the target component within the adjusted data acquisition neighborhood range includes the following:

[0077] The PV module corresponding to the potential abnormal data point is used as the target module, and the corresponding light intensity change rate is calculated in real time. When the light intensity change rate is greater than or equal to the preset threshold, the data collection neighborhood range centered on the target module is narrowed. When the light intensity change rate is less than the preset threshold, the default data collection neighborhood range is restored.

[0078] Based on the adjusted data acquisition neighborhood range, temperature data and power data of the target component within the adjusted data acquisition neighborhood range are extracted, and based on the temperature data and power data, the temperature slope and power change rate of the target component are respectively calculated.

[0079] For example, the default data collection neighborhood range is: a 3x3 matrix centered on the target component (9 components in total). Calculation of the rate of change of light intensity: , t represents the current timestamp, which is used to mark the moment of data collection; Represents the real-time light intensity data at the current time t, Represents the real-time light intensity data at the previous moment t-1, where minutes, preset threshold v th =50W / m 2 / min. when When , the data collection neighborhood is reduced to the target component itself (1x1), to prevent other components in the data collection neighborhood from being disturbed by sudden changes in light. , restore the default data collection neighborhood range.

[0080] Exemplary, temperature slope: ; (Unit: ℃ / minute).

[0081] Power change rate: ;( is the rated power of the component).

[0082] Among them, t represents the current timestamp, which is used to mark the moment of data collection. Represents the real-time temperature data of the photovoltaic module at the current time t; Represents the PV module temperature data 10 minutes before the current time t; Represents the photovoltaic module power data at the current time t; Indicates the PV module power data 10 minutes before the current time t.

[0083] In this example, the data window is the most recent 10 minutes, and data is collected once per second. The temperature slope and power change rate of the target component are calculated after taking the average.

[0084] The data collection neighborhood is dynamically adjusted based on the rate of change in light intensity (e.g., narrowed to a single component during sudden changes in light intensity, and expanded to 3×3 components during stable conditions). This, combined with dynamic characteristic parameters such as temperature slope and power change rate, effectively extracts abnormal signals under varying lighting conditions. For example, in cloudy and fluctuating scenarios, the data collection neighborhood is optimized to avoid interference from other components within the data collection neighborhood, focusing on localized overheating or power drops in the target component, thereby improving the detection sensitivity of hot spot anomalies.

[0085] S5: Determine the neighboring components of the target component within the adjusted data acquisition neighborhood, calculate the mean and standard deviation of the hot spot effect degree of the neighboring components, and determine it as a systematic abnormality when the mean exceeds the abnormal data screening threshold and the standard deviation is lower than the preset fluctuation dynamic threshold.

[0086] It should be noted that a single component anomaly may be individual noise, while a widespread anomaly of components within the data collection neighborhood with small fluctuations is more likely to be a systemic problem (such as local occlusion or circuit failure). Statistical analysis is required to identify such anomalies.

[0087] In one possible implementation, calculating the mean and standard deviation of the hot spot effect levels of adjacent components includes the following:

[0088] Within the adjusted data acquisition neighborhood, obtain the temperature data and power data of the neighboring components of each target component; calculate the temperature slope and power change rate of each neighboring component respectively; calculate the hot spot effect degree of each neighboring component based on the temperature slope, power change rate of each neighboring component and the preset hot spot effect degree calculation model; count the hot spot effect degrees of all neighboring components and calculate their mean and standard deviation.

[0089] Exemplarily, after adjustment, all components within the data collection neighborhood (the default is 3x3 or a reduced range) are collected, and the proximity relationship is determined through the component layout coordinate file (storing the row and column position of each component).

[0090] The degree of hot spot effect is calculated using the hot spot effect model: .

[0091] Among them, the weight coefficient is determined according to the hot spot mechanism experiment, the temperature slope reflects the local overheating trend, and the power change rate reflects the performance abnormality.

[0092] Statistical analysis and judgment, calculate the hot spot effect degree H of all adjacent components within the adjusted data collection neighborhood, and obtain the mean and standard deviation . Preset dynamic fluctuation threshold =0.3. When the mean exceeds the abnormal data screening threshold and the standard deviation falls below the preset dynamic fluctuation threshold, it is determined to be a systemic anomaly (indicating that the components within the data collection neighborhood are generally abnormal with low fluctuations and are not individual noise). This mechanism effectively filters out occasional noise from a single component (such as a transient sensor failure) and identifies consistent anomalies among components within the data collection neighborhood.

[0093] It should be noted that the preset fluctuation dynamic threshold It is a statistical threshold used to measure the degree of discreteness of the hot spot effect of components within the data collection neighborhood. The smaller the value, the more consistent the component status within the data collection neighborhood. The preset fluctuation dynamic threshold It is determined by analyzing the neighborhood standard deviation distribution of the normal operation period and the known fault period, and selecting the best split point that can distinguish the two.

[0094] In summary, the monitoring data set generated by S1 provides input for S2-S5. The environmental classification of S2 determines the basic threshold. S3 dynamically adjusts the threshold based on the correction factor and the degree of temperature deviation, forming the key basis for S5's judgment. S4 dynamically adjusts the data collection neighborhood range and calculates characteristic parameters (temperature slope, power change rate) based on the potential abnormal data points screened by S3, providing core indicators for S5's hot spot effect calculation. Based on the data collection neighborhood range and characteristic parameters of S4, combined with the dynamic threshold of S3, S5 completes the systematic abnormality judgment through statistical analysis, forming a complete detection closed loop.

[0095] The second embodiment provides a photovoltaic power station equipment abnormal state detection system, which is applied to the above photovoltaic power station equipment abnormal state detection method, such as Figure 2 As shown, including:

[0096] A data acquisition module is used to acquire the operating data, ambient temperature data and historical operating data of the photovoltaic module to generate a monitoring data set; wherein the operating data includes the real-time temperature data, real-time power data and real-time light intensity data of the photovoltaic module; the historical operating data includes the historical temperature data, historical power output data and corresponding historical light intensity data of the photovoltaic module since it was put into use.

[0097] The correction factor and threshold matching module is used to generate a correction factor based on real-time light intensity data and historical data; classify the current environment based on ambient temperature data and real-time light intensity data to obtain multiple environmental categories, and match the basic threshold corresponding to each environmental category from a pre-stored environmental category threshold library.

[0098] The threshold adjustment and anomaly screening module is used to calculate the degree of temperature deviation between the real-time temperature data and the average of the historical temperature data of the photovoltaic module; based on the temperature deviation degree and the correction factor, the basic threshold is dynamically adjusted to generate an abnormal data screening threshold; based on the abnormal data screening threshold, potential abnormal data points are screened out from the monitoring data set.

[0099] The target component parameter calculation module is used to take the component corresponding to the potential abnormal data point as the target component, adjust the data acquisition neighborhood range according to the change of real-time light intensity data, and calculate the temperature slope and power change rate of the target component within the adjusted data acquisition neighborhood range.

[0100] A systemic anomaly determination module is used to determine the neighboring components of the target component within the adjusted data acquisition neighborhood, calculate the mean and standard deviation of the hot spot effect degree of the neighboring components, and determine it as a systemic anomaly when the mean exceeds the abnormal data screening threshold and the standard deviation is lower than the preset fluctuation dynamic threshold.

[0101] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A method for detecting abnormal status of photovoltaic power station equipment, characterized in that: The following steps are involved: S1: Acquire the operating data, ambient temperature data, and historical operating data of the photovoltaic module to generate a monitoring data set; the operating data includes the real-time temperature data, real-time power data, and real-time light intensity data of the photovoltaic module; the historical operating data includes the historical temperature data, historical power output data, and corresponding historical light intensity data of the photovoltaic module since it was put into use; S2: Generate a correction factor based on real-time light intensity data and historical data; classify the current environment based on ambient temperature data and real-time light intensity data to obtain multiple environmental categories, and match the basic threshold corresponding to each environmental category from a pre-stored environmental category threshold library; S3: Calculate the temperature deviation between the real-time temperature data and the average historical temperature data of the PV module; dynamically adjust the basic threshold based on the temperature deviation and the correction factor to generate an abnormal data screening threshold; and screen out potential abnormal data points from the monitoring data set based on the abnormal data screening threshold; S4: Taking the component corresponding to the potential abnormal data point as the target component, adjust the data collection neighborhood range according to the change of real-time light intensity data, and calculate the temperature slope and power change rate of the target component within the adjusted data collection neighborhood range; S5: Determine the neighboring components of the target component within the adjusted data collection neighborhood, calculate the mean and standard deviation of the hot spot effect degree of the neighboring components, and determine a systematic abnormality when the mean exceeds the abnormal data screening threshold and the standard deviation is lower than the preset fluctuation dynamic threshold; Generating the correction factor based on the real-time light intensity data and historical data includes the following: The real-time light intensity data is normalized based on the historical minimum and maximum light intensity data of the current season to obtain the normalized value of the real-time light intensity data. Based on the historical power output data and the corresponding historical light intensity data, a photovoltaic module efficiency attenuation model is constructed through polynomial fitting to determine the module aging coefficient that reflects the degree of aging of the photovoltaic module. The normalized value of the real-time light intensity data and the module aging coefficient are combined to generate a correction factor. The current environment is classified based on ambient temperature data and real-time light intensity data, and multiple environmental categories are obtained, including the following: When the ambient temperature data is higher than the preset high temperature threshold and the real-time light intensity data is higher than the preset strong light threshold, it is classified as a high temperature and strong light scene; when the ambient temperature data is lower than the preset low temperature threshold and the real-time light intensity data is lower than the preset weak light threshold, it is classified as a low temperature and weak light scene; when the rate of change of the real-time light intensity data within the preset time period is higher than the preset fluctuation threshold and the ambient temperature data is within the preset temperature range, it is classified as a cloudy and fluctuating scene; The calculating of the temperature deviation degree between the real-time temperature data and the mean value of the historical temperature data of the photovoltaic module comprises the following steps: calculating the absolute value of the difference between the mean values ​​of the real-time temperature data and the historical temperature data within the same light intensity range as the temperature deviation degree; The dynamic adjustment of the basic threshold based on the temperature deviation degree and the correction factor to generate the abnormal data screening threshold includes the following: For the basic threshold values ​​corresponding to the matched environmental categories, a first weight coefficient corresponding to the normalized value of the real-time light intensity data, a second weight coefficient corresponding to the component aging coefficient, and a third weight coefficient corresponding to the temperature deviation degree are set; Multiplying the normalized value of the real-time light intensity data by the first weight coefficient, multiplying the component aging coefficient by the second weight coefficient, and multiplying the temperature deviation degree by the third weight coefficient to obtain respective weighted values; summing the weighted values ​​to obtain the adjustment coefficient; If the adjustment coefficient is greater than zero, the product of the adjustment coefficient and the preset positive adjustment ratio is added to the basic threshold to obtain the adjusted threshold; if the adjustment coefficient is less than zero, the product of the absolute value of the adjustment coefficient and the preset negative adjustment ratio is subtracted from the basic threshold to obtain the adjusted threshold; the adjusted threshold is the abnormal data screening threshold; wherein the first weight coefficient, the second weight coefficient, the third weight coefficient, the positive adjustment ratio and the negative adjustment ratio are all preset constants; Adjustment factor ; Assume that the positive adjustment ratio is 10% and the negative adjustment ratio is 5%, and the adjustment formula is: ； Represents the basic threshold, the generated As the abnormal data screening threshold, Represents the normalized value of real-time light intensity data, Indicates the degree of temperature deviation; represents the first weight coefficient, represents the second weight coefficient, Represents the third weight coefficient.

2. A method for detecting abnormal state of photovoltaic power station equipment according to claim 1, characterized in that: The component corresponding to the potential abnormal data point is used as the target component, the data acquisition neighborhood range is adjusted according to the change of real-time light intensity data, and the temperature slope and power change rate of the target component are calculated within the adjusted data acquisition neighborhood range, including the following: The PV module corresponding to the potential abnormal data point is used as the target module, and the corresponding light intensity change rate is calculated in real time. When the light intensity change rate is greater than or equal to the preset threshold, the data collection neighborhood range centered on the target module is narrowed. When the light intensity change rate is less than the preset threshold, the default data collection neighborhood range is restored. Based on the adjusted data acquisition neighborhood range, temperature data and power data of the target component within the adjusted data acquisition neighborhood range are extracted, and based on the temperature data and power data, the temperature slope and power change rate of the target component are respectively calculated.

3. A method for detecting abnormal state of photovoltaic power station equipment according to claim 1, characterized in that: The calculation of the mean and standard deviation of the hot spot effect degree of adjacent components includes the following: Within the adjusted data acquisition neighborhood, obtain the temperature data and power data of the neighboring components of each target component; calculate the temperature slope and power change rate of each neighboring component respectively; calculate the hot spot effect degree of each neighboring component based on the temperature slope, power change rate of each neighboring component and the preset hot spot effect degree calculation model; count the hot spot effect degrees of all neighboring components and calculate their mean and standard deviation.

4. A photovoltaic power station equipment abnormal state detection system, characterized in that: The method for detecting abnormal conditions of photovoltaic power station equipment according to any one of claims 1 to 3 comprises: A data acquisition module, which is used to acquire operating data, ambient temperature data, and historical operating data of the photovoltaic module to generate a monitoring data set; wherein the operating data includes real-time temperature data, real-time power data, and real-time light intensity data of the photovoltaic module; and the historical operating data includes historical temperature data, historical power output data, and corresponding historical light intensity data of the photovoltaic module since it was put into use; A correction factor and threshold matching module is used to generate a correction factor based on real-time light intensity data and historical data; classify the current environment based on the ambient temperature data and the real-time light intensity data to obtain multiple environmental categories, and match the basic threshold corresponding to each environmental category from a pre-stored environmental category threshold library; A threshold adjustment and anomaly screening module is used to calculate the degree of temperature deviation between real-time temperature data and the mean of the historical temperature data of the photovoltaic module; dynamically adjust the basic threshold based on the temperature deviation and the correction factor to generate an abnormal data screening threshold; and screen out potential abnormal data points from the monitoring data set based on the abnormal data screening threshold; A target component parameter calculation module is used to take the component corresponding to the potential abnormal data point as the target component, adjust the data acquisition neighborhood range according to the change of real-time light intensity data, and calculate the temperature slope and power change rate of the target component within the adjusted data acquisition neighborhood range; A systemic anomaly determination module is used to determine the neighboring components of the target component within the adjusted data acquisition neighborhood, calculate the mean and standard deviation of the hot spot effect degree of the neighboring components, and determine it as a systemic anomaly when the mean exceeds the abnormal data screening threshold and the standard deviation is lower than the preset fluctuation dynamic threshold.

Images