Photovoltaic power generation abnormality diagnosis method, computer equipment, and storage medium

Abstract

To enable accurate evaluation of the operating status of photovoltaic power generation and accurate determination of photovoltaic power generation abnormalities, so as to provide a new perspective on the operation management of a photovoltaic power plant.SOLUTION: In a step S101, operation data of a photovoltaic power plant is acquired, and a preset machine learning model is trained on the basis of the past operation data of the photovoltaic power plant to generate a photovoltaic power generation prediction model. In a step S102, the photovoltaic power generation prediction model is used to obtain a power amount to be generated on a current day, and daily power generation fluctuation evaluation and power generation efficiency evaluation are performed on each of the multiple photovoltaic power plants to determine which photovoltaic power plants have an abnormal power generation situation. In a step S103, the operation data corresponding to the photovoltaic power plant where the abnormal power generation situation exists is compared with diagnosis determination conditions to generate a photovoltaic power generation abnormality diagnosis result.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to the field of photovoltaic power generation technology, and more particularly to a method, apparatus, computer equipment, and storage medium for diagnosing abnormalities in photovoltaic power generation. [Background technology]

[0002] With the development of clean energy transformation, new energy represented by solar power will replace traditional thermal power generation as the main energy source in the future. As more and more solar power plants start production and operation, how to operate them leanly and improve power generation efficiency will become a key concern in production management.

[0003] However, current power plants are based on an operation and maintenance method of manual patrol inspection, which makes it difficult to timely and accurately grasp the production and operation status of equipment, and delays in maintenance operations of some equipment result in reduced operation efficiency. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a photovoltaic power generation abnormality diagnosis method, device, computer equipment and storage medium, which solves the problem that the conventional manual patrol inspection operation and maintenance method cannot timely and accurately grasp the production and operation status by collecting production data, calculating operation indexes and integrating diagnosis rules, and provides accurate equipment maintenance suggestions to improve the operation efficiency of the power plant.

[0005] The technical solutions adopted by the present invention are as follows: In a first aspect, the present invention provides a method for diagnosing abnormalities in photovoltaic power generation, comprising the following steps. Step 1: Obtain the operation data of the solar power plant, train the preset neural network based on the operation data of the solar power plant, and generate a solar power generation prediction model. Step 2: Using the solar power generation forecasting model, daily power generation fluctuation evaluation and power generation efficiency evaluation are performed for multiple solar power plants, and abnormal solar power plants are identified. Step 3: Compare the corresponding operation data of the abnormal PV power plant with the abnormal conditions, and generate the abnormal diagnosis result of the PV power plant.

[0006] In step 1, the operation data of the solar power plant includes daily average irradiance data, battery temperature data and weather data. In step 1, the generation of a solar power generation forecast model includes the following steps: A1: Filter the daily average irradiance data to generate filtered daily average irradiance data. A2: Determine the correction coefficient for the solar power plant based on the filtered daily average irradiance data and battery temperature data. A3: Based on the solar power plant correction coefficient and weather data, a preset neural network is trained to generate a power temperature correction coefficient. A4: Generate a solar power generation forecast model based on the filtered daily average irradiance data, solar power plant correction coefficients and power temperature correction coefficients.

[0007] The A1 includes the following steps: a1: Obtain power generation data corresponding to the daily average irradiance data, and screen the daily average irradiance data and power generation data that are greater than the preset threshold value to determine the power plant's daily power generation and the power plant's daily average irradiance. a2: Identify irradiance abnormal points based on the power plant's daily power generation and the power plant's daily average irradiance, and filter out the irradiance abnormal points. a3: Based on the irradiance data after filtering out irradiance anomalies, perform model pre-training on the preset neural network to generate pre-training results. a4: Based on the pre-training result, determine the outliers in the irradiance data after filtering out the irradiance abnormal points, delete the outliers, and generate filtered daily average irradiance data.

[0008] The a2 step includes the following steps: a2.1: Divide the daily power generation of the power plant into multiple power generation intervals, map the daily average irradiance of the power plant to multiple power generation intervals, and generate the irradiance within each power generation interval. a2.2: Determine a segment box plot based on the irradiance in each power generation interval, and identify and filter out irradiance anomalies based on the segment box plot.

[0009] In A2, a correction factor for the solar power plant is generated using a linear regression method based on the filtered daily average irradiance data and the battery temperature data.

[0010] In A3, the solar power generation forecast model formula is as follows:

number

[0011] The step 2 includes the following steps: B1: Obtain the power generation amount of the solar power plant on the day and the previous day, and calculate the fluctuation value of the daily power generation amount based on the power generation amount of the solar power plant on the day and the previous day. B2: Collect real-time operation data of the solar power plant, input the real-time operation data of the solar power plant into a solar power generation forecasting model, and generate the predicted power generation amount of the solar power plant for that day. B3: Calculate the power generation efficiency of the solar power plant based on the solar power plant's predicted power generation amount for that day and the solar power plant's real-time power generation amount for that day. B4: Identify abnormal solar power plants based on the fluctuation of daily power generation and the power generation efficiency of the solar power plants.

[0012] The step 3 includes the following steps: C1: Calculate the operational index value of the solar power plant based on the corresponding operation data of the abnormal solar power plant. C2: Compare the operational index values ​​of the solar power plant with the abnormal conditions to generate an abnormality diagnosis result for the solar power generation.

[0013] In C1, the performance index values ​​of the PV power station include resource change values, power loss values, and inverter operating status statistics values. The performance index values ​​of the PV power station are calculated according to the corresponding operating data of the abnormal PV power station, and include: c1: Based on the corresponding operation data of the abnormal solar power plant, determine the average daily irradiance of the abnormal solar power plant on the current day and the average daily irradiance of the previous day, and calculate the resource change rate relative to the previous period based on the average daily irradiance of the current day and the average daily irradiance of the previous day. c2: Determine the power loss amount of the solar power plant on the current day and the previous day based on the corresponding operation data of the abnormal solar power plant. c3: Determine the number of inverters in the photovoltaic power station and the unit status data of the inverters in the photovoltaic power station according to the corresponding operation data of the abnormal photovoltaic power station, and calculate the operating status statistics of the inverters on the current day and the previous day according to the number of inverters in the photovoltaic power station and the unit status data of the inverters in the photovoltaic power station. The performance index values ​​of the photovoltaic power station also include inverter output power dispersion rate, photovoltaic power grid energy conversion efficiency, inverter loss rate and inverter shutdown status. The performance index values ​​of the photovoltaic power station are calculated according to the corresponding operation data of the abnormal photovoltaic power station, and further include: c4: Determine the output power of the inverters of the photovoltaic power station according to the corresponding operation data of the abnormal photovoltaic power station, and calculate the discrete rate of the output power of the inverters in the entire photovoltaic power station according to the output power of each inverter of the photovoltaic power station. c5: Based on the corresponding operation data of the abnormal solar power plant, determine the DC power of the solar power plant inverter, the effective area of ​​the solar power grid and the total radiation dose of the slope, and calculate the energy conversion efficiency of the solar power grid according to the DC power of the power plant inverter, the effective area of ​​the solar power grid and the total radiation dose of the slope. c6: Determine the inverter DC power and inverter AC power according to the corresponding operation data of the abnormal solar power plant, and calculate the inverter loss rate according to the inverter DC power and inverter AC power. c7: Determine the daily power generation amount of the inverter of the photovoltaic power station according to the corresponding operation data of the abnormal photovoltaic power station, and determine the stop status of the inverter according to the daily power generation amount of the inverter of the photovoltaic power station.

[0014] In c4, the standard difference and average value of the inverter output power of the photovoltaic power station are determined based on the inverter output power of the photovoltaic power station and the inverter quantity of the photovoltaic power station, and the inverter output power discretization rate in the entire photovoltaic power station is calculated based on the standard difference and average value of the inverter output power of the photovoltaic power station.

[0015] The C2 includes the following steps: c2.1: The resource change rate relative to the previous period is compared with the preset resource threshold, and if the resource change rate relative to the previous period is greater than the preset resource threshold, the abnormality diagnosis result of the photovoltaic power generation is due to resource fluctuation. c2.2: Based on the power loss values ​​of the current day and the previous day and the power generation amount of the solar power plant, the ratio of the power loss amount due to breakdown, the power loss amount due to inspection and repair, the power loss amount due to power limit, and the power loss amount due to accompanying shutdown are determined, as well as the ratio compared to the previous period. c2.3: Based on the inverter operating state statistics, determine the ratio of fault shutdown time to the previous period, normal shutdown time to the previous period, power limit operation shutdown time to the previous period, and external cause shutdown time to the previous period. c2.4: If the proportion of power loss due to a fault is greater than the preset threshold, or if the proportion of the length of time that the fault has stopped or the ratio to the previous period is greater than the preset threshold for the length of time that the fault has stopped, the abnormality diagnosis result for the solar power generation is due to equipment failure. If the proportion of power loss due to inspection and repair is greater than the preset threshold for power loss, or if the proportion of normal shutdown time length or the ratio to the previous period is greater than the preset threshold for shutdown time length, the diagnosis result of the solar power generation abnormality is due to inspection and repair. If the ratio of the power loss amount under the power limit is greater than the preset threshold of the power loss amount, or if the ratio or ratio between the power limit operation and the downtime length is greater than the preset threshold of the downtime length, the diagnosis result of the solar power generation abnormality is due to the power limit. If the proportion of the power loss due to accompanying shutdown is greater than the preset threshold for the power loss, or if the proportion of the duration of the external shutdown or the ratio to the previous period is greater than the preset threshold for the duration of the shutdown, the diagnosis result of the solar power generation abnormality is that it is due to accompanying shutdown.

[0016] In a second aspect, the present invention provides a photovoltaic power generation abnormality diagnosis device, the device including: A training module is used to acquire operation data of the solar power plant, and trains a preset neural network based on the operation data of the solar power plant to generate a solar power generation prediction model and predict power generation for the solar power plant. The evaluation module uses the solar power generation forecast results to evaluate fluctuations in daily power generation and power generation efficiency for each of a plurality of solar power plants, and identifies solar power plants where abnormalities in power generation exist. The abnormality diagnosis module compares the operation data corresponding to the photovoltaic power plant where the power generation abnormality situation exists with the diagnosis determination conditions, and generates an abnormality diagnosis result for the photovoltaic power generation.

[0017] In a third aspect, the present invention provides a computer device, including: The method includes a memory and a processor, the memory and the processor being communicatively connected to each other, computer instructions stored in the memory, and the processor executing the computer instructions to perform the photovoltaic power generation abnormality diagnosis method.

[0018] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored therein, the computer instructions being used to cause a computer to execute a method for diagnosing abnormalities in photovoltaic power generation.

[0019] The present invention provides a method, device, computer equipment and storage medium for diagnosing abnormalities in photovoltaic power generation, and has the following technical advantages. 1) The photovoltaic power generation abnormality diagnosis method provided by the present invention uses a photovoltaic power generation prediction model to evaluate the daily power generation fluctuations and power generation efficiency for each of a plurality of photovoltaic power plants, identify photovoltaic power plants with abnormal power generation, and compare the corresponding operation indicators of the abnormal photovoltaic power plants with the diagnostic judgment criteria to generate photovoltaic power generation abnormality diagnosis results, thereby achieving accurate evaluation of the photovoltaic power generation operation status and accurate judgment of the cause of the photovoltaic power generation abnormality, and providing reference suggestions for improving the photovoltaic power generation operation efficiency. 2) The solar power generation forecasting model provided by the present invention filters the daily average irradiance data and determines the solar power plant correction coefficient and the power temperature correction coefficient, thereby making the solar power generation forecasting model more accurate and improving the accuracy of the solar power generation forecast value. 3) The photovoltaic power generation anomaly diagnosis method provided by the present invention performs model pre-training on a preset machine learning model, and determines outliers in the irradiance data after filtering out irradiance anomaly points based on the Kirk distance of the pre-training result, thereby achieving accurate identification of outliers and improving the accuracy of training data in the photovoltaic power generation prediction model training process. 4) The photovoltaic power generation abnormality diagnosis method provided by the present invention uses segment boxplots to intuitively and accurately identify irradiance abnormalities, making the training data more accurate. 5) The photovoltaic power generation anomaly diagnosis method provided by the present invention takes into consideration that linear regression does not have high requirements for the amount of training data, and linear regression is relatively sensitive to abnormal points, which directly affect the changes in the photovoltaic power plant correction coefficients. Therefore, for daily report data, linear regression can achieve accurate calculation of the photovoltaic power plant correction coefficients, while preset neural networks have stronger tolerance for abnormal points, and a small number of abnormal points will not affect the training effect. The use of linear regression and preset neural networks makes the photovoltaic power generation prediction model more accurate and stable. 6) The photovoltaic power generation abnormality diagnosis method provided by the present invention can timely understand the power generation efficiency status and daily power generation fluctuation status of each photovoltaic power plant by calculating the daily power generation fluctuation value and the power generation efficiency of the photovoltaic power plant, timely identify photovoltaic power plants with low power generation efficiency and significant power generation decline, comprehensively identify a list of photovoltaic power plants that require priority attention every day, and identify targets that require priority attention. 7) The photovoltaic power generation abnormality diagnosis method provided by the present invention, combined with the operational index values ​​of a photovoltaic power plant, can initially find the causes of low power generation efficiency and significant power generation decline, and help the operational department establish a "problem identification, feedback, and improvement" mechanism from a management perspective, thereby improving the overall operation efficiency of photovoltaic power generation. 8) The photovoltaic power generation abnormality diagnosis method provided by the present invention starts from the operational indicators of a photovoltaic power plant, such as the inverter operating status, output power discontinuity rate, wear and tear status, and further diagnoses the specific causes of the photovoltaic power plant's power generation decreasing compared to the previous period or significantly lower than the predicted value. 9) The photovoltaic power generation abnormality diagnosis method provided by the present invention compares the operational index values ​​of a photovoltaic power plant with abnormal conditions, thereby achieving accurate diagnosis of photovoltaic power generation abnormalities, improving the accuracy of photovoltaic power generation abnormality diagnosis results, and laying the foundation for subsequent improvement of photovoltaic power generation operation efficiency. The invention will now be further described with reference to the accompanying drawings and examples. [Brief explanation of the drawings]

[0020] [Figure 1] 3 is a flowchart of a method for diagnosing an abnormality in a photovoltaic power generation system according to an embodiment of the present invention. [Figure 2] 10 is a flowchart of another method for diagnosing an abnormality in a photovoltaic power generation system according to an embodiment of the present invention. [Figure 3] FIG. 1 is a schematic diagram of the basic principles of box plots according to an embodiment of the present invention; [Figure 4] FIG. 1 is a schematic diagram of a segment box plot according to an embodiment of the present invention; [Figure 5] 1 is a schematic diagram of Zhejiang Leqing Letai linear regression fitting effect according to an embodiment of the present invention; FIG. [Figure 6]FIG. 1 is a schematic diagram of the Anhui Huainan Gorge Pan linear regression fitting effect according to an embodiment of the present invention; [Figure 7] 1 is a schematic diagram of the Shaanxi Weinan Jinyu linear regression fitting effect according to an embodiment of the present invention; FIG. [Figure 8] 10 is a flowchart of another method for diagnosing an abnormality in a photovoltaic power generation system according to an embodiment of the present invention. [Figure 9] 10 is a flowchart of yet another method for diagnosing an abnormality in a photovoltaic power generation system according to an embodiment of the present invention. [Figure 10] 1 is a configuration block diagram of a photovoltaic power generation abnormality diagnosis device according to an embodiment of the present invention. [Figure 11] FIG. 2 is a hardware configuration diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0021] In this embodiment, there is provided a method for diagnosing an abnormality in solar power generation that can be used in the above-mentioned mobile terminals such as mobile phones, tablets, etc. Fig. 1 is a flowchart of the method for diagnosing an abnormality in solar power generation according to the embodiment of the present invention, and as shown in Fig. 1, it includes the following steps. Step S101: Obtain operation data of the solar power plant, train a preset neural network based on the operation data of the solar power plant, and generate a solar power generation prediction model. Specifically, experiments are conducted using the solar power plant's daily average irradiance and daily recorded temperature, and the theoretical power generation value for that day is calculated. Alternatively, experiments are conducted using the solar power plant's centrally controlled real-time active power and irradiance, and the theoretical active power at a certain time of the solar power plant can be obtained through model calculations. The solar power plant's operating data is shown in Table 1. (Solar power plant operation data table) [Table 1]

[0022] Furthermore, the preset neural network can employ neural networks such as ANN (Artificial Neural Network, feedforward neural network), CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), etc.

[0023] In step S102, the calculation results of the photovoltaic power generation prediction model are used to perform daily power generation fluctuation evaluation and power generation efficiency evaluation for each of the multiple photovoltaic power plants, and abnormal photovoltaic power plants are identified.

[0024] Step S103: Compare the corresponding operation data of the abnormal photovoltaic power plant with the abnormal conditions to generate an abnormal diagnosis result for the photovoltaic power generation.

[0025] The photovoltaic power generation abnormality diagnosis method provided in this embodiment uses a photovoltaic power generation prediction model to perform daily power generation fluctuation evaluation and power generation efficiency evaluation for multiple photovoltaic power plants, identify abnormal photovoltaic power plants, and then compare the corresponding operation data of the abnormal photovoltaic power plants with the abnormal conditions to generate photovoltaic power generation abnormality diagnosis results, thereby realizing accurate evaluation of the photovoltaic power generation operation status and accurate judgment of photovoltaic power generation abnormalities, and laying the foundation for improving the operation efficiency of photovoltaic power generation.

[0026] This embodiment provides a method for diagnosing abnormalities in solar power generation, which can be used in mobile terminals such as mobile phones, tablets, etc. Figure 2 is a flowchart of the method for diagnosing abnormalities in solar power generation according to an embodiment of the present invention. As shown in Figure 2, the flow includes the following steps: Step S201: Obtain operation data of the photovoltaic power plant, and train the preset neural network based on the operation data of the photovoltaic power plant to generate a photovoltaic power generation prediction model. Specifically, the operation data of the solar power station includes daily average irradiance data, battery temperature data and weather data. In process S2011, filtering is performed on the daily average irradiance data to generate filtered daily average irradiance data.

[0027] In some optional embodiments, the process S2011 includes: Process a1: Obtain the power generation data corresponding to the daily average irradiance data, and screen the daily average irradiance data and power generation data that are greater than the predetermined threshold value, and determine the daily power generation of the solar power plant and the daily average irradiance of the solar power plant. Specifically, to minimize the impact of transmission stability on model training, observation point solar power plants with corresponding daily average irradiance data > 0 and power generation data > 0 are screened.

[0028] In process a2, the irradiance abnormal points are identified based on the daily power generation amount of the photovoltaic power plant and the daily average irradiance of the photovoltaic power plant, and the irradiance abnormal points are filtered. Specifically, the daily power generation of the solar power plant is divided into multiple power generation intervals, and the daily average irradiance of the solar power plant is mapped to the multiple power generation intervals to generate the irradiance within each power generation interval, and a segment box plot is determined based on the irradiance within each power generation interval, and irradiance anomalies are identified and filtered based on the segment box plot.

[0029] Furthermore, as shown in Figure 3, the box plot is composed of five numerical points: minimum observed value (lower end) = Q1 + 1.5IQR, upper quartile (Q1), median, lower quartile (Q3), and maximum observed value (upper end) = Q3 + 1.5IQR, where IQR indicates that the data falls into the middle 50% span, and IQR = lower quartile (Q3) - upper quartile (Q1). If there are outliers in the data, i.e., abnormal values, and the outliers exceed the maximum or minimum observed values, the outliers are shown in the form of "dots."

[0030] Furthermore, as shown in Figure 4, the power generation is divided into intervals, and the irradiance in each interval covers the equidistant intervals of all power generation values. The default number of intervals is 2 or 5 multiplied by a power of 10, and is generally divided into five intervals, namely (0, 2), (2, 4), (4, 6), (6, 8), and (8, 10). Furthermore, the irradiance in each power generation interval is plotted as a box plot, and the corresponding outliers are identified and filtered. The irradiance in each power generation interval is plotted separately, and there are five independent box plots in Figure 4, and then a segment box plot is generated. From the segment box plot, it can be seen that the irradiance tends to increase as the power generation interval becomes higher.

[0031] Step a3: perform model pre-training on the preset neural network based on the irradiance data after filtering out the irradiance anomaly points, and generate a pre-training result. Step a4: Determine the outliers in the irradiance data after filtering out the irradiance abnormalities based on the pre-training results, delete the outliers, and generate filtered daily average irradiance data. Specifically, a model is pre-trained on a preset neural network, and the Cook distance is calculated based on the pre-training results. Based on the Cook distance, solar power plant observation points that have a relatively large impact on the model are filtered out, i.e., outliers are removed.

[0032] Furthermore, Cook's distance can measure whether a given regression model is only affected by a single variable. Cook's distance calculates the influence of each data point on the prediction result. For a solar power plant observation point i corresponding to any irradiance in the irradiance data after filtering out each irradiance anomaly point, Cook's distance measures the change in the fitting value of the actual power generation amount when i is included and when i is not included, and further obtains the influence of the solar power plant observation point on the fitting result. Here, the formula for calculating the Cook distance Di is as follows:

number

[0033] Furthermore, if the Cook distance of a solar power plant observation point is four times greater than the average distance and this solar power plant observation point is an outlier, the irradiance data corresponding to the outlier is deleted to generate filtered daily average irradiance data.

[0034] Step S2012: Determine a correction coefficient for the solar power generation location based on the filtered daily average irradiance data and the battery temperature data. Specifically, a linear regression method is used to generate correction factors for solar power generation locations based on filtered daily average irradiance data and battery temperature data.

[0035] Furthermore, the equipment capacity, daily average irradiance data and battery temperature in the daily report data are used to input them into a preset neural network, and the daily power generation amount is set as the target value, and a correction coefficient for the solar power generation location is obtained by linear regression. For example, as shown in Figure 5, taking the 03Q2 Zhejiang Leqing Letai (160.8978 MW) as an example, the solar power plant correction coefficient is 0.7701711, the amount of training data is 626, and a total of 127 data (20.29%) have a training error greater than 20%. For example, it can be seen that, except for a few outliers, most of the observations are uniformly distributed around the 45° line (i.e., fitted value = actual value). For example, as shown in Figure 6, taking 03A1 Anhui Huainan Xiapan (150 MW) as an example, the solar power plant correction coefficient is 0.8881927, the amount of training data is 1013, and the total number of data with training error greater than 20% is 215 (21.22%). For example, as shown in Figure 7, taking 03D2 Shaanxi Weinan Jinyu (100MW) as an example, the solar power plant correction coefficient is 0.9825449, the amount of training data is 1013, and the total number of data with training error greater than 20% is 125 (12.34%).

[0036] Step S2013: Based on the photovoltaic power station correction coefficient and weather data, a preset neural network is trained to generate a power temperature correction coefficient. Specifically, further research was conducted on solar power plants where the model fitting effect is relatively scattered. Because the effect of the feedback model of some solar power plants is significantly affected by weather, consideration was given to training for different weather conditions to optimize the effect of the linear model. Weather data was divided into sunny, cloudy / cloudy / sunny then cloudy, and rain / snow / cloudy then rain, and the corresponding power temperature correction coefficients were trained based on the above weather data. Taking Shenglin, Huzhou, Zhejiang as an example, the sunny coefficient is somewhat high, while the cloudy / cloudy and rain / snow coefficients are similar, resulting in a certain improvement in model effect (mean square error MSE). The model effect after weather classification is shown in Table 2 below. (Model effect after weather classification) [Table 2]

[0037] In step S2014, a solar power generation forecasting model is generated based on the filtered daily average irradiance data, the solar power plant correction coefficient, and the power temperature correction coefficient. Specifically, the formula for the solar power generation forecasting model is as follows:

number

[0038] The photovoltaic power generation anomaly diagnosis method provided in this embodiment screens and filters outliers through pre-training of a preset machine learning model to improve the data quality of the training set.In addition, considering that the results of linear regression are relatively sensitive to anomalies and that neural network models are relatively inclusive of anomalies, this method adopts a method combining linear regression and neural network to improve the accuracy and stability of the model.

[0039] Step S202: Using the photovoltaic power generation forecasting model, daily power generation fluctuation evaluation and power generation efficiency evaluation are performed for each of the multiple photovoltaic power plants, and abnormal photovoltaic power plants are identified. For details, please refer to step S102 in the embodiment shown in Figure 1, and the details will not be described here. Step S203: Compare the corresponding operation data of the abnormal photovoltaic power plant with the abnormal condition to generate an abnormal diagnosis result for the photovoltaic power generation. For details, please refer to step S103 in the embodiment shown in Figure 1, and will not be described here.

[0040] The solar power generation abnormality diagnosis method provided in this embodiment filters daily average irradiance data and determines a solar power plant correction coefficient and a power temperature correction coefficient, thereby making the generation of a solar power generation prediction model more accurate and improving the accuracy of the solar power generation prediction value.

[0041] The solar power generation abnormality diagnosis method provided in this embodiment can be used in the above-mentioned mobile terminals such as mobile phones, tablets, etc. Figure 8 is a flowchart of the solar power generation abnormality diagnosis method according to the embodiment of the present invention, which, as shown in Figure 8, includes the following steps:

[0042] Step S801: Obtain operation data of the solar power plant, train a preset neural network based on the operation data of the solar power plant, and generate a solar power generation prediction model. For details, refer to step S201 in the embodiment shown in Figure 2, and the details will not be described here.

[0043] In step S802, the photovoltaic power generation prediction model is used to perform daily power generation fluctuation evaluation and power generation efficiency evaluation for each of the multiple photovoltaic power plants, and abnormal photovoltaic power plants are identified.

[0044] Specifically, step S802 includes: Step S8021: Obtain the power generation amount of the solar power plant on the previous day and the real-time power generation amount of the solar power plant on the current day, and calculate the daily power generation fluctuation value based on the power generation amount of the solar power plant on the previous day and the real-time power generation amount of the solar power plant on the current day. Specifically, the daily power generation fluctuation is used to identify solar power plants with a significant drop in daily power generation, and is an index to evaluate the situation in which the power generation volume of the day changes from the power generation volume of the previous day. The calculation formula for the daily power generation fluctuation value is as follows:

number

[0045] Step S8022: Collect real-time operation data of the photovoltaic power plant, input the real-time operation data of the photovoltaic power plant into a photovoltaic power generation prediction model, and generate a predicted power generation amount of the photovoltaic power plant for that day.

[0046] Step S8023: Calculate the power generation efficiency of the photovoltaic power plant based on the predicted power generation amount of the photovoltaic power plant on that day and the real-time power generation amount of the photovoltaic power plant on that day. Specifically, the power generation efficiency of a solar power plant refers to evaluating the difference between the theoretical power generation amount for that day, that is, the daily power generation amount predicted based on machine learning, and the actual power generation amount for that day. The formula for calculating the power generation efficiency of a solar power plant is as follows:

number

[0047] In step S8024, an abnormal photovoltaic power plant is identified based on the daily power generation fluctuation value and the power generation efficiency of the photovoltaic power plant. Specifically, the daily power generation fluctuation value and the power generation efficiency of the solar power plant are compared with preset power generation thresholds, and if they are greater than the preset thresholds, the evaluation result indicates that the power generation efficiency is low and there is a clear decline in the daily power generation.Furthermore, further diagnosis is performed on the scene where the power generation fluctuation value of this day and the power generation efficiency of the solar power plant correspond.

[0048] Step S803: Compare the corresponding operation data of the abnormal photovoltaic power plant with the abnormal conditions to generate an abnormal diagnosis result for the photovoltaic power generation. For details, please refer to step S203 in the embodiment shown in Figure 2, and the details will not be described here.

[0049] The photovoltaic power generation abnormality diagnosis method provided in this embodiment calculates the daily power generation fluctuation value and the power generation efficiency of the photovoltaic power plant, thereby making it possible to understand the power generation efficiency status and daily power generation fluctuation status of each photovoltaic power plant in a timely manner, realizing timely identification of photovoltaic power plants with low power generation efficiency and obvious decline in power generation, realizing identification of a list of photovoltaic power plants that require daily priority attention from a group level, and identifying targets that require priority attention.

[0050] The method for diagnosing abnormalities in solar power generation provided in this embodiment can be used in the above-mentioned mobile terminals such as mobile phones, tablets, etc. Figure 9 is a flowchart of the method for diagnosing abnormalities in solar power generation according to the embodiment of the present invention, and as shown in Figure 9, it includes the following steps:

[0051] In step S901, the operation data of the solar power plant is acquired, and the preset neural network is trained based on the operation data of the solar power plant to generate a solar power generation prediction model. For details, refer to step S801 in the embodiment shown in Fig. 8, and the details will not be described here.

[0052] Step S902: Using the photovoltaic power generation prediction model, evaluate the daily power generation fluctuation and power generation efficiency for each of the multiple photovoltaic power plants, and identify abnormal photovoltaic power plants. For details of step S802, please refer to the embodiment shown in Figure 8 and will not be described here.

[0053] Step S903: Compare the corresponding operation data of the abnormal photovoltaic power plant with the abnormal conditions to generate an abnormal diagnosis result for the photovoltaic power plant.

[0054] Step S903 specifically includes: Step S9031: Calculate the solar power station operation index value according to the corresponding operation data of the abnormal solar power station.

[0055] In some optional embodiments, the solar power plant operation index values ​​include a resource change value, a power loss value, an inverter operating state statistics value, an inverter output power discontinuity rate, a solar power generation grid efficiency, a solar power generation grid efficiency, an inverter loss value, and an inverter stop status.

[0056] Step b1: Determine the average daily irradiance of the abnormal solar power plant on the current day and the average daily irradiance of the previous day based on the corresponding operating data of the abnormal solar power plant, and calculate the resource change value based on the average daily irradiance of the current day and the average daily irradiance of the previous day. Specifically, solar power generation is closely related to resource conditions, and when there is a significant drop in daily power generation, we judge the resource change situation and determine whether there is a direct correlation with resource changes. The resource change value is calculated using the following formula:

number

[0057] Furthermore, the resource change value is calculated by the solar power plant with the day as the dimension. Step b2 determines the power loss amount of each device according to the corresponding operation data of the abnormal photovoltaic power plant, and calculates the power loss amount value according to the power loss amount of each device. Specifically, power losses can be divided into four categories: power limiting, outages, inspection and repair, and breakdowns. Fluctuations in daily power generation or low power generation efficiency can all be related to power losses.

[0058] Step b3: determine the number of inverters in the power station and the unit status data of the inverters in the power station according to the corresponding operation data of the abnormal photovoltaic power station, and calculate the inverter operation status statistics according to the number of inverters in the power station and the unit status data of the inverters in the power station. Specifically, the inverter operating status includes normal operation, power-limited operation, alarm operation, normal stop, fault stop, power-limited stop, external cause stop, and communication abnormality. The operating time under each status can be counted to reflect the overall operating status, and the inverter operating status statistics are calculated using the following formula:

number

number

[0059] Step b4: Determine the output power of the inverter of the power station according to the corresponding operation data of the abnormal photovoltaic power station, and calculate the discrete rate of the inverter output power according to the output power of the inverter of the power station and the number of inverters of the power station. Specifically, the inverter output power discretization rate reflects the degree of discretization of the inverter output power at a certain time, and the smaller the discretization rate value, the more concentrated the inverter's AC power curve, the more consistent and stable the operating status of the entire inverter will be, and the higher the discretization rate, the existence of an inverter with low power generation efficiency.

[0060] Furthermore, the standard deviation and average value of the output power of the inverters of the power station are determined based on the output power of the inverters of the power station and the number of inverters of the power station, and the inverter output power discretization rate is calculated based on the standard deviation and average value of the output power of the inverters of the power station.

[0061] Furthermore, the formulas for calculating the standard deviation and average value of the power plant inverter output power are as follows:

number

[0062]

number

[0063] Step b5: Determine the input power of the inverter of the power station, the effective area of ​​the photovoltaic power grid and the total radiation dose on the slope according to the corresponding operation data of the abnormal photovoltaic power station, and calculate the efficiency of the photovoltaic power grid according to the input power of the inverter of the power station, the effective area of ​​the photovoltaic power grid and the total radiation dose on the slope. Specifically, photovoltaic grid efficiency indicates the energy conversion efficiency of a photovoltaic grid, that is, the ratio of the energy output from the photovoltaic grid to the inverter (inverter input power) to the energy incident on the photovoltaic grid (total radiation dose on the slope calculated from the effective area of ​​the photovoltaic grid). Photovoltaic grid efficiency indicates the energy conversion capacity of a photovoltaic grid, with a higher value indicating a higher energy conversion capacity of the photovoltaic grid. Changes in photovoltaic grid efficiency can also reflect whether the photovoltaic panels are shielded or covered.

number

[0064] Step b6: Determine the inverter DC power and inverter AC power according to the corresponding operation data of the abnormal photovoltaic power generation site, and calculate the inverter loss value according to the inverter DC power and inverter AC power. Specifically, inverter loss refers to the loss during the inverter DC / AC process, and the inverter loss value includes the inverter real-time loss rate and the inverter average loss rate for a day, of which the inverter real-time loss rate calculation formula is as follows:

number

[0065] Furthermore, the inverter real-time loss rate is collected at a fixed frequency every day depending on the inverter level, and the statistical time period is 9:00 to 17:00.

[0066] Furthermore, the average loss rate of the inverter on a given day is calculated as follows:

number

[0067] Furthermore, E ho is calculated as follows:

number

[0068] Furthermore, E hh is calculated as follows:

number

[0069] Furthermore, at the inverter level, inverter loss values ​​are calculated using the day as a dimension, and the statistical time period is 9:00 to 17:00. Step b7: determine the daily power generation amount of the inverter of the power station according to the corresponding operation data of the abnormal photovoltaic power station, and determine the inverter stop state according to the daily power generation amount of the inverter of the power station. Specifically, inverter shutdown refers to analyzing the inverter's daily power generation and determining whether the inverter is in a normal power generation state from the daily power generation. The data information related to inverter shutdown statistics includes the cumulative daily power generation per inverter in the centralized control system. If the daily power generation of a certain inverter is significantly lower than that of other inverters, it can be roughly determined that the inverter is not in a normal power generation state that day. The statistical dimension and frequency depend on the inverter level, and statistics are performed on a day basis.

[0070] Furthermore, based on the calculated solar power plant operational indicator values, the solar power plant type was determined, and specifically, they were classified into two types: plants that require attention and plants with significantly increased power generation. The specific target types are shown in Table 3 below. (Target type table) [Table 3]

[0071] Furthermore, for solar power plants that are recommended for attention, we will focus on explaining the reasons for a significant drop in power generation or low power generation efficiency.For solar power plants where power generation has increased significantly, we will focus on explaining the reasons for the significant increase in power generation compared to the previous period.

[0072] Step S9032: Compare the operational index value of the photovoltaic power plant with the abnormal condition to generate an abnormality diagnosis result for the photovoltaic power generation. In some optional embodiments, step S9032 includes: Step c1: compare the resource variation value with the preset resource threshold value; if the resource variation rate is greater than the preset resource threshold value, the abnormality diagnosis result of the photovoltaic power generation is resource variation abnormality; Step c2: Determine the fault power loss amount, inspection and repair power loss amount, power limit power loss amount and accompanying outage power amount based on the power loss amount value. Step c3: Based on the inverter operating state statistics, determine the ratio of fault shutdown time to the previous period, the ratio of normal shutdown time to the previous period, the ratio of power-limited operation shutdown time to the previous period, and the ratio of externally caused shutdown time to the previous period. Step c4: If the amount of power loss due to the fault is greater than the preset threshold, or if the length of the outage time due to the fault is less than the preset threshold for the length of the outage time compared to the previous period, the abnormality diagnosis result of the photovoltaic power generation is an abnormality due to a fault. Step c5: If the inspection and repair power loss amount is greater than the preset threshold of the power loss amount, or the normal shutdown time length is less than the preset threshold of the shutdown time length compared to the previous period, the abnormality diagnosis result of the photovoltaic power generation is an inspection abnormality. Step c6: If the power-limited loss power amount is smaller than the preset power-loss power amount threshold, or if the power-limited operation and downtime length are smaller than the preset downtime length threshold compared to the previous period, the diagnosis result of the solar power generation abnormality is a power-limited abnormality. Step c7: If the amount of power loss due to accompanying shutdown is less than the preset threshold value of power loss, or if the length of the externally caused shutdown time is less than the preset threshold value of the shutdown time length compared to the previous period, the diagnosis result of the solar power generation abnormality is accompanying shutdown abnormality. Specifically, information such as the resources of the power plant on that day, the amount of power loss, and the inverter operating status was combined, and the causes were initially identified and classified into six types, as shown in Table 4 below. (Photovoltaic power generation abnormality diagnosis result table) [Table 4]

[0073] The photovoltaic power generation abnormality diagnosis method provided in this embodiment combines with the photovoltaic power plant operation index values ​​to initially find the reasons for low power generation efficiency and obvious decline in power generation, and supports the operation department in establishing a "problem identification, feedback and improvement" mechanism for management, thereby improving the overall operation efficiency of photovoltaic power generation. Furthermore, from the photovoltaic power plant operation index values ​​such as inverter operating status, output power discontinuity rate, wear and tear status and suspension status, it is possible to further diagnose the specific reasons why the power generation of the photovoltaic power plant has decreased compared to the previous period or is significantly below the predicted value.

[0074] The procedure for diagnosing abnormalities in photovoltaic power generation will be explained using a specific example. Example 1 The specific steps for training a solar power generation forecasting model are as follows: For power plants that have already accessed the Three Gorges Energy centralized control data, the real-time irradiance data from the centralized control system and the maximum temperature data from the daily report are used, and these data are substituted into the corresponding equation. The real-time active power is used as the target value, and the regression coefficient is obtained by linear regression.

[0075] Based on the Three Gorges Energy power supervision report and the actual access data situation, the centralized control data of Shaanxi Weinan Jinyu (100MW) in October was used as the test point, and a day with no power loss was selected. After processing, one data set was used every 10 minutes (144 data sets per day), and the model coefficient was calculated as 0.977104, which is not significantly different from the daily report coefficient of the power plant, 1.0195779.

[0076] Experimental result verification 1: On October 13th, the coefficient is 0.977104. The daily power generation of the centralized control on that day is 10,285,231.67 million kW, and the model predicted power generation is 9,654,066,812 million kW. The difference between the predicted power generation and the actual power generation is small, and the trends of the predicted active power and the actual active power are generally consistent.

[0077] Experimental result verification 2: On October 22nd, the coefficient was 0.977104. The daily power generation of the centralized control on that day was 325,503,850,000 kW. The model predicted power generation was 509,546,085,700 kW. There was a large difference between the predicted power generation and the actual power generation. During the time period when the deviation between the predicted power and the actual power was large, some units were in a stopped state.

[0078] Furthermore, the above regression coefficients are used to train the preset neural network. For example, the centralized control data for Shaanxi Chunhua Shilitai (100.00MW) Power Plant in September, October, and November are used for training and testing using daytime and nighttime data, and 6,553 training datasets and 1,366 test datasets are determined proportionally. The above training datasets and test datasets are then used to train the preset neural network and generate a solar power generation prediction model.

[0079] Example 2 Taking Guazhou, Gansu Province as an example, during the process of diagnosing abnormalities in solar power generation, it was determined that the power generation efficiency of the power plant in question was low because the power generation volume for the day had increased by 5.93% compared to the previous day, which was lower than the predicted power generation volume of -25.03%. Although the resources for the day had increased by 50.25% compared to the previous day, the increase in power generation volume compared to the previous day was not at this level. All four types of power losses in the power production daily report for that day were zero, and combined with the inverter's operating status, the length of time that the inverter was in normal shutdown and externally caused shutdown states was significantly higher than the previous day. Therefore, it was initially determined that the low power generation efficiency that day was due to inspection anomalies and accompanying shutdown anomalies. Table 5 shows the inverter operation time length (unit: minutes). (Inverter operation time (unit: minutes)) [Table 5] The calculation results showed that almost all inverters maintained normal levels (DC to AC loss ≦4%), but the top 10 loss rates were higher than the power plant's average. The top 10 inverse ranking of inverter daily power generation revealed that the cumulative daily power generation of two inverters (#053, #701) on that day was significantly lower than the other inverters, and that the cumulative daily power generation of the other inverters was slightly lower than the power plant's average for that day. The inverter discrepancy rates of all power plants were low, and the inverter output conditions were relatively consistent.

[0080] Example 3 Taking Shuanggu, Zhongxian, Chongqing as an example, during the process of diagnosing abnormalities in photovoltaic power generation, it was initially determined that the power generation efficiency of the power plant was low because the power generation volume of the plant increased by 500.81% compared to the previous day compared to the previous period, which was lower than the predicted power generation volume of -22.42%. Although the resources of the plant on the day increased by 748.39% compared to the previous day compared to the previous period, the increase in power generation volume compared to the previous period did not reach this level. The failure loss reported in the daily power production report for that day was 0.001 million kW·h, and combined with the inverter's operating state, it was in a normal shutdown state for a relatively long time, so it was initially determined that the low power generation efficiency that day was due to inspection and repair work. Table 6 shows the inverter operation time (unit: minutes). (Inverter operation time (unit: minutes)) [Table 6] The calculation results showed that almost all inverters maintained normal levels (DC to AC loss ≦ 3%), but that the output was slightly higher than the previous day. The top 10 inverter daily power generation rankings show that the inverter daily power generation was higher than the previous day, but the power generation of inverter unit 1883 was clearly lower than the other inverters. The inverter discontinuity rate for the entire power plant was at a high level between 13:30 and 17:00, and some inverters had low output.

[0081] Example 4 Taking Zhaoyuan Ningsheng in Heilongjiang Province as an example, during the abnormality diagnosis of solar power generation, the plant's daily power generation was found to have decreased by -47.22% compared to the previous day, a difference of less than -1.44% from the predicted power generation, indicating a significant decrease in power generation on that day. The resource on that day also decreased by -47.22% compared to the previous day, which is similar to the decrease in power generation on that day compared to the previous day and is consistent with the direct proportional relationship of the physical model. The four types of power losses in the daily power production report on that day were all zero. Combined with the inverter's operating status, the inverter had been in an externally caused shutdown state for a relatively long time, and normal shutdowns had increased compared to the previous day. Therefore, it was initially determined that the low power generation efficiency on that day was due to resource fluctuations, accompanying shutdowns, and inspection and repair. Table 7 shows the inverter operation time length (unit: minutes). (Inverter operation time (unit: minutes)) [Table 7] The calculation results showed that all inverters basically maintained a normal level (DC to AC loss <3%), but that the day was slightly lower than the previous day. The top 10 inverter daily power generation rankings show that the daily power generation of each inverter was lower than the previous day, and that the daily power generation of each inverter was similar. The inverter discontinuity rate for the entire power plant was slightly higher at 11:00, but the daily discontinuity rate was less than 20%, and the inverter output status was relatively consistent.

[0082] Example 5 Taking the Shuangliao Power Station in Jilin as an example, during the abnormality diagnosis of photovoltaic power generation, it was found that the plant's daily power generation increased by 68.43% compared to the previous day, and the corrected power generation was within -2.56% of the predicted value, indicating a clear increase in power generation on that day. The resources on that day increased by 69.5% compared to the previous day, which is similar to the increase in power generation on that day compared to the previous day and is consistent with the direct proportional relationship of the physical model. The limit power loss recorded in the daily power production report on that day was 80,000 kW·h. Combined with the inverter's operating status, the normal shutdown and fault shutdown times on that day were reduced, so it was initially determined that the significant increase in power generation was due to resource fluctuations, faults, and reduced inspection time. Table 8 shows the inverter operation time length (unit: minutes). (Inverter operation time (unit: minutes)) [Table 8] In the above example, linear regression is based on a physical model of solar power generation, while neural networks find patterns through data learning. Linear regression does not require a large amount of training data. Neural networks require sufficient samples to achieve optimal training results. Linear regression is sensitive to anomalies, which directly affect coefficient changes. Neural networks are more inclusive of anomalies, and small anomalies do not affect training results. If training is performed using daily data, linear regression is more suitable than neural networks because the number of samples is limited (one data per day). If centralized control data (which must ensure data stability) is available, neural networks and linear regression can be used simultaneously. Additionally, data information such as irradiance resources, temperature, and weather collected by big data platforms can be used to evaluate solar power generation, identify a comprehensive list of power plants that require daily focus, and identify key targets. Data mining is performed on the power plant's operational data, and specific causes of a decrease in the amount of power generated by a solar power plant compared to the previous period or significantly below the predicted value are diagnosed from business indicators such as the inverter's operating status, output power dispersion rate, wear and tear status, and other operational indicators.

[0083] A photovoltaic power generation abnormality diagnosis device is also provided in this embodiment, which realizes the above-mentioned embodiment and preferred embodiment, and the description already given will be omitted. As used below, the term "module" may refer to a combination of software and / or hardware that realizes a predetermined function. The device described in the following embodiment is preferably realized in software, but it is also possible and envisioned to realize it in hardware or a combination of software and hardware.

[0084] The following embodiment provides an abnormality diagnosis device for photovoltaic power generation shown in FIG. A training module 101 is used to obtain operation data of the solar power plant, train a preset neural network based on the operation data of the solar power plant, and generate a solar power generation prediction model. The evaluation module 102 is used to use the solar power generation forecasting model to perform daily power generation fluctuation evaluation and power generation efficiency evaluation for each of a plurality of solar power plants, and to identify abnormal solar power plants. An abnormality diagnosis module 103 is used to compare the corresponding operation data of the abnormal photovoltaic power generation location with the abnormal conditions, and generate an abnormality diagnosis result for the photovoltaic power generation.

[0085] In some optional embodiments, the training module 101 includes: A filtering unit is used to filter the daily average irradiance data to generate filtered daily average irradiance data. A determination unit is used to determine a correction factor for a solar power plant based on filtered daily average irradiance data and cell temperature data. A training unit is used to train a preset neural network based on the correction coefficients of the solar power plant and weather data to generate power temperature correction coefficients. A first generating unit is used to generate a solar power generation forecasting model based on the filtered daily average irradiance data, the solar power plant correction factor, and the power temperature correction factor.

[0086] In some optional embodiments, the filtration unit comprises: A first determination subunit obtains power generation data corresponding to daily average irradiance data, screens the daily average irradiance data and the power generation data that are greater than a preset threshold, and determines the daily average power generation of the power plant and the daily average irradiance of the power plant. An identification sub-unit is used to filter out irradiance anomalies based on the daily power generation of the power plant and the daily average irradiance of the power plant. A pre-training subunit is used to perform model pre-training on a preset neural network based on the irradiance data after filtering out irradiance anomalies, and generate a pre-training result. a deletion subunit, which is used to determine outliers in the irradiance data after filtering out irradiance abnormal points based on the pre-training result, delete the outliers, and generate filtered daily average irradiance data;

[0087] In some alternative implementations, the identification subunit is specifically used to divide the daily power generation of the power plant into a plurality of power generation intervals, map the daily average irradiance of the power plant into the plurality of power generation intervals, generate the irradiance within each power generation interval, determine a segment box plot according to the irradiance within each power generation interval, and identify and filter irradiance anomalies based on the segment box plot.

[0088] In some alternative implementations, the determination unit is used to generate power plant correction factors using linear regression based on filtered daily average irradiance data and battery temperature data.

[0089] In some alternative implementations, the solar power generation forecasting model at the first generating unit is expressed as follows:

number

[0090] In some optional embodiments, the evaluation module 102 includes: The first calculation unit is used to obtain the power generation amount of the solar power generation location on the previous day and the real-time power generation amount of the solar power plant on the current day, and calculate the fluctuation value of the daily power generation amount based on the power generation amount of the solar power plant on the previous day and the real-time power generation amount of the solar power plant on the current day. A second generation unit is used to collect real-time operation data of the solar power plant, input the real-time operation data of the solar power plant into the solar power generation prediction model, and generate a daily predicted power generation amount of the solar power plant. The second calculation unit is used to calculate the power generation efficiency of the solar power plant according to the predicted power generation amount of the solar power plant on that day and the real-time power generation amount of the solar power plant on that day. The identification unit is used to identify abnormal solar power plants based on the fluctuation value of daily power generation and the power generation efficiency of the solar power plant.

[0091] In some optional embodiments, the anomaly diagnosis module 103 includes: A third calculation unit is used to calculate the performance index value of the photovoltaic power plant according to the corresponding operation data of the abnormal photovoltaic power plant. A comparison unit is used to compare the operational index values ​​of the solar power plant with the abnormal conditions and generate an abnormal diagnosis result for the solar power generation.

[0092] In some optional embodiments, the third computing unit includes: A first calculation sub-unit is used to determine the daily average irradiance of the abnormal solar power plant on the current day and the daily average irradiance of the previous day according to the corresponding operation data of the abnormal solar power plant, and to calculate a resource change value according to the daily average irradiance of the current day and the daily average irradiance of the previous day. The second calculation sub-unit is used to determine the power loss amount of each device according to the corresponding operation data of the abnormal solar power plant, and calculate the power loss amount value according to the power loss amount of each device. The third calculation sub-unit is used to determine the inverter quantity of the power station and the unit status data of the power station inverter according to the corresponding operation data of the abnormal solar power station, and to calculate the inverter operation status statistics according to the inverter quantity of the power station and the unit status data of the power station inverter.

[0093] In some alternative embodiments, the third computing unit further includes: The fourth calculation sub-unit is used to determine the output power of the power station inverter according to the corresponding operation data of the abnormal solar power station, and to calculate the discrete rate of the inverter output power according to the output power of the power station inverter and the number of power station inverters. The fifth calculation sub-unit is used to determine the input power of the power station inverter, the effective area of ​​the photovoltaic power grid and the total radiation dose on the slope according to the corresponding operation data of the abnormal photovoltaic power station, and to calculate the efficiency of the photovoltaic power grid according to the input power of the power station inverter, the effective area of ​​the photovoltaic power grid and the total radiation dose on the slope. A sixth calculation sub-unit is used to determine the inverter DC power and the inverter AC power according to the corresponding operation data of the abnormal solar power plant, and to calculate the inverter loss value according to the inverter DC power and the inverter AC power. A seventh calculation sub-unit is used for determining the daily power generation of the power station inverter according to the corresponding operation data of the abnormal solar power station, and for determining the inverter shutdown state according to the daily power generation of the power station inverter.

[0094] In some optional embodiments, the fourth calculation subunit is specifically used to determine the standard difference and average value of the power station inverter output power based on the output power of the power station inverter and the number of power station inverters, and calculate the discrete rate of the inverter output power based on the standard difference and average value of the power station inverter output power.

[0095] In some optional embodiments, the comparison unit comprises: The comparison subunit is used to compare the resource variation value with a preset resource threshold value, and determine that the abnormality diagnosis result of the photovoltaic power generation is a resource variation abnormality if the resource variation value is greater than the preset resource threshold value. The second determining sub-unit is used to determine the breakdown power loss amount, the inspection and repair power loss amount, the power limit power loss amount and the accompanying shutdown power loss amount according to the power loss amount value. The third determination sub-unit is used to determine the ratio of the length of fault shutdown time to the previous period, the ratio of the length of normal shutdown time to the previous period, the ratio of the length of power limit operation and shutdown time to the previous period, and the ratio of the length of externally caused shutdown time to the previous period based on the inverter operating state statistics. The first diagnostic subunit is used to diagnose the abnormality diagnosis result of the solar power generation as a fault abnormality when the amount of power loss due to the fault is smaller than a preset threshold value for the amount of power loss, or when the length of the fault outage time is smaller than a preset threshold value for the length of the outage time compared to the previous period. The second diagnostic sub-unit is used to diagnose the solar power generation abnormality as an inspection and repair abnormality if the inspection and repair power loss amount is smaller than the preset power loss amount threshold, or the length of normal shutdown time is smaller than the preset shutdown time length threshold compared to the previous period. The third diagnostic sub-unit is used to diagnose the abnormality diagnosis result of the photovoltaic power generation as a power limit abnormality when the amount of power loss under the power limit is smaller than a preset power loss threshold, or the length of the power limit operation and shutdown time is smaller than a preset shutdown time length threshold compared to the previous period. The fourth diagnostic sub-unit is used to diagnose the abnormality diagnosis result of the solar power generation as an accompanying shutdown abnormality when the amount of accompanying shutdown power loss is smaller than a preset power loss threshold, or the length of the externally caused shutdown time is smaller than a preset shutdown time length threshold compared to the previous period.

[0096] The further functional description of each module and unit described above is the same as that of the corresponding embodiment described above, and will not be described here. The photovoltaic power generation abnormality diagnosis in the above-described embodiments is illustrated in the form of functional units, which may refer to ASIC (Application Specific Integrated Circuit) circuits, processors and memories executing one or more software or fixed programs, and / or other devices capable of providing the above-described functionality.

[0097] There is also provided a computer device having the above-described photovoltaic power generation abnormality diagnosis shown in FIG. Referring to FIG. 11, FIG. 11 is a schematic diagram of a computer device according to an embodiment of the present invention. As shown in FIG. 11, the computer device includes one or more processors 10, a memory 20, and interfaces for connecting components, including high-speed and low-speed interfaces. The components are communicatively connected to each other via different buses and may be implemented on a common motherboard or in other ways, as needed. The processor can process instructions executed within the computer device, including instructions stored in or on memory, to display graphical information for a GUI on an external input / output device, such as a display device, coupled to the interface. In some alternative embodiments, multiple processors and / or multiple buses can be used, along with multiple memories, as needed. Similarly, multiple computer devices can be connected, each providing a portion of the required operations (e.g., an array of servers, a set of blade servers, or a multiprocessor system). FIG. 11 illustrates a single processor 10 as an example.

[0098] The processor 10 may be a central processor, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a composite programmable logic device, a field programmable logic gate array, a general purpose array logic, or any combination thereof.

[0099] The aforementioned memory 20 stores instructions executable by the at least one processor 10 to cause the at least one processor 10 to implement the methods illustrated in the above-described embodiments. The memory 20 may include a program storage area and a data storage area. The program storage area may store an operating system and / or application programs required for at least one function. The data storage area may store data generated in response to use of the computer device. The memory 20 may also include high-speed random access memory or non-transitory memory such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may include memory located remotely from the processor 10, and this remote memory may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof. Memory 20 may include volatile memory, such as random access memory. Memory may also include non-volatile storage, such as flash memory, hard disk drives, solid state drives, etc. Memory 20 may also include a combination of the above types of memory.

[0100] The computer device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30, and the output device 40 may be connected via a bus or other methods, and FIG. 11 shows an example of connection via a bus. The input device 30 can receive input numeric or character information and generate key signal inputs associated with user settings and function control of the computing device, such as a touchscreen, keypad, mouse, trackpad, touchpad, indicator stick, one or more mouse buttons, trackball, joystick, etc. The output device 40 can include a display device, auxiliary lighting devices (e.g., LEDs), and haptic feedback devices (e.g., vibration motors), etc. Display devices include, but are not limited to, liquid crystal displays, light-emitting diode displays, and plasma displays. In some alternative embodiments, the display device may be a touchscreen.

[0101] Embodiments of the present invention may also provide a computer-readable storage medium that can implement the methods described above as hardware, firmware, or computer code that can be recorded on a storage medium, or stored on a remote storage medium or non-transitory machine-readable storage medium downloaded via a network, or stored on a local storage medium. Therefore, the methods described herein can be processed by a general-purpose computer, a dedicated processor, or programmable or dedicated hardware using software stored on a storage medium. The storage medium may be a magnetic disk, optical disk, read-only memory, random-access memory, flash memory, hard disk, solid-state drive, or the like. Furthermore, the storage medium may also include a combination of the above types of memory. It is understood that the computer, processor, microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code, and that the software or computer code, when accessed and executed by the computer, processor, or hardware, implements the methods described in the above embodiments. (Addendum) (Appendix 1) A method for diagnosing an abnormality in a solar power generation system, comprising the following steps: Step 1: Obtain the operation data of the solar power plant, train the preset neural network based on the operation data of the solar power plant, and generate a solar power generation prediction model; Step 2: Using the solar power generation forecasting model, evaluate the daily power generation fluctuations and power generation efficiency for multiple solar power plants, and identify abnormal solar power plants. Step 3: Compare the corresponding operation data of the abnormal solar power plant with the abnormal conditions, and generate an abnormal diagnosis result for the solar power plant; In step 1, the operation data of the solar power plant includes daily average irradiance data, battery temperature data and weather data; In the step 1, the generation of the solar power generation forecast model includes the following steps: A1: Filtering the daily average irradiance data to generate filtered daily average irradiance data; A2: Determine the correction coefficient of the solar power plant based on the filtered daily average irradiance data and battery temperature data; A3: Based on the solar power plant correction coefficient and weather data, a preset neural network is trained to generate a power temperature correction coefficient; A4: Generate a solar power generation forecast model based on the filtered daily average irradiance data, the solar power plant correction coefficient and the power temperature correction coefficient; In A4, the formula for the solar power generation prediction model is as follows:

number

Claims

1. A method for diagnosing an abnormality in a solar power generation system, comprising the following steps: Step 1: Obtaining operation data of a solar power plant, training a preset neural network based on the operation data of the solar power plant, and generating a solar power generation prediction model; Step 2: Using the solar power generation forecasting model, a daily power generation fluctuation evaluation and a power generation efficiency evaluation are performed for each of the plurality of solar power plants, and an abnormal solar power plant is identified; Step 3: Comparing the corresponding operation data of the abnormal solar power plant with the abnormal conditions, and generating an abnormal diagnosis result for the solar power plant; In step 1, the operation data of the solar power plant includes daily average irradiance data, battery temperature data and weather data; In the step 1, the generation of the solar power generation prediction model includes the following steps: A1: filtering the daily average irradiance data to generate filtered daily average irradiance data; A2: Determine a correction coefficient for the solar power plant based on the filtered daily average irradiance data and the battery temperature data; A3: Train a preset neural network based on the correction coefficient of the solar power plant and the weather data to generate a power temperature correction coefficient; A4: Generate the solar power generation forecast model based on the filtered daily average irradiance data, the solar power plant correction coefficient, and the power temperature correction coefficient; In A4, the formula of the solar power generation prediction model is as follows: [Equation 1] In the above formula, PE ｄ indicates the predicted power generation amount of the solar power plant on that day, k indicates the correction coefficient of the solar power plant, and P ｐｒｅｄＴａｒｇ indicates the amount of power generated by the solar power generation module under the target environmental conditions, and G ｍｅａｓ indicates the irradiance of the same plane unit area of ​​the solar power generation module, and G ＴＲＣ represents the daily average irradiance corresponding to the amount of power generated by the photovoltaic power generation module under the target environmental conditions, δ represents the power temperature correction coefficient, and T Ｃ indicates the actual measured value of the solar power generation module battery temperature, and T ＴＲＣ indicates the battery temperature value corresponding to the amount of power generated by the photovoltaic power generation module under the target environmental conditions, Step 2 includes the following steps: B1: Obtain the power generation amount of each solar power plant on the day of the event and the day before the day of the event, and calculate the daily power generation fluctuation value based on the power generation amount of each solar power plant on the day of the event and the day before the day of the event; B2: Collect real-time operation data of each solar power plant, input the real-time operation data of each solar power plant into each solar power generation forecast model, and generate the predicted power generation amount of each solar power plant on that day; B3: Calculate the power generation efficiency of each solar power plant based on the predicted power generation amount of each solar power plant on that day and the real-time power generation amount of each solar power plant on that day; B4: Identifying the abnormal photovoltaic power plant based on each daily power generation fluctuation value and the power generation efficiency of each photovoltaic power plant; Step 3 includes the following steps: C1: Calculate the operational index value of the solar power plant according to the corresponding operation data of the abnormal solar power plant; C2: Comparing the operational index value of the solar power plant with the abnormal condition, and generating an abnormality diagnosis result for the solar power generation; In C1, the performance index values ​​of the photovoltaic power station include a resource change value, a power loss value, and an inverter operation status statistic value, and the performance index values ​​of the photovoltaic power station are calculated according to the corresponding operation data of the abnormal photovoltaic power station, and the calculation includes: c1: determining the daily average irradiance of the abnormal solar power plant on the current day and the daily average irradiance of the previous day according to the corresponding operation data of the abnormal solar power plant, and calculating the resource change value relative to the previous day according to the daily average irradiance of the current day and the daily average irradiance of the previous day; c2: Determine the power loss values ​​of the photovoltaic power plant on the current day and the previous day according to the corresponding operation data of the abnormal photovoltaic power plant; c3: determine the number of inverters in the photovoltaic power station and the unit status data of the inverters in the photovoltaic power station according to the corresponding operation data of the abnormal photovoltaic power station, and calculate the inverter operation status statistics of the current day and the previous day according to the number of inverters in the photovoltaic power station and the unit status data of the inverters in the photovoltaic power station; The performance index value of the photovoltaic power station also includes an inverter output power discontinuity rate, a photovoltaic power grid energy conversion efficiency, an inverter loss rate and an inverter stop situation, and calculates the performance index value of the photovoltaic power station according to the corresponding operation data of the abnormal photovoltaic power station, and further includes: c4: determine the output power of the inverters of the abnormal solar power station according to the corresponding operation data of the abnormal solar power station, and calculate the inverter output power discontinuity rate in the solar power station according to the output power of each inverter of the solar power station; c5: determine the DC power amount of the inverter of the solar power station, the effective area of ​​the solar power grid and the total radiation dose of the slope of the solar power grid according to the corresponding operation data of the abnormal solar power station; and calculate the energy conversion efficiency of the solar power grid according to the DC power amount of the inverter of the solar power station, the effective area of ​​the solar power grid and the total radiation dose of the slope of the solar power grid; c6: determine the inverter DC power and the inverter AC power according to the corresponding operation data of the abnormal solar power plant, and calculate the inverter loss rate according to the inverter DC power and the inverter AC power; c7: determine the daily power generation amount of the inverter of the photovoltaic power station according to the corresponding operation data of the abnormal photovoltaic power station, and determine the stop status of the inverter according to the daily power generation amount of the inverter of the photovoltaic power station; In step c4, determining a standard deviation and an average value of the inverter output power of the photovoltaic power station according to the inverter output power of the photovoltaic power station and the number of inverters of the photovoltaic power station, and calculating a discrete rate of the inverter output power in the photovoltaic power station according to the standard deviation and the average value of the inverter output power of the photovoltaic power station; The C2 step includes the following steps: c2.1: comparing the resource change rate relative to the previous day with a preset threshold, and if the resource change rate relative to the previous day is greater than the preset threshold, the abnormality diagnosis result of the photovoltaic power generation is due to resource fluctuation; c2.2: Based on the power loss value and the photovoltaic power plant power generation value of the current day and the previous day, determine the ratio of the power loss amount due to failure, the power loss amount due to inspection and repair, the power loss amount due to power limit, and the power loss amount due to accompanying shutdown, and the ratio compared to the previous day; c2.3: Based on the inverter operation status statistics, determine the length of the fault shutdown time compared to the previous day, the length of the normal shutdown time compared to the previous day, the length of the power limit operation shutdown time compared to the previous day, and the length of the external cause shutdown time compared to the previous day; c2.4: If the proportion of the amount of power loss due to the failure is greater than a preset threshold, or if the proportion of the length of the outage time due to the failure or the ratio to the previous day is greater than a preset threshold for the length of the outage time, the abnormality diagnosis result of the photovoltaic power generation is due to an equipment failure, If the proportion of the power loss amount due to inspection and repair is greater than a preset threshold value of the power loss amount, or if the proportion of the length of the normal shutdown time or the ratio to the previous day is greater than a preset threshold value of the shutdown time length, the diagnosis result of the solar power generation abnormality is that it is due to inspection and repair, If the ratio of the power loss amount of the power limit is greater than a preset threshold value of the power loss amount, or if the ratio of the power limit operation to the length of the stopped time or the ratio to the previous day is greater than a preset threshold value of the length of the stopped time, the diagnosis result of the solar power generation abnormality is due to the power limit, If the proportion of the power loss amount due to the accompanying outage is greater than a preset threshold value for the power loss amount, or if the proportion of the length of the outage time due to the external cause or the ratio to the previous day is greater than a preset threshold value for the length of the outage time, the diagnosis result of the solar power generation abnormality is due to the accompanying outage. A method for diagnosing an abnormality in a solar power generation system.

2. A1 includes the following steps: a1: obtaining power generation data corresponding to the daily average irradiance data, and screening the daily average irradiance data and the power generation data that are greater than a preset threshold to determine the daily power generation of the solar power station and the daily average irradiance of the solar power station; a2: Identifying irradiance abnormal points based on the daily power generation amount of the solar power station and the daily average irradiance of the solar power station, and filtering the irradiance abnormal points; a3: performing pre-training on the preset neural network based on the irradiance data after filtering out the irradiance anomaly points, and generating the pre-training result; a4: based on the result of the pre-training, determine outliers in the irradiance data after filtering the irradiance abnormal points, delete the outliers, and generate the filtered daily average irradiance data; 2. The method for diagnosing an abnormality in a photovoltaic power generation system according to claim 1.

3. a2 includes the following steps: a2.1: Dividing the daily power generation amount of the solar power plant into a plurality of power generation amount intervals, and mapping the daily average irradiance of the solar power plant into the plurality of power generation amount intervals to generate the irradiance within each power generation amount interval; a2.2: determining a segment box plot based on the irradiance in each power generation interval, and identifying and filtering the irradiance anomaly points based on the segment box plot; 3. The method for diagnosing an abnormality in a photovoltaic power generation system according to claim 2.

4. In the step A2, a correction coefficient for the solar power plant is generated based on the filtered daily average irradiance data and the battery temperature data using a linear regression method.

2. The method for diagnosing an abnormality in a photovoltaic power generation system according to claim 1.

5. A computer facility comprising: The method includes a memory and a processor, the memory and the processor are communicatively connected to each other, computer instructions are stored in the memory, and the processor executes the computer instructions to perform the photovoltaic power generation abnormality diagnosis method according to any one of claims 1 to 4.

1. A computer facility characterized by:

6. A computer-readable storage medium, comprising: A computer instruction is stored in the computer-readable storage medium, and the computer instruction causes a computer to execute the photovoltaic power generation abnormality diagnosis method according to any one of claims 1 to 4. A computer-readable storage medium comprising:

Images