Remote monitoring method for photovoltaic power station

Abstract

The invention discloses a remote monitoring method and system for a photovoltaic power station, and a computing device, relates to the technical field of photovoltaic power generation, and solves the problem that in the prior art, dust or accumulated snow covering the surface of a photovoltaic module of the photovoltaic power station is not considered when the generating capacity of the photovoltaic power station is predicted. The light transmittance is reduced; and the generating capacity is reduced. The method comprises the following steps: extracting historical data of a photovoltaic power station in a statistical period and data features of the historical data; inputting the data features into a prediction model, and predicting the generating capacity of the target time period; performing abnormal prediction detection according to a difference relationship between the predicted generating capacity and the actual generating capacity to obtain an evaluation label of the predicted generating capacity; if the evaluation label of the predicted generating capacity is abnormal, correcting the predicted generating capacity by constructing a correction factor to obtain the corrected predicted generating capacity; and if the corrected evaluation label of the predicted generating capacity is normal, remote monitoring of the photovoltaic power station is completed.

Description

Technical Field

[0001] The present invention belongs to the field of photovoltaic power generation, relates to remote monitoring technology, and specifically is a remote monitoring method for a photovoltaic power station. Background Art

[0002] In a photovoltaic power plant, the amount of electricity generated by photovoltaic cells is positively correlated with the light intensity. That is, the higher the light intensity, the greater the current generated by photons exciting electrons, and thus the higher the power generation. At the same time, the power generation is also affected by temperature. Since the efficiency of photovoltaic cells in a photovoltaic power plant decreases with increasing temperature, when low light and high temperature occur in the environment, the power generation efficiency of the power plant will be reduced.

[0003] When predicting the power generation of a photovoltaic power plant, the existing technology does not take into account the possibility that dust or snow covers the surface of the photovoltaic modules of the photovoltaic power plant, causing partial shadows or complete obstruction, thereby reducing the transmittance of the photovoltaic modules. Since the power generation generated by photovoltaic cells is positively correlated with the light intensity, that is, the lower the light intensity, the smaller the current generated by photons exciting electrons, which leads to a technical problem of reduced power generation.

[0004] The present invention provides a remote monitoring method for a photovoltaic power station to solve the above technical problems. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a remote monitoring method for a photovoltaic power station. This method addresses the technical problem that, in the prior art, power generation predictions for a photovoltaic power station fail to take into account dust or snow covering the surface of the photovoltaic modules, causing partial or complete shadowing and thus reducing the light transmittance of the modules. This can lead to reduced power generation.

[0006] To achieve the above objectives, a first aspect of the present invention provides a remote monitoring method for a photovoltaic power station, comprising:

[0007] Extract historical data of photovoltaic power stations within the statistical period and the data characteristics of the historical data;

[0008] Inputting the data features into a prediction model to predict power generation in a target time period; the target time period is after the statistical period; and obtaining actual power generation in the target time period;

[0009] Perform abnormal prediction detection based on the difference between the predicted power generation and the actual power generation to obtain the evaluation label of the predicted power generation;

[0010] If the evaluation label of the predicted power generation is abnormal, the predicted power generation is corrected by constructing a correction factor to obtain the corrected predicted power generation;

[0011] Determine whether the evaluation label of the revised predicted power generation is normal; if yes, complete remote monitoring of the corresponding photovoltaic power station; if not, issue a maintenance signal.

[0012] Preferably, the method of obtaining the data features includes:

[0013] Acquire historical data; wherein the historical data includes: time data, environmental data, and photovoltaic module data; and associate the time data with the timestamps of the environmental data and the photovoltaic module data; wherein the environmental data includes: temperature, rainfall, and light values; and the photovoltaic module data includes: the number and type of photovoltaic modules;

[0014] Set a statistical period; calculate the change characteristics of environmental data and photovoltaic module data within the statistical period, and mark the change characteristics as data characteristics of historical data; the data characteristics include: change characteristics of environmental data and change characteristics of photovoltaic module data; change characteristics of environmental data include: daily average and extreme values ​​of temperature, rainfall and light values ​​in the statistical period, and the average value of power generation in the secondary period with 3 days, 7 days and 30 days respectively in the statistical period; the change characteristics of photovoltaic module data include: the cumulative operating time of the photovoltaic module and the time since the previous maintenance; the number of failures of the photovoltaic module in the statistical period.

[0015] It should be noted that the value of the statistical period is greater than 30 days.

[0016] The present invention constructs a feature system based on multi-dimensional data such as historical meteorology, equipment status, and seasonal patterns, so that the subsequent prediction model can capture the fluctuation pattern of power generation, which is conducive to reducing prediction errors.

[0017] Preferably, the training process of the prediction model satisfies:

[0018] Normalize the data features of the historical data and remove outliers using the quantile method; integrate the data features of the processed historical data into a feature matrix X t×m , integrate the power generation on and before the rth day in the statistical period to form a matrix Y with t rows and 1 column t×1 ; m is the number of changing features;

[0019] Based on the feature matrix X t×12 and the matrix Y 1×t Construct a prediction model; the prediction model satisfies the formula: Z t×1 =(X t×m T X t×m ) -1 X t×m T Y1×t ;

[0020] Among them, Z t×1 is the prediction matrix composed of the power generation of the predicted r+1 day, X t×m T For X t×m The transposed matrix, (X t×m T X t×m ) -1 For X t×m T X t×m The inverse matrix of .

[0021] Preferably, inputting the data features into a prediction model to predict the power generation in a target time period includes:

[0022] Based on the prediction model, the predicted power generation in the target time period is predicted to obtain t prediction results, which form a prediction matrix Z t×1 ;

[0023] By formula Calculate the average predicted power generation on and before the rth day in the statistical period; where n represents the total number of days in the statistical period; the prediction matrix Z t×1 Each value in is substituted into the preset objective function in turn, and the value that satisfies the objective function is taken as the predicted power generation.

[0024] Preferably, the preset objective function includes:

[0025] Retrieve the prediction matrix Z t×1 The t-th predicted power generation z in t , and the average predicted power generation on and before the rth day in the statistical period

[0026] By formula Calculate the objective function.

[0027] Preferably, the abnormality prediction detection based on the difference between the predicted power generation and the actual power generation includes:

[0028] S110: Retrieving the actual power generation and the predicted power generation on the same day, and calculating the difference between the actual power generation and the predicted power generation; presetting a first threshold and a second threshold, wherein the first threshold is less than the second threshold;

[0029] S120: When the difference between the actual power generation and the predicted power generation is greater than a first threshold and less than a preset second threshold, and if the difference is still greater than the first threshold and less than the preset second threshold for more than three days within the corresponding statistical period, a reminder signal is issued to arrange maintenance personnel to perform maintenance;

[0030] When the difference between the actual power generation and the predicted power generation is greater than or equal to the second threshold, an early warning signal is issued and maintenance personnel perform maintenance;

[0031] S130: When the difference between the actual power generation and the predicted power generation is greater than a first threshold, the evaluation label of the predicted power generation obtained by the prediction model is marked as abnormal;

[0032] When the difference between the actual power generation and the predicted power generation is less than or equal to the first threshold, the evaluation label of the predicted power generation obtained by the prediction model is recorded as normal.

[0033] Preferably, the correction of the predicted power generation by constructing a correction factor includes:

[0034] The dust sensor on the photovoltaic module is used to obtain the shielding area S0, and the snow thickness h is obtained by the snow sensor. s ;Record the clean panel efficiency of the PV module as b0;

[0035] By formula The photovoltaic loss rate θ under natural dust obstruction is calculated t ; where t dry is the number of days without rainfall during the statistical period;

[0036] By formula The photovoltaic loss rate δ when snow is blocked is calculated t ;

[0037] Determine whether there is snow on the photovoltaic modules during the statistical period; if so, the photovoltaic loss rate δ when the snow is blocked t As a correction factor, the photovoltaic loss rate θ when the dust is blocked is t Calibrated to 0; otherwise, the photovoltaic loss rate θ when dust is blocked t As a correction factor, the photovoltaic loss rate θ when the snow is blocked is t Calibrated to 0;

[0038] By formula Z t '=Z t (δ t +θ t )+Z t Calculate the revised forecast power generation Z t '.

[0039] It should be noted that the number of days without rainfall in the statistical period is: the number of days between the end of the previous rainfall and the start of the next rainfall in the statistical period; when S0 or t dry When θ is 0, set tThe cleaning panel efficiency of the photovoltaic module is the cleaning efficiency of the cleaner deployed in the photovoltaic module on the obstruction, and the specific value is calibrated through experiments.

[0040] Preferably, the step of determining whether the evaluation label of the revised predicted power generation is normal includes:

[0041] Retrieve the revised predicted power generation and repeat steps S110-S130. If the difference between the actual power generation and the revised predicted power generation is less than or equal to the preset first-level threshold, the evaluation label of the revised predicted power generation is recorded as normal; otherwise, the evaluation label of the revised predicted power generation is recorded as abnormal.

[0042] To achieve the above-mentioned object, a second aspect of the present invention provides a remote monitoring system for a photovoltaic power station, comprising: a data acquisition module, an abnormality monitoring module, and a correction feedback module;

[0043] Data acquisition module: extracts historical data of photovoltaic power stations within the statistical period and the data characteristics of historical data;

[0044] Inputting the data features into a prediction model to predict power generation in a target time period; the target time period is after the statistical period; and obtaining actual power generation in the target time period;

[0045] Abnormal monitoring module: performs abnormal prediction detection based on the difference between predicted power generation and actual power generation, and obtains the evaluation label of the predicted power generation;

[0046] Correction feedback module: If the evaluation label of the predicted power generation is abnormal, the predicted power generation is corrected by constructing a correction factor to obtain the corrected predicted power generation;

[0047] Determine whether the evaluation label of the revised predicted power generation is normal; if yes, complete remote monitoring of the corresponding photovoltaic power station; if not, issue a maintenance signal.

[0048] To achieve the above-mentioned objectives, the third aspect of the present invention provides a remote monitoring computing device for a photovoltaic power station, comprising: a memory and a processor, wherein the memory stores executable instructions of the processor; wherein the processor is configured to execute a remote monitoring method for a photovoltaic power station provided in the first aspect by executing the executable instructions.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] 1. This invention allows operation and maintenance personnel to analyze the type and level of faults and their impact on the power generation of the power station, and conduct selective emergency repairs or batch fault elimination during routine operation and maintenance, making each fault elimination more targeted and reducing blind operation and maintenance costs; it also saves equipment investment and maintenance costs throughout its life cycle: after the system is put into operation, through uninterrupted monitoring and prediction, it can identify and warn equipment failures before they occur, reducing losses from unplanned downtime; combined with real-time power generation analysis, it can accurately grasp the power generation efficiency of each device, and promptly identify and resolve problems with power generation efficiency decline. By optimizing equipment operating parameters, adjusting maintenance plans, and other means, it is possible to effectively increase overall power generation and increase the revenue of photovoltaic power plants.

[0051] 2. The present invention uses historical data to construct environmental parameters (irradiance, temperature), equipment status (attenuation rate of photovoltaic modules) and production data (number of failures) data features, so that the model can capture complex nonlinear relationships, so that the subsequent prediction model can capture the fluctuation pattern of power generation, which is conducive to reducing prediction errors; identifying abnormal patterns through the deviation between predicted values ​​and actual values ​​is conducive to issuing early warning signals before major accidents occur in power plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 Schematic diagram of the specific steps of remote monitoring of the present invention;

[0054] Figure 2 A schematic diagram of the process of constructing the prediction model of the present invention;

[0055] Figure 3 Schematic diagram of the process of abnormal power generation monitoring of the present invention;

[0056] Figure 4 This is a schematic diagram of the relationship between the modules included in the present invention. DETAILED DESCRIPTION

[0057] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0058] See also Figure 1A first embodiment of the present invention provides a remote monitoring method for a photovoltaic power station, comprising:

[0059] Extract historical data of photovoltaic power stations within the statistical period and the data characteristics of the historical data;

[0060] Input the data features into the prediction model to predict the power generation in the target time period; wherein the target time period is after the statistical period; and obtain the actual power generation in the target time period;

[0061] Perform abnormal prediction detection based on the difference between the predicted power generation and the actual power generation to obtain the evaluation label of the predicted power generation;

[0062] If the evaluation label of the predicted power generation is abnormal, the predicted power generation is corrected by constructing a correction factor to obtain the corrected predicted power generation;

[0063] Determine whether the evaluation label of the revised predicted power generation is normal; if yes, complete remote monitoring of the corresponding photovoltaic power station; if not, issue a maintenance signal.

[0064] See also Figure 2 The process of building a prediction model is as follows: obtaining historical data; the historical data includes time data, environmental data, and photovoltaic module data; matching the time data with the timestamps of the environmental data and photovoltaic module data; the environmental data includes temperature, rainfall, and light values; the photovoltaic module data includes the number and type of photovoltaic modules;

[0065] Set a statistical period; calculate the change characteristics of environmental data and photovoltaic module data within the statistical period, and mark the change characteristics as data characteristics of historical data;

[0066] The calculation of the changing characteristics of environmental data includes: calculating the average power generation within the secondary cycle of 3 days, 7 days and 30 days respectively; calculating the daily average and extreme values ​​of temperature, rainfall and sunlight;

[0067] Calculating the changing characteristics of PV module data includes: calculating the cumulative operating time of the PV module, the time since the last maintenance, and the number of PV module failures within the statistical period;

[0068] Normalize the data features of the historical data and remove outliers using the quantile method; integrate the data features of the processed historical data into a feature matrix X t×m , integrate the power generation on and before the rth day in the statistical period to form a matrix Y with t rows and 1 column t×1 ; m is the number of changing features;

[0069] Based on the feature matrix Xt×12 and the matrix Y 1×t Construct a prediction model; the prediction model satisfies the formula: Z t×1 =(X t×m T X t×m ) -1 X t×m T Y 1×t ;

[0070] Among them, Z t×1 is the prediction matrix composed of the power generation of the predicted r+1 day, X t×m T For X t×m The transposed matrix, (X t×m T X t×m ) -1 For X t×m T X t×m The inverse matrix of

[0071] Based on the prediction model, the predicted power generation in the target time period is predicted to obtain t prediction results, which form a prediction matrix Z t×1 ;

[0072] By formula Calculate the average predicted power generation on and before the rth day in the statistical period; where n represents the total number of days in the statistical period;

[0073] The prediction matrix Z t×1 Each value in is substituted into the preset objective function in turn, and the value that satisfies the objective function is taken as the predicted power generation;

[0074] Retrieve the prediction matrix Z t×1 The t-th predicted power generation z in t , and the average predicted power generation on and before the rth day in the statistical period

[0075] By formula Calculate the objective function.

[0076] For example, time data includes: daily weather conditions; environmental data includes: temperature and humidity values; photovoltaic module data includes: photovoltaic module loss rate and actual power generation;

[0077] Data field definition: Time data is the date (2023-10-01 to 2023-10-31), and the timestamp is unified as 12:00 every day (when the sunlight is strongest at noon);

[0078] Environmental data (daily records): Temperature value (℃): average temperature of the day;

[0079] Rainfall (mm): total rainfall for the day;

[0080] Light value (h): number of effective light hours on the day (sunshine ≥ 10% irradiance);

[0081] The changing characteristics of PV module data include:

[0082] Cumulative running time (h): 659h;

[0083] Total running time since the last month's maintenance; Time since the last maintenance (d): the number of days from the last maintenance date to the current day;

[0084] Number of failures (times): whether a failure occurred on that day (0 / 1);

[0085] The statistical period is set to 30 days (2023-10-01 to 2023-10-30) to train the model to predict the power generation on October 31 and further predict the power generation on November 1;

[0086] The changing characteristics of environmental data include:

[0087] Average power generation over three days (e.g. October 29-31): (100 + 98 + 102) / 3 = 100 kWh;

[0088] Average power generation over 7 days (e.g., October 25-31): (95+98+100+99+102+101+100) / 7=99.3kWh;

[0089] 30-day average power generation (full cycle): Where, Pt is the power generation on day r);

[0090] The daily average temperature is 23.5°C, with extreme values ​​including: a maximum of 30°C and a minimum of 18°C;

[0091] Rainfall: daily average 0.2 mm, extreme values ​​(maximum 5 mm, minimum 0 mm);

[0092] Light value: daily average value 5.8h, extreme value (maximum 8h, minimum 3h);

[0093] During the statistical period, the above-mentioned change characteristics are marked as data features, the data of the data features are normalized, and outliers are removed by the quantile method; the processed data features are integrated to form a feature matrix X t×m , integrate the power generation on and before the rth day in the statistical period to form a matrix Y with t rows and 1 column t×1 ; The feature matrix Xt×12 and the matrix Y 1×t As input data to train the prediction model; where m is the number of changing features;

[0094] By formula Z t×1 =(X t×m T X t×m ) -1 X t×m T Y 1×t Calculate the predicted power generation z on day r+1 t The prediction matrix Z consists of t rows and 1 column t×1 Where, X t×m T For X t×m The transposed matrix, (X t×m T X t×m ) -1 For X t×m T X t×m The inverse matrix of

[0095] By formula Calculate the average predicted power generation on and before the rth day in the statistical period; where n represents the total number of days in the statistical period;

[0096] The prediction matrix Z t×1 The tth z in t The values ​​are brought into the preset objective function in turn, and the z that satisfies the objective function is t The estimated power generation is 252KWh.

[0097] See also Figure 3 , the process of abnormal power generation monitoring:

[0098] S110: Retrieving the actual power generation and the predicted power generation on the same day, and calculating the difference between the actual power generation and the predicted power generation; presetting a first threshold and a second threshold, wherein the first threshold is less than the second threshold;

[0099] S120: When the difference between the actual power generation and the predicted power generation is greater than a first threshold and less than a preset second threshold, and if the difference is still greater than the first threshold and less than the preset second threshold for more than three days within the corresponding statistical period, a reminder signal is issued to arrange maintenance personnel to perform maintenance;

[0100] When the difference between the actual power generation and the predicted power generation is greater than or equal to the second threshold, an early warning signal is issued and maintenance personnel perform maintenance;

[0101] S130: When the difference between the actual power generation and the predicted power generation is greater than a first threshold, the evaluation label of the predicted power generation obtained by the prediction model is marked as abnormal;

[0102] When the difference between the actual power generation and the predicted power generation is less than or equal to a first threshold, the evaluation label of the predicted power generation obtained by the prediction model is recorded as normal;

[0103] The dust sensor on the photovoltaic module is used to obtain the shielding area S0, and the snow thickness h is obtained by the snow sensor. s ;Record the clean panel efficiency of the PV module as b0;

[0104] By formula The photovoltaic loss rate θ under natural dust obstruction is calculated t ; where t dry is the number of days without rainfall during the statistical period;

[0105] By formula The photovoltaic loss rate δ when snow is blocked is calculated t ;

[0106] Determine whether there is snow on the photovoltaic modules during the statistical period; if so, the photovoltaic loss rate δ when the snow is blocked t As a correction factor, the photovoltaic loss rate θ when the dust is blocked is t Calibrated to 0; otherwise, the photovoltaic loss rate θ when dust is blocked t As a correction factor, the photovoltaic loss rate θ when the snow is blocked is t Calibrated to 0;

[0107] By formula Z t '=Z t (δ t +θ t )+Z t Calculate the revised forecast power generation Z t ';

[0108] Retrieve the revised predicted power generation and repeat steps S110-S130. If the difference between the actual power generation and the revised predicted power generation is less than or equal to the preset first-level threshold, the evaluation label of the revised predicted power generation is recorded as normal; otherwise, the evaluation label of the revised predicted power generation is recorded as abnormal.

[0109] For example, the power generation predicted before November is 252KWh on November 1, while the actual power generation on November 1 is 240KWh.

[0110] Now set the first threshold (Th1) = 10KWh, the second threshold (Th2) = 20KWh;

[0111] At this time, the difference of 12KWh between the actual power generation of 240KWh and the predicted power generation of 252KWh is greater than the first threshold and less than the preset second threshold. At the same time, the absolute value of the difference between the predicted value and the actual power generation value on November 2, November 3, and November 4 is lower than the first threshold and less than the preset second threshold. A reminder signal is issued and maintenance personnel are arranged to perform maintenance; the evaluation label of the predicted power generation obtained by the prediction model is recorded as abnormal.

[0112] The dust sensor on the photovoltaic module obtains the shading area of ​​2m 2 , and obtain the snow thickness h through the snow sensor s =2mm; the clean panel efficiency of the photovoltaic module is recorded as 56%;

[0113] By formula The photovoltaic loss rate θ under natural dust obstruction is calculated t is 56%; among them, t dry The number of days without rainfall during the statistical period was 12 days;

[0114] By formula The photovoltaic loss rate δ when snow is blocked is calculated t ; Since 0 is less than 2mm and less than 1cm, the photovoltaic loss rate δ when snow is blocked t =-4.2%;

[0115] Since there is snow on the photovoltaic modules during the statistical period, the photovoltaic loss rate of 16.8% when the snow is blocked is used as the correction factor, and the photovoltaic loss rate when the dust is blocked is θ t Calibrated to 0;

[0116] By formula Z t '=Z t (δ t +θ t )+Z t = -4.2% × 252KWh + 252KWh = 294.3KWh The revised forecast power generation Z for November 1 is calculated. t ' is 241.42KWh;

[0117] At this time, the difference between 241.42 KWh and 240 KWh is 1.42 KWh, which is less than the preset first-level threshold of 10 KWh, so the evaluation label of the revised predicted power generation is recorded as normal.

[0118] See also Figure 4 , a second aspect of the present invention provides a remote monitoring system for a photovoltaic power station, comprising: a data acquisition module, an abnormality monitoring module and a correction feedback module;

[0119] Data acquisition module: extracts historical data of photovoltaic power stations within the statistical period and the data characteristics of historical data;

[0120] Input the data features into the prediction model to predict the power generation in the target time period; wherein the target time period is after the statistical period; and obtain the actual power generation in the target time period;

[0121] Abnormal monitoring module: performs abnormal prediction detection based on the difference between predicted power generation and actual power generation, and obtains the evaluation label of the predicted power generation;

[0122] Correction feedback module: If the evaluation label of the predicted power generation is abnormal, the predicted power generation is corrected by constructing a correction factor to obtain the corrected predicted power generation;

[0123] Determine whether the evaluation label of the revised predicted power generation is normal; if yes, complete remote monitoring of the corresponding photovoltaic power station; if not, issue a maintenance signal.

[0124] An embodiment of the third aspect of the present invention provides a remote monitoring computing device for a photovoltaic power station, comprising: a memory and a processor, wherein the memory stores executable instructions of the processor; wherein the processor is configured to execute a remote monitoring method for a photovoltaic power station provided in the first aspect by executing the executable instructions.

[0125] Some of the data in the above formula are calculated by removing the dimensions and taking their numerical values. The formula is a formula that is closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0126] The working principle of the present invention is as follows: by extracting historical data of a photovoltaic power station within a statistical period and data features of the historical data; inputting the data features into a prediction model to predict the power generation in a target time period; obtaining the actual power generation in the target time period; performing abnormal prediction detection based on the difference relationship between the predicted power generation and the actual power generation to obtain an evaluation label of the predicted power generation; if the evaluation label of the predicted power generation is abnormal, the predicted power generation is corrected by constructing a correction factor to obtain a corrected predicted power generation; judging whether the evaluation label of the corrected predicted power generation is normal; if so, completing remote monitoring of the corresponding photovoltaic power station; if not, issuing a maintenance signal.

[0127] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A remote monitoring method for a photovoltaic power station, characterized in that: include: Extract historical data of photovoltaic power stations within the statistical period and the data characteristics of the historical data; Inputting the data features into a prediction model to predict power generation in a target time period; The target time period is after the statistical period; obtaining the actual power generation in the target time period; Perform abnormal prediction detection based on the difference between the predicted power generation and the actual power generation to obtain the evaluation label of the predicted power generation; If the evaluation label of the predicted power generation is abnormal, the predicted power generation is corrected by constructing a correction factor to obtain the corrected predicted power generation; Determine whether the evaluation label of the revised predicted power generation is normal; If yes, the remote monitoring of the corresponding photovoltaic power station is completed; If not, a maintenance signal is issued.

2. A remote monitoring method for a photovoltaic power station according to claim 1, characterized in that: Methods for obtaining the data features include: Acquire historical data; wherein the historical data includes: time data, environmental data, and photovoltaic module data; and associate the time data with the timestamps of the environmental data and the photovoltaic module data; wherein the environmental data includes: temperature, rainfall, and light values; and the photovoltaic module data includes: the number and type of photovoltaic modules; Set a statistical period; calculate the change characteristics of environmental data and photovoltaic module data within the statistical period, and mark the change characteristics as data characteristics of historical data; the data characteristics include: change characteristics of environmental data and change characteristics of photovoltaic module data; change characteristics of environmental data include: daily average and extreme values ​​of temperature, rainfall and light values ​​in the statistical period, and the average value of power generation in the secondary period with 3 days, 7 days and 30 days respectively in the statistical period; the change characteristics of photovoltaic module data include: the cumulative operating time of the photovoltaic module and the time since the previous maintenance; the number of failures of the photovoltaic module in the statistical period.

3. A remote monitoring method for a photovoltaic power station according to claim 1, characterized in that: The training process of the prediction model satisfies: Normalize the data features of the historical data and remove outliers using the quantile method; integrate the data features of the processed historical data into a feature matrix X t×m , integrate the power generation on and before the rth day in the statistical period to form a matrix Y with t rows and 1 column t×1 ; m is the number of changing features; Based on the feature matrix X t×12 and the matrix Y 1×t Construct a prediction model; the prediction model satisfies the formula: Z t×1 =(X t× m T X t×m ) -1 X t×m T Y 1×t ; Among them, Z t×1 is the prediction matrix composed of the power generation of the predicted r+1 day, X t×m T For X t×m The transposed matrix, (X t× m T X t×m ) -1 For X t×m T X t×m The inverse matrix of .

4. A remote monitoring method for a photovoltaic power station according to claim 3, characterized in that: Inputting the data features into a prediction model to predict power generation in a target time period includes: Based on the prediction model, the predicted power generation in the target time period is predicted to obtain t prediction results, which form a prediction matrix Z t×1 ; By formula Calculate the average predicted power generation on and before the rth day in the statistical period; where n represents the total number of days in the statistical period; the prediction matrix Z t×1 Each value in is substituted into the preset objective function in turn, and the value that satisfies the objective function is taken as the predicted power generation.

5. A remote monitoring method for a photovoltaic power station according to claim 4, characterized in that: The preset objective function includes: Retrieve the prediction matrix Z t×1 The t-th predicted power generation z in t , and the average predicted power generation on and before the rth day in the statistical period By formula Calculate the objective function.

6. A remote monitoring method for a photovoltaic power station according to claim 1, characterized in that: The abnormality prediction detection based on the difference between the predicted power generation and the actual power generation includes: S110: Retrieving the actual power generation and the predicted power generation on the same day, and calculating the difference between the actual power generation and the predicted power generation; presetting a first threshold and a second threshold, wherein the first threshold is less than the second threshold; S120: When the difference between the actual power generation and the predicted power generation is greater than a first threshold and less than a preset second threshold, and if the difference is still greater than the first threshold and less than the preset second threshold for more than three days within the corresponding statistical period, a reminder signal is issued to arrange maintenance personnel to perform maintenance; When the difference between the actual power generation and the predicted power generation is greater than or equal to the second threshold, an early warning signal is issued and maintenance personnel perform maintenance; S130: When the difference between the actual power generation and the predicted power generation is greater than a first threshold, the evaluation label of the predicted power generation obtained by the prediction model is marked as abnormal; When the difference between the actual power generation and the predicted power generation is less than or equal to the first threshold, the evaluation label of the predicted power generation obtained by the prediction model is recorded as normal.

7. A remote monitoring method for a photovoltaic power station according to claim 1, characterized in that: The method of correcting the predicted power generation by constructing a correction factor includes: The dust sensor on the photovoltaic module is used to obtain the shielding area S0, and the snow thickness h is obtained by the snow sensor. s ;Record the clean panel efficiency of the PV module as b0; By formula The photovoltaic loss rate θ under natural dust obstruction is calculated t ; where t dry is the number of days without rainfall during the statistical period; By formula The photovoltaic loss rate δ when snow is blocked is calculated t ; Determine whether there is snow on the photovoltaic modules during the statistical period; if so, the photovoltaic loss rate δ when the snow is blocked t As a correction factor, the photovoltaic loss rate θ when the dust is blocked is t Calibrated to 0; otherwise, the photovoltaic loss rate θ when dust is blocked t As a correction factor, the photovoltaic loss rate θ when the snow is blocked is t Calibrated to 0; By formula Z t '=Z t (δ t +θ t )+Z t Calculate the revised forecast power generation Z t '.

8. The remote monitoring method for a photovoltaic power station according to claim 1, characterized in that: The step of determining whether the evaluation label of the revised predicted power generation is normal includes: Retrieve the revised predicted power generation and repeat steps S110-S130. If the difference between the actual power generation and the revised predicted power generation is less than or equal to the preset first-level threshold, the evaluation label of the revised predicted power generation is recorded as normal; otherwise, the evaluation label of the revised predicted power generation is recorded as abnormal.

9. A remote monitoring system for a photovoltaic power station, adapted to the remote monitoring method for a photovoltaic power station according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, abnormality monitoring module and correction feedback module; Data acquisition module: extracts historical data of photovoltaic power stations within the statistical period and the data characteristics of historical data; Inputting the data features into a prediction model to predict power generation in a target time period; The target time period is after the statistical period; obtaining the actual power generation in the target time period; Abnormal monitoring module: performs abnormal prediction detection based on the difference between predicted power generation and actual power generation, and obtains the evaluation label of the predicted power generation; Correction feedback module: If the evaluation label of the predicted power generation is abnormal, the predicted power generation is corrected by constructing a correction factor to obtain the corrected predicted power generation; Determine whether the evaluation label of the revised predicted power generation is normal; If yes, the remote monitoring of the corresponding photovoltaic power station is completed; If not, a maintenance signal is issued.

10. A remote monitoring computing device for a photovoltaic power station, characterized in that: include: A memory and a processor, wherein the memory stores executable instructions of the processor; wherein the processor is configured to execute a remote monitoring method for a photovoltaic power station according to any one of claims 1 to 8 by executing the executable instructions.

Images