Extreme temperature early warning method of wind power plant and computing equipment

Abstract

The invention discloses an extreme temperature early warning method for a wind power plant and computing equipment. The method comprises the following steps: acquiring a first predicted temperature of a first wind power plant in a first time period through at least one weather forecast source; obtaining a temperature prediction deviation based on the historical observation temperature true value corresponding to the first wind power plant and the historical prediction temperature corresponding to the first wind power plant; correcting the first predicted temperature based on the temperature prediction deviation to obtain a second predicted temperature; determining a safety risk parameter of a wind turbine generator in the first wind power plant based on the second predicted temperature; if the safety risk parameter meets the preset condition, extreme temperature early warning information corresponding to the first wind power plant is output, more accurate extreme temperature early warning is provided for the wind turbine generator in the wind power plant, and the safety risk caused by the wind turbine generator in the extreme temperature event is reduced.

Description

Technical Field

[0001] The present application relates to the field of wind power generation, and particularly to an extreme temperature warning method and a computing device for a wind farm. Background Art

[0002] Currently, when predicting extreme temperatures, public meteorological forecast services mainly focus on the human perception of temperature and concentrate on urban areas with dense populations. The fault warning of wind turbines in wind farms mainly relies on the real-time released observation data of wind turbines. At the same time, the warning systems of wind turbines in wind farms usually focus on the wind power and power generation in the next few days, or specific typhoons and special weather events.

[0003] Due to climate change, extreme temperature events occur frequently, but the ability to respond to and deal with extreme temperature events during the development of wind farms was insufficient before. This poses challenges to wind farms in related regions, especially in unattended intelligent operation and maintenance. Wind turbines may face extreme temperature conditions that exceed their tolerable survival temperature range, which may lead to faults and lifespan problems. Moreover, public meteorological services mainly focus on urban areas with dense populations, so the current meteorological forecasts mainly meet the needs of people and lack targeted reference value for wind farms. The current warning thresholds are determined based on the human perception of temperature and cannot directly provide guidance for wind farms on the impact of extreme temperatures. In addition, most current wind farm-level fault warning systems only warn against specific faults and do not give warning prompts 1-2 days in advance for special weather events. This means that wind farms may not be able to make preparations in advance during extreme temperature events, resulting in faults and safety risks.

[0004] In view of this, there is an urgent need to provide a more accurate warning scheme for wind farms in extreme temperature scenarios to reduce the safety risks caused by extreme temperature events to wind turbines. Summary of the Invention

[0005] The present application provides an extreme temperature warning method and a computing device for a wind farm, which helps to provide more accurate extreme temperature warnings for wind turbines in the wind farm and reduce the safety risks caused by extreme temperature events to wind turbines.

[0006] In a first aspect, an extreme temperature warning method for a wind farm is provided. The method includes:

[0007] Obtaining a first predicted temperature of a first wind farm in a first time period through at least one meteorological forecast source;

[0008] Obtaining a temperature prediction deviation based on the historical observed temperature truth value corresponding to the first wind farm and the historical predicted temperature corresponding to the first wind farm;

[0009] Correct the first predicted temperature based on the temperature prediction deviation to obtain the second predicted temperature;

[0010] Determine the safety risk parameter of the wind turbine in the first wind farm based on the second predicted temperature;

[0011] If the safety risk parameter meets the preset condition, output the extreme temperature warning information corresponding to the first wind farm.

[0012] In a second aspect, an extreme temperature warning device for a wind farm is provided. The device includes:

[0013] An acquisition unit, configured to acquire the first predicted temperature of the first wind farm in the first time period through at least one weather forecast source; and obtain the temperature prediction deviation based on the true value of the historical observed temperature corresponding to the first wind farm and the historical predicted temperature corresponding to the first wind farm;

[0014] A correction unit, configured to correct the first predicted temperature based on the temperature prediction deviation to obtain the second predicted temperature;

[0015] A determination unit, configured to determine the safety risk parameter of the wind turbine in the first wind farm based on the second predicted temperature;

[0016] An output unit, configured to output the extreme temperature warning information corresponding to the first wind farm if the safety risk parameter meets the preset condition.

[0017] In a third aspect, a computing device is provided. The computing device includes a processor and a memory. The processor is coupled to the memory. At least one computer program instruction is stored in the memory. The at least one computer program instruction is loaded and executed by the processor so that the computing device implements the method provided in the first aspect or any optional manner of the first aspect. For the specific details of the computing device provided in the third aspect, reference can be made to the first aspect or any optional manner of the first aspect, which will not be elaborated here.

[0018] In a fourth aspect, a computer-readable storage medium is provided. At least one instruction is stored in the storage medium. When the instruction runs on a computer, the computer is made to execute the method provided in the first aspect or any optional manner of the first aspect.

[0019] In a fifth aspect, a computer program product is provided. The computer program product includes one or more computer program instructions. When the computer program instructions are loaded and run on a computer, the computer is made to execute the method provided in the first aspect or any optional manner of the first aspect.

[0020] In a sixth aspect, a chip is provided, including a memory and a processor. The memory is used to store computer instructions, and the processor is used to call and run the computer instructions from the memory to execute the method in the first aspect and any possible implementation manners of the first aspect as described above.

[0021] Thus, the embodiments of the present application have the following beneficial effects:

[0022] In the embodiments of the present application, by obtaining a temperature prediction deviation based on the historical observed temperature true value and the historical predicted temperature, and then correcting the predicted temperature, the temperature condition of the wind farm can be predicted more accurately. By determining the safety risk parameter of the wind turbine in the wind farm based on the corrected predicted temperature, if the safety risk parameter meets the preset condition, an extreme temperature warning message is output, which helps to early warn of extreme temperature events, thereby reducing the safety risk caused by extreme temperature to the wind turbine. Description of the Drawings

[0023] Figure 1 is a flowchart of an extreme temperature warning method for a wind farm provided by an embodiment of the present application;

[0024] Figure 2 is a schematic diagram of the logical structure of an extreme temperature warning system provided by an embodiment of the present application;

[0025] Figure 3 is a schematic flowchart of data preprocessing and true value selection provided by an embodiment of the present application;

[0026] Figure 4 shows a schematic diagram of the warning deviation and the extreme low temperature altitude of the nearest meteorological station;

[0027] Figure 5 shows a schematic diagram of the warning deviation and the extreme low temperature surface roughness of the nearest meteorological station on day C of month B of year A;

[0028] Figure 6 shows a schematic diagram of obtaining the predicted temperature through multiple meteorological forecast sources;

[0029] Figure 7 shows that the deviation of the maximum and minimum values of the previous several days and the true value is used as the deviation input of the extreme temperature event and is updated day by day in a rolling manner;

[0030] Figure 8 shows a schematic diagram of the observed temperature true value sequence and the predicted temperature sequence;

[0031] Figure 9 shows a schematic diagram of the statistical result of the temperature prediction deviation in a certain cold air area;

[0032] Figure 10 shows a schematic diagram of an extreme temperature warning example;

[0033] Figure 11 Shows extreme warning case information and review examples. Specific embodiments

[0034] To make the objectives, technical solutions, and advantages of this application clearer, the following will further describe the embodiments of this application in detail with reference to the accompanying drawings.

[0035] The following gives examples of the application scenarios of the embodiments of this application.

[0036] The embodiments of this application are applied to the scenario of predicting and warning extreme high and low temperatures in a wind farm.

[0037] Currently, both public meteorological forecast services and wind turbine fault warnings have the following problems.

[0038] First, climate change has led to frequent extreme temperature events. In the past, when developing wind farms, the ability to respond to and deal with extreme temperature events was insufficient. For wind farms in relevant regions, frequent extreme temperature events have brought some challenges, especially in unattended intelligent operation and maintenance. On the one hand, current frequent extreme weather disaster events may cause wind turbines to face extreme temperature conditions beyond the survival temperature range that the wind turbines can withstand. On the other hand, extreme temperature events have an impact on the faults and service life of wind turbines, and it is also necessary to consider planning arrangements such as the safe travel of personnel. On the other hand, the characteristics of extreme temperatures in different regions are quite different, which means that different extreme temperature situations need to be considered for wind farms in different regions, and corresponding preventive and coping measures need to be taken.

[0039] Second, wind farms need to forecast extreme temperatures, but currently public meteorological services are mainly designed for the needs of people and lack targeted reference value for wind farms. Current public meteorological services mainly focus on urban areas with dense populations. The thresholds for extreme temperature warnings are mainly determined based on the human perception of temperature and cannot directly provide guidance for wind farms regarding the impact of extreme high and low temperatures.

[0040] Third, most current wind farm-level fault warning systems only give warnings for specific faults and do not give warning prompts 1-2 days in advance for special weather events.

[0041] Most of the fault warnings for wind turbines in a wind farm are based on the temperatures recorded by the temperature sensors of the wind turbines to give extreme temperature fault warnings. Based on this, there is an urgent need to provide an extreme temperature warning system for wind farm equipment designed with targeted thresholds based on wind farm-level temperature prediction combined with the characteristics of wind turbines.

[0042] Based on this, in some embodiments of the present application, early warnings for extreme temperature events (including extreme high temperature events and extreme low temperature events) that may occur in the next few days are provided for wind turbines in a wind farm, and the warnings are accurate for different regions. In some embodiments, different levels of early warnings such as blue, yellow, orange, and red are triggered based on the impacts of extreme temperature events on the faults, operating states, and service lives of wind turbines and on personnel safety, making the early warnings more refined and flexible, and providing support for the intelligent operation and maintenance of the wind farm under extreme temperatures, the exploration of wind turbines under extreme temperatures, and the improvement of equipment reliability.

[0043] Some term concepts related to the embodiments of the present application are explained below.

[0044] (1) Extreme threshold value

[0045] The extreme threshold value refers to the critical value at which a certain statistic reaches an extreme condition.

[0046] (2) Amplitude of daily temperature

[0047] The value when the daily minimum temperature is lower than the daily minimum temperature of the previous day.

[0048] (3) Amplitude of consecutive daily temperature drop

[0049] The amplitude of consecutive daily temperature drop refers to the sum of the amplitudes of daily temperature within the consecutive temperature drop period.

[0050] (4) Extreme high temperature

[0051] Extreme high temperature refers to the daily maximum temperature greater than or equal to the extreme threshold value. For example, statistical analysis is performed on the long-term observation data of a weather station to determine the extreme high temperature. Exemplarily, a measurement height of 2 meters is selected, and the average value of the observed temperatures is recorded at an observation cycle of every 10 minutes to determine the extreme high temperature. Exemplarily, for a wind turbine, the extreme high temperature is the daily maximum temperature in the temperature forecast for the next few days that is greater than or equal to the extreme threshold value affecting the unit.

[0052] (5) Extreme low temperature

[0053] The extreme low temperature refers to the daily minimum temperature less than or equal to the extreme low temperature threshold. For example, the long-term observed data of a weather station is statistically analyzed to determine the extreme high temperature. Exemplarily, 2 meters is selected as the measurement height, and the average value of the observed temperature is recorded every 10 minutes as an observation cycle to determine the extreme low temperature. Exemplarily, the extreme low temperature is the daily minimum temperature in the temperature forecast results for the next few days that is less than or equal to the extreme threshold affecting the unit.

[0054] (6) Wind turbine ambient temperature

[0055] The wind turbine ambient temperature refers to the ambient temperature outside the nacelle recorded by the temperature sensor located at the hub height of the wind turbine.

[0056] Appendix Figure 1 is a flowchart of an extreme temperature warning method for a wind farm provided by an embodiment of the present application. Appendix Figure 1 The method shown involves using data provided by multiple wind farms. To distinguish different wind farms, the wind farm used as the prediction target is called the first wind farm, and the wind farm referred to when predicting the temperature of the first wind farm is called the second wind farm. Appendix Figure 1 The method shown involves using a historical time period to predict the temperature of a wind farm in a future time period. To distinguish different time periods, the future time period used as the prediction target is called the first time period. Appendix Figure 1 The method shown includes the following steps.

[0057] S110, obtain the first predicted temperature of the first wind farm in the first time period through at least one weather forecast source.

[0058] In some embodiments, the weather forecast source includes a mesoscale weather forecast model. The mesoscale weather forecast model is, for example, a numerical weather model based on the atmosphere and ocean. The atmosphere model mainly simulates the movement of the atmosphere, while the ocean model simulates the movement of the ocean. The mesoscale weather forecast model performs numerical calculations based on the initial conditions and boundary conditions by running the atmosphere model and the ocean model to obtain the first predicted temperature of the first wind farm in the first time period.

[0059] In some embodiments, based on the historical predicted temperature of the target area where the first wind farm is located and the true value of the historical observed temperature corresponding to the target area in the season to which the first time period belongs for at least one weather forecast source, determine the temperature prediction deviation of each weather forecast source; based on the temperature prediction deviation of each weather forecast source, determine the target forecast source from at least one weather forecast source; obtain the first predicted temperature of the first wind farm in the first time period through the target forecast source.

[0060] The target area where the first wind farm is located is, for example, the intersection of the administrative area where the first wind farm is located and the climate zone where the first wind farm is located. Exemplarily, for each of at least one weather forecast source, the historical predicted temperature provided by the weather forecast source for the target area and the true value of the historical observed temperature provided by the weather forecast source for the target area are compared to determine the difference between the historical predicted temperature and the true value of the historical observed temperature as the temperature prediction deviation of the weather forecast source. The at least one weather forecast source is sorted in ascending order of the temperature prediction deviation, and the target forecast source with the smallest temperature prediction deviation is determined from the at least one weather forecast source; the first predicted temperature of the first wind farm in the first time period is obtained through the target forecast source.

[0061] By analyzing the temperature prediction deviations of different weather forecast sources, the target forecast source can be selected according to the magnitude of the deviation. By comparing the temperature prediction deviations of different weather forecast sources, a more accurate forecast source (target forecast source) can be determined to obtain the predicted temperature data of the first wind farm in the first time period, thereby improving the temperature prediction accuracy of the first wind farm.

[0062] In some embodiments, the predicted temperatures of the first wind farm in the first time period are obtained through each of at least one weather forecast source to obtain a set of predicted temperatures; the mode of the set of predicted temperatures is determined to obtain the first predicted temperature.

[0063] The first predicted temperature is the data with the highest frequency of occurrence in the set of predicted temperatures, representing the consensus of most weather forecast sources on the predicted temperature. In other words, if most weather forecast sources predict the same extreme temperature, it can be considered that the probability of the occurrence of the extreme temperature event is relatively high. Therefore, the first predicted temperature can more accurately represent the predicted temperature of the wind farm. In some embodiments, in the case where the true value of the observed temperature is not obtained, the mode is used as the initially used predicted temperature. When the true value of the observed temperature is obtained subsequently, the predicted temperature is adjusted based on the true value of the observed temperature.

[0064] S120, obtain the true value of the historical observed temperature corresponding to the first wind farm and the historical predicted temperature corresponding to the first wind farm.

[0065] In some embodiments, the similarity between at least one weather station in the target area where the first wind farm is located and the first wind farm is determined; based on the similarity between each weather station and the first wind farm, a target weather station is determined from the at least one weather station, and the similarity between the target weather station and the first wind farm meets the condition; based on the true value of the historical observed temperature of the target weather station, the true value of the historical observed temperature corresponding to the first wind farm is obtained.

[0066] In the case where the first wind farm itself lacks historical observed temperature true values, the historical observed temperature true values of the area where the wind farm is located can be obtained through the observed data of the target weather station, so as to supplement the missing historical observed temperature true values: By selecting a weather station similar to the first wind farm, the accuracy of the historical observed temperature true values can be improved. Because similar weather stations usually have similar weather and climate characteristics, and thus can better reflect the temperature conditions of the area where the first wind farm is located.

[0067] In some embodiments, the process of determining a similar weather station includes the following steps 1 to 4.

[0068] Step 1, based on the location of each weather station and the location of the first wind farm, obtain the distance between each weather station and the first wind farm.

[0069] For example, based on the location coordinates of each weather station and the location coordinates of the first wind farm, use latitude and longitude calculation or Euclidean distance to obtain the distance between each weather station and the first wind farm, so as to be able to determine their spatial relationship geographically, which helps to determine whether the historical observed temperature true values provided by the weather station are applicable to the temperature prediction of the wind farm.

[0070] Step 2, based on the elevation value of each weather station and the elevation value of the first wind farm, obtain the elevation difference between each weather station and the first wind farm.

[0071] By comparing the elevation value of each weather station with the elevation value of the first wind farm, the elevation difference between each weather station and the first wind farm can be obtained, which helps to determine the altitude difference between the weather station and the wind farm and further measure their relative position relationship in the vertical direction. Considering the distance and elevation difference can more accurately judge the similarity between the weather station and the wind farm. Geographical factors have an important impact on weather and climate, so it helps to determine whether the historical observed temperature true values provided by the weather station are applicable to the temperature prediction of the wind farm.

[0072] Step 3, based on the historical temperature distribution of each weather station and the historical temperature distribution of the first wind farm, obtain the correlation of the historical temperature distribution between each weather station and the first wind farm.

[0073] For example, based on the historical temperature distribution of each weather station and the first wind farm, calculate the correlation of the historical temperature distribution between each weather station and the first wind farm. For example, use a correlation analysis method (such as Pearson correlation coefficient) to quantify the linear correlation degree between the historical temperature distributions of each weather station and the first wind farm, so as to evaluate the temperature similarity between the weather station and the wind farm.

[0074] Step 4: Based on the weights of the distance dimension, the elevation dimension, and the historical temperature distribution dimension, perform a weighted sum of the distance between each weather station and the first wind farm, the elevation difference between each weather station and the first wind farm, and the correlation of the historical temperature distribution between each weather station and the first wind farm, to obtain the similarity between each weather station and the first wind farm.

[0075] For example, multiply the weight of the distance dimension by the distance value between each weather station and the first wind farm, the weight of the elevation dimension by the elevation difference value between each weather station and the first wind farm, and the weight of the historical temperature distribution dimension by the correlation value of the historical temperature distribution between each weather station and the first wind farm. Then sum up the weighted values obtained from the three dimensions to get the similarity between each weather station and the first wind farm. By means of weighted summation, the information of the three dimensions of distance, elevation difference, and historical temperature distribution is comprehensively considered, so as to more completely and comprehensively determine the similarity between the weather station and the first wind farm. The similarity obtained by calculating the weighted sum can provide a specific value to quantitatively measure the similarity between each weather station and the first wind farm.

[0076] In some embodiments, based on the historical meteorological data of each wind farm in the target area where the first wind farm is located, the elevation values of each wind farm in the target area, and the locations of each wind farm in the target area, determine a second wind farm that can represent the target area from each wind farm in the target area; based on the true value of the historical observed temperature of the second wind farm, obtain the true value of the historical observed temperature corresponding to the first wind farm.

[0077] Analyze the weather conditions through the historical meteorological data of each wind farm in the target area, including the true value of the historical observed temperature, and use it as the basis for evaluating the first wind farm and the representative second wind farm. This helps to determine the similarity or correlation between the two wind farms. By comparing the elevation values of each wind farm in the target area, it helps to understand the altitude difference between each wind farm. Considering the elevation difference can provide more accurate regional characteristics and help to evaluate the similarity between different wind farms. By using the location information of each wind farm in the target area, it helps to determine their geographical distribution, which is of great significance for considering geographical factors, climate change and other factors. By using the true value of the historical observed temperature of the second wind farm, the true value of the historical observed temperature corresponding to the first wind farm can be obtained. It is equivalent to referring to and inferring the historical temperature change in the area where the first wind farm is located through the observed data of the second wind farm representing the whole area, which helps to provide more accurate and reliable wind farm temperature information.

[0078] In some embodiments, based on the true historical observed temperatures in the temperature records of multiple wind turbines in the first wind farm, the standard deviation of the temperature records of the multiple wind turbines is obtained; from the temperature records of the multiple wind turbines, the temperature records with a deviation exceeding a threshold between the true historical observed temperature and the standard deviation are filtered out; the average value of the true historical observed temperatures in the filtered temperature records is determined to obtain the true historical observed temperature corresponding to the first wind farm.

[0079] S130, based on the true historical observed temperature corresponding to the first wind farm and the historical predicted temperature corresponding to the first wind farm, the temperature prediction deviation is obtained.

[0080] In some embodiments, based on the administrative region and the climate region of the wind farm management, regional division is performed to determine the target region where the first wind farm is located; based on the true historical observed temperature and the historical predicted temperature in the historical extreme temperature events that occurred in the target region, the temperature prediction deviation is obtained.

[0081] In some embodiments, based on the time difference between the occurrence time of multiple historical extreme temperature events that occurred in the target region and the first time period, the weight corresponding to each historical extreme temperature event is determined, and the weight corresponding to the historical extreme temperature event is negatively correlated with the time difference; based on the weight corresponding to each historical extreme temperature event, the deviations between the true historical temperature and the historical predicted temperature in each historical extreme temperature event are weighted and summed to obtain the temperature prediction deviation.

[0082] By comparing the time difference between the occurrence time of each historical extreme temperature event and the first time period, the weight corresponding to each historical extreme temperature event can be determined. The greater the time difference between the occurrence time of the historical extreme temperature event and the first time period, the smaller the weight corresponding to the historical extreme temperature event; conversely, the smaller the time difference between the occurrence time of the historical extreme temperature event and the first time period, the greater the weight corresponding to the historical extreme temperature event. By determining the weight of each historical extreme temperature event based on the time difference, the influence of historical extreme temperature events on temperature prediction can be considered more accurately. The smaller the time difference indicates that the historical extreme temperature event is closer to the target time period. Therefore, the influence weight of the historical extreme temperature event on the prediction is greater. By weighting and summing the deviations between the true historical temperature and the historical predicted temperature according to the weight corresponding to each historical extreme temperature event. The deviation corresponding to the event with a larger weight will have an impact on the predicted temperature, reducing the impact of those events with larger fluctuations or less reliability in the deviation direction on the predicted temperature.

[0083] In some embodiments, a temperature prediction deviation is obtained based on the deviation between the true historical observed temperature of the first wind farm on multiple previous dates and the historical predicted temperature of the first wind farm on the multiple previous dates. The previous dates are the dates before the dates within the first time period. The multiple previous dates are, for example, the previous 9 days. For example, for each previous date, the true historical observed temperature and the historical predicted temperature of the first wind farm are obtained, the true historical observed temperature and the historical predicted temperature are compared, and the deviation therebetween is calculated. For example, the deviation is obtained by subtracting the predicted temperature from the true observed temperature. The temperature prediction deviation is obtained based on the deviation between the true historical observed temperature and the historical predicted temperature of the multiple previous dates. For example, the temperature prediction deviation is obtained by taking an average value, a weighted average value, or other statistical methods based on the deviation between the true historical observed temperature and the historical predicted temperature of the multiple previous dates.

[0084] In some embodiments, curve fitting is performed based on the true historical observed temperature of the first wind farm to obtain a first temperature distribution curve; curve fitting is performed based on the historical predicted temperature of the first wind farm to obtain a second temperature distribution curve; and a temperature prediction deviation is obtained based on the difference between the first temperature distribution curve and the second temperature distribution curve. Specifically, based on the true historical observed temperature and the true temperature data of the first wind farm, a statistical curve can be generated, where the horizontal axis represents temperature and the vertical axis represents the frequency of temperature occurrence. This curve describes the temperature distribution. Since attention is paid to the extreme temperature distribution, it is based on the fact that the extreme temperature distribution conforms to a certain predetermined probability distribution, such as the Weibull distribution or the Gumbel distribution, etc. Then, curve fitting methods are used to fit the distribution curves of the true temperature and the predicted temperature respectively, and the deviation of the extreme temperature samples is measured by the difference between these two curves. Exemplarily, under the condition of meeting the extreme temperature (for example, the forecast exceeds 40 degrees Celsius), the deviation between the predicted temperature and the true temperature is found in the temperature distribution curve, and then this deviation is used to optimize the forecast source.

[0085] S140, correct the first predicted temperature based on the temperature prediction deviation to obtain a second predicted temperature.

[0086] S150, determine the safety risk parameter of the wind turbine in the first wind farm based on the second predicted temperature.

[0087] In some embodiments, the second predicted temperature is input into the equipment failure model to obtain the failure risk parameter of the wind turbine. The equipment failure model is established based on historical extreme temperature events and the historical operation records of the wind turbine. The historical operation records are the records of the failures or changes in the operating states of the wind turbine when historical extreme temperature events occur.

[0088] Collect historical extreme temperature events of different models for all collectible records, and record the cases that cause unit failures or changes in the operating state of the unit as the target data set. Take the extreme temperature intensity, duration, relationship with wind speed, etc. as independent variables, and use a correlation reaching a specific threshold, such as above 0.5, as a reference variable. Analyze the probability of different characteristics caused by extreme events with the failure or change in the operating state of the unit as the target. After each extreme event, enrich the case set.

[0089] For the establishment process of the equipment failure model, exemplarily, collect historical extreme temperature events of wind turbines of each model, and record the historical extreme temperature events that cause failures or changes in the operating state of the wind turbines. Analyze the correlation between factors such as the intensity, duration, and wind speed of extreme temperature events and failures or changes in the operating state of the unit. When the correlation reaches a specific threshold (such as above 0.5), take the failure or change in the operating state of the unit as the target, and analyze the probability of different characteristics occurring due to extreme events, thereby establishing an equipment failure model.

[0090] In some embodiments, input the second predicted temperature and the model of the wind turbines in the first wind farm into the life load model to obtain the life loss risk parameters of the wind turbines. The life load model is established based on extreme temperature, fatigue loads of wind turbines of various models under extreme temperature, ultimate loads of wind turbines of various models under extreme temperature, and shutdown condition loads of wind turbines of various models due to shutdown caused by extreme temperature.

[0091] For the establishment process of the life load model, exemplarily, collect historical operation data of wind turbines and meteorological observation data under different working conditions. The meteorological observation data includes temperature, temperature change, wind speed, air density, etc., and at the same time record the ultimate load and fatigue load of the wind turbines operating under these meteorological observation data. According to the collected historical operation data of the wind turbines and meteorological observation data, statistical analysis methods or physical mechanism-based modeling methods can be used to establish the life load model. The life load model will consider the influence of different working condition factors (such as temperature, temperature change, wind speed, and air density) on the ultimate load and fatigue load of the unit. By performing sensitivity analysis on the established life load model, such as evaluating the influence of extreme temperature processes on the ultimate load and fatigue load of wind turbines under different working conditions, the life load model is established.

[0092] S160, if the safety risk parameter meets the preset condition, output the extreme temperature warning information corresponding to the first wind farm.

[0093] The method provided in this embodiment can obtain the temperature prediction deviation based on the historical observed temperature true value and the historical predicted temperature, and then correct the predicted temperature, so as to more accurately predict the temperature situation of the wind farm. By determining the safety risk parameters of the wind turbines in the wind farm based on the corrected predicted temperature, if the safety risk parameters meet the preset conditions, extreme temperature warning information is output, which helps to early warn of extreme temperature events and thus reduce the safety risks caused by extreme temperatures to the wind turbines.

[0094] Figure 2 It is a schematic diagram of the logical structure of an extreme temperature warning system provided by an embodiment of the present application. Figure 2 The extreme temperature warning system shown includes a meteorological prediction module, an ensemble optimization module, a device impact assessment module, and a warning release module. The extreme temperature warning system is used to establish an extreme temperature warning model for wind farms in different regions based on reliable multiple mesoscale future multi-day (such as 7-day) meteorological forecast sources, so as to realize different-level warnings of daily extreme temperature events and the impact of extreme temperatures on wind turbines.

[0095] The meteorological prediction module is used to predict the meteorological data of multiple mesoscale future days in different regions. The input data of the meteorological prediction module includes the predicted meteorological data provided by multiple reliable meteorological forecast sources to improve the accuracy and reliability of the prediction. Mesoscale refers to the scale range between small scale (such as a few kilometers to dozens of kilometers) and large scale (such as hundreds of kilometers to thousands of kilometers). The meteorological data includes temperature, and optionally also includes wind speed, humidity, wind direction, and air pressure.

[0096] The ensemble optimization module is used to collect the SCADA data observed by the wind turbines, the data observed by the meteorological stations around the wind turbines, and the historical prediction data; based on the existing historical observed data and historical prediction data, determine the optimal method applicable to the target area from multiple optimization methods.

[0097] The device impact assessment module is used to establish a device failure model and a life load model based on the optimal predicted meteorological data, the current state of the wind turbines, and the historical characteristics of the wind turbines.

[0098] The warning release module is used to evaluate the level of extreme temperature and the degree of influence of extreme temperature, and send extreme temperature warnings to the managers of the wind turbines and the wind farm.

[0099] Module 1: Meteorological prediction module

[0100] The functions of the meteorological prediction module include providing meteorological data of the next 3 days for the target wind farm at a specific moment every day according to the wind farm location. The meteorological prediction module uses multiple meteorological forecast sources to predict meteorological data.

[0101] The reason for adopting the meteorological prediction module is that mesoscale weather prediction models use atmospheric models and ocean models to predict the weather. With the increasing power of supercomputers, the prediction ability of numerical weather models can only be maintained for about 6 days, and the length of the prediction time is related to the location, season, and type of weather pattern. Currently, in the selection of the 7-day weather forecast, the occurrence of extreme temperature events in the next 3 days is mainly referred to as the basic reminder information for the possible impact of extreme temperature events at the wind farm site, so as to prepare spare parts and carry out relevant maintenance work in advance.

[0102] The technical problems that the meteorological prediction module can solve include: Wind turbines can usually only record the currently observed temperature or make short-term predictions (e.g., 0 - 4 hours) based on the observed temperature. However, for multi-day future predictions, the predicted temperature provided by mesoscale forecast sources is required as input data. In addition, a single meteorological forecast source may have insufficient prediction of extreme temperature events. Therefore, using multiple independent meteorological forecast sources can improve the recognition success rate of extreme temperature events.

[0103] The application effects of the meteorological prediction module mainly include: Through the mesoscale prediction time series data at the station level, it can provide multi-day forecasts for the wind farm, and this kind of forecast has timeliness. At the same time, using multiple meteorological forecast sources can provide guarantee for the identification of extreme temperature events in the wind farm.

[0104] Module 2: Ensemble optimization module

[0105] Explain the technical problems solved by the ensemble optimization module.

[0106] Data-related problems include the following aspects.

[0107] (1) Data reliability: There are temperature observation deviations among different devices in the reliability of meteorological data observed at the hub height of wind farm equipment. At the same time, the consistency of regional temperature within the wind farm, the impact of temperature detection equipment failures, and the impact of the operation of wind turbines on temperature also need to be considered.

[0108] (2) Data quality: The observed temperature may have situations such as data missing, mutation, and gross error.

[0109] Data missing means that the observed temperature may be incomplete or lost at some time points, resulting in the inability to obtain the corresponding observed temperature during that time period. Mutation means that the observed temperature suddenly and abnormally changes at some moments, which may be caused by measurement instrument failures, environmental changes, etc. Gross error means that there may be some obviously incorrect values in the observed temperature, and these error values may be caused by instrument failures, human misoperations, or other reasons.

[0110] (3) Data area coverage: Only operating wind farms can collect the true values of historical observed temperatures. Since under-construction wind farms are not powered on, they cannot actively respond to extremely harsh conditions, so extreme temperature prediction is also required. However, these under-construction wind farms cannot collect the true values of historical observed temperatures. Therefore, not all wind farms can collect the true values of historical observed temperatures.

[0111] (4) Regional distribution: Wind farms with accessible SCADA data show regional clustering. For example, the geographical locations of these wind farms tend to be concentrated, and they may be located in adjacent or nearby areas. At the same time, there are also spatial distribution differences in the regional concentration of some batch wind power projects. For example, wind power projects are unevenly distributed geographically and may be concentrated in a specific area.

[0112] (5) Other reference data

[0113] Historical meteorological station data has a long time series and covers long-term observed meteorological data. However, the locations of different meteorological stations usually deviate from the locations of wind farms, and the distances between different meteorological stations and wind farms may vary. Therefore, the temperatures observed at different meteorological stations may deviate from the actual temperatures of specific wind farms. In addition, there are differences in the altitude between different meteorological stations and wind turbines. Meteorological stations are usually located on the ground, while wind turbines are usually installed at higher positions, so there is an altitude difference between meteorological stations and wind turbines. This means that the temperatures observed at meteorological stations may not accurately reflect the temperature conditions of wind farms. In addition, meteorological station temperature observations are usually carried out at a height of 2 meters, while the hub height of wind turbines may be much higher than 2 meters. Therefore, the temperature data observed at meteorological stations may differ from the temperatures observed at the actual hub height of wind turbines.

[0114] (6) Limited historical prediction data for wind farms.

[0115] Due to the continuous update and iteration of prediction models, the historical prediction meteorological data available for optimizing the models is only for the recent few years. In contrast, the sample size of the historical observed meteorological data collected by meteorological stations over the years is larger.

[0116] Regionally optimizable problems

[0117] (1) Non-uniformity of extreme temperature events. In recent years, the frequency of extreme temperature events has increased. However, for extreme temperature events that have a significant impact on regional wind farms, the number of occurrences per year is limited, and the distribution of these extreme temperature events in time and space is not uniform. For example, the frequency of extreme temperature events in some regions is relatively high, while the frequency of extreme temperature events in other regions is relatively low.

[0118] (2) Extreme temperature moments are limited. For a wind farm or a region, the time of extreme temperature events each year is limited, resulting in limited samples.

[0119] The set optimization module specifically includes three sub-modules: data preprocessing and truth value selection module, optimization algorithm set module and optimization search module.

[0120] The data preprocessing and truth value selection module is used to form relatively reliable observed temperature truth values ​​for wind farms in different regions based on the current actual data status; according to the distribution characteristics of wind farms in each region, a meteorological station optimization model is established to obtain the most representative observed temperature truth value for the region.

[0121] The optimization algorithm set module is used to expand different optimizable methods from the perspectives of sample expansion, regional similarity, and time series similarity for limited extreme samples. The optimization algorithm set module is used to enrich the optimizable sample set in as many dimensions as possible from the perspectives of different weather forecast sources and different optimization algorithms.

[0122] The optimization module is used to evaluate and perform rolling optimization on the original weather forecast source and multiple optimized weather forecast sources to obtain the optimal solution. For example, a rolling scoring system and rolling optimization are established.

[0123] The technical problems that the data preprocessing and truth value selection module is used to solve include the following.

[0124] Equipment ambient temperature observation problem.

[0125] (1) Data reliability: The reliability of the temperature data observed at the hub height of the wind turbine in the wind farm is due to the deviation of the temperature observed by different wind turbines. In addition, the temperature consistency of the area within the wind farm, the failure of the detection equipment and the possible impact of the wind turbine are taken into consideration.

[0126] (2) Data quality: Ambient temperature observation data may contain missing data, sudden changes, gross errors, etc.

[0127] The spatial distribution of observation values ​​is uneven and the effective observation time is insufficient.

[0128] (3) Data area coverage: Ambient temperature data can only be collected from wind farms that are already in operation. Wind farms under construction may not be powered on and cannot actively respond to extreme conditions, so extreme temperature prediction is also required. Therefore, not all wind farms can collect ambient temperature data.

[0129] (4) Regional distribution: Wind farms with available SCADA data are clustered in different regions, and some batch projects have relatively concentrated spatial distribution differences.

[0130] (5) Other reference data: Historical meteorological station data, with a long time series, but different distances from the wind farm, different altitudes from the wind farm, and most temperature observations are at 2m, different from the temperature at the hub height.

[0131] (6) Limited historical prediction data of the wind farm: Since the prediction model is constantly updated and iterated, the historical prediction data that can be collected for optimizing the model is only for the recent few years, which is still a smaller sample compared to the historical meteorological station data over the years.

[0132] Figure 3 It is a schematic flow chart of data preprocessing and true value selection provided by an embodiment of the present application. Figure 3 The method shown includes the following steps.

[0133] S310, Coarse error rejection and selection of effective devices.

[0134] To control the quality of temperature data from the source, the temperature data is usually detected at the hub height of typical wind turbines. However, due to the problem of equipment ambient temperature observation, the temperature data detected at the positions of each wind turbine in the entire wind farm is used. For each season, the daily extreme temperature averaged every 15 minutes is calculated. For example, the daily maximum temperature averaged every 15 minutes in summer, and the daily minimum temperature averaged every 15 minutes in winter. Perform 3σ judgment, and for wind turbines exceeding 3σ, give a temperature maintenance warning and exclude them from the mean calculation. Select the reliable mean of temperature observation as the benchmark value of the average temperature of the wind farm;

[0135] The 3σ judgment rejection process includes the following steps 1 to 3.

[0136] Step 1, Construct a sample set. Take the average temperature data of the wind farm every 15 minutes as the sample set, and take the temperature record of each wind turbine as a sample in the sample set.

[0137] Step 2, Calculate the standard deviation of the sample set at each moment of every 15 minutes. By statistically analyzing the sample set, calculate the standard deviation of the sample set at the current moment.

[0138] Step 3, For each wind turbine, calculate the absolute value of the difference between the temperature record of the wind turbine and the mean of the sample set, and compare the absolute value of the difference with 3 times the standard deviation. If the difference exceeds 3 times the standard deviation, that is, deviates too much from the mean, then determine that the temperature record of the wind turbine is an outlier, and filter out the temperature record of the wind turbine from the sample set.

[0139] Step 4, Repeat steps 1 to 3 until no outliers are excluded in this round of loop.

[0140] The formula is as follows:

[0141]

[0142] Among them, represents the average temperature of the wind farm within 15 minutes, n represents the number of temperature records of the wind turbines in the sample set, and x i represents the temperature of the wind turbine in the i-th record.

[0143] To ensure the sample size, multiple wind farms within a nearby area such as within 20 km can be considered integrally. However, for some wind farms with large single-unit capacities and few turbine samples, the difference between the highest and lowest average temperature values per 15 minutes of the wind farm and the average value of the field can be calculated. Those with a deviation higher than 1.5 °C are excluded as deviations.

[0144] S320, perform data quality control on the observed temperature true values provided by each selected wind turbine.

[0145] For those with fluctuations above a specific temperature N °C (such as 3 °C) in the 12 15-minute temperature average values per 3 hours, outlier marking is performed. For the observed temperature true values with a data availability rate less than 50%, they are recorded as missing measurements; for the observed temperature true values with a data availability rate greater than or equal to 50%, interpolation is performed, and the interpolation method can be linear interpolation or non-linear interpolation.

[0146] S330, regional representativeness evaluation.

[0147] For representativeness evaluation, regional ranking can be carried out with reference to historical meteorological data, elevation data, longitude and latitude data. Taking elevation as an example, if the coverage of the existing observed wind power projects in the region is between 15% and 85%, then through this element test, it is considered that the regional representativeness of this evaluation dimension is reliable; otherwise, surrounding meteorological stations need to be supplemented as a reference.

[0148] S340, determine the meteorological station most similar to the first wind farm from each region through the surrounding meteorological station evaluation model.

[0149] Modeling is carried out from the existing SCADA observation data, classified by region, elevation, etc., and regional meteorological station similarity tests are carried out. The test methods include but are not limited to the closest distance, the closest elevation, inverse distance weighting within a certain radius, historical temperature distribution correlation of historical long-term sequence data, and combinations thereof, to obtain regional similarity.

[0150] Exemplarily, there are meteorological station 1, meteorological station 2... meteorological station s and wind farm 1, wind farm 2... wind farm n deployed around the first wind farm. The formula for selecting the meteorological station most similar to the first wind farm in each region one by one is as follows.

[0151] f = t1 * v1 + t2 * v2 + t3 * v3.

[0152] Among them, f represents the similarity between the meteorological station and the first wind farm. The influencing factors include distance similarity, elevation similarity, and the correlation of historical long-term temperature distributions. The distance similarity is t1. The weight of the distance similarity is v1. The elevation similarity is t2, and the weight of the elevation similarity is v2. The correlation of the historical long-term temperature distribution is t3, and the weight of the historical temperature distribution is v3. The similarity between each meteorological station in the area and the first wind farm can be determined, and the meteorological station with the maximum similarity (f) is selected from multiple meteorological stations in the area as the optimal meteorological station.

[0153] S350. Obtain the true value of the historical observed temperature of the meteorological station most similar to the first wind farm as the true value of the historical observed temperature of the first wind farm.

[0154] For the ambient temperature data of wind farms with limited data quality, the optimal historical true value sequence can be obtained. For forecast wind farms without ambient temperature observations, combined with surrounding meteorological stations and wind farms, the true value sequence of the target point is obtained.

[0155] Exemplarily, please refer to Figure 4 , Figure 4 shows the schematic diagram of the warning deviation released on January 11, 2023 and the extreme low temperature altitude of the nearest meteorological station. Please refer to Figure 5 , Figure 5 shows the schematic diagram of the warning deviation released on January 11, 2023 and the extreme low temperature surface roughness of the nearest meteorological station.

[0156] Sub-module 2: Optimization algorithm set module

[0157] The technical problems solved by the optimization algorithm set module include: Even if the true values of multiple target wind farms have been obtained, there are still the following sample asymmetries and limitations in extreme temperature events.

[0158] (1) Inhomogeneity of extreme temperature events: For extreme temperature events that have a significant impact on regional wind farms, the number of extreme temperature events per year is limited and unevenly distributed.

[0159] (2) Limited extreme temperature moments: From the perspective of a single wind farm or a region, the number of days in extreme temperature events per year is limited, resulting in limited samples.

[0160] The optimization algorithm set module is used to establish a targeted optimization method for the asymmetry characteristics of existing extreme temperature events.

[0161] (1) Insufficient historical observed case data: Establish a regional experience database from a spatial perspective and enrich the cumulative data volume from the perspective of regional consistency.

[0162] (2) Limited number of extreme case samples: Expand the case samples from a time series perspective.

[0163] (3) Limited extreme case temperature thresholds: Expand the sample from the perspective of extreme temperature distribution.

[0164] Sub-module 2.1: Spatial similarity method

[0165] Divide the area respectively according to the administrative region and climate zoning of wind farm management. Accumulate the historical extreme temperature time and conduct event statistics for different regional scopes. Preferentially select the number of wind farms covered by extreme temperature time, the similarity of occurrence months, and the system deviation characteristics within the characteristic area for regional optimization; The regional statistics are shown in Table 1 below.

[0166] Table 1 Statistics of extreme high temperature events in Region A

[0167]

[0168] Sub-module 2.2: Temporal similarity method

[0169] Considering that the development of extreme temperature events has a temporal process, for typical wind farms with historical prediction and true value areas, reproduce the extreme temperature prediction, compare the different development cycles of different extreme temperature degrees, and select the maximum and minimum temperatures of the previous days of the target extreme temperature cycle as the reference quantity for prediction. The duration of the previous date may vary from 3 days to 3 months, subject to the optimal training of this area.

[0170] Taking an extreme low temperature in the north as an example, the daily extreme temperature threshold is the previous 9 days. That is, the deviation between the daily low temperature observation and forecast in the previous 9 days during the occurrence process of extreme low temperature is used as the deviation to optimize the forecast extreme low temperature on the current day.

[0171] Sub-module 2.3: Extreme temperature distribution method

[0172] Due to the limited sample of extreme temperature time occurrences, accumulate the distribution curves of the predicted temperature and true value temperature of all historical data of a wind farm, and use the extrapolation of the difference of this curve as the deviation of the extreme temperature sample. That is, adopt the forecast deviation of the near-extreme temperature event as the forecast deviation of the extreme temperature event in this area. Taking the extreme high temperature in the North China Plain as an example, to obtain the forecast error above 40°C, extrapolate the forecast deviation amount from all extreme events above 35°C;

[0173] The effects achieved after the application of the optimization algorithm set module include enriching the reference sample volume from the time and space perspectives and exploring more optimization space for extreme temperature forecasting;

[0174] Sub-module 3: Ensemble optimization evaluation module

[0175] The technical problems to be solved by the ensemble optimization evaluation module include combining the extreme temperature true value and selecting the most suitable meteorological forecast source from multiple meteorological forecast sources.

[0176] The functions and reasons for adopting the set optimization evaluation module include that different prediction sources and different regional optimization methods can obtain a variety of optimized prediction data, and the initial value of the forecast result finally defaulted to is preferably the mode of the extreme temperature forecast set; in subsequent iterations, the true value is gradually used as the optimal evaluation for each wind farm model; finally, the forecast model with the most optimal occurrences in this season is used as the extreme temperature model for this region. Taking the North China Plain region as an example, there are forecast sources from different channels, etc., and the results shown in Figure 6 can be obtained by combining regional optimization methods. For example, in regions where there is no existing observed data available for evaluation, the initial selection is the mode in the result set of all forecast methods as the forecast result. In regions where there are observed results for evaluation, the optimization for this region and this season is used as the evaluation, as shown in Table 2 below.

[0177] Table 2 Statistical example of extreme low temperature events

[0178] Serial number Date Number of affected wind farms Excellent Good Medium Poor Overall excellent and good ratio 1 2023.1.11 33 4 10 4 15 42.4％ 2 2023.1.21 51, 17 key concerns 25 15 5 6 78.4％

[0179] The specific extreme low temperature forecast effects for each wind farm and the characteristics of the nearest meteorological station are as shown in Table 3 below, where Table 2 above shows some station examples of extreme low temperature events released on January 11, 2023.

[0180] Table 3 Summary of wind farms affected by typical extreme low temperature events in Region A

[0181]

[0182] Evaluation method:

[0183] Regional forecast ranking: Establish a rolling scoring system and perform rolling optimization;

[0184] (1) Spatial similarity method

[0185] This method assumes that the extreme temperature distribution in the selected region has spatial similarity, that is, the systematic deviation of similar events in the region is the optimization result of the deviation of this event; the regional deviation distribution is divided according to historical cases. When there are newly added and adjacent extreme temperature events in this year, the proportion of the deviation optimization of adjacent extreme temperature events is higher than that of historical similar extreme temperature events; for example, the systematic deviation of historical extreme temperature events in 2021 is a1℃, the systematic deviation of historical extreme temperature events in 2022 is a2℃, the systematic deviation of historical extreme temperature events in this year is a3℃, and the systematic deviation of the extreme temperature event on the previous day of this event is a4℃, then the selection rule for the daily systematic deviation is a4 * 60% + (a3 + a2 + a1) / 3 * 40%, and it is updated daily by rolling, that is, the daily systematic deviation is calculated according to the regional deviation distribution of historical cases and the newly added extreme temperature events in this year, as shown in Table 4 below.

[0186] Table 4 Characteristics of a Certain Historical Extreme Temperature Event in Region A

[0187]

[0188] Based on Table 4, the prediction deviation of this historical extreme temperature event is determined to be 0.78.

[0189] (2) Temporal Similarity Method

[0190] The temporal similarity assumes that the deviation of the maximum and minimum temperatures of the previous days has coherence, that is, the reference sample is expanded, and even non-extreme temperature events are included in the reference range. To select the previous several days of the wind farm (the number of days varies for each wind farm, ranging from 3 days to 30 days. Taking the extreme low temperature event forecast of a wind farm in Xinjiang as an example, the deviation of the maximum and minimum temperature forecast results of the previous 9 days is used as the forecast deviation of this extreme event). After determining that the window for the wind power project is 9 days, for the time series of this project, it is updated rollingly, and the average deviation of the maximum and minimum temperature forecasts of the previous 9 days is used as the correction amount for the forecast deviation of the current day rollingly. Although the previous 9 days did not reach the extreme high temperature warning threshold, the temporal similarity method verifies that the previous 9 days are the optimal screening period for this wind farm. The deviation between the maximum and minimum values of the previous 9 days and the true value is selected as the deviation input for the extreme temperature event and updated day by day, as Figure 7 shown.

[0191] (3) Extreme Temperature Distribution Method

[0192] The iteration of the historical extreme temperature distribution method lies in updating the true value and the prediction sample day by day. After the update, the deviation statistical means (mean, median) of different wind speed segments of the updated temperature distribution can be obtained. Figure 8 shows a schematic diagram of the observed true temperature sequence and the predicted temperature sequence. Figure 9 shows a schematic diagram of the regional deviation statistics of a certain cold air mass.

[0193] After applying the extreme temperature distribution method, select the optimal model for wind farms in different regions from the best; for wind farms without observed true values, the representativeness of the meteorological station and the consistency with surrounding wind farms can be comprehensively considered for selection; that is: if the wind farm has a true value, use the true value of the wind farm as the basis for optimization; if the wind farm has no true value, comprehensively consider the representativeness of the true values of surrounding meteorological stations and the optimization models of other wind farms in this region for reference.

[0194] Module 3: Equipment Impact Assessment Module

[0195] The technical problems solved by the equipment impact assessment module include the following problems.

[0196] (1) Equipment Reliability: Extreme temperatures can bring risks of equipment failure and malfunction to wind turbines, such as overheating of generators caused by extreme high temperatures and low-temperature impacts on gearboxes caused by extreme low temperatures.

[0197] (2) Equipment operation status: Extreme temperatures can affect the change of the unit operation status. For example, extreme temperatures can cause the unit to shut down due to overheating or low temperature. There are differences in the self-protection settings of the unit in the shutdown state and the operation state.

[0198] (3) Equipment power generation: To increase the power generation of the unit during operation, it is desired to explore the design margin of the unit and explore the controllable operation boundary under extreme temperatures.

[0199] The functions and reasons for adopting the equipment impact assessment module include coping with extreme temperature processes of different severity levels for the unit through the equipment extreme temperature failure model and the equipment operation life model.

[0200] The effects achieved after the application of the equipment impact assessment module include setting targeted thresholds according to different equipment states and the characteristics of equipment failure life, and selecting different impact response measures for extreme temperature processes.

[0201] Sub-module 1: Equipment failure model

[0202] The technical problems solved by the equipment failure model include the failure occurrence frequency of historical extreme temperature events and the historical records of failures, which can be used as a reference for spare parts.

[0203] The functions and reasons for adopting the equipment failure model include collecting historical extreme temperature events of all collectible records of different models, recording the cases that cause the unit to report failures or the change of the unit operation status, using them as the target data set, taking the relationship between extreme temperature intensity, duration, and wind speed as independent variables, and taking the correlation reaching a specific threshold such as above 0.5 as the reference variable. Analyze the probability of different characteristics caused by extreme events with the failure or the change of the unit operation status as the target. After each extreme event, enrich the case set.

[0204] 3. Effects achieved after application:

[0205] If the extreme high or low temperature reaches a specific threshold, it can cause the unit to shut down; the superposition of extreme temperature, wind speed change, and duration can cause the unit to fail to a certain extent;

[0206] Sub-module 2: Equipment life load model

[0207] The technical problems solved by the equipment life load model include quantitatively evaluating whether the unit can explore lower extreme temperatures and the life loss of the unit caused by extreme temperature processes.

[0208] The functions and reasons for adopting the equipment life load model include selecting typical direct-drive and semi-direct-drive models, evaluating the sensitivity list of the impact of extreme temperature processes (different temperatures, temperature changes, wind speeds, air densities) on the ultimate and fatigue loads of units under different operating conditions (this sensitivity list needs to be combined with different models and different designs); according to the degree of superposition of extreme temperatures and different elements, give suggestions for looking up tables. Exemplarily, the sensitivity list is shown in Table 5 below.

[0209] Table 5

[0210]

[0211] During an extreme weather process, the specific function can be defined as a safety threshold function. The inputs are the extreme temperature, wind speed, and environmental parameters of this extreme low-temperature event. The function statistically analyzes the space of the fatigue and ultimate load thresholds and design thresholds of the unit through looking up tables of historical temperature cases. Combining with the early warning results, if the space is higher than 10% and the indirect impact is lower than 1%, then adjust the extreme temperature shutdown threshold of the unit. For example: for the first item, although there is a design space, the failure probability after shutdown is high, so cancel the adjustment of the operation threshold of this unit and only issue an extreme temperature early warning. For the second item, if there is a design space and the failure probability after shutdown is controllable, then adjust the operation temperature threshold of the unit. For example, adjust the low-temperature type unit from -40°C to -42°C. Finally, with the goal of optimal safe operation and power generation over the entire life cycle, determine whether it is possible to operate under super-extreme temperatures during this extreme temperature event.

[0212] The effects achieved after applying the equipment life load model include that the operation boundaries can be adjusted for some regions and some extreme temperature events. Taking the extreme high-temperature events of batch projects in the Xinjiang region as an example, in areas where the extreme high-temperature duration lasts for 12 hours and does not exceed 42°C, attempts can be made to operate and generate electricity.

[0213] The problems to be solved by the early warning release module include that for wind turbines, the early warning information needs to be linked with the equipment information of the wind turbines. For equipment management, not only the extreme temperature events of each wind farm in the region are worthy of attention, but also the severity, impact range, and time of the extreme temperature events are worthy of attention.

[0214] The functions and reasons for adopting the early warning release module include, through the input of extreme temperature early warning levels, the impact of equipment failure models, and the impact on equipment life, clarify the impact range, impact degree, and early warning level of extreme temperature events, as well as differentiated prompt information for different equipment types (new and old models, different extreme survival temperatures in different regions, different sensitive temperatures for different technical routes) for the needs of different levels of personnel in wind farm equipment and regional companies. The effects achieved after applying the early warning release module include presenting the early warning results for each wind farm and each region.

[0215] In some embodiments, the dynamic working process or working principle of the early warning release method is shown in Table 6 below.

[0216] Table 6

[0217]

[0218]

[0219] The following provides several specific examples of extreme temperature early warnings.

[0220] Figure 10 An example diagram of an extreme temperature early warning is shown. Figure 11 An example of individual extreme warning information and review is shown.

[0221] In the temperature prediction module of this embodiment, methods such as machine learning and deep learning can also be applied instead of the mesoscale model for predicting extreme temperatures and related elements.

[0222] In the ensemble optimization module of this embodiment, methods such as machine learning, wavelet analysis, and similarity optimization can be applied to correct the deviations of meteorological elements of extreme temperatures and related forecasts. Specifically, in order to reduce the difference between the predicted extreme temperature and the true value of the historical observed temperature, the difference between the predicted temperature and the true value of the historical observed temperature in the case of extreme temperatures is analyzed. Taking similarity optimization as an example, the true values of observed temperatures and predicted temperatures for many years of history can be collected, and then individual data points most similar to the current time point can be found. For example, if the temperature rise is defined as 8 degrees, the temperature has risen by 8 degrees in the previous 3 hours, and the previous temperature is 10 degrees, such a condition can be set. In the data of the past five years, regional values similar to this condition can be found. These similar individual data points are gathered and subjected to averaging or statistical processing.

[0223] After applying the extreme temperature early warning, extreme temperature events in multiple regions are successfully identified. The operation and maintenance implementation plan for the impact of extreme temperatures on wind farms includes material reserves, provides guidance for the dispatch of intelligent operation and maintenance work orders and personnel arrangements in wind farms; provides early warning prompts for the safety of wind farm personnel.

[0224] Generally speaking, a method for optimizing extreme high and low temperature forecasts based on wind farm SCADA environmental temperature data is provided. In addition, a method for determining the impact of extreme high and low temperatures on the unit is provided.

[0225] Each embodiment in this specification is described in a progressive manner. The same or similar parts between each embodiment can be referred to each other, and the key points of each embodiment are the differences from other embodiments.

[0226] A referring to B means that A is the same as B or A is a simple deformation of B.

[0227] In the description and claims of the embodiments of the present application, terms such as "first" and "second" are used to distinguish different objects, rather than to describe a specific order of the objects, nor can they be construed as indicating or implying relative importance. For example, the first wind farm and the second wind farm are used to distinguish different wind farms, rather than to describe a specific order of the wind farms, nor can it be construed that the first wind farm is more important than the second wind farm.

[0228] The information (including but not limited to user equipment information, user personal information, etc.), data (including but not limited to data for analysis, stored data, displayed data, etc.), and signals involved in the embodiments of the present application are all authorized by the user or fully authorized by all parties, and the collection, use, and processing of the relevant data need to comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the predicted temperatures involved in the present application are all obtained under full authorization.

[0229] In the embodiments of the present application, unless otherwise specified, "at least one" means one or more, and "a plurality" means two or more. For example, a plurality of historical extreme temperature events means two or more historical extreme temperature events.

[0230] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid-state disk (SSD)).

[0231] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. An extreme temperature warning method for a wind farm, characterized in that, The method includes: Obtaining a first predicted temperature of a first wind farm in a first time period through at least one weather forecast source; Obtaining a temperature prediction deviation based on the historical observed temperature true value corresponding to the first wind farm and the historical predicted temperature corresponding to the first wind farm; Correcting the first predicted temperature based on the temperature prediction deviation to obtain the second predicted temperature; Determining a safety risk parameter of a wind turbine in the first wind farm based on the second predicted temperature; If the safety risk parameter meets a preset condition, outputting an extreme temperature warning message corresponding to the first wind farm.

2. The method according to claim 1, wherein The obtaining of the temperature prediction deviation based on the historical observed temperature true value corresponding to the first wind farm and the historical predicted temperature corresponding to the first wind farm includes: Performing regional division based on the administrative region and climate region of wind farm management to determine a target region where the first wind farm is located; Obtaining a temperature prediction deviation based on the historical observed temperature true value and the historical predicted temperature in historical extreme temperature events that occurred in the target region.

3. The method according to claim 2, wherein The obtaining of the temperature prediction deviation based on the historical observed temperature true value and the historical predicted temperature in historical extreme temperature events that occurred in the target region, obtaining the temperature prediction deviation, includes: Determining a weight corresponding to each historical extreme temperature event based on the time difference between the occurrence time of a plurality of historical extreme temperature events that occurred in the target region and the first time period, where the weight corresponding to the historical extreme temperature event is negatively correlated with the time difference; Performing weighted summation on the deviation between the historical temperature true value and the historical predicted temperature in each historical extreme temperature event based on the weight corresponding to each historical extreme temperature event to obtain a temperature prediction deviation.

4. The method according to claim 1, wherein The obtaining of the temperature prediction deviation based on the historical observed temperature true value corresponding to the first wind farm and the historical predicted temperature corresponding to the first wind farm includes: Obtaining the temperature prediction deviation based on the deviation between the historical observed temperature true value of the first wind farm on a plurality of previous dates and the historical predicted temperature of the first wind farm on the plurality of previous dates, where the previous dates are the dates before the date where the first time period is located.

5. The method according to claim 1, wherein The obtaining of the temperature prediction deviation based on the historical observed temperature true value corresponding to the first wind farm and the historical predicted temperature corresponding to the first wind farm includes: Performing curve fitting on the historical observed temperature true value of the first wind farm to obtain a first temperature distribution curve; Performing curve fitting on the historical predicted temperature of the first wind farm to obtain a second temperature distribution curve; Obtaining a temperature prediction deviation based on the difference between the first temperature distribution curve and the second temperature distribution curve.

6. The method according to claim 1, wherein The method further includes: Determining the similarity between at least one weather station in the target region where the first wind farm is located and the first wind farm; Determining a target weather station from the at least one weather station based on the similarity between each weather station and the first wind farm, where the similarity between the target weather station and the first wind farm meets the condition. Based on the historical observed true temperature of the target weather station, obtain the historical observed true temperature corresponding to the first wind farm.

7. The method according to claim 6, wherein The determining the similarity between at least one weather station in the target area where the first wind farm is located and the first wind farm includes: Based on the location of each weather station and the location of the first wind farm, obtain the distance between each weather station and the first wind farm; Based on the elevation value of each weather station and the elevation value of the first wind farm, obtain the elevation difference between each weather station and the first wind farm; Based on the historical temperature distribution of each weather station and the historical temperature distribution of the first wind farm, obtain the correlation of the historical temperature distribution between each weather station and the first wind farm; Based on the weights in the distance dimension, the elevation dimension, and the historical temperature distribution dimension, perform a weighted sum of the distance between each weather station and the first wind farm, the elevation difference between each weather station and the first wind farm, and the correlation of the historical temperature distribution between each weather station and the first wind farm, to obtain the similarity between each weather station and the first wind farm.

8. The method according to claim 1, characterized in that, The method further includes: Based on the historical meteorological data of each wind farm in the target area where the first wind farm is located, the elevation value of each wind farm in the target area, and the location of each wind farm in the target area, determine a second wind farm in the target area that can represent the target area from each wind farm in the target area; Based on the historical observed true temperature of the second wind farm, obtain the historical observed true temperature corresponding to the first wind farm.

9. The method according to claim 1, wherein The method further includes: Based on the historical observed true temperature in the temperature records of multiple wind turbines in the first wind farm, obtain the standard deviation of the temperature records of the multiple wind turbines; From the temperature records of the multiple wind turbines, filter out the temperature records whose deviation between the historical observed true temperature and the standard deviation exceeds the threshold; Determine the average value of the historical observed true temperature in the filtered temperature records, to obtain the historical observed true temperature corresponding to the first wind farm.

10. The method according to claim 1, wherein The obtaining the first predicted temperature of the first wind farm in the first time period through at least one weather forecast source includes: Based on the historical predicted temperature of the target area where the first wind farm is located and the historical observed true temperature corresponding to the target area in the season to which the first time period belongs by the at least one weather forecast source, determine the temperature prediction deviation of each weather forecast source; Based on the temperature prediction deviation of each weather forecast source, determine a target forecast source from the at least one weather forecast source; Obtain the first predicted temperature of the first wind farm in the first time period through the target forecast source.

11. The method according to claim 1, characterized in that, The obtaining the first predicted temperature of the first wind farm in the first time period through at least one weather forecast source includes: Obtain the predicted temperature for the first wind farm in the first time period through each weather forecast source in the at least one weather forecast source, to obtain a set of predicted temperatures; Determine the mode in the set of predicted temperatures, to obtain the first predicted temperature.

12. The method according to claim 1, characterized in that Determining the safety risk parameter of the wind turbine in the first wind farm based on the second predicted temperature includes: Inputting the second predicted temperature into the equipment failure model to obtain the failure risk parameter of the wind turbine. The equipment failure model is established based on historical extreme temperature events and the historical operation records of the wind turbine. The historical operation records are the records of the failures or changes in the operating states of the wind turbine when the historical extreme temperature events occur.

13. The method according to claim 1, characterized in that, Determining the safety risk parameter of the wind turbine based on the second predicted temperature and the operating state of the wind turbine in the first wind farm includes: Inputting the second predicted temperature and the model type of the wind turbine in the first wind farm into the life load model to obtain the life loss risk parameter of the wind turbine. The life load model is established based on extreme temperature, fatigue loads of wind turbines of various model types under extreme temperature, ultimate loads of wind turbines of various model types under extreme temperature, and shutdown condition loads of wind turbines of various model types caused by shutdown due to extreme temperature.

14. A computing device, characterized in that, The computing device includes: a processor, the processor is coupled to a memory, and at least one computer program instruction is stored in the memory. The at least one computer program instruction is loaded and executed by the processor so that the computing device implements the method according to any one of claims 1-13.

Images