Wind power plant energy prediction method and system based on big data analysis

Abstract

The invention discloses a big data analysis-based wind power plant energy prediction method and system. The method comprises the steps of collecting multivariate environment data and equipment data of a wind power plant; simplifying the wind field energy in the multivariate environment data; performing wind power plant energy prediction by using the simplified wind power plant energy and equipment data; and obtaining a final prediction result of the energy of the wind power plant by adding interference of probability anomaly to the prediction result. By analyzing multivariate environment data and equipment data of a wind power plant, energy simplification and prediction are realized by utilizing space gridding and neural network technologies, and by introducing a probability anomaly interference model, the accuracy and robustness of energy prediction can be improved. The grid size is dynamically adjusted in real time, different environment conditions are adapted, the wind power plant operation efficiency is remarkably improved, the uncertainty influence is reduced, the maintenance cost is reduced, and the power grid stability is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of wind farm energy prediction, and in particular to a wind farm energy prediction method and system based on big data analysis. Background Art

[0002] In the context of global energy transformation, wind energy, as a clean and renewable energy, has become an important way to solve the energy crisis and reduce greenhouse gas emissions. Wind farms are increasingly used around the world, especially in areas with abundant wind resources, where the construction and operation of wind farms are expanding. However, wind energy is essentially a highly random and volatile energy source. It is affected by many factors such as climate conditions and geographical environment, making the energy output of wind farms highly uncertain. This brings many challenges to the operation and management of wind farms, energy dispatching, and the stability of the power grid. Summary of the invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a wind farm energy prediction method based on big data analysis, comprising:

[0005] Collect multiple environmental data and equipment data of wind farms;

[0006] Simplifying the wind field energy in the multivariate environmental data;

[0007] Use simplified wind farm energy and equipment data to forecast wind farm energy;

[0008] By adding the interference of probability anomaly to the prediction result, the final prediction result of wind farm energy is obtained.

[0009] As a preferred solution of the wind farm energy prediction method based on big data analysis described in the present invention, the multivariate environmental data includes meteorological data, wind speed and direction data, geographic information data, atmospheric high altitude data, and light and solar radiation data;

[0010] The equipment data includes equipment operation data, power grid load data, maintenance and fault data.

[0011] As a preferred solution of the wind farm energy prediction method based on big data analysis described in the present invention, wherein: simplifying the wind farm energy in the multivariate environmental data includes obtaining the position coordinates of each wind turbine in the system, dividing the wind farm into a plurality of spatial grids, each grid may contain one or more wind turbines;

[0012] The wind turbines in each grid are considered to be under the same wind field microclimate conditions, thus simplifying the analysis of the spatial wind field;

[0013] The multivariate environmental data is input through the trained neural network, and the energy available in each grid at the predicted time T is output;

[0014] During the training process, the energy available in each grid is: the cluster center value obtained by clustering in the energy obtained by the fan; the energy obtained by the fan = the energy produced by the fan ÷ the conversion rate.

[0015] As a preferred solution of the wind farm energy prediction method based on big data analysis described in the present invention, the spatial grid also includes: assuming that the initial grid is W0, if the difference between the available energy in the grid and the energy obtained by the wind turbine exceeds a preset threshold value Y, the grid size is reduced until there is no situation where the difference between the available energy and the energy obtained by the wind turbine exceeds the preset threshold value Y;

[0016] Real-time update of grid status W new , the updated mesh is applied to the simplified process of wind farm energy.

[0017] As a preferred solution of the wind farm energy prediction method based on big data analysis described in the present invention, the wind farm energy prediction includes: taking the available energy in the grid as the energy obtained by each wind turbine therein, the energy produced by the wind turbine = the energy obtained by the wind turbine × the conversion rate, and calculating the energy produced by each wind turbine in the grid according to the conversion rate of each wind turbine;

[0018] The output of each wind turbine is integrated to obtain the predicted wind farm energy:

[0019]

[0020] Where i represents the fan index, M represents the number of fans, and E i represents the energy produced by the i-th wind turbine, and FD represents the energy forecast of the wind farm.

[0021] As a preferred solution of the wind farm energy prediction method based on big data analysis described in the present invention, wherein: the interference of the probabilistic abnormality includes but is not limited to the interference of environmental abnormality, the interference of equipment abnormality, and the interference of integration abnormality;

[0022] Using the abnormal environmental interference, the energy obtained by the wind turbine in each grid is revised;

[0023] J new =J×α

[0024] After interference E i =J new×ε i ;

[0025] Using the abnormal interference of the equipment, the energy produced by each wind turbine in each grid is revised;

[0026] E i,new =E i +β i

[0027] Using the interference of the integration anomaly, the integration process is revised;

[0028]

[0029] Among them, J new represents the energy obtained by the wind turbine in each grid after the interference; J represents the energy obtained by the wind turbine in each grid before the interference; E i，new represents the energy produced by the i-th wind turbine in the grid after the disturbance; FD new Energy prediction of wind farm after disturbance; i represents the conversion rate of the i-th fan;

[0030] α represents the interference factor of environmental anomalies, and its value is a random number within the range, α∈(-ρ,ρ); ρ represents the limit of the interference value of environmental fluctuations;

[0031] β i Indicates the interference factor of the abnormality of the i-th wind turbine in the grid, according to the max The probability of abnormality is U, otherwise it is 0; U = τ × C; ω represents the probability of abnormality of the device itself, Indicates the probability of abnormality of the device model, max Indicates taking the maximum value of the two; C indicates the maximum value of the impact of equipment abnormality; τ indicates the proportion of the impact when an abnormality occurs, which is obtained through the abnormal impact model of probability distribution;

[0032] γ i represents the interference factor of the abnormal integration of the i-th wind turbine in the grid, describing the loss during the integration process, γ i =f+Δ; f is the historical average loss of each wind turbine in the integrated process; Δ represents the random interference term, Δ∈(-μ,μ), and μ represents the limit of the random interference term of the integrated process.

[0033] As a preferred solution of the wind farm energy prediction method based on big data analysis described in the present invention, wherein: the final prediction result includes taking the revised integrated result as the final prediction result of the wind farm energy.

[0034] A wind farm energy prediction system based on big data analysis using the method described in the present invention is characterized by:

[0035] The collection unit collects the multi-dimensional environmental data and equipment data of the wind farm;

[0036] A simplification unit, which simplifies the wind field energy in the multivariate environmental data;

[0037] A prediction unit, which uses simplified wind farm energy and equipment data to perform wind farm energy prediction;

[0038] The output unit obtains the final prediction result of the wind farm energy by adding the interference of probability anomaly to the prediction result.

[0039] A computer device comprises: a memory and a processor; the memory stores a computer program, wherein: the processor implements the steps of any one of the methods of the present invention when executing the computer program.

[0040] A computer-readable storage medium stores a computer program, wherein: when the computer program is executed by a processor, the steps of any one of the methods of the present invention are implemented.

[0041] Beneficial effects of the invention: The wind farm energy prediction method based on big data analysis provided by the invention analyzes the multivariate environmental data and equipment data of the wind farm, uses spatial gridding and neural network technology to achieve energy simplification and prediction, and introduces a probabilistic abnormal interference model to improve the accuracy and robustness of energy prediction. The grid size is adjusted dynamically in real time to adapt to different environmental conditions, significantly improve the operational efficiency of the wind farm, reduce the impact of uncertainty, reduce maintenance costs, and ensure the stability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0043] Figure 1 An overall flow chart of a wind farm energy prediction method based on big data analysis provided for the first embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0045] Example 1, reference Figure 1 , as an embodiment of the present invention, provides a wind farm energy prediction method based on big data analysis, comprising:

[0046] S1: Collect multi-dimensional environmental data and equipment data of the wind farm;

[0047] The multivariate environmental data includes meteorological data, wind speed and direction data, geographic information data, atmospheric high altitude data, and light and solar radiation data; the equipment data includes equipment operation data, power grid load data, and maintenance and fault data.

[0048] The generation of wind energy is closely related to meteorological conditions. Meteorological data such as temperature, humidity, and air pressure directly affect atmospheric movement, thereby affecting changes in wind speed and direction. Therefore, meteorological data is an important basis for wind farm energy prediction. Wind speed determines the energy output of the wind turbine, while wind direction affects the angle adjustment of the wind turbine and the overall wind energy utilization efficiency. Accurate wind speed and direction data can provide real-time wind energy resource conditions for wind farms and help optimize energy output predictions. Solar radiation affects the distribution of surface temperature and air pressure, thereby indirectly affecting the generation and movement trajectory of wind. Therefore, light data can provide a reference for changes in wind speed and direction, which helps to improve the comprehensiveness of the prediction model.

[0049] The operating status of wind turbines directly affects the power generation capacity. Through equipment data, we can understand the health status, power generation efficiency and operation status of each wind turbine, and ensure the consistency of energy forecast and actual output. The grid load reflects the fluctuation of power demand. Collecting this data helps to better coordinate the balance between the power generation output of wind farms and grid demand and optimize the dispatching strategy. The maintenance and failure of wind turbines will affect their energy output. By collecting this data, the abnormal status of the equipment can be considered in the forecast, thereby improving the accuracy of the forecast and the feasibility in actual operation.

[0050] S2: Simplify the wind field energy in the multivariate environmental data.

[0051] Simplifying the wind farm energy in the multivariate environmental data includes obtaining the position coordinates of each wind turbine in the system, dividing the wind farm into a plurality of spatial grids, each of which may contain one or more wind turbines. The wind turbines in each grid are considered to be under the same wind farm microclimate conditions, thereby simplifying the analysis of the spatial wind farm.

[0052] The multivariate environmental data is input through the trained neural network, and the energy available in each grid at the predicted time T is output. During the training process, the energy available in each grid is: the cluster center value obtained by clustering in the energy obtained by the fan; the energy obtained by the fan = the energy produced by the fan ÷ conversion rate.

[0053] Assume that the initial grid is W0. If the difference between the available energy in the grid and the energy obtained by the wind turbine exceeds the preset threshold value Y, the grid size is reduced until there is no situation where the difference between the available energy and the energy obtained by the wind turbine exceeds the preset threshold value Y. Update the grid status W in real time new , the updated mesh is applied to the simplified process of wind farm energy.

[0054] The wind farm is divided into multiple grids, and the wind turbines in each grid are considered to be under similar microclimate conditions. By classifying and simplifying similar areas, the amount of raw data that needs to be processed is reduced, and the overall computational complexity is reduced. Compared with processing fine-grained data for each wind turbine, the gridding method greatly improves computational efficiency. Gridding can effectively capture the differences in wind speed and direction in different areas of the wind farm, so that the model can take into account local wind field changes in the wind farm to a certain extent, while avoiding excessive redundant information caused by independent calculations of all wind turbines. The grid size can be dynamically adjusted according to the prediction accuracy. If it is found that the predicted energy of a grid is significantly different from the actual energy (exceeding the preset threshold Y), the grid is reduced to improve the prediction accuracy of the area. This adaptive design ensures that the model provides more detailed predictions in key areas and avoids wasting computing resources in low-importance areas.

[0055] By processing and predicting energy output through neural networks, historical data can be used to continuously train and optimize the model, giving it better generalization ability and prediction accuracy. Especially in complex, multivariate data scenarios, neural networks can extract implicit nonlinear relationships and improve prediction results. During the training process, a clustering algorithm is used to centralize the energy output in each grid. Simplifying the data dimension by clustering the center value can not only effectively eliminate the interference of some outliers, but also further reduce the amount of calculation and improve the robustness of the model.

[0056] S3: Use simplified wind farm energy and equipment data to perform wind farm energy forecasting.

[0057] The available energy in the grid is taken as the energy obtained by each wind turbine. The energy produced by the wind turbine = the energy obtained by the wind turbine × the conversion rate. According to the conversion rate of each wind turbine, the energy produced by each wind turbine in the grid is calculated. The output of each wind turbine is integrated to obtain the predicted wind farm energy:

[0058]

[0059] Where i represents the fan index, M represents the number of fans, and E i represents the energy produced by the i-th wind turbine, and FD represents the energy forecast of the wind farm.

[0060] S4: By adding the interference of probability anomaly to the prediction result, the final prediction result of the wind farm energy is obtained.

[0061] Furthermore, the probabilistic abnormal interference includes, but is not limited to, interference caused by environmental abnormality, interference caused by equipment abnormality, and interference caused by integration abnormality.

[0062] It should be said that the energy output of a wind farm is directly dependent on environmental conditions, such as wind speed, wind direction, and meteorological conditions. However, environmental conditions are highly random and volatile, especially in extreme weather or sudden climate changes, wind energy resources may fluctuate greatly, resulting in abnormal wind turbine energy acquisition. Therefore, abnormal environmental interference is to reflect the impact of environmental conditions on wind turbine energy acquisition. When typhoons, storms or other extreme weather occur, the wind speed may exceed the normal range, or the wind direction may change suddenly. These environmental factors will have a significant impact on the energy output of the wind turbine, so it is necessary to correct it through the interference factor of environmental abnormality.

[0063] The health status and operating efficiency of wind farm equipment (such as wind turbines) have a direct impact on energy output. Equipment may have abnormal energy output due to mechanical failure, inadequate maintenance, etc. Equipment abnormal interference is designed to simulate the failure or efficiency reduction that wind turbine equipment may encounter during operation, so as to be closer to actual operation. When a wind turbine's drive system fails or the wind turbine speed control system fails, the wind turbine's energy output may be significantly lower than normal. Equipment abnormal interference can revise the wind turbine's energy output based on its health status and equipment failure probability.

[0064] In a wind farm, the energy output of each wind turbine needs to be aggregated to the grid through an integrated system. However, there may be energy loss, reduced transmission efficiency or other systemic problems during the integration process. This abnormality at the integration level affects the overall energy output of the wind farm, so the integrated abnormal interference is to simulate the loss and random fluctuations in the entire integration process.

[0065] The energy obtained by the wind turbines in each grid is revised by utilizing the interference of the abnormal environment.

[0066] J new =J×α

[0067] After interference E i =J new ×ε i .

[0068] The energy produced by each wind turbine in each grid is revised using the abnormal interference of the equipment.

[0069] E i，new =E i +β i

[0070] The integration process is revised by utilizing the interference of the integration anomaly.

[0071]

[0072] Among them, J new represents the energy obtained by the wind turbine in each grid after the interference; J represents the energy obtained by the wind turbine in each grid before the interference; E i，new represents the energy produced by the i-th wind turbine in the grid after the disturbance; FD new Energy prediction of wind farm after disturbance; i represents the conversion rate of the i-th fan.

[0073] α represents the interference factor of environmental anomalies, and its value is a random number within the range. Indicates the limit of the interference value of environmental fluctuations. This limit value is matched by meteorological characteristics. For example, this value will be larger in typhoon weather and smaller in stable weather.

[0074] β i Indicates the interference factor of the abnormality of the i-th wind turbine in the grid, according to the max The probability of abnormality is U, otherwise it is 0; U = τ × C; ω represents the probability of abnormality of the device itself, Indicates the probability of abnormality of the device model, max Indicates the maximum value of the two; C indicates the maximum value of the impact of the equipment abnormality; τ indicates the proportion of the impact when the abnormality occurs, which is obtained through the abnormal impact model of probability distribution. It should be noted that the "abnormal impact model of probability distribution" can use normal distribution, based on historical data, to count the impact proportion when abnormal events occur, and calculate the mean and standard deviation. A normal distribution model is established to describe the probability of occurrence of different impact proportions. Finally, according to the probability of occurrence of different impact proportions, an impact proportion is generated for calculation. (Fan equipment abnormalities do not occur regularly, nor do they have the same impact on energy output in all cases. By generating an abnormal interference factor U (U when an abnormality occurs, otherwise 0) through a probability model, the randomness of equipment abnormalities can be simulated in the model, reflecting the abnormal conditions that may occur in the equipment under different equipment states and operating conditions. By using the normal distribution model to calculate the abnormal impact ratio, the prediction results can be dynamically adjusted according to historical data. This design takes into account the volatility of energy output when equipment abnormalities occur, and gives the magnitude of the impact based on statistical data. Using the normal distribution to calculate the mean and standard deviation can well describe the impact distribution and ensure that the impact level is consistent with the actual operating status of the equipment. When the equipment has experienced multiple small-scale abnormalities in history and the impact ratio is concentrated in a specific interval, the normal distribution model can capture this feature and generate a reasonable impact ratio for calculation, thereby reflecting the true impact of the equipment abnormality)

[0075] γ i represents the interference factor of the abnormal integration of the i-th wind turbine in the grid, describing the loss during the integration process, γ i =f+Δ; f is the historical average loss of each fan in the integration process; Δ represents the random interference term, Δ∈(-μ,μ), and μ represents the limit of the random interference term in the integration process. This interference term is related to the energy output of each fan. In different output ranges, this limit value will match different values.

[0076] It should be said that the interference form of the integration process is related to the weather, and the operation method of using the interference term is similar to that of the environmental anomaly interference term. The reason is that both interferences reflect the impact of some random fluctuations on the energy output of the wind farm. Whether it is the fluctuation of power transmission loss in the integration process or the abnormal changes in environmental conditions, their influence mechanisms have similar characteristics, that is, deviations or losses caused by some random or uncertain factors. Environmental anomaly interference and integration anomaly interference are essentially energy deviations caused by random changes in external factors. For example, weather conditions in environmental factors bring uncertainty (such as typhoons or other extreme weather), and in the integration process, power transmission, inverter efficiency fluctuations or line losses also carry uncertainty. This randomness and fluctuation can be described by similar mathematical models (such as normal distribution, random interference terms, etc.). The interference of environmental anomalies and the interference in the integration process are both revised by introducing random interference factors to revise the model output. In environmental anomalies, the interference term is a random number generated according to the extreme values ​​of weather conditions (for example, the interference factor of typhoon weather is larger and that of stable weather is smaller). Similarly, the randomness of power transmission or equipment loss in the integration process can also be handled by similar interference terms. The two methods are consistent in dealing with uncertainty. Both define the range of random fluctuations through limit values ​​and generate random interference in combination with historical data models. When designing a model, it is a common method to describe the impact of abnormal interference through probability distribution. Whether it is environmental anomalies or possible loss fluctuations during the integration process, probability models such as normal distribution can well describe the impact of these random fluctuations and determine the size of the interference through the mean and standard deviation. Therefore, the interference terms of the two methods are generated in a similar way, and both reflect the impact of uncertainty on energy output through the random number generation mechanism.

[0077] It is also important to know that during the integration process, the energy loss of the wind turbine is often not fixed, but fluctuates with the change of the wind turbine output power. The degree of loss of the wind turbine will also vary in different output power ranges. By associating the limit value of the interference term with the wind turbine output, the actual loss situation can be more accurately reflected, thereby avoiding the prediction error caused by the use of fixed loss values. When the wind turbine output power is high, the loss in the power transmission and conversion process may increase (for example, the heating effect of the transmission line increases), so the limit value of the interference term should be increased; when the wind turbine output is low, the loss is relatively small, and the corresponding limit value of the interference term should be reduced. This design can capture the trend of loss changing with output power. By dynamically adjusting the limit value of the interference term, the model can more accurately simulate the change of energy loss at different wind turbine output power levels, and improve the accuracy of the prediction results. The energy output of each wind turbine is different, and the loss behavior will also change accordingly, and the fixed loss parameter cannot reflect this dynamic characteristic. By distinguishing the limit values ​​of the interference term in different output ranges, the energy prediction can be adjusted more finely. When the wind turbine is at low wind speed, the energy output is less, and the loss ratio in power transmission is smaller; when the wind speed is high and the wind turbine output is large, the loss ratio may increase. By adjusting the interference item limit value, the model can better reflect these loss differences and improve the accuracy of the overall prediction. During the integration process, each wind turbine may have different loss characteristics due to equipment status, geographical location, operating load, etc. By adjusting the interference item limit value in different energy output ranges, the impact of individual differences of each wind turbine on the integrated loss can be reflected, making the prediction more personalized and accurate. A wind turbine on flat terrain has a shorter power transmission line and less loss; while a wind turbine located in a mountainous or offshore wind farm has a longer transmission line and greater loss. By adjusting the interference item limit value of each wind turbine in combination with historical data, the loss impact caused by individual differences can be better reflected.

[0078] The revised integrated results are taken as the final prediction results of wind farm energy.

[0079] On the other hand, this embodiment also provides a wind farm energy prediction system based on big data analysis, which includes:

[0080] The collection unit collects the multi-dimensional environmental data and equipment data of the wind farm.

[0081] A simplification unit simplifies the wind field energy in the multivariate environmental data.

[0082] The prediction unit uses simplified wind farm energy and equipment data to perform wind farm energy prediction.

[0083] The output unit obtains the final prediction result of the wind farm energy by adding the interference of probability anomaly to the prediction result.

[0084] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0085] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0086] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0087] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0088] Example 2 is an embodiment of the present invention, which provides a wind farm energy prediction method based on big data analysis. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0089] The test was conducted in a simulated wind farm system, with different environmental conditions, equipment status and abnormal factors in the integration process. The simulated wind farm area consists of 10 wind turbines, located in different geographical conditions (including plains and mountains), and the environmental conditions change frequently. The test evaluates the energy forecast by collecting multi-environmental data (including meteorological data, wind speed and direction data, light and solar radiation data) and equipment data (equipment operation data, grid load data, maintenance and fault data) of the wind farm.

[0090] During the experiment, the wind farm was first gridded and divided into five grids, each containing two wind turbines. The wind turbines in each grid were considered to be under the same microclimate conditions, simplifying the analysis and energy calculation of the wind farm. Next, the neural network model was used to predict the wind farm energy in the grid. The input of the neural network model was the collected multivariate environmental data and equipment operation data, and the output was the wind turbine energy acquisition at the predicted time T.

[0091] During the model training process, historical wind turbine operation data was used for fitting, and the energy output center value in each grid was obtained by combining the clustering algorithm. The energy acquisition calculation formula of the wind turbine is: energy obtained by the wind turbine = energy produced by the wind turbine ÷ conversion rate. The experiment adjusted the size of the grid several times to ensure that the difference between the predicted energy and the actual measured energy was controlled within the preset threshold range.

[0092] To verify the effectiveness of the anomaly handling mechanism, three interference test scenarios were designed: environmental anomalies (such as typhoons and low wind speed weather), equipment anomalies (such as wind turbine operation failure or maintenance), and integrated process anomalies (such as losses and line failures in power transmission). These abnormal conditions are simulated according to different probability distributions, where the limit values ​​of the environmental anomaly interference factor are set to larger and smaller values ​​according to typhoon weather or calm weather, respectively. The equipment anomaly interference factor generates failure probabilities based on historical data, and the integrated anomaly interference factor is generated by simulating fluctuations in power transmission.

[0093] During the test, the wind farm energy output data under normal conditions, abnormal environment, abnormal equipment and abnormal integration were recorded, and multiple experiments were conducted to ensure the reliability of the data. The data collection interval was once an hour, and the experiment lasted for 24 hours.

[0094] Table 1 Experimental data table

[0095]

[0096]

[0097] It can be seen from the above table data that the inventive scheme of this embodiment performs well in energy prediction under normal and abnormal conditions, and through the multiple interference revision mechanism, it can effectively improve the accuracy and robustness of the prediction results. Under the condition that the wind turbines 1, 3, 5, 6 and other equipment are operating normally, the predicted energy acquisition value is slightly different from the actual energy acquisition value, and the average error rate is controlled at about 5%, indicating that under normal conditions, the energy prediction model based on big data analysis can better reflect the actual situation and has high accuracy.

[0098] When abnormalities occurred in fans 2, 4, and 10, although the operating status of the equipment affected the energy acquisition of the fan, the prediction model accurately predicted the energy output under abnormal conditions by adjusting the abnormal interference factor of the equipment. In particular, when fans 4 and 10 failed, although the equipment conversion rate was greatly reduced, the model still adjusted the prediction results according to the probability distribution of the abnormal impact, so that the predicted value was not much different from the actual value, proving the effectiveness of the abnormal interference mechanism of the equipment.

[0099] For abnormal environmental conditions (such as fluctuations in wind speed and direction), the model predicts that the energy acquisition value can adapt to environmental changes by adjusting the environmental interference factors. Wind turbines 5 and 7 are respectively operated in environments with higher wind speeds (9.0m / s) and lower wind speeds (6.0m / s), and the predicted values ​​are 165kWh and 105kWh, which are not much different from the actual values, indicating that the environmental interference mechanism effectively responds to changes in meteorological conditions and improves the accuracy of energy prediction.

[0100] In terms of integrated abnormal interference, the actual energy acquisition of wind turbine 9 is 140kWh and the predicted energy is 138kWh, which shows that during the integration process, the model can accurately handle transmission loss and integration abnormality problems and maintain a high prediction accuracy. By using random interference terms, the model can cope with the uncertainty in the integration process, making the overall energy prediction more robust.

[0101] In general, compared with the prior art, the present invention has significant creativity and novelty. By introducing gridding, neural networks and multi-level interference correction mechanisms, accurate prediction of wind farm energy can be achieved, especially when dealing with complex environments and equipment anomalies. The invention content greatly improves the reliability and practicality of the prediction, which is significantly better than traditional time series analysis or physical model prediction. By combining historical data with real-time data, the present invention can adaptively adjust the prediction results and has strong practical application potential.

[0102] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A wind farm energy prediction method based on big data analysis, characterized in that: include: Collect multiple environmental data and equipment data of wind farms; Simplifying the wind field energy in the multivariate environmental data; Use simplified wind farm energy and equipment data to forecast wind farm energy; By adding the interference of probability anomaly to the prediction result, the final prediction result of wind farm energy is obtained.

2. The wind farm energy prediction method based on big data analysis according to claim 1, characterized in that: The multivariate environmental data includes meteorological data, wind speed and direction data, geographic information data, atmospheric high altitude data, and light and solar radiation data; The equipment data includes equipment operation data, power grid load data, maintenance and fault data.

3. The wind farm energy prediction method based on big data analysis according to claim 2, characterized in that: Simplifying the wind farm energy in the multivariate environmental data includes obtaining the position coordinates of each wind turbine in the system, dividing the wind farm into a plurality of spatial grids, each grid may contain one or more wind turbines; The wind turbines in each grid are considered to be under the same wind field microclimate conditions, thus simplifying the analysis of the spatial wind field; The multivariate environmental data is input through the trained neural network, and the energy available in each grid at the predicted time T is output; During the training process, the energy available in each grid is: the cluster center value obtained by clustering in the energy obtained by the fan; the energy obtained by the fan = the energy produced by the fan ÷ the conversion rate.

4. The wind farm energy prediction method based on big data analysis according to claim 3, characterized in that: The spatial grid also includes, assuming that the initial grid is W0, if the difference between the available energy in the grid and the energy obtained by the wind turbine exceeds a preset threshold value Y, then reducing the grid size until there is no situation where the difference between the available energy and the energy obtained by the wind turbine exceeds the preset threshold value Y; Real-time update of grid status W new , the updated mesh is applied to the simplified process of wind farm energy.

5. The wind farm energy prediction method based on big data analysis according to claim 4, characterized in that: The wind farm energy prediction includes taking the available energy in the grid as the energy obtained by each wind turbine therein, the energy produced by the wind turbine = the energy obtained by the wind turbine × the conversion rate, and calculating the energy produced by each wind turbine in the grid according to the conversion rate of each wind turbine; The output of each wind turbine is integrated to obtain the predicted wind farm energy: Where i represents the fan index, M represents the number of fans, and E i represents the energy produced by the i-th wind turbine, and FD represents the energy forecast of the wind farm.

6. The wind farm energy prediction method based on big data analysis according to claim 5, characterized in that: The probabilistic abnormal interference includes, but is not limited to, interference caused by abnormal environment, abnormal equipment, and abnormal integration; Using the abnormal environmental interference, the energy obtained by the wind turbine in each grid is revised; J new =J×α After interference E i =J new ×ε i ; Using the abnormal interference of the equipment, the energy produced by each wind turbine in each grid is revised; AND i，new =And i +β i Using the interference of the integration anomaly, the integration process is revised; Among them, J new represents the energy obtained by the wind turbine in each grid after the interference; J represents the energy obtained by the wind turbine in each grid before the interference; E i，new represents the energy produced by the i-th wind turbine in the grid after the disturbance; FD new Energy forecasting of wind farms after disturbances; ε i represents the conversion rate of the i-th fan; α represents the interference factor of environmental anomalies, and its value is a random number within the range, α∈(-ρ,ρ); ρ represents the limit of the interference value of environmental fluctuations; β i Represents the interference factor of the abnormality of the i-th wind turbine in the grid, according to The probability of abnormality is U, otherwise it is 0; U = τ × C; ω represents the probability of abnormality of the device itself, Indicates the probability of abnormality of the device model. Indicates taking the maximum value of the two; C indicates the maximum value of the impact of equipment abnormality; τ indicates the proportion of the impact when an abnormality occurs, which is obtained through the abnormal impact model of probability distribution; γ i represents the interference factor of the abnormal integration of the i-th wind turbine in the grid, describing the loss during the integration process, γ i =f+Δ; f is the historical average loss of each wind turbine in the integrated process; Δ represents the random interference term, Δ∈(-μ,μ), and μ represents the limit of the random interference term of the integrated process.

7. The wind farm energy prediction method based on big data analysis according to claim 6, characterized in that: The final prediction result includes taking the revised integrated result as the final prediction result of the wind farm energy.

8. A wind farm energy prediction system based on big data analysis using the method according to any one of claims 1 to 7, characterized in that: The collection unit collects the multi-dimensional environmental data and equipment data of the wind farm; A simplification unit, which simplifies the wind field energy in the multivariate environmental data; A prediction unit, which uses simplified wind farm energy and equipment data to perform wind farm energy prediction; The output unit obtains the final prediction result of the wind farm energy by adding the interference of probability anomaly to the prediction result.

9. A computer device comprising: A memory and a processor; the memory stores a computer program, wherein the processor implements the steps of any method as claimed in claim 1 when executing the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images