Tower scanning risk terrain identification method, model training method, device and equipment

Abstract

The present application discloses a method for identifying tower sweeping risk terrain, a model training method, a device and equipment, and the above-mentioned tower sweeping risk terrain identification method includes: obtaining original terrain information of a target area, the target area is an area within a preset distance range of a target position; preprocessing the original terrain information to obtain target terrain information; inputting the target terrain information into a trained tower sweeping risk terrain identification model to obtain a risk identification result, and the risk identification result is used to indicate whether there is a tower sweeping risk of a wind turbine at the target position. The embodiments of the present application can effectively improve the recognition efficiency and accuracy of terrain with tower sweeping risks.

Description

Technical Field

[0001] The present application belongs to the field of wind power generation technology, and in particular relates to a tower scanning risk terrain identification method, a model training method, a device and equipment. Background Art

[0002] As we all know, in the process of selecting a site for wind turbine installation, the terrain information of the selected site is usually determined to further identify the impact of the terrain on the operation of the wind turbine. For example, some terrains are prone to negative wind shear, that is, the wind speed decreases with increasing height, and negative wind shear may cause the wind turbine blade to sweep the tower, resulting in huge quality losses.

[0003] In the prior art, when identifying risky terrain where negative wind shear may occur and cause tower sweeping accidents, the identification is usually done based on manual experience, which has the problems of low efficiency and poor accuracy. Summary of the invention

[0004] The embodiments of the present application provide a method for identifying tower sweeping risk terrain, a model training method, a device and equipment to solve the problem that the prior art relies on manual experience to identify risk terrain where negative wind shear may occur and cause tower sweeping accidents, resulting in low recognition efficiency and poor accuracy.

[0005] In a first aspect, an embodiment of the present application provides a tower sweep risk terrain identification method, comprising: obtaining original terrain information of a target area, where the target area is an area within a preset distance range of a target position;

[0006] Preprocess the original terrain information to obtain target terrain information;

[0007] The target terrain information is input into the trained tower sweep risk terrain identification model to obtain the risk identification result. The risk identification result is used to indicate whether there is a wind turbine tower sweep risk at the target location.

[0008] In a second aspect, an embodiment of the present application provides a tower sweep risk terrain recognition model training method, comprising:

[0009] Obtain terrain information samples and historical wind measurement data of the sample area;

[0010] Annotate the terrain information samples according to the historical wind measurement data to obtain risk terrain annotations associated with the terrain information samples, wherein the risk terrain annotations are used to indicate whether there is a risk of wind turbine tower sweeping;

[0011] According to the terrain information samples and risk terrain annotations, the tower-sweeping risk terrain recognition model is trained to obtain a trained tower-sweeping risk terrain recognition model.

[0012] In a third aspect, an embodiment of the present application provides a tower sweep risk terrain identification device, the device comprising:

[0013] A first acquisition module is used to acquire original terrain information of a target area, where the target area is an area within a preset distance range of a target position;

[0014] A preprocessing module is used to preprocess the original terrain information to obtain target terrain information;

[0015] The identification module is used to input the target terrain information into the trained tower sweep risk terrain identification model to obtain a risk identification result, and the risk identification result is used to indicate whether there is a wind turbine tower sweep risk at the target location.

[0016] In a fourth aspect, an embodiment of the present application provides a tower sweep risk terrain identification model training device, comprising:

[0017] The second acquisition module is used to obtain terrain information samples and historical wind measurement data of the sample area;

[0018] The first labeling module is used to label the terrain information sample according to the historical wind measurement data to obtain the risk terrain labeling associated with the terrain information sample, wherein the risk terrain labeling is used to indicate whether there is a risk of wind turbine tower sweeping;

[0019] The training module is used to train the tower-sweeping risk terrain recognition model according to the terrain information samples and the risk terrain annotations to obtain the trained tower-sweeping risk terrain recognition model.

[0020] In a fifth aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: a processor and a memory storing computer program instructions;

[0021] When the processor executes the computer program instructions, it implements the above-mentioned tower-sweeping risk terrain identification method, or implements the above-mentioned tower-sweeping risk terrain identification model training method.

[0022] In a sixth aspect, an embodiment of the present application provides a computer storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the above-mentioned tower scanning risk terrain identification method or the above-mentioned tower scanning risk terrain identification model training method is implemented.

[0023] The tower sweep risk terrain identification method, model training method and device, equipment and storage medium of the embodiment of the present application, after determining the target position and target area, pre-processes the original terrain information of the target area to obtain the target terrain information, and inputs the target terrain information into the trained tower sweep risk terrain identification model to obtain a risk identification result, and based on the risk identification result, it can be further identified whether there is a tower sweep risk of the wind turbine at the target position. The embodiment of the present application uses the trained tower sweep risk terrain identification model to identify the terrain information of the area within the preset range of the target position to obtain a risk identification result, so as to further identify whether there is a risk of tower sweep accidents caused by terrain factors when the wind turbine is installed at the target position. Compared with the existing identification method based on manual experience, it can effectively improve the identification efficiency; at the same time, it also reduces the influence of subjective judgment factors, and can improve the accuracy of identifying risk terrain that may cause tower sweep accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0025] Figure 1 It is a flowchart of a tower sweep risk terrain identification model training method provided in an embodiment of the present application;

[0026] Figure 2 is an example diagram of the division of the region of interest in an embodiment of the present application;

[0027] Figure 3 It is a flow chart of a tower sweeping risk terrain identification model training method provided in an embodiment of the present application in a specific application example;

[0028] Figure 4 It is a flowchart of a tower-sweeping risk terrain identification method provided by an embodiment of the present application;

[0029] Figure 5 is a flow chart of a tower-sweeping risk terrain identification method provided by another embodiment of the present application;

[0030] Figure 6 It is a flow chart marking the risk identification results on the map in an embodiment of the present application;

[0031] Figure 7 It is an example diagram of risk identification results marked on a map in an embodiment of the present application;

[0032] Figure 8 It is a flow chart of a tower sweeping risk terrain identification method provided by an embodiment of the present application in a specific application example;

[0033] Fig. 9 It is a structural schematic diagram of a tower sweeping risk terrain identification device provided in an embodiment of the present application;

[0034] Fig.10 It is a structural schematic diagram of a tower sweeping risk terrain identification device provided in an embodiment of the present application;

[0035] Fig.11 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0037] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0038] In order to solve the problems of the prior art, the embodiments of the present application provide a tower sweeping risk terrain identification method, device, equipment and computer storage medium. The following first introduces the tower sweeping risk terrain identification model training method provided by the embodiments of the present application.

[0039] Figure 1 A schematic diagram of a flow chart of a tower sweep risk terrain identification model training method provided by an embodiment of the present application is shown. Figure 1 As shown, the method includes:

[0040] Step 101, obtaining terrain information samples and historical wind measurement data of a sample area;

[0041] Step 102, annotating the terrain information sample according to the historical wind measurement data to obtain the risk terrain annotation associated with the terrain information sample, wherein the risk terrain annotation is used to indicate whether there is a risk of wind turbine tower sweeping;

[0042] Step 103 , training a tower-sweeping risk terrain recognition model according to the terrain information samples and the risk terrain annotations to obtain a trained tower-sweeping risk terrain recognition model.

[0043] In this embodiment, the sample area can be generally considered to be some areas that actually exist in geography, and these areas usually have specific corresponding terrain environments, for example, the sample area can be a mountain, a canyon, a hill or a plain, etc. These terrain environments can be reflected by terrain information, and the terrain information can be obtained through a digital elevation model (DEM) or a topographic map, or can also be obtained through on-site surveying and mapping, which is not specifically limited here.

[0044] It is easy to understand that DEM usually refers to the digital simulation of the ground terrain through limited terrain elevation data, that is, the digital expression of the terrain surface morphology; DEM can be considered as a physical ground model that represents the ground elevation in the form of a set of ordered numerical arrays. Based on DEM, it is relatively convenient to obtain the above terrain information.

[0045] In this embodiment, the terrain information of the sample area can be considered as a component of the training sample, so it can be called a terrain information sample. In some other embodiments, it can also be called other names, which are not limited in this application. In the sample area, there may be some historical wind measurement data. For example, a wind measurement tower can be set up in the sample area, or wind measurement equipment can be set up on the existing infrastructure in the sample area to obtain these historical wind measurement data.

[0046] The historical wind measurement data may include some data that may cause wind turbine tower sweep accidents, such as wind direction, wind speed, and wind shear data. Based on the historical wind measurement data, the terrain information samples may be annotated to obtain risk terrain annotations associated with the terrain information samples, and the risk terrain annotations may be used to indicate whether there is a risk of wind turbine tower sweeps.

[0047] For example, based on historical wind measurement data, the wind shear data of the time points when the wind speed in a certain sample area is greater than the wind speed threshold within a period of history can be counted. When the number of negative wind shear data in these wind shear data meets specific conditions, the sample area can be marked as "1". The number 1 can be considered as the risk terrain label associated with the terrain information sample of the sample area, which is used to indicate the risk of wind turbine tower sweeping.

[0048] The tower scanning risk terrain identification model can be a deep learning model, for example: it can be a neural network model, or it can be a logistic regression model or a support vector machine, etc., which is not specifically limited here. For the sake of simplicity, the following will mainly take the tower scanning risk terrain identification model as a neural network model as an example for explanation.

[0049] Based on the terrain information samples and risk terrain annotations, the tower-sweeping risk terrain recognition model is trained to obtain a trained tower-sweeping risk terrain recognition model;

[0050] It is easy to understand that deep learning models usually have corresponding loss functions. When training the deep learning model, the terrain information samples can be used as the input of the deep learning model, and the loss value of the loss function can be obtained based on the output of the deep learning model and the risk terrain annotation. The network parameters of the deep learning model are continuously adjusted according to the loss value. When the loss value of the loss function is less than the preset value, or the accuracy of the output result of the deep learning model meets the requirements, the training of the deep learning model can be completed to obtain the above-mentioned trained tower scanning risk terrain recognition model.

[0051] For the trained tower sweep risk terrain identification model, its input and output are similar to the tower sweep risk terrain identification during training. Specifically, when using the tower sweep risk terrain identification model for risk identification, the input can be the terrain information of a certain area, and the output result may be the above-mentioned "1" indicating the existence of a wind turbine tower sweep risk, or it may be the probability of the output being "1", and so on.

[0052] The tower-sweeping risk terrain identification model training method provided in the embodiment of the present application obtains terrain information samples and historical wind measurement data of the sample area, uses the historical wind measurement data to annotate the terrain information samples, obtains risk terrain annotations associated with the terrain information samples, and uses the terrain information samples and the risk terrain annotations to train the tower-sweeping risk terrain identification model to obtain a trained tower-sweeping risk terrain identification model; subsequently, when identifying terrain that may bring tower-sweeping risks, it is only necessary to input the terrain information into the trained tower-sweeping risk terrain identification model to obtain the tower-sweeping risk terrain identification result, thereby helping to improve the identification efficiency of risk terrain, while also reducing the influence of subjective judgment factors, and helping to improve the identification accuracy of risk terrain.

[0053] In one example, the terrain information sample includes terrain information of a sample region of interest (ROI), and the historical wind measurement data is measured at a sample position in the sample region;

[0054] The sample ROI corresponds to a preset wind direction; the boundary of the sample ROI is formed by the first side, the second side, the third side and the fourth side being connected and enclosed in sequence, and the sample position is located within the boundary enclosed area of ​​the sample ROI;

[0055] The extension direction of the first side and the extension direction of the third side are both consistent with the preset wind direction corresponding to the sample ROI;

[0056] The extension direction of the second side and the extension direction of the fourth side are both perpendicular to the preset wind direction corresponding to the sample ROI, and in the preset wind direction corresponding to the sample ROI, the second side is located upwind of the fourth side; the vertical distance between the sample position and the second side is greater than the vertical distance between the sample position and the fourth side.

[0057] It is easy to understand that, in general, ROI can refer to an area to be processed that is outlined from the processed image in the form of a box, circle, ellipse or irregular polygon in machine vision and image processing. In this embodiment, the sample ROI can be simply understood as an area outlined from a map or digital model with elevation information by a rectangular box. The specific method of obtaining the sample ROI will be described below.

[0058] For the sample location, it can be considered as a wind tower, or the geographical location of other types of wind measurement equipment such as radar. For the sample ROI, see Figure 2 ,exist Figure 2 In the figure, point O can be considered as the sample position, the dotted rectangle can be considered as the boundary of the sample ROI, and the curve can be considered as the contour line, or terrain information. There is an enclosed area within the boundary of the sample ROI. Combined with the contour line, the elevation value of each position point in the enclosed area can be obtained, and the elevation values ​​of all position points in the enclosed area constitute an elevation value matrix, which can be used as the terrain information of the sample ROI. Of course, in practical applications, the above elevation values ​​can be obtained through digital elevation models or topographic maps.

[0059] The sample ROI corresponds to a preset wind direction. For example, if the north direction is taken as the reference direction 0°, the preset wind direction corresponding to a sample ROI can be in the angle range of -11.25° to 11.25°. In practical applications, the sample ROI corresponds to a preset wind direction, and the enclosed area of ​​the sample ROI boundary can be matched with the preset wind direction. For example, the enclosed area is placed more in the upwind area of ​​the sample position, or the long side of the rectangular enclosed area is kept consistent with the extension direction of the center line of the preset wind direction.

[0060] To achieve the effect of matching the preset wind direction in the enclosed area of ​​the sample ROI boundary, continue to refer to Figure 2, the boundary of the sample ROI can be a rectangular frame, corresponding to the dotted frame where the four points A, B, C, and D in the figure are located. These four points can be the midpoints of the edges respectively; the sample position O can be located within the boundary enclosed area of ​​the above sample ROI. For the sample ROI, the following parameters can be associated: the length length in front of the incoming flow direction, the length rLength behind the incoming flow direction, the width width, and the azimuth alpha of the incoming flow. Among them, the OA length in the figure can correspond to length, or the vertical distance between the above sample position and the second edge; the OB length can correspond to rLength, or the vertical distance between the above sample position and the fourth edge; the CD length can correspond to width; and the azimuth alpha corresponds to a preset angle, which can be measured with due north as the reference direction 0°, in Figure 2 In the figure, N can refer to the north direction, and accordingly, ∠NCA in the figure corresponds to alpha.

[0061] In order to facilitate the acquisition of the terrain information of the sample ROI, the longitude long and latitude lat of the sample location can also be associated with the sample ROI. In this way, the elevation values ​​of the relevant location points can be extracted more conveniently from the DEM or topographic map based on the longitude and latitude of the sample location and the enclosed area of ​​the sample ROI.

[0062] In an optional solution, the vertical distance between the target position and the first side may be equal to the vertical distance between the target position and the third side.

[0063] In another optional solution, since the sample ROI corresponds to a preset wind direction, when using historical wind measurement data to mark the terrain information of the sample ROI, historical wind measurement data matching the preset wind direction corresponding to the sample ROI can also be used for marking, so that the marking result can be more reasonable.

[0064] Optionally, more specifically, the terrain information sample may correspond to a preset wind direction. Accordingly, the above step 102, marking the terrain information sample according to the historical wind measurement data to obtain the risk terrain marking associated with the terrain information sample, includes:

[0065] A plurality of target wind shear data are obtained from the historical wind measurement data, and the wind direction corresponding to each target wind shear data matches the preset wind direction corresponding to the terrain information sample;

[0066] When the proportion of negative wind shear data in multiple target wind shear data is greater than a proportion threshold, the terrain information sample is marked as a positive sample, and the positive sample is used to indicate the risk of wind turbine tower sweeping.

[0067] In this embodiment, wind shear data is mainly used as the basis for judging whether the terrain of the sample area will bring the risk of tower sweeping. At the same time, the data sorting and annotation method of wind direction can be adopted. For example, for a terrain information sample of a sample area, the corresponding preset wind direction is the angle range of -11.25° to 11.25° shown in the above embodiment, and the angle range can be 0° with due north as the reference direction; at this time, the wind shear data at the time point when the wind direction in the sample area is -11.25° to 11.25° can be counted, that is, the above-mentioned target wind shear data; in this way, the wind direction corresponding to the target wind shear data can be matched with the preset wind direction corresponding to the terrain information sample.

[0068] Of course, in practical applications, other factors may also be considered when determining the target wind shear data. For example, the target wind shear data may be determined in combination with the wind speed, and the wind speed is further determined as the target wind shear data only when it is greater than a wind speed threshold.

[0069] Generally speaking, when negative wind shear occurs, it is easy to cause tower sweep accidents, and negative wind shear may be affected by both meteorological conditions and geographical environment. In this embodiment, in order to reduce the impact of meteorological conditions on the identification of tower sweep risk terrain, for a terrain information sample, when the proportion of negative wind shear data in the corresponding multiple target wind shear data is greater than the proportion threshold, the terrain information sample can be marked as a positive sample to indicate that the terrain corresponding to the terrain information sample has a tower sweep risk; further, in practical applications, label 1 can also be set for positive samples; relatively, there can also be negative samples, that is, the corresponding terrain information samples whose proportion of negative wind shear data is less than a proportion threshold, and label 0 can be set for negative samples. It is worth emphasizing that the proportion thresholds used in determining positive and negative samples can be the same or different. For example, the proportion threshold used in determining negative samples can be made smaller, so that positive and negative samples can be effectively distinguished.

[0070] like Figure 3 As shown, Figure 3 The flowchart of the tower sweep risk terrain identification model training method provided in the embodiment of the present application in a specific application example is shown, including:

[0071] Step 301, preprocessing the wind measurement data based on the wind tower observation data;

[0072] The wind tower data content can include wind direction and wind speed time series at different altitudes with a resolution of 10 minutes, and can include more than 4,000 wind towers located at different wind measurement locations.

[0073] For the preprocessing of wind measurement data, firstly, the data can be divided into 16 sectors according to the wind direction of the wind tower closest to the hub height. Each sector corresponds to a 22.5° interval, with the north direction as 0°. After that, each sector is screened according to the wind speed and the amount of data with strong winds. In this application example, 6m / s can be used as the wind speed threshold. When the wind speed at the hub height exceeds the wind speed threshold, it is considered to be strong wind. In order to ensure that the sample size in each sector is sufficient to avoid the instability of proportional statistics, it can be defined as a sufficient sample size of more than 10,000.

[0074] For a sector with enough strong wind samples, if the proportion of wind negative shear samples is greater than 33.3%, the sector is considered a positive sample. 33.3% corresponds to the above-mentioned proportion threshold, which can be determined based on post-evaluation experience. To ensure a significant difference between positive and negative samples, negative samples can be set to a proportion of wind negative shear samples less than 0.1%. Finally, the positive samples are set to label 1 and the negative samples are set to label 0.

[0075] Step 302, extracting terrain information from the digital elevation model for preprocessing according to the coordinates of the wind tower;

[0076] When discussing how terrain affects wind turbine flow, the wind direction can be set, and the sample ROI associated with it is the area within a certain distance in the direction of the wind. The sample ROI can be a rectangular area to facilitate the use of 2D convolution operations.

[0077] For the sample ROI, five parameters can be defined: the longitude long, latitude lat, length length in front of the incoming flow direction, length rLength behind the incoming flow direction, width, and azimuth alpha of the incoming flow. Figure 2 As shown in the figure, the curves are contour lines, indicating the terrain undulations. Point O is the location of the wind tower or wind turbine site, corresponding to the above-mentioned point of interest, defined by longitude and latitude. N is the north direction, and ∠NCA is the azimuth alpha of the incoming flow. The lengths of AO and OB correspond to the length and rLength parameters, and CD corresponds to the width parameter, all in meters. The dotted box in the figure is the sample ROI, and the DEM elevation values ​​in this area will be extracted.

[0078] For each point of interest, the longitude and latitude and the azimuth of the incoming flow direction are read from the statistical data of the wind tower, and the length and width of the sample ROI are uniformly set. After the rectangular area of ​​interest is delineated, the digital elevation model or topographic map is cropped and the elevation values ​​within the boundary are extracted using the regional boundary and stored in the form of a matrix. To ensure that the relationship between any extracted matrix and the point of interest is consistent, all matrices can be adjusted to take the left boundary as the incoming flow direction. After data preprocessing, all matrices should have the same size, such as a matrix of 40×30 elements.

[0079] Finally, the preprocessed terrain matrix is ​​paired with the label of the corresponding sector (i.e., positive or negative sample 0 / 1) as training samples. 20% of the training samples are randomly selected as the test set, and the rest are the training set.

[0080] Step 303, using the pre-processed wind measurement data and terrain information to train the deep learning model;

[0081] After preprocessing the wind data and terrain information of the wind tower, the deep learning model can be trained and verified. The deep learning model here can correspond to the above-mentioned tower scanning risk terrain identification model. The input training sample of the deep learning model is a matrix of size N×L×W, where N is the number of training samples, and L and W are the length and width of the sample ROI (or the number of elements in each row and column of the matrix). The terrain matrix label input to the deep learning model can be a one-dimensional vector with N elements.

[0082] In one example, the above-mentioned deep learning model can be an 8-layer deep convolutional neural network, in which the 1st, 3rd, and 5th layers are convolutional layers, the 2nd, 4th, and 6th layers are pooling layers, and the 7th and 8th layers are fully connected layers. The size of the convolution kernel in the convolutional layer is 3×3, the step size is 1, and the pooling layer uses maximum pooling with a step size of 2. The convolutional layer and the first fully connected layer use the ReLU function as the activation function, and the last fully connected layer uses the Sigmoid function as the activation function. In order to avoid overfitting of the model on the test set, the 7th fully connected layer is set to randomly discard 20% of the neurons.

[0083] The training of deep learning models can use TensorFlow2.0 as the computing framework, stochastic gradient descent as the optimization algorithm, and binary cross entropy as the loss function. The training is carried out for 100 epochs, and the training samples are traversed once in each epoch.

[0084] Step 304, verify whether the tower sweeping risk terrain recognition result reaches the preset value. If so, end the training of the deep learning model to obtain the trained tower sweeping risk terrain recognition model. If not, execute step 305;

[0085] If the deep learning model identifies risk terrain on the test set, or the accuracy of identifying high wind negative shear meets the requirements, such as 75%, the deep learning model is saved to obtain the above-mentioned trained tower scanning risk terrain identification model.

[0086] Step 305, debug the deep learning model, and return to execute step 303;

[0087] Deep learning models can be optimized by adjusting the neural network structure and hyperparameters.

[0088] The trained tower-sweeping risk terrain identification model obtained by the above-mentioned tower-sweeping risk terrain identification model training method, in subsequent applications, only needs to input the terrain information that needs to be identified for risk into it, and the identification results of the risk terrain that may cause tower-sweeping accidents can be obtained. The time cost and computing resources consumed are both low, and the terrain that is prone to tower-sweeping risks for wind turbines can be efficiently identified.

[0089] Figure 4 A schematic diagram of a process flow of a tower sweep risk terrain identification method provided by an embodiment of the present application is shown. Figure 4 As shown, the method includes:

[0090] Step 401, obtaining original terrain information of a target area, where the target area is an area within a preset distance range of a target position;

[0091] Step 402, preprocessing the original terrain information to obtain target terrain information;

[0092] Step 403: input the target terrain information into the trained tower sweep risk terrain identification model to obtain a risk identification result, which is used to indicate whether there is a wind turbine tower sweep risk at the target location.

[0093] In this embodiment, the target location can be considered as the geographical location of the wind turbine, or the candidate installation location of the wind turbine. The target location often has a corresponding terrain environment, for example, the target location may be located in terrain such as a mountain top, a hill or a canyon; and the terrain environment will have an impact on the operation of the wind turbine. For example, in some terrain environments, negative wind shear is prone to occur, that is, the wind speed decreases with increasing height. At this time, when the wind speed is greater than a certain value, a tower sweep accident will occur; specifically, under the condition of negative wind shear, the wind speed at the lower blade tip of the wind turbine may be greater than the wind speed at the center of the hub, causing the deformation of the blade to exceed the design allowable range and hit the tower.

[0094] The terrain environment of the target location can be reflected by the terrain information of the target area where the target location is located. Specifically, the target area can be a geographical area within a preset distance range of the target location, and the preset distance range can be set according to actual needs and is not specifically limited here; and the terrain information here can be considered as the original terrain information mentioned above, and the original terrain information can be obtained through an existing digital elevation model or topographic map, or can also be obtained through on-site surveying and mapping, which is also not specifically limited here.

[0095] The original terrain information can usually be used to obtain the elevation values ​​of each location point in the target area. However, the above original terrain information generally needs to be preprocessed. For example, the original terrain information may only have contour lines, and it may be necessary to interpolate the elevation values ​​of each point in the target area; or, it may only be necessary to obtain the elevation values ​​of the location points in a certain angle range of the target location, etc.

[0096] Of course, preprocessing the original terrain information can not only be used to obtain terrain information of a specific area, but also be used to process the original terrain information into a format that can be received by the trained tower-sweeping risk terrain recognition model. It can be seen that by preprocessing the original terrain information, the terrain information of a specific area can be extracted according to specific needs, and the processing requirements of the trained tower-sweeping risk terrain recognition model for terrain information can be reduced, thereby reducing the difficulty of developing the tower-sweeping risk terrain recognition model.

[0097] By inputting the target terrain information into the trained tower-sweeping risk terrain identification model, the risk identification result can be obtained; for the trained tower-sweeping risk terrain identification model, it can be obtained by training the tower-sweeping risk terrain identification model based on training samples, and the tower-sweeping risk terrain identification model can be a deep learning model, and the deep learning model here can be a neural network model, or a logistic regression model or a support vector machine, etc., which is not specifically limited here. For the sake of simplicity of description, unless otherwise specified below, the trained tower-sweeping risk terrain identification model will be referred to as the tower-sweeping risk terrain identification model, and the tower-sweeping risk terrain identification model before training will be referred to as the deep learning model; at the same time, the deep learning model will be mainly explained as a neural network model.

[0098] The above-mentioned training samples may include terrain information samples, and risk terrain annotations associated with the terrain information samples. It is easy to understand that for terrain information samples, their specific format may be similar to the above-mentioned target terrain information, so that subsequent target terrain information can be processed and identified by the tower sweep risk terrain recognition model. For risk terrain annotations, they can be obtained by terrain information samples, mainly used to indicate whether there is a wind turbine tower sweep risk in the terrain environment corresponding to the terrain information sample; for example, historical wind measurement data in the terrain environment corresponding to the terrain information sample can be collected, and by sorting out the historical wind measurement data, it can be determined whether a specific wind negative shear situation has occurred. When it occurs, the terrain information sample can be annotated so that its associated risk terrain annotation indicates the existence of a wind turbine tower sweep risk. Of course, this is just an example of how to obtain training samples, and the specific methods of the above annotations can be adjusted according to actual needs.

[0099] Corresponding to the training process of the deep learning model, the output of the tower sweep risk terrain recognition model of this embodiment can be a risk recognition result, that is, the recognition result of the tower sweep risk terrain recognition model for the target terrain information can be used to indicate whether there is a wind turbine tower sweep risk at the target location. For example, the output of the tower sweep risk terrain recognition model can be expressed as "the probability value of a wind turbine tower sweep is 80%", and if the probability value is greater than a probability threshold of 50%, it can be considered that the risk recognition result is that there is a wind turbine tower sweep risk at the target location.

[0100] The tower sweep risk terrain identification method provided in the embodiment of the present application, after determining the target position and the target area, pre-processes the original terrain information of the target area to obtain the target terrain information, and inputs the target terrain information into the trained tower sweep risk terrain identification model to obtain the risk identification result, and based on the risk identification result, it can be further identified whether there is a tower sweep risk of the wind turbine at the target position. The embodiment of the present application uses the trained tower sweep risk terrain identification model to identify the terrain information of the area within the preset range of the target position to obtain the risk identification result, so as to further identify whether there is a risk of tower sweep accidents caused by terrain factors when the wind turbine is installed at the target position. Compared with the existing identification method based on manual experience, it can effectively improve the identification efficiency; at the same time, it also reduces the influence of subjective judgment factors, and can improve the accuracy of identifying risk terrain that may cause tower sweep accidents.

[0101] See also Figure 5 , Figure 5 A flow chart of a tower sweep risk terrain identification method provided by another embodiment of the present application is shown. The difference between this embodiment and the previous embodiment is that the above step 402 can be refined into step 502 and step 503, and correspondingly, the above step 403 can be refined into step 504.

[0102] Specifically, in this embodiment, the tower sweep risk terrain identification method includes:

[0103] Step 501, obtaining original terrain information of a target area, where the target area is an area within a preset distance range of a target position;

[0104] Step 502, dividing the target area to obtain M first ROIs, wherein each first ROI corresponds to a preset wind direction, and M is a positive integer;

[0105] Step 503, obtaining terrain information of each first ROI respectively according to the original terrain information; wherein the target terrain information includes the terrain information of each first ROI;

[0106] Step 504, respectively input the terrain information of each first ROI into the trained tower sweep risk terrain identification model to obtain the risk identification result associated with each first ROI; wherein the risk identification result associated with the first ROI is used to indicate whether there is a wind turbine tower sweep risk in the preset wind direction corresponding to the first ROI at the target location.

[0107] In this embodiment, step 501 is the same as step 401 in the previous embodiment and will not be described again here.

[0108] In order to facilitate the tower scanning risk terrain recognition model to receive and identify the target terrain information, in this embodiment, a method of dividing ROI according to the target area is adopted, and the ROI can be subsequently input into the tower scanning risk terrain recognition model, thereby helping to ensure the consistency of the input of the tower scanning risk terrain recognition model.

[0109] By dividing the target area, at least one first ROI can be obtained, and each first ROI can correspond to a preset wind direction. For ease of understanding, the following uses several examples to illustrate the method of dividing the target area:

[0110] In one example, the preset wind direction may correspond to a preset angle range centered on the target position. For example, with due north as the reference direction 0°, a preset wind direction may correspond to an angle range of -11.25° to 11.25°. The first ROI corresponding to the preset wind direction may be matched with the preset wind direction in the enclosed area of ​​the first ROI boundary. For example, the enclosed area may be placed more in the upwind area of ​​the target position, or the long side of the rectangular enclosed area may be kept consistent with the extension direction of the center line of the preset wind direction.

[0111] Accordingly, combined with the above division of the preset angle range, the target area can actually be divided into 16 first ROIs, each of the first ROIs corresponds to a preset angle range of 22.5°, and the 16 first ROIs are symmetrical about the center of the target position.

[0112] In another example, the preset wind direction may also be all wind directions centered at the target position, that is, corresponding to an angle range of -180° to 180°. Accordingly, the number of the first ROIs obtained by division may be one.

[0113] Of course, in practical applications, the division of the angle range corresponding to the preset wind direction, or the selection of the number of first ROIs and the corresponding preset wind direction, can be set as needed.

[0114] Since the first ROI usually needs to be consistent or similar to the sample ROI used in the training process of the tower scanning risk terrain recognition model, for the first ROI and its terrain information, you can also refer to Figure 2, at this time, Figure 2 In the figure, point O can be considered as the target position, the dotted rectangle can be considered as the boundary of the first ROI, and the curve can be considered as the contour line, which can correspond to the above-mentioned original terrain information. There is an enclosed area within the boundary of the first ROI. Combined with the original terrain information, the elevation value of each position point in the enclosed area can be obtained, and the elevation values ​​of all position points in the enclosed area constitute an elevation value matrix, which can be used as the terrain information of the first ROI.

[0115] The first ROI is input into the tower sweep risk terrain identification model, and the risk identification result associated with the first ROI can be obtained. Since the first ROI corresponds to a preset wind direction, the risk identification result obtained in this implementation can be used to indicate whether there is a wind turbine tower sweep risk at the target location in the preset wind direction corresponding to the first ROI. For example, if the preset wind direction corresponding to a first ROI is -11.25°~11.25°, and the first ROI is input into the tower sweep risk terrain identification model, the risk identification result obtained indicates the existence of a tower sweep risk, which means that if the wind turbine is installed at the target location, a tower sweep accident may occur under the influence of the due north wind.

[0116] It can be seen that in this embodiment, on the one hand, the target area is divided into at least one first ROI, which can ensure the consistency of the input of the tower sweep risk terrain identification model; on the other hand, each first ROI is corresponded to a preset wind direction, so that the risk identification results of tower sweep accidents occurring at the target location under various wind directions can be obtained.

[0117] In one example, the boundary of the first ROI is formed by the first side, the second side, the third side, and the fourth side being connected and enclosed in sequence, and the target position is located within the boundary enclosed area of ​​the first ROI;

[0118] The extension direction of the first side and the extension direction of the third side are both consistent with the preset wind direction corresponding to the first ROI;

[0119] An extension direction of the second side and an extension direction of the fourth side are both perpendicular to a preset wind direction corresponding to the first ROI, and in the preset wind direction corresponding to the first ROI, the second side is located upwind of the fourth side; a vertical distance between the target position and the second side is greater than a vertical distance between the target position and the fourth side.

[0120] See also Figure 2 The boundary of the first ROI may be a rectangular frame, corresponding to the dotted frame where the four points A, B, C, and D in the figure are located, and these four points may be the midpoints of the edges respectively; the target position O may be located within the boundary enclosed area of ​​the first ROI.

[0121] For the first ROI, the following parameters can be associated: length in front of the incoming flow direction, length rLength behind the incoming flow direction, width width, and alpha of the incoming flow. The OA length in the figure can correspond to length, or the vertical distance between the above target position and the second side; the OB length can correspond to rLength, or the vertical distance between the above target position and the fourth side; the CD length can correspond to width; and the alpha azimuth angle corresponds to a preset angle, which can be measured with due north as the reference direction 0°. Figure 2 In the figure, N can refer to the north direction, and accordingly, ∠NCA in the figure corresponds to alpha.

[0122] In addition, as shown above, Figure 2 The curves in the image may be contour lines, which can generally be used to represent terrain undulations. In this embodiment, these contour lines may be used to extract elevation values ​​in the enclosed area of ​​the first ROI boundary, thereby obtaining terrain information of the first ROI. Of course, in practical applications, the elevation values ​​may be obtained through a digital elevation model (DEM) or a topographic map.

[0123] In combination with a specific application scenario, the size of the first ROI can be fixed to L×W, where L and W are the length and width of the first ROI, or the number of elements in the length direction and the number of elements in the width direction, and each element can correspond to an elevation value. For example, the size of the first ROI can be 40×30, and the digital elevation model or topographic map can be cropped using the first ROI boundary, and the elevation values ​​within the boundary can be extracted and stored in a matrix form. In order to ensure that the relative position relationship of each element in any first ROI is consistent, the left boundary of the rectangular boundary of each first ROI, that is, the direction of the side where point C or point D is located, can be adjusted to the incoming flow direction, that is, the preset wind direction corresponding to the first ROI.

[0124] In an optional solution, the vertical distance between the target position and the first side may be equal to the vertical distance between the target position and the third side.

[0125] This embodiment limits the setting method of the first ROI boundary, so that each first ROI can have a high consistency in the boundary extension direction, the relative position of the internal elements, etc., which helps to improve the quality of the input of the risk terrain prediction model and improve the accuracy of identifying the risk terrain that is prone to tower sweeping accidents. At the same time, the limitation of the distance between the target position and the two edges distributed in sequence in the incoming flow direction in the first ROI boundary also makes the first ROI boundary enclosed area and the corresponding preset wind direction more matched, so that the risk of tower sweeping accidents at the target position under each preset wind direction can be more accurately identified.

[0126] Optionally, the target terrain information corresponds to a preset wind direction; accordingly, Figure 6 As shown, in the above step 403, the target terrain information is input into the trained tower-sweeping risk terrain identification model, and after the risk identification result is obtained, the tower-sweeping risk terrain identification method further includes:

[0127] Step 601, obtaining a target wind direction, where the target wind direction is a preset wind direction at a target location where there is a risk of wind turbine tower sweeping;

[0128] Step 602, obtaining a target sector area corresponding to the target wind direction according to a preset correspondence relationship between wind directions and sector areas;

[0129] Step 603: Mark the target sector area in the map showing the target location.

[0130] In combination with the above embodiment, the original terrain information is preprocessed to obtain terrain information of M first ROIs, each of which corresponds to a preset wind direction; accordingly, the terrain information of the first ROI is input into the tower sweep risk terrain identification model, and the result is an identification result of whether there is a wind turbine tower sweep risk at the target location in the corresponding preset wind direction. In other words, the target terrain information can actually correspond to a preset wind direction; in order to simplify the description, the following mainly takes the example that the target terrain information includes the terrain information of the first ROI, and the terrain information of the first ROI corresponds to a preset wind direction.

[0131] For the target wind direction, it may refer to the preset wind direction in which there is a tower sweep risk at the target location; and for ease of explanation, the first ROI corresponding to the target wind direction may be defined as the second ROI. It is easy to understand that when an identification result indicating the existence of a tower sweep risk appears, the above-mentioned second ROI may be obtained according to the input-output relationship of the tower sweep risk terrain identification model, and the second ROI corresponds to a preset wind direction, and the preset wind direction corresponding to the second ROI may be considered as the target wind direction. For example, in combination with the example of dividing the target area into 16 first ROIs in the above text, when the target wind direction is 0°, the corresponding target sector area may be the sector area corresponding to -11.25° to 11.25° with the target location as the circle.

[0132] Of course, it is easy to understand that the above is only an example of the specific usage of the correspondence between the preset wind direction and the fan-shaped area. In the preset correspondence, the number of the first ROI divisions, and the angle of the preset wind direction, the reference direction, etc. can be set according to actual needs.

[0133] The map showing the target location may be a map such as Google Earth or Baidu Map, in which the target location and the target area can be displayed. The marking method for the target sector area may be to mark the outline of the target sector area, or to load a semi-transparent layer on the target sector area, etc., which is not specifically limited here.

[0134] In one example, in order to facilitate the above-mentioned target wind direction that may bring the risk of wind turbine tower sweep, it can be marked on Google Maps, and a Google Maps KML file with the target location and target wind direction marked can be output, so that when the user opens the file, he can clearly and intuitively observe which wind directions at the target location have the risk of tower sweep. Figure 7 As shown in the figure, the pin position P represents the target position, that is, the location of the wind turbine or the candidate installation location of the wind turbine; and the area S in the figure is the sector range where the tower sweep risk may occur, which is identified based on the tower sweep risk terrain recognition model.

[0135] As can be seen from the above, this embodiment helps the user to more clearly and intuitively know the wind direction where there is a tower sweep risk at the target location by marking the target sector area in the map showing the target location.

[0136] Optionally, the above step 401, obtaining original terrain information of the target area, includes:

[0137] Get the target latitude and longitude information of the target area;

[0138] Based on the target longitude and latitude information and the preset elevation information, the original terrain information of the target area is obtained, and the preset elevation information records the corresponding relationship between the longitude and latitude information and the terrain elevation.

[0139] It is easy to understand that the target location can refer to the installation location of the wind turbine, or the candidate installation location; the target area is the area within the preset distance range of the target location. In practical applications, the target area is usually an actual geographical area; therefore, the above target area can also have a target longitude and latitude information, which can include the longitude and latitude of the target location and other locations in the target area; wherein the longitude in the longitude and latitude can be represented by long, and the latitude can be represented by lat.

[0140] The preset elevation information may exist in, for example, a digital elevation model or a topographic map, and is used to record the correspondence between longitude and latitude information and terrain elevation; the longitude and latitude information here can be recognized as corresponding to the longitude and latitude position, such as the various positions in the target area mentioned above; therefore, based on the target longitude and latitude information and the preset elevation information, the terrain elevation of each position point in the target area is obtained, and the original terrain information of the target area can be obtained.

[0141] This embodiment acquires the original terrain information of the target area based on the target longitude and latitude information, which can be implemented using the existing preset elevation information. The acquisition method is relatively convenient and efficient.

[0142] like Figure 8 As shown, Figure 8 The flowchart of the tower sweep risk terrain identification method provided in the embodiment of the present application in a specific application example is shown, including:

[0143] Step 801, read demand information;

[0144] The demand information can be considered as the target location specified by the user for tower scanning risk terrain identification; when the target location and distance range are determined, the target area can be determined;

[0145] Step 802, preprocessing the original terrain information of the target area based on the digital elevation model;

[0146] The purpose of this step is to obtain target terrain information that can be input into the tower scanning risk terrain identification model. For example, the first ROI can be obtained by dividing the target area, and the terrain information of the first ROI can be extracted according to the digital elevation model.

[0147] Step 803, inputting the target terrain information obtained by preprocessing into the tower-sweeping risk terrain identification model to perform tower-sweeping risk terrain identification;

[0148] For any target terrain information, the tower-sweeping risk terrain identification model can output a value between 0 and 1, representing the probability that the terrain within the relevant angle range of the target position is a risk terrain that may cause a tower-sweeping accident; for example, when the output value is greater than 0.5, it can be considered as risk terrain; if the terrain information of multiple first ROIs obtained by dividing a target area is input at one time, the risk terrain probability values ​​of all 16 sectors corresponding to the target area should be output.

[0149] Step 804, displaying the tower scanning risk terrain identification result;

[0150] The result of identifying the risk terrain for tower sweeping can be a probability table of the risk terrain composition corresponding to each sector. For the convenience of display, a Google Earth KML file with the target location and the direction where the tower sweeping risk exists can also be output. When the user opens the file, he can clearly and intuitively observe which target locations and directions have the risk of tower sweeping.

[0151] Based on the use of the above-mentioned tower sweeping risk terrain identification model, the risk terrain that may cause tower sweeping accidents can be identified relatively quickly; the time cost and computing resources consumed are less, and compared with manual identification, the efficiency and accuracy of tower sweeping risk terrain identification can be effectively improved.

[0152] like Fig. 9 As shown, the embodiment of the present application also provides a tower sweep risk terrain identification device, including:

[0153] The first acquisition module 901 is used to acquire original terrain information of a target area, where the target area is an area within a preset distance range of a target position;

[0154] A preprocessing module 902 is used to preprocess the original terrain information to obtain target terrain information;

[0155] The identification module 903 is used to input the target terrain information into the trained tower sweep risk terrain identification model to obtain a risk identification result, and the risk identification result is used to indicate whether there is a wind turbine tower sweep risk at the target location.

[0156] Optionally, the preprocessing module 902 includes:

[0157] A division unit, used for dividing the target area to obtain M first regions of interest ROIs, wherein each first ROI corresponds to a preset wind direction, and M is a positive integer;

[0158] A first acquisition unit is used to acquire the terrain information of each first ROI respectively according to the original terrain information; wherein the target terrain information includes the terrain information of each first ROI;

[0159] Accordingly, the identification module 903 includes:

[0160] The identification unit is used to input the terrain information of each first ROI into the terrain identification model respectively to obtain the risk identification result associated with each first ROI; wherein the risk identification result associated with the first ROI is used to indicate whether there is a risk of wind turbine tower sweeping in the preset wind direction corresponding to the first ROI at the target location.

[0161] Optionally, the boundary of the first ROI is formed by the first side, the second side, the third side and the fourth side being connected and enclosed in sequence, and the target position is located within the boundary enclosed area of ​​the first ROI;

[0162] The extension direction of the first side and the extension direction of the third side are both consistent with the preset wind direction corresponding to the first ROI;

[0163] An extension direction of the second side and an extension direction of the fourth side are both perpendicular to a preset wind direction corresponding to the first ROI, and in the preset wind direction corresponding to the first ROI, the second side is located upwind of the fourth side; a vertical distance between the target position and the second side is greater than a vertical distance between the target position and the fourth side.

[0164] Optionally, the target terrain information corresponds to a preset wind direction;

[0165] Accordingly, the above-mentioned tower sweep risk terrain identification device may further include:

[0166] The third acquisition module is used to acquire a target wind direction, where the target wind direction is a preset wind direction at a target location where there is a risk of wind turbine tower sweeping;

[0167] A fourth acquisition module is used to acquire a target sector area corresponding to the target wind direction according to a preset correspondence between the wind direction and the sector area;

[0168] The second marking module is used to mark the target sector area in the map showing the target location.

[0169] Optionally, the first acquisition module 901 includes:

[0170] A second acquisition unit is used to acquire target longitude and latitude information of a target area;

[0171] The third acquisition unit is used to acquire original terrain information of the target area based on the target longitude and latitude information and preset elevation information, wherein the preset elevation information records the corresponding relationship between the longitude and latitude information and the terrain elevation.

[0172] It should be noted that the tower sweeping risk terrain identification device is a device corresponding to the above-mentioned tower sweeping risk terrain identification method. All implementation methods in the above-mentioned tower sweeping risk terrain identification method embodiment are applicable to the embodiment of the tower sweeping risk terrain identification device and can achieve the same technical effect.

[0173] like Fig.10 As shown, the embodiment of the present application also provides a tower sweep risk terrain identification model training device, which is characterized by comprising:

[0174] The second acquisition module 1001 is used to acquire terrain information samples and historical wind measurement data of the sample area;

[0175] The first labeling module 1002 is used to label the terrain information sample according to the historical wind measurement data to obtain the risk terrain label associated with the terrain information sample, wherein the risk terrain label is used to indicate whether there is a risk of wind turbine tower sweeping;

[0176] The training module 1003 is used to train the tower-sweeping risk terrain recognition model according to the terrain information samples and the risk terrain annotations to obtain a trained tower-sweeping risk terrain recognition model.

[0177] Optionally, the terrain information sample includes terrain information of a sample ROI, and the historical wind measurement data is measured at a sample position in the sample area;

[0178] The sample ROI corresponds to a preset wind direction; the boundary of the sample ROI is formed by the first side, the second side, the third side and the fourth side being connected and enclosed in sequence, and the sample position is located within the boundary enclosed area of ​​the sample ROI;

[0179] The extension direction of the first side and the extension direction of the third side are both consistent with the preset wind direction corresponding to the sample ROI;

[0180] The extension direction of the second side and the extension direction of the fourth side are both perpendicular to the preset wind direction corresponding to the sample ROI, and in the preset wind direction corresponding to the sample ROI, the second side is located upwind of the fourth side; the vertical distance between the sample position and the second side is greater than the vertical distance between the sample position and the fourth side.

[0181] Optionally, the terrain information sample corresponds to a preset wind direction;

[0182] Accordingly, the first marking module 1002 includes:

[0183] A fourth acquisition unit is used to acquire a plurality of target wind shear data from the historical wind measurement data, wherein the wind direction corresponding to each target wind shear data matches the preset wind direction corresponding to the terrain information sample;

[0184] The labeling unit is used to label the terrain information sample as a positive sample when the proportion of negative wind shear data in multiple target wind shear data is greater than a proportion threshold, and the positive sample is used to indicate the existence of a wind turbine tower sweep risk.

[0185] It should be noted that the tower-sweeping risk terrain identification model training device is a device corresponding to the above-mentioned tower-sweeping risk terrain identification model training method, and all implementation methods in the above-mentioned tower-sweeping risk terrain identification model training method embodiment are applicable to the embodiment of the tower-sweeping risk terrain identification model training device, and can also achieve the same technical effect.

[0186] Fig.11 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application is shown.

[0187] The electronic device may include a processor 1101 and a memory 1102 storing computer program instructions.

[0188] Specifically, the processor 1101 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0189] The memory 1102 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 1102 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 1102 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 1102 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 1102 is a non-volatile solid-state memory.

[0190] The processor 1101 reads and executes computer program instructions stored in the memory 1102 to implement any one of the tower-sweeping risk terrain identification methods or tower-sweeping risk terrain identification model training methods in the above-mentioned embodiments.

[0191] In one example, the electronic device may further include a communication interface 1103 and a bus 1104. Fig.11 As shown, the processor 1101, the memory 1102, and the communication interface 1103 are connected via a bus 1104 and communicate with each other.

[0192] The communication interface 1103 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0193] Bus 1104 includes hardware, software or both, and the parts of online data flow billing equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front-end bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 1104 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the present application considers any suitable bus or interconnection.

[0194] In addition, in combination with the tower sweeping risk terrain identification method and the tower sweeping risk terrain identification model training method in the above embodiments, the present application embodiment can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by the processor, any one of the tower sweeping risk terrain identification method or tower sweeping risk terrain identification model training method in the above embodiments is implemented.

[0195] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0196] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0197] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0198] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0199] The above are only specific implementation methods of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. A tower sweep risk terrain identification method, characterized in that: include: Acquire original terrain information of a target area, where the target area is an area within a preset distance range of a target position; Preprocessing the original terrain information to obtain target terrain information; Inputting the target terrain information into a trained tower sweep risk terrain identification model to obtain a risk identification result, wherein the risk identification result is used to indicate whether there is a wind turbine tower sweep risk at the target location; The preprocessing of the original terrain information to obtain target terrain information includes: Dividing the target area to obtain M first regions of interest ROIs, wherein each of the first ROIs corresponds to a preset wind direction, and M is a positive integer; According to the original terrain information, respectively obtain terrain information of each first ROI; wherein the target terrain information includes the terrain information of each first ROI; The tower sweep risk terrain recognition model is obtained by training based on terrain information samples of the sample area and risk terrain annotations associated with the terrain information samples, and the risk terrain annotations are used to indicate whether there is a tower sweep risk for wind turbines; The terrain information sample corresponds to a preset wind direction, and the risk terrain annotation is obtained in the following manner: Acquire a plurality of target wind shear data from the historical wind measurement data of the sample area, wherein the wind direction corresponding to each of the target wind shear data matches the preset wind direction corresponding to the terrain information sample; When the proportion of negative wind shear data in the multiple target wind shear data is greater than a proportion threshold, the terrain information sample is marked as a positive sample, and the positive sample is used to indicate the existence of a wind turbine tower sweep risk.

2. The method according to claim 1, characterized in that The target terrain information is input into the trained tower scanning risk terrain identification model to obtain a risk identification result, including: The terrain information of each of the first ROIs is respectively input into the terrain recognition model to obtain a risk identification result associated with each of the first ROIs; wherein the risk identification result associated with the first ROI is used to indicate whether there is a risk of wind turbine tower sweeping in the preset wind direction corresponding to the first ROI at the target location.

3. The method according to claim 1, characterized in that: The boundary of the first ROI is formed by the first side, the second side, the third side and the fourth side being connected and enclosed in sequence, and the target position is located within the boundary enclosed area of ​​the first ROI; The extension direction of the first side and the extension direction of the third side are both consistent with the preset wind direction corresponding to the first ROI; An extension direction of the second side and an extension direction of the fourth side are both perpendicular to a preset wind direction corresponding to the first ROI, and in the preset wind direction corresponding to the first ROI, the second side is located upwind of the fourth side; and a vertical distance between the target position and the second side is greater than a vertical distance between the target position and the fourth side.

4. The method according to claim 1, characterized in that: The target terrain information corresponds to a preset wind direction; After inputting the target terrain information into the trained tower-sweeping risk terrain identification model to obtain a risk identification result, the method further includes: Obtaining a target wind direction, where the target wind direction is a preset wind direction at the target location where there is a risk of wind turbine tower sweeping; According to the preset correspondence between wind direction and sector area, obtaining the target sector area corresponding to the target wind direction; The target sector area is marked on the map showing the target location.

5. The method according to any one of claims 1 to 4, characterized in that The obtaining of original terrain information of the target area includes: Obtaining target longitude and latitude information of the target area; The original terrain information of the target area is acquired based on the target longitude and latitude information and the preset elevation information, wherein the preset elevation information records the corresponding relationship between the longitude and latitude information and the terrain elevation.

6. A tower sweep risk terrain identification model training method, characterized in that: include: Obtain terrain information samples and historical wind measurement data of the sample area; The terrain information sample is labeled according to the historical wind measurement data to obtain a risk terrain label associated with the terrain information sample, wherein the risk terrain label is used to indicate whether there is a risk of wind turbine tower sweeping; According to the terrain information sample and the risk terrain annotation, a tower-sweeping risk terrain recognition model is trained to obtain a trained tower-sweeping risk terrain recognition model; The terrain information sample includes terrain information of a sample ROI, and the sample ROI corresponds to a preset wind direction; the historical wind measurement data is measured at a sample position in the sample area.

7. The method according to claim 6, characterized in that The boundary of the sample ROI is formed by the first side, the second side, the third side and the fourth side being connected and enclosed in sequence, and the sample position is located within the boundary enclosed area of ​​the sample ROI; The extension direction of the first side and the extension direction of the third side are both consistent with the preset wind direction corresponding to the sample ROI; The extension direction of the second side and the extension direction of the fourth side are both perpendicular to the preset wind direction corresponding to the sample ROI, and in the preset wind direction corresponding to the sample ROI, the second side is located upwind of the fourth side; the vertical distance between the sample position and the second side is greater than the vertical distance between the sample position and the fourth side.

8. The method according to claim 6, characterized in that The terrain information sample corresponds to a preset wind direction; The step of labeling the terrain information sample according to the historical wind measurement data to obtain risk terrain labeling associated with the terrain information sample includes: Acquire a plurality of target wind shear data from the historical wind measurement data, wherein the wind direction corresponding to each of the target wind shear data matches the preset wind direction corresponding to the terrain information sample; When the proportion of negative wind shear data in the multiple target wind shear data is greater than a proportion threshold, the terrain information sample is marked as a positive sample, and the positive sample is used to indicate the existence of a wind turbine tower sweep risk.

9. A tower sweep risk terrain identification device, characterized in that: include: A first acquisition module is used to acquire original terrain information of a target area, where the target area is an area within a preset distance range of a target position; A preprocessing module, used for preprocessing the original terrain information to obtain target terrain information; An identification module is used to input the target terrain information into a trained tower sweep risk terrain identification model to obtain a risk identification result, wherein the risk identification result is used to indicate whether there is a wind turbine tower sweep risk at the target location; The preprocessing module comprises: A division unit, configured to divide the target area into M first regions of interest ROIs, wherein each of the first ROIs corresponds to a preset wind direction, and M is a positive integer; A first acquisition unit, configured to acquire terrain information of each of the first ROIs respectively according to the original terrain information; wherein the target terrain information includes terrain information of each of the first ROIs; The tower sweep risk terrain recognition model is obtained by training based on terrain information samples of the sample area and risk terrain annotations associated with the terrain information samples, wherein the risk terrain annotations are used to indicate whether there is a tower sweep risk for a wind turbine generator; The terrain information sample corresponds to a preset wind direction, and the risk terrain annotation is obtained in the following manner: Acquire a plurality of target wind shear data from the historical wind measurement data of the sample area, wherein the wind direction corresponding to each of the target wind shear data matches the preset wind direction corresponding to the terrain information sample; When the proportion of negative wind shear data in the multiple target wind shear data is greater than a proportion threshold, the terrain information sample is marked as a positive sample, and the positive sample is used to indicate the existence of a wind turbine tower sweep risk.

10. A tower sweep risk terrain identification model training device, characterized in that: include: The second acquisition module is used to obtain terrain information samples and historical wind measurement data of the sample area; A first labeling module is used to label the terrain information sample according to the historical wind measurement data to obtain a risk terrain label associated with the terrain information sample, wherein the risk terrain label is used to indicate whether there is a risk of wind turbine tower sweeping; A training module, used for training a tower-sweeping risk terrain recognition model according to the terrain information sample and the risk terrain annotation, to obtain a trained tower-sweeping risk terrain recognition model; The terrain information sample includes terrain information of a sample ROI, and the sample ROI corresponds to a preset wind direction; the historical wind measurement data is measured at a sample position in the sample area.

11. An electronic device, characterized in that: The electronic device comprises: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the tower sweeping risk terrain identification method as described in any one of claims 1 to 5, or implements the tower sweeping risk terrain identification model training method as described in any one of claims 6 to 8.

12. A computer storage medium, characterized in that: The computer storage medium stores computer program instructions, which, when executed by a processor, implement the tower-sweeping risk terrain identification method as described in any one of claims 1 to 5, or implement the tower-sweeping risk terrain identification model training method as described in any one of claims 6 to 8.

Images