Laser radar wind field inversion method, device, electronic device and storage medium

Abstract

The present invention provides a laser radar wind field inversion method, device, electronic device and storage medium, and relates to the field of laser radar detection technology. The method includes: obtaining the original power spectrum data of the laser radar in multiple target scanning directions within a first preset scanning height; applying a coding network to divide the original power spectrum data according to multiple preset measurement heights included in the first preset scanning height, and adding the power spectrum data sequence corresponding to the change in the preset measurement height obtained by the division and the first position code to obtain the power spectrum encoding data of the target scanning direction corresponding to the change in each preset measurement height; inputting all the power spectrum encoding data within the first preset scanning height into a decoder network to obtain wind field inversion data based on the vector wind in the target scanning direction. The present invention improves the height of wind field detection based on laser radar.

Description

Technical Field

[0001] The present invention relates to the field of laser radar detection technology, and more specifically, to a laser radar wind field inversion method, device, electronic equipment and storage medium. Background Art

[0002] The atmospheric wind field in the troposphere is closely related to the site selection of wind power plants, airport wind shear detection, and prediction of pollutant propagation paths. For the detection of the wind field in the troposphere, lidar is a very important measurement tool.

[0003] Since the echo signal of the lidar depends on the Mie scattering of aerosols, the radar signal decays exponentially with the increase of the detection distance. In addition, the concentration of aerosols in the high altitude is low, which makes the signal-to-noise ratio of the radar echo signal in the high altitude low. When the wind field is inverted by traditional wind field inversion algorithms such as the fast Fourier transform spectrum method and the least squares method, the effective wind field inversion height is limited. Summary of the invention

[0004] In view of this, the present invention provides a lidar wind field inversion method, device, electronic device and storage medium.

[0005] One aspect of the present invention provides a laser radar wind field inversion method, including: obtaining original power spectrum data of multiple target scanning directions of the laser radar within a first preset scanning height; applying a coding network to divide the above original power spectrum data according to multiple preset measurement heights included in the above first preset scanning height, and adding the power spectrum data sequence corresponding to the change in the above preset measurement height obtained by the division and a first position code to obtain power spectrum encoded data of the above target scanning direction corresponding to each change in the above preset measurement height; the above first position code is obtained by encoding the position information of the original power spectrum data of the above target scanning direction corresponding to the change in the above preset measurement height in the above power spectrum data sequence; all power spectrum encoded data within the above first preset scanning height are input into a decoder network to obtain wind field inversion data based on the vector wind in the above target scanning direction.

[0006] According to an embodiment of the present invention, the above-mentioned decoder network includes a multilayer perceptron and a Kolmogorov-Arnold network; the above-mentioned multilayer perceptron is used to convert all power spectrum encoding data within the above-mentioned preset scanning height into encoding data in a preset form; the above-mentioned Kolmogorov-Arnold network is used to output wind field inversion data of vector wind based on multiple target scanning directions according to the above-mentioned encoding data.

[0007] According to an embodiment of the present invention, the encoding network and the decoder network are obtained through joint training.

[0008] According to an embodiment of the present invention, the joint training process of the encoding network and the decoder network includes: obtaining original power spectrum sample data based on the original power spectrum data of the target scanning direction collected by the laser radar within a second preset scanning height; obtaining wind field inversion sample data of the vector wind based on the target scanning direction corresponding to the original power spectrum sample data; applying the encoding network to be trained to divide the original power spectrum sample data according to the multiple measurement sample heights included in the second preset scanning height, and adding the power spectrum sample data sequence corresponding to the change in the measurement sample height obtained by the division and the second position code to obtain each of the above-mentioned measurement sample heights. The power spectrum encoded sample data of the above-mentioned target scanning direction corresponding to the change in the height; the above-mentioned second position code is obtained by encoding the position information of the original power spectrum sample data of the above-mentioned target scanning direction corresponding to the change in the above-mentioned measured sample height in the above-mentioned power spectrum sample data sequence; all the power spectrum encoded sample data within the above-mentioned second preset scanning height are input into the decoder network to be trained, and the wind field inversion prediction data based on the vector wind in the above-mentioned target scanning direction is output; the loss result between the above-mentioned wind field inversion prediction data and the above-mentioned wind field inversion sample data is calculated, and when the above-mentioned loss result meets the preset conditions, the trained encoding network and the trained decoder network are obtained.

[0009] According to an embodiment of the present invention, the loss result is the mean absolute error between the wind field inversion prediction data and the wind field inversion sample data.

[0010] According to an embodiment of the present invention, the above-mentioned acquisition of wind field inversion sample data of the vector wind based on the above-mentioned target scanning direction corresponding to the above-mentioned original power spectrum sample data includes: using the centroid method to process the above-mentioned original power spectrum sample data of each of the above-mentioned target scanning directions to obtain the radial wind speed and radial distance of each of the above-mentioned target scanning directions; obtaining the elevation information and azimuth information when the above-mentioned laser radar collects the above-mentioned original power spectrum sample data; and determining the wind field inversion sample data of the vector wind based on multiple target scanning directions according to the above-mentioned elevation information, the above-mentioned azimuth information, the above-mentioned radial wind speed and the above-mentioned radial distance of each target scanning direction.

[0011] Another aspect of the present invention provides a laser radar wind field inversion device, including: a target power spectrum data determination module, used to obtain the original power spectrum data of the laser radar in multiple target scanning directions within a first preset scanning height; a power spectrum encoding module, used to apply a coding network to divide the above-mentioned original power spectrum data according to multiple preset measurement heights included in the above-mentioned first preset scanning height, and add the power spectrum data sequence corresponding to the change in the above-mentioned preset measurement height obtained by the division and the first position code to obtain the power spectrum encoded data of the above-mentioned target scanning direction corresponding to each change in the above-mentioned preset measurement height; the above-mentioned first position code is obtained by encoding the position information of the original power spectrum data of the above-mentioned target scanning direction corresponding to the change in the above-mentioned preset measurement height in the above-mentioned power spectrum data sequence; a wind field data determination module, used to input all the power spectrum encoded data within the above-mentioned first preset scanning height into a decoder network, and obtain wind field inversion data based on the vector wind in the above-mentioned target scanning direction.

[0012] Another aspect of the present invention provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned method.

[0013] Another aspect of the present invention provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the above method when executed.

[0014] Another aspect of the present invention provides a computer program product, the computer program product comprising computer executable instructions, and the instructions are used to implement the above method when executed.

[0015] According to an embodiment of the present invention, by applying a coding network and a decoder network to process the original power spectrum data, the wind field inversion data of the vector wind based on the target scanning direction can be obtained relatively quickly, thereby improving the efficiency of wind field detection. The coding network is first applied to divide the original power spectrum data according to a plurality of preset measurement heights included in the first preset scanning height, and obtain the power spectrum encoding data of the target scanning direction corresponding to the change amount of each preset measurement height. Since the coding network does not directly encode the power spectrum data corresponding to different preset measurement heights, but encodes the power spectrum data corresponding to the change amount of the preset measurement height, the influence of the attenuation of the high-altitude radar signal on the power encoding data can be weakened. Moreover, each preset measurement height corresponds to a wind field detection height. Therefore, the coding network can obtain more accurate power spectrum encoding data of the target scanning direction at the height when the signal-to-noise ratio of the high-altitude radar echo signal is low. Compared with the traditional wind field inversion algorithm that directly performs wind field inversion based on the coherent Doppler laser radar signal, since the traditional wind field inversion algorithm has a weak processing capability for the radar echo signal with a low signal-to-noise ratio, the coding network improves the height of wind field detection. Therefore, the technical problem of limited effective wind field inversion height of the traditional wind field inversion algorithm of coherent Doppler lidar is at least partially overcome, thereby achieving the technical effect of improving the height of wind field detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0017] Figure 1 An exemplary system architecture to which the lidar wind field inversion method of the present invention can be applied is shown.

[0018] Figure 2 A flow chart of a laser radar wind field inversion method according to an embodiment of the present invention is shown.

[0019] Figure 3 A schematic diagram of an encoding network and a decoder network according to an embodiment of the present invention is shown.

[0020] Figure 4 A flow chart of a laser radar wind field inversion method according to another embodiment of the present invention is shown.

[0021] Figure 5 A block diagram of a laser radar wind field inversion device according to an embodiment of the present invention is shown.

[0022] Figure 6 A block diagram of an electronic device suitable for implementing a laser radar wind field inversion method according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0023] Below, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of embodiments of the present invention. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of concepts of the present invention.

[0024] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0025] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0026] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0027] In the embodiments of the present invention, the collection, updating, analysis, processing, use, transmission, provision, disclosure, storage and other aspects of the data involved (for example, including but not limited to user personal information) are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information and network security.

[0028] Compared with in-situ measurements, remote sensing can obtain wind field information over a wide range, increasing the richness of the information obtained. As a remote sensing technology, LiDAR can achieve high temporal and spatial resolution wind field monitoring compared to other remote sensing technologies, even reaching meter-level and second-level resolutions.

[0029] Coherent Doppler wind lidar based on Mie scattering can measure the full range of atmospheric wind fields from the surface to the top of the troposphere, and has the advantages of high precision, high temporal and spatial resolution, and a large detection range. Due to the low concentration of aerosols at high altitudes, the signal-to-noise ratio of the radar echo signal of the coherent Doppler wind lidar at high altitudes is low. Since the traditional wind field inversion algorithm based on coherent Doppler lidar has a weak processing capability for radar echo signals with low signal-to-noise ratio, the effective wind field inversion is highly limited.

[0030] An embodiment of the present invention provides a laser radar wind field inversion method, including: obtaining original power spectrum data of a laser radar in multiple target scanning directions within a first preset scanning height; applying a coding network to divide the original power spectrum data according to multiple preset measurement heights included in the first preset scanning height, and adding a power spectrum data sequence corresponding to the change in the preset measurement height obtained by the division and a first position code to obtain power spectrum encoded data of the target scanning direction corresponding to each change in the preset measurement height; the first position code is obtained by encoding the position information of the original power spectrum data of the target scanning direction corresponding to the change in the preset measurement height in the power spectrum data sequence; all power spectrum encoded data within the first preset scanning height are input into a decoder network to obtain wind field inversion data based on the vector wind in the target scanning direction.

[0031] Figure 1 An exemplary system architecture 100 to which the laser radar wind field inversion method according to an embodiment of the present invention can be applied is shown. It should be noted that: Figure 1 The examples shown are merely examples of system architectures to which the embodiments of the present invention may be applied, to help those skilled in the art understand the technical content of the present invention, but do not mean that the embodiments of the present invention may not be used in other devices, systems, environments or scenarios.

[0032] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a laser radar 101, a first terminal device 1021, a second terminal device 1022, a third terminal device 1023, a network (not shown in the figure) and a server 103. The network is used to provide a medium for a communication link between the first terminal device 1021, the second terminal device 1022, the third terminal device 1023 and the laser radar 101 or the server 103. The network may include various connection types, such as wired and / or wireless communication links, etc.

[0033] The user can use the first terminal device 1021, the second terminal device 1022, and the third terminal device 1023 to interact with the server 103 or the laser radar 101 through the network to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 1021, the second terminal device 1022, and the third terminal device 1023, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (only for example).

[0034] The first terminal device 1021, the second terminal device 1022, and the third terminal device 1023 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0035] The server 103 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 1021, the second terminal device 1022, and the third terminal device 1023. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0036] It should be noted that the laser radar wind field inversion method provided in the embodiment of the present invention can generally be executed by the server 103. Accordingly, the laser radar wind field inversion device provided in the embodiment of the present invention can generally be set in the server 103. The laser radar wind field inversion method provided in the embodiment of the present invention can also be executed by a server or server cluster that is different from the server 103 and can communicate with the first terminal device 1021, the second terminal device 1022, the third terminal device 1023 and / or the server 103. Correspondingly, the laser radar wind field inversion device provided in the embodiment of the present invention can also be set in a server or server cluster that is different from the server 103 and can communicate with the first terminal device 1021, the second terminal device 1022, the third terminal device 1023 and / or the server 103. Alternatively, the laser radar wind field inversion method provided in the embodiment of the present invention can also be performed by the first terminal device 1021, the second terminal device 1022 or the third terminal device 1032, or by other terminal devices different from the first terminal device 1021, the second terminal device 1022 or the third terminal device 1023. Correspondingly, the laser radar wind field inversion device provided in the embodiment of the present invention can also be set in the first terminal device 1021, the second terminal device 1022 or the third terminal device 1023, or in other terminal devices different from the first terminal device 1021, the second terminal device 1022 or the third terminal device 1023.

[0037] For example, the original power spectrum data may be originally stored in any one of the first terminal device 1021, the second terminal device 1022 or the third terminal device 1023 (for example, the first terminal device 1021, but not limited thereto), or stored on an external storage device and may be imported into the first terminal device 1021. Then, the first terminal device 1021 may locally execute the laser radar wind field inversion method provided in the embodiment of the present invention, or send the original power spectrum data to other terminal devices, servers, or server clusters, and the other terminal devices, servers, or server clusters that receive the original power spectrum data may execute the laser radar wind field inversion method provided in the embodiment of the present invention.

[0038] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0039] Figure 2 A flow chart of a laser radar wind field inversion method according to an embodiment of the present invention is shown.

[0040] like Figure 2 As shown, the method includes operations S210 to S230.

[0041] In operation S210, original power spectrum data of the laser radar in multiple target scanning directions within a first preset scanning height is obtained.

[0042] In operation S220, a coding network is applied to divide the original power spectrum data according to a plurality of preset measurement heights included in a first preset scanning height, and a power spectrum data sequence corresponding to a change in the preset measurement height obtained by the division is added to a first position code to obtain power spectrum coded data in a target scanning direction corresponding to each change in the preset measurement height; the first position code is obtained by encoding position information of the original power spectrum data in the target scanning direction corresponding to the change in the preset measurement height in the power spectrum data sequence.

[0043] In operation S230, all power spectrum encoding data within the first preset scanning height are input into a decoder network to obtain wind field inversion data based on the vector wind in the target scanning direction.

[0044] According to an embodiment of the present invention, in operation 210, the laser radar may use a coherent wind laser radar with multiple scanning directions. The coherent wind laser radar performs cyclic scanning in multiple scanning directions to obtain original power spectrum data in multiple scanning directions; the target scanning direction is a scanning direction preset according to actual needs, and multiple target scanning directions are determined according to actual needs. For example, the multiple target scanning directions may include the east-west direction, the north-south direction, and the vertical direction, and may also include the X-axis direction, the Y-axis direction, and the Z-axis direction after constructing a three-dimensional coordinate system, and may also include multiple directions selected after constructing a three-dimensional coordinate system. Determine the original power spectrum data of multiple target scanning directions from the original power spectrum data of multiple scanning directions.

[0045] According to an embodiment of the present invention, since all laser radars have a scanning height, original power spectrum data of a first preset scanning height is selected from the original power spectrum data of the laser radar according to actual needs, wherein the first preset scanning height is less than or equal to the scanning height of the laser radar. Original power spectrum data of multiple target scanning directions are selected from the original power spectrum data of the first preset scanning height.

[0046] According to an embodiment of the present invention, the coherent wind laser radar collects raw power spectrum data at different distances through multiple range gates to detect wind fields at different distances, i.e., at scanning heights. Since the raw power spectrum data detected in each scanning direction is the raw power spectrum data of the scalar wind, the raw power spectrum data of the vector wind in the target scanning direction can be obtained at the same distance, i.e., at the scanning height, by combining the raw power spectrum data of the scalar wind in multiple scanning directions.

[0047] According to an embodiment of the present invention, in operation S220, a coding network is applied to divide the original power spectrum data of multiple target scanning directions within the first preset scanning height obtained in operation S210. When dividing, the data is divided according to the multiple preset measurement heights included in the first preset scanning height. For example, the first preset scanning height is 1000 meters, so the original power spectrum data obtained is data at an altitude of 1000 meters. The difference between each preset measurement height and the adjacent preset measurement height is 10 meters. Then the first preset measurement height obtained corresponds to the original power spectrum data with a scanning height of 10 meters, and the second preset measurement height corresponds to the original power spectrum data with a scanning height of 20 meters, and so on. The divided original power spectrum data includes multiple groups of power spectrum data, each group of power spectrum data corresponds to a change in a preset measurement height, the first group of power spectrum data includes a scanning height range of (0, 10], in meters, the second group of power spectrum data includes a scanning height range of (10, 20], in meters, and so on.

[0048] According to an embodiment of the present invention, each group of power spectrum data is a power spectrum sequence. The power spectrum sequence corresponding to each change in preset measurement height and the first position code corresponding to the change in the preset measurement height are added. For example, each group of power spectrum data includes original power spectrum data of multiple target scanning directions, and the power spectrum data of each target scanning direction is encoded according to its position in the power spectrum sequence to obtain a first position code, and each power spectrum data is added to the corresponding code in the first position code to obtain the added power spectrum data, thereby obtaining multiple groups of added power spectrum data, and each group of added power spectrum data is encoded to obtain multiple groups of encoded power spectrum data, wherein each group of encoded power spectrum data is the power spectrum encoded data of the target scanning direction corresponding to each change in preset measurement height.

[0049] According to an embodiment of the present invention, in operation S230, all groups of encoded power spectrum data are input into a decoder network, and all groups of encoded power spectrum data are decoded to obtain wind field inversion data based on the vector wind in the target scanning direction.

[0050] According to an embodiment of the present invention, by applying a coding network and a decoder network to process the original power spectrum data, the wind field inversion data of the vector wind based on the target scanning direction can be obtained relatively quickly, thereby improving the efficiency of wind field detection. By applying a coding network, the original power spectrum data is divided according to a plurality of preset measurement heights included in the first preset scanning height, and the original power spectrum data corresponding to each target scanning direction corresponding to each preset measurement height is obtained. That is to say, the processing object of the present invention is the preset measurement height change amount, rather than directly processing the preset scanning height, thereby increasing the range of the preset scanning height. Since the scanning height corresponds to the height of wind field detection, the height of wind field detection is increased. Therefore, the technical problem of the limited effective wind field inversion height of the traditional wind field inversion algorithm of the coherent Doppler laser radar is at least partially overcome, thereby achieving the technical effect of increasing the height of wind field detection. Compared with the traditional wind field inversion method, the new method not only improves the accuracy of wind field detection, but also improves the inversion height of the effective wind field, breaking through the limitations of the traditional inversion algorithm.

[0051] According to an embodiment of the present invention, the encoding network can select a Transformer encoder. The traditional ViT (Vision Transformer) architecture is used to process common images in life. For the task of inverting the lidar power spectrum, the embedding layer (embedding layer) of the Transformer encoder architecture adopted by the present invention cuts the input power spectrum into a series of embedding vectors, that is, a power spectrum data sequence, each embedding vector represents a power spectrum signal of a certain scanning height layer, and then the obtained embedding vector is added with a position code, wherein the position code is the first position code, and the encoded vector is obtained through several layers of multi-head self-attention layers and feedforward neural networks, combined with residual connections and layer normalization.

[0052] According to an embodiment of the present invention, the decoder network includes a multilayer perceptron and a Kolmogorov-Arnold network; the multilayer perceptron is used to convert all power spectrum encoding data within a preset scanning height into encoding data in a preset form; the Kolmogorov-Arnold network is used to output wind field inversion data of vector wind based on multiple target scanning directions according to the encoding data.

[0053] According to an embodiment of the present invention, the decoder network includes an MLP (Multilayer Perceptron) layer for converting the shape of the intermediate variable and a KAN (Kolmogorov-Arnold Networks) layer connected to the MLP layer as a decoder. The wind field information is obtained through the decoder network. The KAN network is composed of multiple KAN layers connected in series. Each KAN layer can be represented as a The matrix of and In turn, they represent the dimensions of the input and output of the layer. Each element of the matrix is ​​an activation function By setting multiple KAN layers, the nonlinear mapping capability of the decoder can be improved to improve the inversion accuracy of the wind field.

[0054] Figure 3 A schematic diagram of an encoding network and a decoder network according to an embodiment of the present invention is shown.

[0055] like Figure 3As shown in the figure, the encoding network and the decoder network are obtained through joint training. The encoding network can be selected from the Transformer encoding network, and the decoder network can be selected from the multi-layer perceptron and the KAN network. After the Transformer encoding network, the multi-layer perceptron used to transform the shape of the intermediate variable, and the KAN network are connected in series, the overall model is obtained. When training the model, the input end is used to input the original power spectrum sample data, and the output end is used to output the corresponding wind field inversion sample data. The original power spectrum sample data and the corresponding wind field inversion sample data are used to train the overall model.

[0056] According to an embodiment of the present invention, the joint training process of the encoding network and the decoder network includes: obtaining original power spectrum sample data based on the original power spectrum data of multiple target scanning directions collected by the laser radar within a second preset scanning height; obtaining wind field inversion sample data based on the vector wind in the target scanning direction corresponding to the original power spectrum sample data; applying the encoding network to be trained to divide the original power spectrum sample data according to the multiple measurement sample heights included in the second preset scanning height, and adding the power spectrum sample data sequence corresponding to the change in the measurement sample height obtained by the division and the second position code to obtain the power spectrum encoded sample data of the target scanning direction corresponding to each change in the measurement sample height; the second position code is obtained by encoding the position information of the original power spectrum sample data of the target scanning direction corresponding to the change in the measurement sample height in the power spectrum sample data sequence; inputting all the power spectrum encoded sample data within the second preset scanning height into the decoder network to be trained, and outputting wind field inversion prediction data based on the vector wind in the target scanning direction; calculating the loss result between the wind field inversion prediction data and the wind field inversion sample data, and when the loss result meets the preset conditions, obtaining the trained encoding network and the trained decoder network.

[0057] According to an embodiment of the present invention, the second preset scanning height is set according to actual conditions. There is no direct connection with the first preset scanning height. The laser radars that obtain the original power spectrum sample data are all coherent wind measurement laser radars. The target scanning direction is consistent with the target scanning direction set by the trained encoding network and the trained decoder network in actual application, that is, when the encoding network and the decoder network are trained, the target scanning direction is set, and when the trained encoding network and the trained decoder network are actually applied, it is necessary to collect the original power spectrum data of the set target scanning direction.

[0058] According to an embodiment of the present invention, the original power spectrum sample data is screened data, that is, the data that may affect the training quality of the encoding network and the decoder network, such as obviously erroneous data and noise data, are removed from the obtained original power spectrum sample data to ensure the training accuracy of the encoding network and the decoder network.

[0059] According to an embodiment of the present invention, wind field inversion sample data of vector wind based on the target scanning direction corresponding to the original power spectrum sample data is obtained, including: using the center of gravity method to process the original power spectrum sample data of each target scanning direction to obtain the radial wind speed and radial distance of each target scanning direction; obtaining the elevation information and azimuth information when the laser radar collects the original power spectrum sample data; and determining the wind field inversion sample data of vector wind based on multiple target scanning directions according to the elevation information, azimuth information, radial wind speed and radial distance of each target scanning direction.

[0060] According to an embodiment of the present invention, the radial wind speed in each detection direction is obtained by using a centroid method algorithm for the original power spectrum of the laser radar in each scanning direction of n directions, and then the wind speed information in n (n>=3) directions (the wind speed in the specified direction, i.e., the target scanning direction) is converted into a three-dimensional vector wind and the obtained vector wind data is used as a label, thereby obtaining several groups of power spectrum and vector wind data label pairs.

[0061] According to an embodiment of the present invention, for a coherent wind laser radar with multiple scanning directions, the traditional center of gravity method is first used to measure the wind speed information in each scanning direction. Afterwards, the obtained multiple radial wind speeds are used to calculate the three-dimensional vector wind or horizontal wind field, horizontal wind direction and vertical wind field data in the three-dimensional coordinate system. The laser radar power spectrum data in each scanning direction are combined to obtain training data, and the vector wind field data is the label data. Generally speaking, the number of scanning directions n is greater than or equal to 3, otherwise these radial wind speeds will not be converted into vector winds.

[0062] According to an embodiment of the present invention, for a coherent wind laser radar with multiple scanning directions, the traditional centroid method is first used to measure the wind speed information in each scanning direction. Afterwards, the obtained multiple radial wind speeds are used to calculate the three-dimensional vector wind or horizontal wind field, horizontal wind direction and vertical wind field data in the three-dimensional coordinate system. The laser radar power spectrum data in each scanning direction are merged to obtain training data, and the vector wind field data is label data. Generally speaking, the number of scanning directions n is greater than or equal to 3, otherwise these radial wind speeds will not be converted into vector wind. Assuming that the power spectrum of each scanning direction is a matrix of size 220×128, and there are 3 scanning directions, then training data of size 3×220×128 will be obtained. The label data is 3 vectors of length 220, that is, a matrix of size 3×220, where 3 represents the three axes of X, Y and Z.

[0063] According to an embodiment of the present invention, assuming that the length of the embedding vector is 128, the embedding layer of the present invention converts the training data of size 3×220×128 into 220 embedding vectors of length 128, that is, an embedding matrix of size 220×128. Then, the obtained embedding vector is positionally encoded, and passed through several layers of multi-head self-attention layers and feedforward neural networks, combined with residual connections and layer normalization, to obtain the encoded vector.

[0064] According to an embodiment of the present invention, the MLP layer converts the embedding matrix of size 220×128 into a new matrix of size 660×128. Assuming that the input and output layer dimensions of the KAN network are 128 and 1 respectively, a vector of length 660 can be obtained, which corresponds to the three-axis wind speed components of the wind field of the 220 range gates in turn.

[0065] According to an embodiment of the present invention, the input of the encoding network is the original power spectrum data, the label is the three-dimensional wind field data, and the loss result is the mean absolute error between the wind field inversion prediction data and the wind field inversion sample data. Through multiple rounds of training, a wind field inversion model based on the new Transformer and KAN network can be obtained, that is, a trained encoding network and a trained decoder network. In order to improve the accuracy of the inverted wind field, the accuracy of the inverted wind field results, and taking into account the continuity characteristics of the wind field, a median filter is also used to process the high-frequency noise of the wind field.

[0066] Figure 4 A flow chart of a laser radar wind field inversion method according to another embodiment of the present invention is shown.

[0067] like Figure 4 As shown, the laser radar wind field inversion method provided by the present invention includes operations S401~S405.

[0068] In operation S401, an overall model for lidar wind field inversion is constructed; the overall model includes a Transformer encoder, a multi-layer perceptron and a KAN network. The Transformer encoder is connected to the KAN network through the multi-layer perceptron.

[0069] In operation S402, a training data set of the overall model is obtained; the training data set includes original power spectrum sample data and corresponding wind field inversion sample data. The original power spectrum sample data is original power spectrum data of multiple target scanning directions collected by the laser radar within the second preset scanning height, and the wind field inversion sample data is wind field inversion sample data of vector wind based on the target scanning direction corresponding to the original power spectrum sample data.

[0070] In operation S403, the training data set is used to train the overall model to obtain the trained overall model. Exemplarily, the original power spectrum sample data is input into the overall model to obtain wind field inversion prediction data, and the mean absolute error between the wind field inversion prediction data and the wind field inversion sample data is calculated. When the mean absolute error is less than a preset error value, the trained overall model is obtained.

[0071] In operation S404, original power spectrum data of the laser radar in multiple target scanning directions within a first preset scanning height is obtained.

[0072] In operation S405, the original power spectrum data is input into the trained overall model to obtain wind field inversion data based on the vector wind in the target scanning direction.

[0073] According to an embodiment of the present invention, the sounding balloon data is used as a benchmark for result analysis. As shown in Table 1, the root mean square error (RMSE), mean absolute error (MAE) and Pearson correlation coefficient (PCC) are used to describe the overall performance of the two methods in the low-altitude range and the entire laser radar detection range. The values ​​in the table represent the average values ​​of the experiment. Since the index performance of the centroid method in the entire laser radar detection range is poor, it is not shown in the table. The wind field data obtained by the traditional centroid method will fluctuate greatly at high altitudes, and the calculated value and the observed data of the sounding balloon have a large deviation. However, the model using the Transformer encoding network and the KAN network of the present invention still has a good match with the sounding balloon data at the inversion of the high-altitude wind field, and the error is within the meteorological standard range. This result shows that the method improves the effective detection height of the laser radar. In addition, in the low-altitude range, the accuracy of the wind field inverted by the model based on the new Transformer and KAN network is also higher than that of the wind field inverted by the centroid method, and the error with the sounding balloon data is smaller, which also shows that the method effectively improves the accuracy of wind field detection in the low-altitude range.

[0074] Table 1 Statistical table of results analysis based on sounding balloon data

[0075]

[0076] Figure 5 A block diagram of a laser radar wind field inversion device according to an embodiment of the present invention is shown.

[0077] like Figure 5 As shown, the laser radar wind field inversion device 500 includes a target power spectrum data determination module 510, a power spectrum encoding module 520 and a wind field data determination module 530.

[0078] The target power spectrum data determination module 510 is used to obtain the original power spectrum data of the laser radar in multiple target scanning directions within a first preset scanning height.

[0079] The power spectrum encoding module 520 is used to apply a coding network to divide the original power spectrum data according to a plurality of preset measurement heights included in a first preset scanning height, and add a power spectrum data sequence corresponding to the change in the preset measurement height obtained by the division and a first position code to obtain power spectrum encoding data in the target scanning direction corresponding to each change in the preset measurement height; the first position code is obtained by encoding the position information of the original power spectrum data in the target scanning direction corresponding to the change in the preset measurement height in the power spectrum data sequence.

[0080] The wind field data determination module 530 is used to input all power spectrum encoding data within the first preset scanning height into the decoder network to obtain wind field inversion data based on the vector wind in the target scanning direction.

[0081] According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits, or at least part of the functions of any one of them can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, submodules, units, and subunits can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging the circuit, or by any one of the three implementation methods of software, hardware, and firmware, or by a proper combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, submodules, units, and subunits can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be executed.

[0082] For example, any multiple of the target power spectrum data determination module 510, the power spectrum encoding module 520, and the wind field data determination module 530 can be combined in one module / unit / subunit for implementation, or any one of the modules / units / subunits can be split into multiple modules / units / subunits. Alternatively, at least part of the functions of one or more of these modules / units / subunits can be combined with at least part of the functions of other modules / units / subunits and implemented in one module / unit / subunit. According to an embodiment of the present invention, at least one of the target power spectrum data determination module 510, the power spectrum encoding module 520, and the wind field data determination module 530 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, at least one of the target power spectrum data determination module 510, the power spectrum encoding module 520 and the wind field data determination module 530 may be at least partially implemented as a computer program module, and when the computer program module is executed, a corresponding function may be performed.

[0083] It should be noted that the laser radar wind field inversion device part in the embodiment of the present invention corresponds to the laser radar wind field inversion method part in the embodiment of the present invention. The description of the laser radar wind field inversion device part specifically refers to the laser radar wind field inversion method part, which will not be repeated here.

[0084] Figure 6 A block diagram of an electronic device suitable for implementing a laser radar wind field inversion method according to an embodiment of the present invention is shown. Figure 6 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0085] like Figure 6As shown, the electronic device 600 according to an embodiment of the present invention includes a processor 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 to a random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include an onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0086] In RAM 603, various programs and data required for the operation of electronic device 600 are stored. Processor 601, ROM 602 and RAM 603 are connected to each other via bus 604. Processor 601 performs various operations of the method flow according to an embodiment of the present invention by executing the programs in ROM 602 and / or RAM 603. It should be noted that the program can also be stored in one or more memories other than ROM 602 and RAM 603. Processor 601 can also perform various operations of the method flow according to an embodiment of the present invention by executing the programs stored in the one or more memories.

[0087] According to an embodiment of the present invention, the electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to the bus 604. The electronic device 600 may further include one or more of the following components connected to the input / output (I / O) interface 605: an input portion 606 including a keyboard, a mouse, etc.; an output portion 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 608 including a hard disk, etc.; and a communication portion 609 including a network interface card such as a LAN card, a modem, etc. The communication portion 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage portion 608 as needed.

[0088] According to an embodiment of the present invention, the method flow according to an embodiment of the present invention can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, the above-mentioned functions defined in the system of the embodiment of the present invention are executed. According to an embodiment of the present invention, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0089] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present invention is implemented.

[0090] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device.

[0091] For example, according to an embodiment of the present invention, the computer-readable storage medium may include the ROM 602 and / or the RAM 603 described above and / or one or more memories other than the ROM 602 and the RAM 603 .

[0092] An embodiment of the present disclosure also includes a computer program product, which includes a computer program, which contains program code for executing the method provided by the embodiment of the present invention. When the computer program product runs on an electronic device, the program code is used to enable the electronic device to implement the lidar wind field inversion method provided by the embodiment of the present invention.

[0093] When the computer program is executed by the processor 601, the above functions defined in the system / device of the embodiment of the present invention are executed. According to the embodiment of the present invention, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0094] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 609, and / or installed from a removable medium 611. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0095] According to an embodiment of the present invention, the program code for executing the computer program provided by the embodiment of the present invention can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0096] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments of the present invention can be combined and / or combined in various ways, even if such a combination or combination is not explicitly recorded in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features recorded in the various embodiments of the present invention can be combined and / or combined in various ways. All these combinations and / or combinations fall within the scope of the present invention.

[0097] The embodiments of the present invention are described above. However, these embodiments are only for the purpose of illustration, and are not intended to limit the scope of the present invention. Although each embodiment is described above, it does not mean that the measures in each embodiment cannot be used in combination advantageously. Without departing from the scope of the present invention, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present invention.

Claims

1. A laser radar wind field inversion method, characterized in that: The method comprises: Acquire original power spectrum data of the laser radar in multiple target scanning directions within a first preset scanning height; A coding network is applied to divide the original power spectrum data according to a plurality of preset measurement heights included in the first preset scanning height, and a power spectrum data sequence corresponding to the change in the preset measurement height obtained by the division is added to a first position code to obtain power spectrum coded data of the target scanning direction corresponding to each change in the preset measurement height; the first position code is obtained by encoding position information of the original power spectrum data of the target scanning direction corresponding to the change in the preset measurement height in the power spectrum data sequence; All power spectrum encoding data within the first preset scanning height are input into a decoder network to obtain wind field inversion data based on the vector wind in the target scanning direction.

2. The method according to claim 1, characterized in that The decoder network includes a multi-layer perceptron and a Kolmogorov-Arnold network; The multi-layer perceptron is used to convert all power spectrum coded data within the preset scanning height into coded data in a preset form; The Kolmogorov-Arnold network is used to output wind field inversion data based on vector winds in multiple target scanning directions according to the encoded data.

3. The method according to claim 1, characterized in that The encoding network and the decoder network are obtained by joint training.

4. The method according to claim 3, characterized in that The joint training process of the encoding network and the decoder network includes: Based on the original power spectrum data of the plurality of target scanning directions collected by the laser radar within the second preset scanning height, original power spectrum sample data is obtained; Acquire wind field inversion sample data based on the vector wind in the target scanning direction corresponding to the original power spectrum sample data; The encoding network to be trained is applied to divide the original power spectrum sample data according to a plurality of measurement sample heights included in the second preset scanning height, and the power spectrum sample data sequence corresponding to the change in the measurement sample height obtained by the division is added with the second position code to obtain the power spectrum encoded sample data of the target scanning direction corresponding to each change in the measurement sample height; the second position code is obtained by encoding the position information of the original power spectrum sample data of the target scanning direction corresponding to the change in the measurement sample height in the power spectrum sample data sequence; Inputting all power spectrum encoded sample data within the second preset scanning height into a decoder network to be trained, and outputting wind field inversion prediction data based on the vector wind in the target scanning direction; The loss result between the wind field inversion prediction data and the wind field inversion sample data is calculated, and when the loss result meets a preset condition, a trained encoding network and a trained decoder network are obtained.

5. The method according to claim 4, characterized in that The loss result is the mean absolute error between the wind field inversion prediction data and the wind field inversion sample data.

6. The method according to claim 4, characterized in that The obtaining of wind field inversion sample data corresponding to the original power spectrum sample data and based on the vector wind in the target scanning direction comprises: Processing the original power spectrum sample data in each target scanning direction by using the centroid method to obtain the radial wind speed and radial distance in each target scanning direction; Obtaining elevation information and azimuth information when the laser radar collects the original power spectrum sample data; Wind field inversion sample data based on vector winds in multiple target scanning directions are determined according to the elevation information, the azimuth information, the radial wind speed in each target scanning direction, and the radial distance.

7. A laser radar wind field inversion device, characterized in that: The device comprises: A target power spectrum data determination module is used to obtain original power spectrum data of multiple target scanning directions of the laser radar within a first preset scanning height; A power spectrum encoding module, for applying an encoding network to divide the original power spectrum data according to a plurality of preset measurement heights included in the first preset scanning height, and adding a power spectrum data sequence corresponding to the change in the preset measurement height obtained by the division and a first position code to obtain power spectrum encoding data of the target scanning direction corresponding to each change in the preset measurement height; the first position code is obtained by encoding position information of the original power spectrum data of the target scanning direction corresponding to the change in the preset measurement height in the power spectrum data sequence; The wind field data determination module is used to input all power spectrum encoding data within the first preset scanning height into a decoder network to obtain wind field inversion data based on the vector wind in the target scanning direction.

8. An electronic device, characterized in that: include: one or more processors; a memory for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: Executable instructions are stored thereon, and when the instructions are executed by a processor, the processor implements the method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images