Navigation scene recognition method, system, device and medium based on satellite distribution

Abstract

The present invention provides a navigation scene recognition method, system, device and medium based on satellite distribution, the method comprising: presetting multiple navigation scenes, and respectively collecting multi-constellation GNSS measured data corresponding to each navigation scene; constructing a navigation scene recognition data set according to the multi-constellation GNSS measured data of all navigation scenes, and dividing the navigation scene recognition data set into a training set and a verification set according to a preset ratio; training a CNN network according to the training set and the verification set to obtain a navigation scene prediction model; obtaining the multi-constellation GNSS measured data to be identified, and obtaining a scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model. The present invention directly uses measured GNSS data for scene recognition, which not only consumes less hardware resources, but also can mine the deep features of GNSS data, effectively improve recognition accuracy and generalization ability, and provide reliable guarantee for navigation tasks in complex urban environments.

Description

Technical Field

[0001] The present invention relates to the field of navigation and positioning technology, and in particular to a navigation scene recognition method, system, computer equipment and storage medium based on satellite distribution. Background Art

[0002] In navigation tasks in complex urban environments, the interference characteristics of different environments to GNSS signals are significantly different. Navigation scene recognition can help analyze various interference characteristics, and can effectively assist the algorithm development and performance testing of GNSS terminals, thereby providing reliable guarantees for the accuracy of navigation recognition.

[0003] The existing navigation scene recognition methods mainly include image recognition method and GNSS signal discrimination method. However, in the application of image recognition method that extracts pixel features from surrounding environment photos to identify scene types, the image recognition method based on artificially defined features to extract low-level features cannot fully describe the scene, resulting in low scene recognition accuracy. Although the image recognition method based on CNN to extract high-level semantic features has improved accuracy, it requires real-time acquisition of environmental image information and consumes a lot of hardware resources. At the same time, the GNSS signal discrimination method that selects a variety of suitable GNSS feature parameters for machine learning recognition not only has a small number of identifiable categories, but also faces a complex and changeable navigation environment with blurred boundaries between different scenes. The existing terminal epoch’s scalar observation information spatial feature representation ability is weak, scene classification is difficult, and generalization ability is insufficient, resulting in low recognition accuracy.

[0004] Therefore, there is an urgent need to provide a navigation scene recognition method that can utilize GNSS measured data for modeling and recognition while taking into account high scene recognition accuracy, strong generalization ability and less hardware resource consumption. Summary of the invention

[0005] The purpose of the present invention is to provide a navigation scene recognition method based on satellite distribution. By combining satellite spatial distribution with observation information, a CNN network model for navigation scene prediction and recognition is constructed using GNSS measured data to mine the deep features of multi-constellation GNSS data, effectively improving scene recognition accuracy and generalization ability while reducing hardware resource usage.

[0006] In order to achieve the above objectives, it is necessary to provide a navigation scene recognition method, system, computer device and storage medium based on satellite distribution in response to the above technical problems.

[0007] In a first aspect, an embodiment of the present invention provides a navigation scene recognition method based on satellite distribution, the method comprising the following steps:

[0008] Preset multiple navigation scenarios, and collect multi-constellation GNSS measured data corresponding to each navigation scenario respectively; the multi-constellation GNSS measured data includes satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange;

[0009] Constructing a navigation scene recognition data set based on the multi-constellation GNSS measured data of all navigation scenes, and dividing the navigation scene recognition data set into a training set and a validation set according to a preset ratio; the navigation scene recognition data set includes the multi-constellation GNSS measured data of each navigation scene and the corresponding scene labels;

[0010] According to the training set and the validation set, a CNN network is trained to obtain a navigation scene prediction model;

[0011] Acquire the multi-constellation GNSS measured data to be identified, and obtain a scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model.

[0012] Furthermore, the navigation scenarios include urban canyons, urban-rural fringe areas, suburbs, tree-lined areas, and under elevated roads.

[0013] Furthermore, the step of constructing a navigation scene recognition data set based on multi-constellation GNSS measured data of all navigation scenes includes:

[0014] Converting each multi-constellation GNSS measured data into corresponding scene image data respectively; the scene image data includes a satellite distribution image, a carrier-to-noise ratio image, a Doppler frequency shift image and a pseudorange image corresponding to the satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange respectively;

[0015] The scene image data is added with corresponding scene labels to obtain the navigation scene recognition data set.

[0016] Furthermore, the step of converting each multi-constellation GNSS measured data into corresponding scene image data includes:

[0017] Projecting the satellite positions of the multi-constellation GNSS measured data onto a circular image plane and performing pixel adjustment to obtain the satellite distribution image;

[0018] The carrier-to-noise ratio, Doppler frequency shift and pseudorange of the multi-constellation GNSS measured data are normalized and von Lonoy diagram interpolated in turn to obtain corresponding carrier-to-noise ratio images, Doppler frequency shift images and pseudorange images, respectively.

[0019] Furthermore, the step of projecting the satellite positions of the multi-constellation GNSS measured data onto a circular image plane and performing pixel adjustment to obtain the satellite distribution image comprises:

[0020] According to the following projection algorithm, the satellite position is projected onto the circular image plane to obtain an initial circular image:

[0021] x=R×cosφ×cosθ+R

[0022] y=R×cosφ×sinθ+R

[0023] Where x and y represent the X-axis and Y-axis coordinate values ​​of the circular plane rectangular coordinate system respectively; φ and θ represent the satellite altitude angle and azimuth respectively; R represents the circular image radius;

[0024] The circular image of the satellite position is pixel-adjusted according to a preset rule to obtain an updated circular image; the preset rule includes setting the pixel value of the position where the satellite is distributed to 255 and setting the pixel value of the position where the satellite is not distributed to 0;

[0025] The inner area of ​​the circumscribed square of the updated circular image is used as the generation area of ​​the satellite distribution image, and the area outside the updated circular image in the generation area is filled with a gray value of 0 to obtain the satellite distribution image.

[0026] Furthermore, the CNN network includes an input layer, a convolutional layer, a pooling layer, a plurality of fully connected layers and an output layer connected in sequence; the number of convolutional layers and pooling layers are set according to the navigation scenario;

[0027] The step of training the CNN network according to the training set and the validation set to obtain the navigation scene prediction model comprises:

[0028] The CNN network is trained using cross entropy as the loss function and stochastic gradient descent as the optimizer until a preset iteration termination condition is reached to obtain the navigation scene prediction model; the preset iteration termination condition is that both the training accuracy and the verification accuracy reach the corresponding preset accuracy standards, and the difference between the training accuracy and the verification accuracy is within a preset range.

[0029] Furthermore, the step of obtaining a scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model includes:

[0030] Converting the multi-constellation GNSS measured data to be identified into corresponding scene image data to be identified; the scene image data to be identified includes a satellite distribution image to be identified, a carrier-to-noise ratio image to be identified, a Doppler frequency shift image to be identified and a pseudorange image to be identified, which correspond to the satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange respectively;

[0031] The scene image data to be identified is input into the navigation scene prediction model for scene recognition to obtain the scene recognition result.

[0032] In a second aspect, an embodiment of the present invention provides a navigation scene recognition system based on satellite distribution, the system comprising:

[0033] Preset multiple navigation scenarios, and collect multi-constellation GNSS measured data corresponding to each navigation scenario respectively; the multi-constellation GNSS measured data includes satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange;

[0034] Constructing a navigation scene recognition data set based on the multi-constellation GNSS measured data of all navigation scenes, and dividing the navigation scene recognition data set into a training set and a validation set according to a preset ratio; the navigation scene recognition data set includes the multi-constellation GNSS measured data of each navigation scene and the corresponding scene labels;

[0035] According to the training set and the validation set, a CNN network is trained to obtain a navigation scene prediction model;

[0036] Acquire the multi-constellation GNSS measured data to be identified, and obtain a scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model.

[0037] In a third aspect, an embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0038] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0039] The present application provides a navigation scene recognition method, system, computer device and storage medium based on satellite distribution. Through the method, it is realized to collect multi-constellation GNSS measured data corresponding to a plurality of preset navigation scenes, and after corresponding image conversion processing, a navigation scene recognition data set is constructed, and after dividing it into a training set and a validation set according to a preset ratio, a CNN network is trained to obtain a navigation scene prediction model, which is used to predict navigation scenes according to the multi-constellation GNSS measured data to be identified, and obtain a technical solution for scene recognition results. Compared with the prior art, the navigation scene recognition method based on satellite distribution does not require auxiliary equipment of other sensors (cameras, etc.) to collect and store data. By combining satellite spatial distribution with observation information, the GNSS measured data is directly used to construct a CNN network model for navigation scene prediction and recognition for intelligent navigation scene recognition. It not only consumes less hardware resources, but also can automatically mine the deep features of GNSS data, effectively improve recognition accuracy and generalization ability, and provide reliable guarantee for navigation tasks in complex urban environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a schematic diagram of an application scenario of a navigation scenario recognition method based on satellite distribution in an embodiment of the present invention;

[0041] Figure 2 is a flow chart of a navigation scene recognition method based on satellite distribution in an embodiment of the present invention;

[0042] Figure 3 is a schematic diagram of a satellite distribution image, a carrier-to-noise ratio image, a Doppler frequency shift image, and a pseudorange image in an embodiment of the present invention;

[0043] Figure 4 is a schematic diagram of the structure of a CNN network in an embodiment of the present invention;

[0044] Figure 5 is a schematic structural diagram of a navigation scene recognition system based on satellite distribution in an embodiment of the present invention;

[0045] Figure 6 It is a diagram of the internal structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical scheme and beneficial effects of the present application clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Obviously, the embodiments described below are part of the embodiments of the present invention and are only used to illustrate the present invention, but are not used to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] The navigation scene recognition method based on satellite distribution provided by the present invention can be applied to Figure 1 The terminal or server shown in the figure. The terminal can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices. The server can be implemented as an independent server or a server cluster composed of multiple servers. The server or terminal can be implemented as follows Figure 2 The navigation scene recognition architecture shown executes the artificial intelligence navigation scene recognition method of the present invention that combines satellite spatial distribution and observation information, performs accurate and effective navigation scene recognition in real time, and uses the obtained navigation scene recognition results for navigation tasks in complex urban environments; the following embodiments will explain the navigation scene recognition method based on satellite distribution of the present invention in detail.

[0048] In one embodiment, Figure 2 As shown, a navigation scene recognition method based on satellite distribution is provided, comprising the following steps:

[0049] S11, presetting multiple navigation scenarios, and collecting multi-constellation GNSS measured data corresponding to each navigation scenario respectively; the multi-constellation GNSS measured data includes satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange; wherein, the navigation scenario can be set according to actual application requirements. In this embodiment, five typical scenarios are selected as identification objects for complex urban environment navigation, including urban canyons, urban-rural fringe areas, suburbs, tree-lined areas and under elevated roads;

[0050] After determining the type of navigation scenario, a mobile terminal capable of acquiring GNSS data can be used to collect multi-constellation GNSS measured data corresponding to each type of navigation scenario for subsequent modeling and analysis. In order to ensure the training effect of the subsequent scene recognition model, as much measured data as possible is collected for each type of navigation scenario. The specific sampling frequency and number of sampling times are not restricted. For example, sampling can be performed at a sampling interval of 10 minutes in the above five scenarios. A large amount of multi-constellation GNSS measured data with scene labels can be obtained through multi-day collection, so as to construct the data set required for model training according to the following method.

[0051] S12. Construct a navigation scene recognition data set based on the multi-constellation GNSS measured data of all navigation scenes, and divide the navigation scene recognition data set into a training set and a validation set according to a preset ratio; the navigation scene recognition data set includes the multi-constellation GNSS measured data of each navigation scene and the corresponding scene labels; wherein, the navigation scene recognition data set does not directly use the collected multi-constellation GNSS measured data as data set information that can be used for input into the recognition model, but needs to perform image conversion processing on the satellite position, carrier-to-noise ratio, Doppler frequency shift, pseudorange and other data in the multi-constellation GNSS measured data to obtain image data that can be used for CNN network recognition; the corresponding scene label can be understood as marking the corresponding multi-constellation GNSS measured data as any one of the corresponding labels of urban canyon, urban-rural fringe area, suburb, tree-lined area and under the elevated road according to the actual data collection scene, for subsequent CNN network scene recognition training; the preset ratio of dividing the navigation scene recognition data set into the training set and the validation set can be determined according to the actual application requirements. In this embodiment, the division ratio is preferably set to training set: validation set = 7:3;

[0052] Specifically, the step of constructing a navigation scene recognition data set based on multi-constellation GNSS measured data of all navigation scenes includes:

[0053] Each multi-constellation GNSS measured data is converted into corresponding scene image data respectively; the scene image data includes a satellite distribution image, a carrier-to-noise ratio image, a Doppler frequency shift image and a pseudorange image corresponding to the satellite position, the carrier-to-noise ratio, the Doppler frequency shift and the pseudorange respectively; wherein the scene image data can be understood as 4-channel image data consisting of a satellite distribution image, a carrier-to-noise ratio image, a Doppler frequency shift image and a pseudorange image; specifically, the steps of converting each multi-constellation GNSS measured data into corresponding scene image data respectively include:

[0054] Projecting the satellite positions of the multi-constellation GNSS measured data onto a circular image plane and performing pixel adjustment to obtain the satellite distribution image specifically includes:

[0055] According to the following projection algorithm, the satellite position is projected onto the circular image plane to obtain an initial circular image:

[0056] x=R×cosφ×cosθ+R

[0057] y=R×cosφ×sinθ+R

[0058] Among them, x and y represent the X-axis and Y-axis coordinate values ​​of the circular plane rectangular coordinate system with the center of the circle as the origin; φ and θ represent the satellite altitude angle and azimuth respectively; R represents the radius of the circular image, and the specific value can be set according to actual needs;

[0059] The circular image of the satellite position is pixel-adjusted according to a preset rule to obtain an updated circular image; the preset rule includes setting the pixel value of the position where the satellite is distributed to 255 and setting the pixel value of the position where the satellite is not distributed to 0;

[0060] The inner area of ​​the circumscribed square of the updated circular image is used as the generation area of ​​the satellite distribution image, and the area outside the updated circular image in the generation area is filled with a gray value of 0 to obtain Figure 3 Satellite distribution image shown in a; Figure 3 The square image shown in a is a satellite distribution image obtained by projecting the satellite positions of the multi-constellation GNSS measured data in a certain navigation scenario according to R=50 (pixels); considering that the visible number and spatial distribution of satellites in different navigation scenarios are quite different, and cannot be used as the basis for scene recognition alone, the satellite distribution image of each group of multi-constellation GNSS measured data is used as a feature channel and combined with other data for scene recognition to improve the accuracy of scene recognition;

[0061] The carrier-to-noise ratio, Doppler frequency shift and pseudorange of the multi-constellation GNSS measured data are normalized and von Lonoy diagram interpolated in turn to obtain the following: Figure 3b, 3c and 3d show the carrier-to-noise ratio image, Doppler frequency shift image and pseudorange image; it should be noted that, Figure 3 The carrier-to-noise ratio image, Doppler frequency shift image and pseudo-range image shown in 3b, 3c and 3d are only exemplary descriptions in a certain navigation scenario and do not specifically limit the protection scope of the present invention;

[0062] Adding corresponding scene labels to the scene image data to obtain the navigation scene recognition data set. The navigation scene recognition data set can be understood as an image data set consisting of multiple groups of 4-channel image data of different navigation scenes;

[0063] S13, according to the training set and the validation set, training the CNN network to obtain a navigation scene prediction model; wherein the CNN network is as follows Figure 4 As shown, it includes an input layer, a convolution layer, a pooling layer, a plurality of fully connected layers and an output layer connected in sequence; the number of the convolution layer and the pooling layer are set according to the navigation scenario;

[0064] The step of training the CNN network according to the training set and the validation set to obtain the navigation scene prediction model comprises:

[0065] The CNN network is trained using cross entropy as the loss function and stochastic gradient descent as the optimizer until a preset iteration termination condition is reached to obtain the navigation scene prediction model; the preset iteration termination condition is that both the training accuracy and the verification accuracy reach the corresponding preset accuracy standard, and the difference between the training accuracy and the verification accuracy is within a preset range; wherein, the method for obtaining the training accuracy and the verification accuracy is as follows:

[0066] Training accuracy = number of samples correctly identified by CNN in the training set / total number of samples in the training set

[0067] Verification accuracy = number of samples correctly identified by CNN in the verification set / total number of samples in the verification set

[0068] The parameter settings in the CNN network training process in this embodiment are shown in Table 1. The corresponding preset iteration termination condition can be understood as the scene recognition effect of the trained CNN network meets the application requirements, that is, the higher the accuracy obtained on the training set and the verification set, the better, and the closer the two are, the better; correspondingly, the preset accuracy standard and the preset range can be determined according to the actual application requirements, and no specific restrictions are made here;

[0069] Table 1 CNN parameter settings

[0070] serial number parameter value 1 Loss function Cross Entropy 2 Optimizer Stochastic Gradient Descent 3 Dropout 0.26 4 Number of conv layers 1 5 Number of dense layers 3 6 Number of Epochs 250 7 Input dimension size 100×100×4 8 Output dimension size 5 9 Size of Batch 64 10 Validation split size 7:3 11 Learning rate 1e-7

[0071] S14, obtaining the multi-constellation GNSS measured data to be identified, and obtaining a scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model; wherein the multi-constellation GNSS measured data to be identified is the same as the multi-constellation GNSS measured data of each preset navigation scene collected when constructing the navigation scene recognition data set, and when performing actual prediction, it is also necessary to use the same data processing method when constructing the data set to obtain 4-channel image data that can be directly input into the navigation scene prediction model for identification; specifically, the step of obtaining the scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model comprises:

[0072] Converting the multi-constellation GNSS measured data to be identified into corresponding scene image data to be identified; the scene image data to be identified includes a satellite distribution image to be identified, a carrier-to-noise ratio image to be identified, a Doppler frequency shift image to be identified and a pseudorange image to be identified, which correspond to the satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange respectively;

[0073] The scene image data to be identified is input into the navigation scene prediction model for scene recognition to obtain the scene recognition result.

[0074] The embodiment of the present application generates a satellite distribution image by directly projecting the satellite positions of multi-constellation GNSS measured data, and normalizes the carrier-to-noise ratio, Doppler frequency shift and pseudorange of the multi-constellation GNSS measured data. Then, the von Lonoy diagram interpolation method is used to obtain the carrier-to-noise ratio image, the Doppler frequency shift image and the pseudorange image, which together constitute the corresponding 4-channel image data. The navigation scene prediction model is obtained in combination with CNN network training, so as to realize a technical solution for real-time and accurate prediction of navigation scenes according to the multi-constellation GNSS measured data to be identified. This solution realizes the collection and storage of data without the need for auxiliary equipment of other sensors (cameras, etc.), and intelligent navigation scene recognition can be performed only by combining satellite spatial distribution with observation information. This solution not only consumes less hardware resources, but also can automatically mine the deep features of GNSS data, effectively improve the recognition accuracy and generalization ability, and provide reliable guarantee for navigation tasks in complex urban environments.

[0075] It should be noted that although the steps in the above flowchart are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders.

[0076] In one embodiment, Figure 5 As shown, a navigation scene recognition system based on satellite distribution is provided, and the system includes:

[0077] The data acquisition module 1 is used to preset multiple navigation scenarios and respectively collect multi-constellation GNSS measured data corresponding to each navigation scenario; the multi-constellation GNSS measured data includes satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange;

[0078] The data set construction module 2 is used to construct a navigation scene recognition data set based on the multi-constellation GNSS measured data of all navigation scenes, and divide the navigation scene recognition data set into a training set and a validation set according to a preset ratio; the navigation scene recognition data set includes the multi-constellation GNSS measured data of each navigation scene and the corresponding scene label;

[0079] A model training module 3 is used to train a CNN network according to the training set and the validation set to obtain a navigation scene prediction model;

[0080] The scene recognition module 4 is used to obtain the multi-constellation GNSS measured data to be identified, and obtain the scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model.

[0081] For the specific definition of a navigation scene recognition system based on satellite distribution, please refer to the definition of a navigation scene recognition method based on satellite distribution above, which will not be repeated here. Each module in the above-mentioned navigation scene recognition system based on satellite distribution can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0082] Figure 6 FIG. 1 shows an internal structure diagram of a computer device in an embodiment, and the computer device may specifically be a terminal or a server. Figure 6 As shown, the computer device includes a processor, a memory, a network interface, a display and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a navigation scene recognition method based on satellite distribution is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0083] It can be understood by those skilled in the art that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the present application scheme, and does not constitute a limitation on the computer device to which the present application scheme is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have the same component arrangement.

[0084] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the computer program.

[0085] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0086] In summary, the embodiments of the present invention provide a navigation scene recognition method based on satellite distribution, a system, a computer device and a storage medium. The navigation scene recognition method based on satellite distribution realizes the collection of multi-constellation GNSS measured data corresponding to a plurality of preset navigation scenes, and the satellite distribution image generated by projecting the satellite positions of the multi-constellation GNSS measured data, and the carrier-to-noise ratio image, the Doppler frequency shift image and the pseudorange image obtained by normalizing the carrier-to-noise ratio, the Doppler frequency shift and the pseudorange of the multi-constellation GNSS measured data in turn and performing von Lonoy diagram interpolation processing, so as to construct a navigation scene recognition data set, and train a CNN network based on the data to obtain a navigation scene prediction model, which is used to perform real-time navigation scene prediction according to the multi-constellation GNSS measured data to be identified, and obtain a technical solution for scene recognition results. The method realizes the collection and storage of data without the need for auxiliary equipment of other sensors (cameras, etc.), and intelligent navigation scene recognition can be performed only by combining satellite spatial distribution with observation information. It not only consumes less hardware resources, but also can automatically mine the deep features of GNSS data, effectively improve the recognition accuracy and generalization ability, and provide reliable guarantee for navigation tasks in complex urban environments.

[0087] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above-mentioned embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0088] The above-mentioned embodiments only express several preferred implementation modes of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in the technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be based on the protection scope of the claims.

Claims

1. A navigation scene recognition method based on satellite distribution, characterized in that: The method comprises the following steps: Preset multiple navigation scenarios, and collect multi-constellation GNSS measured data corresponding to each navigation scenario respectively; the multi-constellation GNSS measured data includes satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange; Constructing a navigation scene recognition data set based on the multi-constellation GNSS measured data of all navigation scenes, and dividing the navigation scene recognition data set into a training set and a validation set according to a preset ratio; the navigation scene recognition data set includes the multi-constellation GNSS measured data of each navigation scene and the corresponding scene labels; According to the training set and the validation set, a CNN network is trained to obtain a navigation scene prediction model; Acquire the multi-constellation GNSS measured data to be identified, and obtain a scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model; The step of constructing a navigation scene recognition data set based on multi-constellation GNSS measured data of all navigation scenes includes: Converting each multi-constellation GNSS measured data into corresponding scene image data respectively; the scene image data includes a satellite distribution image, a carrier-to-noise ratio image, a Doppler frequency shift image and a pseudorange image corresponding to the satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange respectively; Adding corresponding scene labels to the scene image data to obtain the navigation scene recognition data set; The step of converting each multi-constellation GNSS measured data into corresponding scene image data comprises: Projecting the satellite positions of the multi-constellation GNSS measured data onto a circular image plane and performing pixel adjustment to obtain the satellite distribution image; The carrier-to-noise ratio, Doppler frequency shift and pseudorange of the multi-constellation GNSS measured data are normalized and interpolated using a von Lonoy diagram to obtain corresponding carrier-to-noise ratio images, Doppler frequency shift images and pseudorange images, respectively; The step of projecting the satellite positions of the multi-constellation GNSS measured data onto a circular image plane and performing pixel adjustment to obtain the satellite distribution image comprises: According to the following projection algorithm, the satellite position is projected onto the circular image plane to obtain an initial circular image: in, and Respectively represent the X-axis and Y-axis coordinate values ​​of the circular plane rectangular coordinate system; and Respectively represent the satellite altitude angle and azimuth angle; Indicates the radius of the circular image; The circular image of the satellite position is pixel-adjusted according to a preset rule to obtain an updated circular image; the preset rule includes setting the pixel value of the position where the satellite is distributed to 255 and setting the pixel value of the position where the satellite is not distributed to 0; The inner area of ​​the circumscribed square of the updated circular image is used as the generation area of ​​the satellite distribution image, and the area outside the updated circular image in the generation area is filled with a gray value of 0 to obtain the satellite distribution image.

2. The navigation scene recognition method based on satellite distribution according to claim 1, characterized in that: The navigation scenarios include urban canyons, urban-rural fringe areas, suburbs, tree-lined areas, and under elevated roads.

3. The navigation scene recognition method based on satellite distribution according to claim 1, characterized in that: The CNN network includes an input layer, a convolutional layer, a pooling layer, a plurality of fully connected layers and an output layer connected in sequence; the number of convolutional layers and pooling layers are set according to the navigation scenario; The step of training the CNN network according to the training set and the validation set to obtain the navigation scene prediction model comprises: The CNN network is trained using cross entropy as the loss function and stochastic gradient descent as the optimizer until a preset iteration termination condition is reached to obtain the navigation scene prediction model; the preset iteration termination condition is that both the training accuracy and the verification accuracy reach the corresponding preset accuracy standards, and the difference between the training accuracy and the verification accuracy is within a preset range.

4. The navigation scene recognition method based on satellite distribution according to claim 1, characterized in that: The step of obtaining a scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model comprises: Converting the multi-constellation GNSS measured data to be identified into corresponding scene image data to be identified; the scene image data to be identified includes a satellite distribution image to be identified, a carrier-to-noise ratio image to be identified, a Doppler frequency shift image to be identified and a pseudorange image to be identified, which correspond to the satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange respectively; The scene image data to be identified is input into the navigation scene prediction model for scene recognition to obtain the scene recognition result.

5. A navigation scene recognition system based on satellite distribution, characterized in that: The method for identifying a navigation scene based on satellite distribution as claimed in any one of claims 1 to 4 is applied, wherein the system comprises: A data acquisition module is used to preset multiple navigation scenarios and respectively collect multi-constellation GNSS measured data corresponding to each navigation scenario; the multi-constellation GNSS measured data includes satellite position, carrier-to-noise ratio, Doppler frequency shift and pseudorange; A data set construction module is used to construct a navigation scene recognition data set based on the multi-constellation GNSS measured data of all navigation scenes, and divide the navigation scene recognition data set into a training set and a validation set according to a preset ratio; the navigation scene recognition data set includes the multi-constellation GNSS measured data of each navigation scene and the corresponding scene label; A model training module, used to train a CNN network based on the training set and the validation set to obtain a navigation scene prediction model; The scene recognition module is used to obtain the multi-constellation GNSS measured data to be identified, and obtain the scene recognition result according to the multi-constellation GNSS measured data to be identified and the navigation scene prediction model.

6. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images