Satellite signal identification method and device, electronic equipment and readable storage medium
By extracting features such as satellite system, carrier-to-noise ratio, satellite altitude angle and pseudorange residual, the satellite signal recognition model is trained, which solves the problem of NLOS signal recognition in GNSS positioning and improves the positioning accuracy.
Patent Information
- Application Number
- CN202310548860.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-15
- Publication Date
- 2024-11-15
AI Technical Summary
In urban environments, GNSS positioning accuracy is affected by non-line-of-sight signals and multi-path signals. It is difficult for the prior art to effectively identify NLOS signals in satellite signals, resulting in a decrease in positioning accuracy.
By obtaining the sample data set, signal characteristics including satellite system, carrier-to-noise ratio, satellite altitude angle and pseudorange residual are extracted, satellite signal recognition model is trained, and signal characteristics of the target satellite signal are identified to improve the recognition accuracy of NLOS signals.
It improves the recognition ability and accuracy of satellite signal recognition models, enhances the recognition ability of NLOS signals, and improves the positioning accuracy of GNSS in urban environments.
Smart Images

Figure CN118962739A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to a satellite signal identification method and device, electronic equipment, computer-readable storage medium, and computer program product. Background Art
[0002] At present, with the rapid development of global navigation satellite systems, global navigation satellite systems (GNSS) have been applied in many technical fields, and can be mainly used for navigation, positioning, measurement, etc. In urban canyon scenes, satellite signals are affected by obstacles such as high-rise buildings and viaducts during space propagation, and their obstruction will generate non-line of sight (NLOS) signals and multipath signals, which will affect the accuracy of GNSS positioning. Therefore, how to effectively identify NLOS signals in satellite signals is of great significance to improving the positioning accuracy of GNSS in urban environments. Summary of the invention
[0003] The embodiments of the present application provide a satellite signal identification method and device, electronic equipment, computer-readable storage medium and computer program product, which can improve the satellite identification model's ability to identify NLOS signals in satellite signals and improve the accuracy and stability of classification and identification.
[0004] A satellite signal identification method is applied to an electronic device, wherein the electronic device includes a receiver for receiving satellite signals, and the method includes:
[0005] Acquire a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model; wherein the sample data set includes a plurality of sample data collected within the same preset time period, and each of the sample data includes a signal feature and a type label of the satellite signal; wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite elevation angle, and a pseudorange residual;
[0006] Receiving a target satellite signal to be identified;
[0007] Analyzing the target satellite signal and extracting signal features;
[0008] The signal characteristics of the target satellite signal are input into a satellite signal recognition model to obtain a recognition result of the target satellite signal.
[0009] A satellite signal identification device is applied to an electronic device, wherein the electronic device includes a receiver for receiving satellite signals, and the device includes:
[0010] A model training module, used to obtain a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model; wherein the sample data set includes a plurality of sample data collected within the same preset time period, and each of the sample data includes a signal feature and a type label of the satellite signal; wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite altitude angle, and a pseudorange residual;
[0011] A signal receiving module, used for receiving a target satellite signal to be identified;
[0012] A feature extraction module, used to analyze the target satellite signal and extract signal features;
[0013] The type recognition module is used to input the signal characteristics of the target satellite signal into a satellite signal recognition model to obtain a recognition result of the target satellite signal.
[0014] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the following steps:
[0015] Acquire a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model; wherein the sample data set includes a plurality of sample data collected within the same preset time period, and each of the sample data includes a signal feature and a type label of the satellite signal; wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite elevation angle, and a pseudorange residual;
[0016] Receiving a target satellite signal to be identified;
[0017] Analyzing the observation data of the target satellite signal to extract the signal features;
[0018] The signal characteristics of the target satellite signal are input into a satellite signal recognition model to obtain a recognition result of the target satellite signal.
[0019] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:
[0020] Acquire a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model; wherein the sample data set includes a plurality of sample data collected within the same preset time period, and each of the sample data includes a signal feature and a type label of the satellite signal; wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite elevation angle, and a pseudorange residual;
[0021] Receiving a target satellite signal to be identified;
[0022] Analyzing the observation data of the target satellite signal to extract the signal features;
[0023] The signal characteristics of the target satellite signal are input into a satellite signal recognition model to obtain a recognition result of the target satellite signal.
[0024] A computer program product, comprising a computer program, which, when executed by a processor, implements the following steps:
[0025] Acquire a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model; wherein the sample data set includes a plurality of sample data collected within the same preset time period, and each of the sample data includes a signal feature and a type label of the satellite signal; wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite elevation angle, and a pseudorange residual;
[0026] Receiving a target satellite signal to be identified;
[0027] Analyzing the observation data of the target satellite signal to extract the signal features;
[0028] The signal characteristics of the target satellite signal are input into a satellite signal recognition model to obtain a recognition result of the target satellite signal.
[0029] The above-mentioned satellite signal recognition method, device, electronic device and computer-readable storage medium obtain a sample data set, and train based on the sample data set to obtain a satellite signal recognition model, receive the target satellite signal to be identified, analyze the observation data of the target satellite signal, extract the signal features related to visibility, and input the signal features of the target satellite signal into the satellite signal recognition model to obtain the recognition result of the target satellite signal; wherein the satellite signal recognition model is obtained by training based on the sample data set, and the sample data set includes multiple sample data collected within the same preset time period, each sample data includes the signal features and type label of the satellite signal, wherein the signal features include at least the satellite system, carrier-to-noise ratio, satellite altitude angle and pseudorange residual, and the full The signal features that affect satellite signals are mined and the signal features that are not related to visibility are excluded as sample data for training. Compared with the problem of low accuracy in identifying NLOS signals by only using directly obtained principle features (such as C / N0 carrier-to-noise ratio threshold) in the related art, the signal features in the present application also include indirect features that need to be obtained by calculation (such as pseudorange residuals), which increases the number of signal features related to visibility in the sample data and improves the visibility of the sample data, so that the recognition ability of the trained satellite signal recognition model is stronger and the recognition result is more accurate, that is, a satellite signal recognition model with the ability to comprehensively recognize multiple signal features can be used to identify satellite signals, thereby improving the recognition accuracy of NLOS signals in satellite signals. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0031] Figure 1 is a flowchart of a satellite signal identification method in one embodiment;
[0032] Figure 2 is a schematic diagram of communication between a satellite and a receiver in one embodiment;
[0033] Figure 3 is a flowchart of a method for determining a type label of sample data in one embodiment;
[0034] Figure 4 A schematic diagram of a process of training based on the sample data set to obtain a satellite signal recognition model in an embodiment;
[0035] Figure 5 is a schematic diagram of the structure of an initial neural network module in one embodiment;
[0036] Figure 6 is a schematic flow chart of a satellite signal identification method in another embodiment;
[0037] Figure 7 is a structural block diagram of a satellite signal identification device in one embodiment;
[0038] Figure 8 FIG. 1 is a schematic diagram of the internal structure of an electronic device in an embodiment. DETAILED DESCRIPTION
[0039] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0040] Figure 1 Flowchart of a satellite signal identification method in one embodiment. The satellite signal identification method in this embodiment is described by taking the method running on an electronic device as an example. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the electronic device can be any terminal device including a mobile phone, a tablet computer, a PDA (Personal Digital Assistant), a POS (Point of Sales), a vehicle-mounted computer, a wearable device, etc. The server can be implemented as an independent server or a server cluster consisting of multiple servers.
[0041] In one embodiment, Figure 1 FIG. 1 is a flow chart of a satellite signal identification method in an embodiment. A satellite signal identification method is applied to an electronic device, where the electronic device is a receiver for receiving satellite signals. A satellite signal identification method includes steps 102 to 108.
[0042] Step 102: Obtain a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model.
[0043] Wherein, the sample data set includes multiple sample data collected within the same preset time period, and each of the sample data includes the signal characteristics and type label of the satellite signal. The characteristic signal in the sample data is based on the correlation analysis and processing of the observation data of the received satellite signal by the electronic device, and the signal characteristics related to visibility are extracted. Wherein, the receiver can be built-in or external to the electronic device. The satellite signal may include the observation data of the satellite signal received by the receiver. The satellite signal may be a Global Navigation Satellite System (GNSS) signal, and the observation data of the satellite signal may be GNSS observation data.
[0044] Optionally, in an embodiment of the present application, the correlation analysis can be performed using three methods: Spearman rank correlation coefficient method, Pearson correlation coefficient method and Kendall method. The Spearman rank correlation coefficient method is a non-parametric indicator for measuring the dependence of two variables, and uses a monotone equation to evaluate the correlation of two statistical variables. The Pearson correlation coefficient method is to calculate the quotient of the covariance and standard deviation between two variables; the Kendall correlation coefficient method is a method for expressing the degree of correlation of multiple column rank variables. If n similar statistical objects are sorted by specific attributes, other attributes are usually disordered, and the ratio of the difference between the same order pair and the different order pair to the total number of pairs [n*(n-1) / 2] is defined as the Kendall coefficient. The correlation coefficients of these three methods are between [-1,1]. The closer the absolute value of the correlation coefficient is to 1, the higher the correlation is; if the absolute value is equal to 0, it means that there is no correlation between the two variables. The features whose absolute values of the correlation coefficients of the three methods are all greater than 0.5 are selected as the basis for feature selection. By performing correlation analysis on the observation data of satellite signals, signal features related to satellite visibility can be obtained. The signal features include at least a satellite system, a carrier-to-noise ratio, a satellite elevation angle, and a pseudorange residual. The signal features of each sample data are the same. For example, the signal features in each sample data can be understood as a signal feature point set including at least four sub-features: a satellite system, a carrier-to-noise ratio, a satellite elevation angle, and a pseudorange residual.
[0045] Satellite systems may include China's BeiDou Navigation Satellite System (BDS), the Global Positioning System (GPS), the Global Navigation Satellite System (GLONASS) and the Galileo satellite navigation system (Galileo).
[0046] The carrier-to-noise ratio (also called the signal-to-noise ratio) is a standard measurement scale used to indicate the relationship between the carrier and the carrier noise, usually recorded as CNR or C / N (dB). A high carrier-to-noise ratio can provide better network reception rate, better network communication quality, and better network reliability. In the carrier-to-noise ratio, the power of the carrier is represented by Pc, and the power of the noise is represented by Pn. The decibel unit formula of the carrier-to-noise ratio is: C / N = 10lg (Pc / Pn).
[0047] Satellite altitude angle refers to the angle between the line connecting the satellite and the receiver and the horizontal plane. For example, if the receiver is built into an electronic device, the receiver can be understood as a GNSS receiving chip in the electronic device, and the electronic device can directly obtain signal characteristics such as the satellite system, carrier-to-noise ratio and satellite altitude angle by calling relevant interfaces.
[0048] In the embodiment of the present application, the pseudorange residual can be obtained by the following formula. The calculation formula of the pseudorange residual can be:
[0049]
[0050] Where s represents satellite, i represents epoch, represents pseudo range, represents the geometric distance between the satellite and the receiver (such as the user), c represents the speed of light, and dt r Indicates the receiver clock error; dT s represents the satellite clock error, represents the ionospheric delay, represents the process delay, ε P represents the receiver observation noise.
[0051] Pseudorange The signal propagation time can be obtained based on the transmission time of the satellite signal and the reception time of the signal by the receiver, and then multiplied by the propagation speed of the signal (for example, the speed of light).
[0052] The satellite clock error can be eliminated by the difference of the pseudorange residuals between satellites.
[0053] The calculation formula of pseudorange can be expressed as:
[0054]
[0055] In the formula, t Tx is the transmission time of the satellite signal, in nanoseconds; t RxThe receiving time of the satellite signal is in nanoseconds. For example, if the electronic device includes an Android system, the electronic device can directly obtain the transmission time and the receiving time through ReceivedSvTimeNanos in the GNSSMeasurement class. It should be noted that for different satellite systems, the time period is different. For example, for GPS, Galileo and BDS, the time is the time of the week, and for GLONASS, it is the time of the day. For the receiving time t Rx Taking GPS as an example, the following formula can be used to first calculate the number of nanoseconds from GPS zero time to the current time, and then deduct the number of nanoseconds from GPS zero time to the current week, weekNumberNanos.
[0056] t Rx =TimeNanos-(FullBiasNanos+BiasNanos)-weekNumberNanos
[0057] Among them, weekNumberNanos=round(FullBiasNanos / (604800*1E9)).
[0058] Among them, the user position is pre-calibrated, and the satellite position can be calculated based on the launch time and broadcast ephemeris; dT s It is obtained by fitting the satellite clock error coefficient in the broadcast ephemeris; Calculated using the Krobucher model; The Saastamoinen model is used for calculation. It should be noted that in an urban canyon environment, since the satellite is blocked by high-rise buildings, multipath and non-line of sight (NLOS) reception will occur, where the receiver observation noise includes multipath or non-line of sight errors.
[0059] The presentation of multiple sub-features in the signal feature point set can be sorted and presented in a preset order, such as a signal feature point set (A1, A2, A3, ..., An), where n≥4, Ai, n≥i≥1 represents any sub-feature. Each sub-feature can be represented by a feature annotation value in the signal feature point set. Among them, the feature annotation values of different sub-features can be set according to the physical meaning of the sub-features, and the feature annotation values of each sub-feature can be the same or different. For example, the satellite system is a BDS system, and the corresponding feature annotation value is 0; the satellite system is a GPS system, and the corresponding feature annotation value is 1; the satellite system is a GLONASS system, and the corresponding feature annotation value is 2; the satellite system is a Galileo system, and the corresponding feature annotation value is 3. The feature annotation values of the carrier-to-noise ratio, satellite altitude angle, and pseudorange residual can all be represented by the observed values or calculated values of each sub-feature. In the embodiment of the present application, the representation method of the feature annotation value of each sub-feature is not limited, nor is it limited to the above examples.
[0060] Satellite signal labels include non-line of sight (NLOS) satellite signals and line of sight (LOS) satellite signals. In a wireless communication system, the propagation conditions can be divided into line of sight (LOS) and non-line of sight (NLOS) environments. Under line of sight conditions, wireless signals propagate in a straight line between the satellite and the receiver without obstruction, which requires that there are no objects that block the radio waves in the first Fresnel zone. The type label can be represented by a label annotation value in the sample data. The label annotation value refers to the correct type label annotation corresponding to the sample data. For example, the category is non-line of sight satellite signal, and the corresponding label annotation value is 0; the category is line of sight satellite signal, and the corresponding label annotation value is 1. In an embodiment of the present application, the sample data may include a signal feature point set and a type label. For example, the sample data can be represented as (A1, A2, A3, ..., An, B); where B is the label annotation value.
[0061] In an embodiment of the present application, the type label corresponding to the satellite signal can be determined by the signal characteristics obtained above. Exemplarily, the electronic device can determine its type label according to the satellite altitude angle and pseudorange residual in the signal characteristics, or the electronic device can determine its type label according to the pseudorange residual in the signal characteristics.
[0062] In an embodiment of the present application, an initial neural network model can be trained based on the acquired sample data set to obtain a satellite signal recognition model. The initial neural network model refers to an initial recognition model that has not been trained. The initial neural network model can be used to identify satellite signals. The initial satellite signal recognition model can specifically be a convolutional neural network, etc., but is not limited thereto. The satellite signal recognition model refers to a neural network model that has been trained based on a sample data set, which can be used to obtain a satellite signal recognition result based on the target satellite signal recognition.
[0063] Step 104: Receive a target satellite signal to be identified.
[0064] The electronic device can identify the type of satellite signal based on the target satellite signal received by the receiver. The number of target satellite signals is not limited. The target satellite signal can be a satellite signal sent by any satellite system.
[0065] Step 106, analyzing the observation data of the target satellite signal to extract signal features.
[0066] The electronic device analyzes the observation data of the target satellite signal to extract the signal features related to visibility. The signal features include at least the satellite system, carrier-to-noise ratio, satellite altitude angle and pseudorange residual. Optionally, in the embodiment of the present application, the correlation analysis can be performed using three methods: Spearman rank correlation coefficient method, Pearson correlation coefficient method and Kendall method.
[0067] Step 108: input the signal characteristics of the target satellite signal into a satellite signal recognition model to obtain a recognition result of the target satellite signal.
[0068] Among them, the satellite signal recognition model is obtained by training the initial neural network model according to the sample data set, and the sample data set includes multiple sample data collected within the same preset time period, and each sample data includes signal features and type labels, and the type labels include non-line-of-sight satellite signals and line-of-sight satellite signals. The satellite signal recognition model is a trained model. The recognition result is used to characterize the category of satellite signals, such as non-line-of-sight satellite signals and line-of-sight satellite signals. The target satellite signal recognition result can be embodied in the form of values, text, audio, etc., and is not limited to this. It should be noted that the signal features included in the sample data used in the satellite signal recognition model training process are the same as the signal features extracted in step 106 related to visibility.
[0069] The electronic device can obtain the signal characteristics of the target satellite signal based on the analysis, and input the signal characteristics into the satellite signal recognition model. The satellite signal recognition model performs recognition based on the signal characteristics to obtain a category recognition result of the satellite signal.
[0070] The satellite signal recognition method in this embodiment obtains a sample data set, and performs training based on the sample data set to obtain a satellite signal recognition model, receives a target satellite signal to be identified, analyzes the observation data of the target satellite signal, extracts signal features related to visibility, and inputs the signal features of the target satellite signal into the satellite signal recognition model to obtain a recognition result of the target satellite signal; wherein the satellite signal recognition model is obtained by training based on the sample data set, and the sample data set includes multiple sample data collected within the same preset time period, each sample data includes a signal feature and a type label of the satellite signal, wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite altitude angle, and a pseudorange residual, which fully mines the factors affecting the satellite signal. The signal characteristics of the satellite signal and the signal characteristics that are not related to visibility are used as sample data for training. Compared with the problem of low accuracy in identifying NLOS signals by only using directly obtained principle characteristics (such as C / N0 carrier-to-noise ratio threshold) in the related art, the signal characteristics in the present application also include indirect characteristics that need to be obtained by calculation (such as pseudorange residuals), which increases the number of signal characteristics related to visibility in the sample data and improves the visibility of the sample data, so that the recognition ability of the trained satellite signal recognition model is stronger and the recognition result is more accurate, that is, a satellite signal recognition model with the ability to comprehensively recognize multiple signal characteristics can be used to identify satellite signals, thereby improving the recognition accuracy of NLOS signals in satellite signals.
[0071] In one embodiment, based on the above embodiment, the signal characteristics in the embodiment of the present application may also include carrier frequency, pseudorange uncertainty, pseudorange rate consistency and phase availability. It should be noted that the signal characteristics of the target satellite signal are the same as the signal characteristics of the sample data.
[0072] The carrier frequency can be understood as the frequency corresponding to the satellite's communication frequency band. Satellite communication frequencies generally operate in the microwave frequency band, with a frequency range of 1GHz to 40GHz. According to the frequency band, it can be divided into: L, S, C, X, Ku, K, Ka. For different satellite systems, the carrier frequency band is different, which can be shown in Table 1.
[0073] Table 1 is a table of carrier frequency bands for different satellite signals
[0074] Serial number GNSS Frequency band 1 China BeiDou BDS B1I+B1C+B2a 2 US GPS L1+L5 3 Russian GLONASS G1 4 EU Galileo E1+E5a
[0075] For example, if the receiver is built into an electronic device, the receiver can be understood as a GNSS receiving chip in the electronic device, and the electronic device can directly obtain the satellite carrier frequency by calling a related interface.
[0076] The uncertainty of the pseudorange can be expressed as the uncertainty of the satellite launch time.
[0077] Pseudorange rate consistency can also be understood as pseudorange rate Doppler consistency; where pseudorange rate Doppler consistency can be understood as the difference between the pseudorange rate represented by the pseudorange and the pseudorange rate represented by the Doppler frequency shift. Pseudorange rate Doppler consistency is expressed as:
[0078]
[0079] The pseudorange rate is the change in pseudorange of the same satellite between two epochs. is the pseudorange rate represented by the Doppler frequency shift. The pseudorange rate is represented by the change of pseudorange as:
[0080]
[0081] The pseudorange rate is expressed by the Doppler frequency shift as:
[0082]
[0083] in Expressed as the carrier wavelength, D i is the Doppler shift of satellite s, which is generated by the self-locked frequency loop without the influence of carrier cycle jump and NLOS error and multipath error. Therefore, the difference in pseudorange rate reflects the influence of NLOS error or multipath error. It should be noted that Doppler shift also has gross errors, and the gross errors of Doppler shift can be identified and eliminated to improve the accuracy of Doppler shift. Optionally, the identification of Doppler shift gross errors can be performed through the quadratic difference between Doppler epochs or the carrier-to-noise ratio model.
[0084] Phase availability is determined by the phase tracking state. When the state is cycle slip, half cycle slip and phase reset, it is recorded as phase unavailable, and other states are phase available.
[0085] The signal features in each sample data can be understood as a signal feature point set including at least eight sub-features, namely, satellite system, carrier frequency, carrier-to-noise ratio, satellite altitude angle, pseudorange residual, pseudorange uncertainty, pseudorange rate consistency and phase availability. The presentation of the eight sub-features in the signal feature point set can be presented in a preset order, such as a signal feature point set (A1, A2, A3, ..., A8). Each sub-feature can be represented by a feature annotation value in the signal feature point set. Among them, the feature annotation values of different sub-features can be set according to the physical meaning of the sub-features, and the feature annotation values of each sub-feature can be the same or different. For example, if the phase availability is phase unavailable, the corresponding feature annotation value is 0; if the phase availability is phase available, the corresponding feature annotation value is 1. The feature annotation value of the uncertainty of the pseudorange can be represented by a calculated value. The feature annotation value of the consistency of the pseudorange rate can be represented by a calculated value. In the embodiment of the present application, the representation method of the feature annotation value of each sub-feature is not limited, nor is it limited to the above examples.
[0086] In this embodiment, the electronic device extracts signal features based on the results of the correlation analysis, which may include eight signal features related to the visibility of satellite signals, namely, satellite system, carrier frequency, carrier-to-noise ratio, satellite altitude angle, pseudorange residual, pseudorange uncertainty, pseudorange rate consistency and phase availability. It performs a correlation analysis between the GNSS signal type and the features, and fully explores the signal features that affect the GNSS signal type. The signal features in this application also include indirect features that need to be obtained by calculation (such as pseudorange residual, pseudorange uncertainty, pseudorange rate consistency and phase availability), which increases the number of signal features related to visibility in the sample data, improves the visibility of the sample data, and excludes irrelevant signal features from entering the initial neural network model, so that the trained satellite signal recognition model can have stronger recognition ability and higher accuracy of the recognition result, thereby improving the recognition accuracy of NLOS signals in satellite signals.
[0087] In one of the embodiments, the type label of the sample data is determined in the following manner: determining whether a pseudorange double difference residual exists according to the pseudorange residual in the signal feature, and determining the type label according to the pseudorange double difference residual if the pseudorange double difference residual exists.
[0088] The electronic device can calculate the pseudo-range double difference residuals of all satellites based on the observation data of the satellite signals received by the receiver. Figure 2 , the pseudorange double difference residual formula is:
[0089]
[0090] Where i represents the epoch; r represents the mobile station receiver, for example, electronic devices such as mobile phones; b represents the base station receiver; j represents the satellite observed by both receivers r and b at the same time; j≠k; represents the quadratic difference of pseudoranges between satellite j and reference satellite k between receiver r and reference receiver b, Represents the quadratic difference in the geometric distance between satellite j and reference satellite k between receiver r and reference receiver b; reference satellite k is the satellite with the largest average carrier-to-noise ratio within the preset time period; ε P represents the receiver observation noise.
[0091] In the process of traversing all satellites to calculate the pseudorange double difference residuals, there may be some satellites that do not have their pseudorange double difference residuals. Exemplarily, receiver r and reference receiver b can observe satellite j respectively; if receiver r can observe satellite j, but reference receiver b does not observe satellite j, it can be understood that the pseudorange double difference residuals of satellite j do not exist. It can be understood that in the process of obtaining pseudorange double difference residuals, if the common-view satellites between receiver r and reference receiver b are inconsistent, it may cause a small number of satellites received by receiver r to be unable to be identified, leading to the non-existence of its pseudorange double difference residuals. If both receiver r and reference receiver b can observe satellite j, it can be considered that the pseudorange double difference residuals of satellite j exist.
[0092] The electronic device can determine whether the pseudorange double difference residual of the satellite corresponding to the satellite signal exists according to the above-mentioned pseudorange double difference residual formula. If the pseudorange double difference residual of the satellite exists, the type label of the satellite corresponding to the satellite signal can be determined based on the determined pseudorange double difference residual. Compared with the related art, in the process of calibrating the type label, only the height calibration of the 3D building model or the fisheye camera image calibration is used to improve the accuracy of the type label of the satellite signal, thereby improving the accuracy of the sample data, thereby making the recognition ability of the trained satellite signal recognition model stronger, thereby improving the recognition accuracy of the NLOS signal in the satellite signal.
[0093] In one of the embodiments, when a pseudorange double difference residual exists, the electronic device may determine a type label based on the pseudorange double difference residual, which may specifically include: when the pseudorange double difference residual is greater than a preset threshold, determining that the type label is a non-line-of-sight satellite signal; when the pseudorange double difference residual is less than or equal to a preset threshold, determining that the type label is a line-of-sight satellite signal.
[0094] Among them, the preset threshold can be determined according to parameters such as the geometric distance of the satellite. Exemplarily, the preset threshold can be 1-5 meters. In the embodiment of the present application, the value range of the preset threshold is not limited to the above example. Exemplarily, the preset threshold is 1 meter as an example. When the absolute value of the pseudorange double difference residual is greater than the preset threshold (for example, 1m), the electronic device can determine that the type label of its satellite signal is an NLOS satellite signal. Otherwise, it can determine that the type label of its satellite signal is a LOS satellite signal.
[0095] The satellite signal recognition method in this embodiment determines the type label of the satellite signal through the pseudo-range double difference residual, thereby avoiding the influence of the accuracy of the data type label on the accuracy of the 3D building model or the pixel accuracy of the fisheye camera in the related technology, and can improve the calibration accuracy of the type label, thereby ensuring the accuracy of the sample data, and further improving the recognition accuracy of the NLOS signal in the satellite signal.
[0096] like Figure 3 As shown, in one embodiment, the type label of the sample data is determined as follows:
[0097] Step 302: Determine whether a pseudorange double difference residual exists according to the pseudorange residual.
[0098] Step 304: When the pseudorange double difference residual does not exist, respectively obtain the building elevation angle and the satellite elevation angle.
[0099] Among them, the building elevation angle is the elevation angle of the building at a preset azimuth within the preset range of the receiver. Exemplarily, the electronic device can obtain the coordinates of the corner points of the bottom and top of the building within the preset range of the receiver, and calculate the building elevation angle based on the preset reference direction and the preset step length. In the embodiment of the present application, the specific values of the reference direction and the preset step length are not limited. For ease of explanation, in the embodiment of the present application, the electronic device can calculate the building elevation angle in a clockwise direction based on the geographic north direction and at intervals of 0.5°.
[0100] Furthermore, the calculation method of the elevation angle of a building at a certain azimuth angle can be adopted as follows:
[0101] Determine the plane equation of a building based on three arbitrary points on the plane (for example, the coordinates of the three corner points of the bottom of a building):
[0102] ax+by+cz+d=0
[0103] Calculate the horizontal distance d between the preset acquisition point (x0, y0, z0) and the building plane:
[0104]
[0105] Get the height h of the current building, then the height angle of the building at this azimuth is:
[0106] θ = arctan(h / d).
[0107] The electronic device can obtain the altitude angle at each azimuth according to the preset step size. For example, if the preset step size is 0.5°, 720 azimuths and 720 building altitude angles corresponding to the 720 azimuths can be obtained. The electronic device can construct a corresponding relationship between the azimuth and the building altitude angle. It should be noted that the building altitude angle can be understood as an altitude angle threshold, which can be used as an altitude angle threshold for type label calibration. In an embodiment of the present application, m altitude angle thresholds can be constructed based on the preset step size; wherein m=360 / preset step size. In other words, the smaller the preset step size, the more altitude angle thresholds it corresponds to, and the higher the calibration accuracy of its type label calibration.
[0108] In the absence of the pseudorange residual double difference of the current satellite, the type tag can be calibrated based on the building elevation angle and the satellite elevation angle. Specifically, the electronic device can compare the satellite elevation angle with the building elevation angle of the same azimuth, and calibrate the type tag according to the comparison result.
[0109] Step 306: If the satellite elevation angle is greater than the building elevation angle at the same azimuth, the type label is determined to be a non-line-of-sight satellite signal.
[0110] Step 308: If the satellite elevation angle is less than or equal to the building elevation angle, determine that the type label is a line-of-sight satellite signal.
[0111] In addition to obtaining the satellite altitude angle, the electronic device can also obtain the satellite azimuth based on the observation data of the satellite signal. Based on the obtained satellite azimuth and the correspondence between the azimuth and the building altitude angle constructed above, the electronic device can determine the target altitude angle threshold that matches the satellite azimuth. Among them, the azimuth corresponding to the target altitude angle threshold is the same as the satellite azimuth. The electronic device can determine the type label by comparing the satellite altitude angle and the target altitude angle threshold.
[0112] Specifically, if the satellite elevation angle is greater than the building elevation angle at the same azimuth, for example, the satellite elevation angle is greater than the target elevation angle threshold, the type label is determined to be a non-line-of-sight satellite signal. If the satellite elevation angle is less than or equal to the building elevation angle at the same azimuth, for example, the satellite elevation angle is less than or equal to the target elevation angle threshold, the type label is determined to be a line-of-sight satellite signal.
[0113] The satellite signal recognition method in this embodiment, in the process of determining the type label, when determining that the pseudorange residual double difference of the signal feature does not exist, calibrates the type label based on the building altitude angle and the satellite altitude angle. It uses a double cross-validation method to determine the type label, which can ensure the correctness of the type label, thereby improving the accuracy of the sample data, so that the recognition ability of the trained satellite signal recognition model is stronger, and thus the recognition accuracy of the NLOS signal in the satellite signal can be improved.
[0114] In one embodiment, obtaining a sample data set includes: obtaining a plurality of sample data, and preprocessing signal features of the plurality of sample data to obtain the sample data set.
[0115] The sample data set includes multiple sample data collected in the same time period, and each sample data may include the signal characteristics and type label of the observation data of the GNSS signal (hereinafter referred to as GNSS observation data). Specifically, the electronic device may obtain multiple GNSS observation data collected by the receiver in the same time period, and obtain corresponding sample data based on the multiple GNSS observation data. Multiple GNSS observation data have the same timestamp. Each sample data may include the signal characteristics and type label of the GNSS observation data.
[0116] The electronic device may preprocess the signal features in the acquired multiple sample data to obtain the corresponding sample data set. It can be understood that the signal features of each sample data in the sample data set are all preprocessed sample data. Among them, the preprocessing includes at least one of outlier processing and normalization processing. In the process of preprocessing the signal features, the electronic device may use corresponding preprocessing for different sub-features. The preprocessing method of the same sub-feature in the signal features in each sample data is the same. The same preprocessing method means that the same preprocessing method is used for processing. For example, the electronic device may perform normalization processing on the carrier-to-noise ratio, pseudorange uncertainty, etc. in the signal features. The electronic device may perform outlier processing on the pseudorange in the signal features used to obtain the pseudorange residual double difference to delete the pseudorange with abnormal values or the sample data without pseudorange. It should be noted that in the process of outlier processing and normalization processing of each sub-feature, the preprocessing parameters of each sub-feature can be customized according to the attributes of the sub-feature. The preprocessing parameters can be understood as the relevant parameters involved in the preprocessing process, for example, the reference value range involved in the outlier processing, and the mean, mean square error and other parameters involved in the normalization processing.
[0117] Optionally, the electronic device can also perform outlier processing and normalization processing on all sub-features respectively. If the electronic device performs outlier processing on each sub-feature in the process of preprocessing, it can first perform outlier processing to eliminate abnormal sample data, and then perform normalization processing on the signal features in the remaining sample data to improve the processing efficiency of signal feature preprocessing.
[0118] The satellite signal recognition method in this embodiment may first obtain multiple sample data in the process of obtaining a sample data set, and perform at least one of outlier processing and normalization processing on the signal features of each sample data, thereby improving the accuracy of the signal features of each sample data and ensuring that the signal features of each sample data have the same measurement scale. The satellite signal recognition model obtained by training is trained based on the preprocessed sample data, and thus the recognition ability is stronger, thereby improving the recognition accuracy of NLOS signals in satellite signals.
[0119] In one embodiment, when the electronic device performs outlier processing on the signal feature, it can be performed in the following manner: obtaining the reference value range corresponding to each sub-feature in the signal feature; when the feature value of any sub-feature of the sample data is not within the corresponding reference value range, deleting the sample data. It should be noted that each sub-feature in the signal feature can be any one of the satellite system, carrier frequency, carrier-to-noise ratio, satellite altitude angle, pseudorange residual, pseudorange uncertainty, pseudorange rate consistency and phase availability.
[0120] For each sub-feature, a reference value range can be set accordingly, and the reference value range of each sub-feature can be defined according to the attributes of the corresponding sub-feature so that it can meet the normal satellite communication requirements. Exemplarily, the signal features include eight sub-features, including satellite system, carrier frequency, carrier-to-noise ratio, satellite altitude angle, pseudo-range residual, pseudo-range uncertainty, pseudo-range rate consistency and phase availability. For these eight sub-features, eight reference value ranges can be set accordingly. For these eight sub-features, it can be judged whether the characteristic value of each sub-feature is within the corresponding reference value range respectively. If the characteristic value of any sub-feature is not within the corresponding preset range, the sample data including the sub-feature is deleted. Exemplarily, for the sub-feature of pseudo-range residual, the reference value range of pseudo-range residual can be set accordingly. If the characteristic value of the pseudo-range residual of a sample data (for example, the pseudo-range residual value) is not within the reference value range, or the pseudo-range residual does not exist, the pseudo-range residual can be considered to be abnormal, and the sample data including the pseudo-range residual can be deleted.
[0121] The satellite signal recognition method in this embodiment, when performing outlier processing on the signal features in the sample data, obtains the reference value range corresponding to each sub-feature in the signal feature, and deletes the sample data when the feature value of any sub-feature of the sample data is not in the corresponding reference value range, so as to eliminate the sample data with abnormal sub-features. This can make all the sample data in the sample data set correct and valid, thereby improving the validity of the sample data set. The satellite signal recognition model obtained by training based on the preprocessed sample data has a stronger recognition ability, thereby improving the recognition accuracy of NLOS signals in satellite signals.
[0122] In one embodiment, when the electronic device performs normalization processing on the signal feature, it can be performed in the following manner: obtaining a reference value range corresponding to each sub-feature in the signal feature; if the feature value of any sub-feature of the sample data is not within the corresponding reference value range, deleting the sample data.
[0123] In one embodiment, when the electronic device performs normalization processing on the signal feature, it can be performed in the following manner: respectively obtaining the mean and variance of each sub-feature in the signal feature; for each sub-feature, normalizing the sub-feature according to the mean and variance of the sub-feature.
[0124] Among them, the size range of the characteristic values of each sub-feature included in the signal feature is different. The characteristic value of the sub-feature can be a calculated value, a measured value, etc. of each sub-feature. Exemplarily, the size range of the carrier-to-noise ratio characteristic value is between 20-50, and the size range of the pseudorange uncertainty characteristic value is 0-1000. In the embodiment of the present application, the mean and variance of each sub-feature in the sample data set can be obtained accordingly, and for each sub-feature, the sub-feature can be normalized according to the corresponding mean and variance. Specifically, when the electronic device normalizes the sub-feature according to the mean and variance of the sub-feature, the normalization is achieved by de-meaning the characteristic value and scaling it to unit variance. For example, the calculation method of (eigenvalue-mean) / variance can be used to normalize each sub-feature separately. It should be noted that in the process of normalizing each sub-feature, the normalization method in the relevant technology can also be adopted. In the embodiment of the present application, the specific implementation method of the normalization is not specifically limited.
[0125] The satellite signal recognition method in this embodiment, when normalizing the signal features in the sample data, obtains the mean and variance of each sub-feature in the signal features respectively; for each sub-feature, the sub-feature is normalized according to the mean and variance of the sub-feature, so that the same sub-feature of all sample data in the sample data set has the same measurement scale, so that when training is performed based on the normalized sample data set, the gradient descent can converge quickly, and the accuracy of the satellite signal recognition model obtained by training will increase with the increase in the number of iterations, thereby making the recognition ability of the satellite signal recognition model stronger, and thus the recognition accuracy of NLOS signals in satellite signals can be improved.
[0126] In one embodiment, before inputting the signal features of the target satellite signal into the satellite signal recognition model to obtain the satellite signal recognition result of the object to be recognized, the method further includes: a step of normalizing the sub-features in the signal features of the target satellite signal.
[0127] Wherein, the processing parameters of the normalization processing of each sub-feature are respectively the same as the processing parameters of the normalization processing of the same sub-feature in the sample data. The electronic device may pre-set the processing parameters of each sub-feature; wherein the processing parameters may include the mean and variance corresponding to each sub-feature. The mean and variance corresponding to each sub-feature pre-stored by the electronic device are the mean and variance of each sub-feature obtained based on the sample data set. Exemplarily, if in the sample data set, the mean of the carrier-to-noise ratio of each sample data is μ and the variance is σ, then the electronic device may store the processing parameters of the carrier-to-noise ratio, which include the mean μ and the variance σ. Before the electronic device inputs the signal features of the target satellite signal into the satellite signal recognition model, it may first normalize the carrier-to-noise ratio of the target satellite signal based on the mean μ and the variance σD.
[0128] In the satellite signal recognition method of this embodiment, the electronic device may normalize the signal characteristics of the target satellite signal before inputting the signal characteristics of the target satellite signal into the satellite signal recognition model. For example, normalization is achieved by de-meaning the characteristic values and scaling them to unit variance, so that the satellite signal recognition model can quickly recognize the signal characteristics of the target satellite signal, thereby improving recognition efficiency.
[0129] In one embodiment, based on any of the foregoing embodiments, the satellite signal identification method further includes a step of performing equalization processing on a plurality of sample data according to type labels of the sample data to obtain a sample data set.
[0130] In the process of acquiring a sample data set, there may be an imbalance among multiple sample data, for example, the amount of sample data of different types of labels is quite different. The electronic device may correspondingly acquire multiple sample data and a type label for each sample data. If the number of sample data with non-line-of-sight satellite signals is different from the number of sample data with line-of-sight satellite signals, the electronic device may balance the multiple sample data so that the number of sample data with non-line-of-sight satellite signals is the same as the number of sample data with line-of-sight satellite signals, or the difference between the two is maintained within a preset range. The preset range may be set based on the capacity of the sample data set. In an embodiment of the present application, the preset range is not limited.
[0131] The satellite signal recognition method in this embodiment can perform equalization processing on multiple sample data according to the type labels of the sample data, so that the number of sample data with non-line-of-sight satellite signals is the same as the number of sample data with line-of-sight satellite signals, or the difference between the two is maintained within a preset range. In this way, the sample data with different type labels in the sample data set are in a balanced state, thereby improving the balance of the sample data set. The satellite signal recognition model obtained by training based on the sample data after equalization has stronger recognition ability, thereby improving the recognition accuracy of NLOS signals in satellite signals.
[0132] The acquisition of multiple sample data is based on multiple satellite signals collected in an open scene and a scene blocked by a high-rise building in the same preset time period. In this process, it cannot be ensured that the first sample data with a non-line-of-sight satellite signal and the second sample data with a line-of-sight satellite signal in the multiple sample data acquired are the same or substantially the same. The satellite signal recognition model training method in the embodiment of the present application can perform equalization processing on the multiple sample data acquired, so that the first sample data and the second sample data in the sample data set are the same or substantially the same.
[0133] Specifically, when the electronic device performs equalization processing on the sample data, it can be performed in the following manner: based on multiple sample data, obtain first sample data with a type label of non-line-of-sight satellite signal and second sample data with a type label of line-of-sight satellite signal. In the case where the number of first sample data and the number of second sample data are different, the first sample data or the second sample data with a larger number are eliminated so that the number of remaining first sample data is the same as the number of remaining second sample data. If the number of first sample data is less than the number of second sample data, part of the second sample data can be eliminated, or if the number of second sample data is less than the number of first sample data, part of the first sample data can be eliminated so that the number of remaining first sample data is the same as the number of remaining second sample data, or is approximately the same.
[0134] The satellite signal identification method in this embodiment can respectively obtain the number of first sample data with a type label of non-line-of-sight satellite signal and the number of second sample data with a type label of line-of-sight satellite signal, and eliminate the large number of sample data based on the small number of sample data to ensure that the number of first sample data and the number of second sample data in the sample data set are the same, or approximately the same. In this way, the balance of the sample data set can be improved, and the recognition ability of the trained satellite signal identification model can be strengthened, thereby improving the recognition accuracy of NLOS signals in satellite signals.
[0135] like Figure 4 As shown, in one embodiment, training is performed based on a sample data set to obtain a satellite signal recognition model, including steps 402 to 404.
[0136] Step 402: input each sample data in the sample data set into the initial neural network model for recognition to obtain a label prediction value for each sample data.
[0137] In one embodiment, the neural network of the initial neural network model is composed of a multi-layer neural network, each layer of the network is composed of a plurality of neurons. Each layer of the network is composed of a plurality of neurons. The signal y output from the neuron is represented by the input element x and the corresponding weight w and bias b.
[0138]
[0139] In the formula, f represents the activation function, which is used to determine the input of the next neuron. When the weighted sum of all neurons exceeds the threshold, the neuron is activated. Common activation functions are sigmoid, ReLU, and softmax. The weights w and biases b are adjusted using the back-propagation method in the neural network. For the back-propagation method, a loss function is defined for calculating the difference between the output and the correct value, and the gradient between the output layer and the input layer is optimized to minimize.
[0140] The neural network includes an input layer, a first output layer, a second output layer and multiple hidden layers, and the multiple hidden layers are located between the input layer and the output layer; wherein the number of neurons in each hidden layer decreases exponentially from the input layer to the output layer, and the first output layer and the second output layer are used to output recognition results with different satellite signals. The input layer is the signal feature extracted by the feature extraction module, which may include, for example, satellite system, carrier frequency, carrier-to-noise ratio, satellite altitude angle, pseudorange residual, pseudorange uncertainty, pseudorange rate consistency and phase availability. The hidden layer is an intermediate layer, for example, using ReLU as an activation function, designed as a multi-layer perceptron, and the layers are fully connected. The first output layer can be used to output label prediction values with non-line-of-sight satellite signals; correspondingly, the first output layer can be used to output label prediction values with line-of-sight satellite signals. Exemplarily, a softmax function can be used in the output layer, and the softmax function is a function of the probability that the total number of output classes is 100%. As Figure 5 As shown, the embodiment of the present application can provide an initial neural network model with a six-layer structure, and the number of neurons in each layer is: 2048, 2048, 512, 256, 32, and 2 respectively.
[0141] The label prediction value refers to the value obtained after the initial neural network model performs recognition. The label prediction value is not necessarily correct. Specifically, each time training is performed, each sample data can be input into the initial neural network model for recognition, thereby obtaining the label prediction value of each sample data.
[0142] Step 404, according to the cross entropy loss between the label prediction value and the label annotation value, update the parameters in the initial neural network model until the preset conditions are met to obtain the satellite signal recognition model.
[0143] Among them, the cross entropy loss is used to characterize the difference between the label prediction value and the label annotation value. The cross entropy loss can be calculated by the loss function. For example, based on the TensorFlow framework, the optimizer uses adam, the initial learning rate is set to 1E-6, and the loss function uses the cross entropy loss function, the formula is:
[0144]
[0145] In the formula, y is the label value in the sample data, Predicted value for the predicted label.
[0146] Optionally, the loss function may be a mean square error, a mean absolute error, a cross entropy loss function, etc., but not limited thereto. The preset condition may be at least one of the following: the training reaches a preset number of times, and the cross entropy loss is less than a preset loss value, but not limited thereto. The satellite signal recognition model refers to a trained neural network recognition model, which can be used to obtain a satellite signal type recognition result based on sample data.
[0147] Specifically, the electronic device calculates the loss value according to the label prediction value and the label annotation value. The electronic device performs back propagation on the entire neural network according to the cross entropy loss, adjusts the parameter values in the initial neural network model, and obtains a trained satellite signal recognition model until at least one of the preset number of training times and the cross entropy loss is less than the preset loss value is satisfied.
[0148] In the satellite signal recognition method of this embodiment, during the process of training the initial neural network model, each sample data in the sample data set is input into the initial neural network model for recognition, and a label prediction value of each sample data is obtained. According to the cross entropy loss between the label prediction value and the label annotation value, the parameters in the initial neural network model are updated until the preset conditions are met to obtain a satellite signal recognition model. In this way, based on the sample data set obtained above, the multi-layer initial neural network model can be trained to obtain a satellite signal recognition model, which can improve the recognition accuracy of NLOS signals in satellite signals.
[0149] In one embodiment, Figure 6 As shown, a satellite signal identification method includes:
[0150] Step 602: Offline training.
[0151] During the offline training process, the extraction of signal features and the determination of the type label of satellite visibility may be included. The extraction of signal features may be obtained based on the observation data of satellite signals, the location information of the receiver, and features such as satellite ephemeris. The determination of the type label of satellite visibility may be determined based on signal features, satellite ephemeris, and building elevation angle. The extraction of signal features and the determination of the type label of satellite visibility may refer to the satellite signal recognition model training method in any of the aforementioned embodiments, and will not be repeated here.
[0152] Based on the acquired signal features and type labels, a sample data set can be correspondingly acquired, and the sample data set is input into an initial neural network model to train it.
[0153] Step 604: online prediction.
[0154] During the online prediction process, the electronic device can obtain the observation data of the target satellite signal and analyze the observation data to extract the signal features related to visibility. Among them, the signal features include at least the satellite system, carrier-to-noise ratio, satellite altitude angle and pseudorange residual. The extracted signal features are input into the classifier in the trained satellite signal recognition model to obtain the classification results of the target satellite signal (for example, non-line-of-sight satellite signals and line-of-sight satellite signals).
[0155] The satellite signal recognition method in this embodiment performs satellite signal and feature correlation analysis on the observation data of the satellite signal, and extracts multiple sub-features related to visibility, for example, at least including the satellite system, carrier-to-noise ratio, satellite altitude angle and pseudorange residual, etc., which fully mines the signal features that affect the satellite signal and excludes the signal features that are not related to visibility as sample data to enter the initial neural network model. Compared with the problem of low accuracy of identifying NLOS signals by only using directly obtained principle features (such as C / N0 carrier-to-noise ratio threshold) in the related art, the signal features in the present application also include indirect features that need to be calculated (such as pseudorange residuals), which increases the number of signal features related to visibility in the sample data and improves the visibility of the sample data; in addition, the determination of the type label in the sample data is also based on the signal feature to ensure the accuracy of the type label, and a highly parallel and adaptable initial neural network model is used for training to train the obtained satellite signal recognition model. Furthermore, the electronic device receives the target satellite signal to be identified, analyzes the observation data of the target satellite signal, extracts signal features related to visibility, inputs the signal features of the target satellite signal into a trained satellite signal recognition model, and obtains the recognition result of the target satellite signal, which can make the satellite signal recognition model more capable of recognition and the recognition result more accurate.
[0156] It should be understood that although Figure 1, Figure 3-4 , Figure 6 The steps in the flowchart are shown in sequence as indicated by the arrows, but 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. Moreover, Figure 1 , Figure 3-4 , Figure 6 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.
[0157] like Figure 7 This is a structural block diagram of a satellite signal identification device in an embodiment. In one embodiment, the present application also provides a satellite signal identification device, which is applied to an electronic device. The electronic device includes a receiver for receiving satellite signals. The satellite signal identification device includes: a model training module 710, a signal receiving module 720, a feature extraction module 730 and a type identification module 740.
[0158] The model training module 710 is used to obtain a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model; wherein the sample data set includes a plurality of sample data collected within the same preset time period, and each of the sample data includes a signal feature and a type label of the satellite signal; wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite altitude angle, and a pseudorange residual;
[0159] A signal receiving module 720 is used to receive a target satellite signal to be identified;
[0160] The feature extraction module 730 is used to analyze the observation data of the target satellite signal and extract the signal features;
[0161] The type identification module 740 is used to input the signal characteristics of the target satellite signal into the satellite signal identification model to obtain the identification result of the target satellite signal.
[0162] The satellite signal recognition device in this embodiment receives a target satellite signal to be identified, analyzes observation data of the target satellite signal, extracts signal features related to visibility, inputs the signal features of the target satellite signal into a satellite signal recognition model, and obtains a recognition result of the target satellite signal; wherein the satellite signal recognition model is obtained by training an initial neural network model based on a sample data set, and the sample data set includes multiple sample data collected within the same preset time period. A satellite signal recognition model with the ability to comprehensively recognize multiple signal features can be used to perform satellite signal recognition, thereby improving the recognition accuracy of NLOS signals in satellite signals.
[0163] In one embodiment, the signal characteristics of the target satellite signal are the same as the signal characteristics of the sample data, wherein the signal characteristics further include at least one of carrier frequency, pseudorange uncertainty, pseudorange rate consistency and phase availability.
[0164] In one embodiment, the model training module is further used to determine whether a pseudorange double difference residual exists according to the pseudorange residual in the signal feature, and to determine a type label according to the pseudorange double difference residual if the pseudorange double difference residual exists.
[0165] In one embodiment, the model training module is also used to determine that the type label is a non-line-of-sight satellite signal when the pseudorange double difference residual is greater than a preset threshold; and to determine that the type label is a line-of-sight satellite signal when the pseudorange double difference residual is less than or equal to a preset threshold.
[0166] In one embodiment, the model training module is also used to obtain the building elevation angle and the satellite elevation angle respectively when the pseudorange double difference residual does not exist; wherein the building elevation angle is the elevation angle of a building within a preset range of the receiver at a preset azimuth; the type label determination module is also used to determine that the type label is a non-line-of-sight satellite signal when the satellite elevation angle is greater than the building elevation angle at the same azimuth; the type label determination module is also used to determine that the type label is a line-of-sight satellite signal when the satellite elevation angle is less than or equal to the building elevation angle at the same azimuth.
[0167] In one embodiment, the model training module is further used to obtain multiple sample data, preprocess the signal features of the multiple sample data, and obtain a sample data set, wherein the preprocessing includes at least one of outlier processing and normalization processing.
[0168] In one embodiment, the model training module is further used to obtain reference value ranges corresponding to each sub-feature in the signal feature, and delete the sample data when the feature value of any sub-feature of the sample data is not within the corresponding reference value range.
[0169] In one embodiment, the model training module is further used to respectively obtain the mean and variance of each sub-feature in the signal feature, and for each sub-feature, normalize the sub-feature according to the mean and variance of the sub-feature.
[0170] In one embodiment, the model training module is further used to perform equalization processing on the sample data according to the type labels of the sample data to obtain a sample data set.
[0171] In one embodiment, the model training module is also used to obtain first sample data with a type label of non-line-of-sight satellite signal and second sample data with a type label of line-of-sight satellite signal based on multiple sample data. When the number of first sample data and the number of second sample data are different, the first sample data or the second sample data with a larger number are eliminated so that the number of remaining first sample data and the number of remaining second sample data are the same.
[0172] In one embodiment, the model training module is also used to input each sample data in the sample data set into the initial neural network model for identification, obtain the label prediction value of each sample data, and update the parameters in the initial neural network model according to the cross entropy loss between the label prediction value and the type label until the preset conditions are met to obtain the satellite signal recognition model.
[0173] In one embodiment, the satellite signal identification device also includes a preprocessing module for normalizing the sub-features in the signal features of the target satellite signal, wherein the processing parameters of the normalization processing of each sub-feature are the same as the processing parameters of the normalization processing of the same sub-feature in the sample data.
[0174] The division of each module in the above-mentioned satellite signal recognition device is only for illustration. In other embodiments, the satellite signal recognition model training device and the satellite signal recognition device may be divided into different modules as needed to complete all or part of the functions of the above-mentioned satellite signal recognition model training device and the satellite signal recognition device.
[0175] For the specific definition of the satellite signal identification device, please refer to the definition of the satellite signal identification method above, which will not be repeated here. Each module in the above-mentioned satellite signal identification device 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.
[0176] Figure 8 FIG. 1 is a schematic diagram of the internal structure of an electronic device in one embodiment. Figure 8As shown, the electronic device includes a processor and a memory connected via a system bus. The processor is used to provide computing and control capabilities to support the operation of the entire electronic device. The memory may include a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The computer program can be executed by the processor to implement a method provided in each of the following embodiments. The internal memory provides a cached operating environment for the operating system computer program in the non-volatile storage medium. The electronic device can be any terminal device such as a mobile phone, a tablet computer, a PDA (Personal Digital Assistant), a POS (Point of Sales), a vehicle-mounted computer, a wearable device, etc.
[0177] The implementation of each module in the satellite signal identification device provided in the embodiment of the present application can be in the form of a computer program. The computer program can be run on a terminal or a server. The program module constituted by the computer program can be stored in the memory of an electronic device. When the computer program is executed by a processor, the steps of the satellite signal identification method described in any embodiment of the present application are implemented.
[0178] The embodiment of the present application also provides a computer-readable storage medium. One or more non-volatile computer-readable storage media containing computer-executable instructions, when the computer-executable instructions are executed by one or more processors, the processors execute the steps of the satellite signal identification method described in any embodiment of the present application.
[0179] A computer program product comprising instructions, when running on a computer, enables the computer to execute the steps of the satellite signal identification method described in any embodiment of the present application.
[0180] Any reference to memory, storage, database or other medium used in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM), which is used as an external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0181] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A satellite signal identification method, characterized in that: Applied to an electronic device, the electronic device includes a receiver for receiving satellite signals, the method includes: Acquire a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model; wherein the sample data set includes a plurality of sample data collected within the same preset time period, and each of the sample data includes a signal feature and a type label of the satellite signal; wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite elevation angle, and a pseudorange residual; Receiving a target satellite signal to be identified; Analyzing the observation data of the target satellite signal to extract the signal features; The signal characteristics of the target satellite signal are input into a satellite signal recognition model to obtain a recognition result of the target satellite signal.
2. The method according to claim 1, characterized in that The signal characteristics of the target satellite signal are the same as the signal characteristics of the sample data, wherein the signal characteristics further include at least one of carrier frequency, pseudorange uncertainty, pseudorange rate consistency and phase availability.
3. The method according to claim 1, characterized in that The type label of the sample data is determined as follows: Determining whether a pseudorange double difference residual exists according to the pseudorange residual in the signal feature; In a case where the pseudorange double difference residual exists, the type label is determined according to the pseudorange double difference residual.
4. The method according to claim 3, characterized in that The determining the type label according to the pseudorange double difference residual comprises: When the pseudorange double difference residual is greater than a preset threshold, determining that the type label is the non-line-of-sight satellite signal; When the pseudorange double difference residual is less than or equal to the preset threshold, the type label is determined to be the line-of-sight satellite signal.
5. The method according to claim 3, characterized in that: The method further comprises: In the case where the pseudorange double difference residual does not exist, respectively obtaining the building elevation angle and the satellite elevation angle; wherein the building elevation angle is the elevation angle of the building within the preset range of the receiver at a preset azimuth; If the satellite elevation angle is greater than the building elevation angle at the same azimuth, determining that the type label is the non-line-of-sight satellite signal; If the satellite elevation angle is less than or equal to the building elevation angle at the same azimuth, the type label is determined to be the line-of-sight satellite signal.
6. The method according to claim 1, characterized in that The obtaining of the sample data set comprises: Get multiple sample data; Preprocessing is performed on the signal features of the plurality of sample data to obtain the sample data set, wherein the preprocessing includes at least one of an outlier processing and a normalization processing.
7. The method according to claim 6, characterized in that The singularization process is performed in the following manner: Respectively obtaining a reference value range corresponding to each sub-feature in the signal feature; wherein the sub-feature is one of the satellite system, the carrier-to-noise ratio, the satellite altitude angle, and the pseudorange residual; When the feature value of any of the sub-features of the sample data is not within the corresponding reference value range, the sample data is deleted.
8. The method according to claim 6, characterized in that The normalization is performed as follows: Respectively obtain the mean and variance of each sub-feature in the signal feature; For each of the sub-features, the sub-feature is normalized according to the mean and variance of the sub-feature.
9. The method according to claim 8, characterized in that Before inputting the signal feature of the target satellite signal into the satellite signal recognition model to obtain the satellite signal recognition result of the object to be recognized, the method further includes: The sub-features in the signal feature of the target satellite signal are respectively normalized, wherein the processing parameters of the normalization processing of each sub-feature are respectively the same as the processing parameters of the normalization processing of the same sub-feature in the sample data.
10. The method according to claim 6, characterized in that The method further comprises: The sample data is balanced according to the type labels of the sample data to obtain the sample data set.
11. The method according to claim 9, characterized in that The equalization process is performed in the following manner: Based on the plurality of sample data, acquiring first sample data whose type label is the non-line-of-sight satellite signal and second sample data whose type label is the line-of-sight satellite signal; When the number of the first sample data is different from the number of the second sample data, the first sample data or the second sample data with a larger number is eliminated to make the number of the remaining first sample data and the number of the remaining second sample data the same.
12. The method according to claim 1, characterized in that The training based on the sample data set to obtain a satellite signal recognition model includes: Input each of the sample data in the sample data set into the initial neural network model for identification, and obtain a label prediction value for each of the sample data; According to the cross entropy loss between the label prediction value and the type label, the parameters in the initial neural network model are updated until the preset conditions are met to obtain a satellite signal recognition model.
13. A satellite signal recognition device, characterized in that: Applied to electronic equipment, the electronic equipment includes a receiver for receiving satellite signals, and the satellite signal identification device includes: A model training module, used to obtain a sample data set, and perform training based on the sample data set to obtain a satellite signal recognition model; wherein the sample data set includes a plurality of sample data collected within the same preset time period, and each of the sample data includes a signal feature and a type label of the satellite signal; wherein the signal feature includes at least a satellite system, a carrier-to-noise ratio, a satellite altitude angle, and a pseudorange residual; A signal receiving module, used for receiving a target satellite signal to be identified; A feature extraction module, used to analyze the observation data of the target satellite signal and extract signal features related to visibility; The type recognition module is used to input the signal characteristics of the target satellite signal into a satellite signal recognition model to obtain a recognition result of the target satellite signal.
14. An electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the computer program is executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 12.
15. 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 12 are implemented.
16. A computer program product comprising a computer program, 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 12 are implemented.
