A method and device for airborne multi-source integrity monitoring of aviation navigation operation network

Abstract

The present invention discloses an airborne multi-source integrity monitoring method and device for an aviation navigation operation network, which belongs to the field of satellite navigation and constructs the overall design architecture of an airborne multi-source high-performance navigation receiver. An ionospheric gradient monitoring algorithm based on carrier phase differential is proposed, the statistical error characteristics of IGM detection of carrier phase are studied, an ionospheric gradient monitoring threshold is established, and an airborne DFree to IFree mode switching mechanism is realized. An aviation navigation multipath mitigation suppression algorithm is studied, and an observation data feature matrix is ​​established in combination with multivariate feature extraction of observation data. A parallel time convolutional neural network architecture is designed, and weights are updated by stochastic gradient descent. The present invention provides an airborne multi-source integrity monitoring method and device for an aviation navigation operation network, which realizes accurate monitoring of ionospheric gradients, optimizes the switching mechanism between DFree and IFree modes, and improves the integrity monitoring capability and system availability of the aviation navigation operation network.

Description

Technical Field

[0001] The present invention relates to the field of satellite navigation technology, and in particular to a method and device for airborne multi-source integrity monitoring of an aviation navigation operation network. Background Art

[0002] In today's globalized aviation environment, the reliability and efficiency of aviation navigation systems are crucial. With technological advancements and increased air traffic, traditional land-based and satellite-based navigation systems are gradually demonstrating their limitations. While land-based navigation is stable and reliable, its coverage is limited; while satellite-based navigation offers wide-area coverage, it is often affected by various interference issues. These issues not only limit the improvement of aviation navigation services but also impact aviation safety and operational efficiency. Therefore, there is an urgent need to transition from traditional land-based navigation systems to more advanced aviation navigation operational networks that support the high safety requirements of civil aviation and provide high-precision, high-integrity navigation guidance information.

[0003] In a complex aviation environment, the implementation of an airborne multi-source integrity monitoring method and device for an aviation navigation operation network requires solving the following problems: First, the ground-based monitoring of satellite navigation ionospheric anomalies has low resolution and a conservative threat model. Existing ionospheric anomaly monitoring mainly relies on simple models, which lacks real-time and accuracy. Second, in a complex environment, multipath effects occur frequently and have spatiotemporal complexity, and are the main error source of carrier phase differential positioning. Existing multipath mitigation and suppression mainly considers single data features and fails to fully utilize observation data to mitigate and suppress multipath errors.

[0004] To address the first issue of ionospheric gradient monitoring, in 2017, DLR, KAIST, and TUM jointly proposed an ionospheric gradient monitoring (IGM) system for the DFM CGBAS. This system bases the positioning algorithm on single-frequency observations and uses a second navigation frequency for ionospheric monitoring. In 2022, DLR proposed a dual-frequency, space-based ionospheric gradient monitoring method. Building on the existing monitoring concept, it derived a new unbiased test statistic, successfully eliminating the impact of common biases on ionospheric estimates. The current DFM CGBAS architecture uses an ionospheric-free smoothing mode (IFree) and a divergence-free smoothing mode (DFree) to minimize the impact of ionospheric anomalies. IFree completely eliminates the impact of ionospheric gradients, eliminating the need for ionospheric gradient monitoring. However, IFree has a high noise level, resulting in larger navigation system errors. In contrast, DFree has a lower noise level but is affected by ionospheric gradients. Therefore, it is necessary to study the carrier phase gradient monitoring algorithm and optimize the switching mode from DFree to IFree.

[0005] To address the second multipath mitigation and suppression issue, Feng Shen et al. in 2020 considered multipath and NLOS signals as additive sparse errors, established a linear collaborative navigation observation model, and addressed the navigation problem by studying sparse estimation algorithms. However, this method is limited by multipath interference and the increasing number of NLOS signal channels, and its use of only range and Doppler observations has certain limitations. In 2021, Chao Liu et al. used a CNN-LSTM neural network to mine multipath features in the coordinate and frequency domains, achieving better multipath mitigation performance than traditional methods. However, the potential for multipath modeling in the spatial domain using machine learning has not been studied. Therefore, it is necessary to use machine learning, a data-driven approach, to explore intelligent identification and mitigation methods for air-ground collaborative multipath effects, thereby improving the network integrity monitoring capabilities and system availability of aviation navigation operations. Summary of the Invention

[0006] The purpose of the present invention is to provide an onboard multi-source integrity monitoring method and device for an aviation navigation operation network to solve the problems mentioned in the background technology.

[0007] To achieve the above-mentioned object, the present invention provides an airborne multi-source integrity monitoring device for an aviation navigation operation network, comprising a navigation source signal receiving module, a pan-source navigation information processing module, and a communication receiving module. The navigation source signal receiving module comprises:

[0008] GNSS module, used to receive signals transmitted by the global positioning system to determine the position and time information of the aircraft;

[0009] VDB module, used to receive navigation correction information sent from the ground;

[0010] an ILS / VHF omnidirectional range beacon (VOR) receiver module to provide additional navigation signals;

[0011] Inertial navigation system module, which calculates position and velocity by measuring the device's own acceleration;

[0012] Dual-frequency multi-mode receiver for simultaneous reception and processing of multiple navigation signals;

[0013] The pan-source navigation information processing module includes:

[0014] GBAS module and SBAS module are used to provide additional navigation correction information;

[0015] GNSS / VOR / DME / INS / LDACS data fusion processing module, integrating information from multiple sources;

[0016] The communication receiving module includes:

[0017] ARINC429 interface and RS422 interface are used to complete the data communication standards between avionics components to transmit information between devices;

[0018] RJ45 port, used for network connection, for device configuration or maintenance communication.

[0019] The present invention also provides an airborne multi-source integrity monitoring method for an aviation navigation operation network, comprising the following steps:

[0020] S1. Propose an ionospheric gradient monitoring algorithm based on carrier phase difference, study the statistical error characteristics of IGM detection of carrier phase, establish the ionospheric gradient monitoring threshold, and implement the switching mechanism from DFree to IFree mode onboard;

[0021] S2. An aviation navigation multipath mitigation and suppression algorithm is proposed. Combined with the multivariate feature extraction of observation data, an observation data feature matrix is ​​established, and a parallel time convolutional neural network architecture is designed. The weights are updated through the stochastic gradient descent method to achieve the prediction and mitigation of multipath errors.

[0022] Preferably, S1 specifically includes:

[0023] S11. Optimize multiple navigation sources according to different flight phases of route operation and design a flexible integrated navigation algorithm;

[0024] S12, take the carrier phase observations at the L1 and L5 frequency points, and use the carrier phase difference to estimate the ionospheric gradient on the airborne side;

[0025] S13, carrier phase difference and pseudo-range residual are combined to eliminate ionospheric delay term, and time average is used to eliminate ;

[0026] S14, project the receiver differential code bias to the clock error term, eliminate the satellite differential code bias by differentiating it with the ground observation, and then use the carrier phase to code balance CCL to estimate the ionospheric delay ;

[0027] S15, using the carrier phase observation, by combining pseudorange and carrier phase Variable averaging, constructing the test statistic for ionospheric gradient monitoring, and calculating using recursive filters In the iteration Noise variance and error standard deviation, and calculate the monitoring threshold.

[0028] Preferably, the content of S11 is as follows:

[0029] The ground station uploads the original observation data to the airborne terminal instead of the differential correction value, and uses the pure carrier phase for positioning on the airborne terminal. The observation values ​​broadcast by the ground station to the airborne terminal are as follows:

[0030] (1)

[0031] (2)

[0032] in, Represents a specific frequency Pseudorange measurement under ; Represents the carrier phase measurement at this frequency; Indicates the direct path LOS; and are the satellite and receiver clock biases, respectively; and represent the ionospheric and tropospheric delays, respectively; Indicates that the ionospheric divergence is at a frequency The cumulative error of the filter under DFree and IFree smoothing is 0; and is a term related to multipath error; and It represents the receiver thermal noise factor; Indicates the frequency Antenna group delay under ; and Respectively represent the code deviation of the receiver and satellite RF front end at this frequency; Indicates the frequency Changes in the antenna phase center under ; and Indicates the phase deviation between the receiver and the satellite RF front end; Represents the carrier phase ambiguity.

[0033] Preferably, S12 and S13 are as follows:

[0034] Taking the difference in carrier phase measurement between the L1 and L5 frequencies yields:

[0035] (3)

[0036] in represents the sum of carrier phase thermal noise and multipath error; and represent the differential phase deviations of the receiver and the satellite respectively; Indicates the differential phase center change; represents the combined form of the ambiguity after dual-frequency carrier differencing; the difference between the carrier phase measurements at L1 and L5 frequencies is used to derive an expression for the airborne ionospheric gradient, which is used to separate the observables and isolate the noise term on the right side of the equation:

[0037]

[0038] (4)

[0039] Carrier phase ambiguity , Differential Phase Center Change , and the differential phase deviation are included as known quantities in the measurement equation;

[0040] The differential observation is simplified as follows:

[0041] (5)

[0042] (6)

[0043] Error term 、 Random and zero mean, and represent the sum of thermal noise and multipath error of pseudorange and carrier phase respectively; combining carrier difference with pseudorange difference yields:

[0044]

[0045] (7)

[0046] ;

[0047] in, represents the antenna group delay variation; Indicates the receiver differential code deviation; Indicates the satellite differential code deviation;

[0048] The ionospheric term is eliminated and the average term By averaging the above formula over time, we can get:

[0049]

[0050] (8).

[0051] Preferably, the content of S14 is as follows:

[0052] According to the calculation method of ionospheric observation in CCL, the estimated ionospheric delay is:

[0053]

[0054] (9)

[0055] Estimated Used to make differential estimates of the ionospheric differences between the airborne terminal and the ground station:

[0056]

[0057] (10)

[0058] in, is the receiver differential airborne code deviation, the receiver differential ground code deviation is the noise term, is the ground-side antenna group delay characteristic, is the group delay characteristic of the airborne antenna, is the carrier phase differential ground-side ionospheric gradient estimation, The ionospheric gradient estimation of the airborne side at the L1 frequency point;

[0059] Since the receiver DCB term in the ionospheric delay is common across all satellites, the DCB term is solved jointly using weighted least squares or multi-epoch filters;

[0060] Using the VHF Data Broadcast (VDB), the airborne system can estimate the combined ionospheric delay of each satellite observed by the ground reference receiver using the equation for ionospheric delay, which can be reconstructed as follows:

[0061] (11)

[0062] in, It represents the observed value of the wave phase at the L5 frequency. It represents the observed value of the wave phase at the L1 frequency;

[0063] The airborne equipment calculates an estimate of the ionospheric delay seen by the ground station using the following expression:

[0064] (12)

[0065] The airborne system uses the CCL process described above to estimate the ionospheric delay in the air. .

[0066] Preferably, the content of S15 is as follows:

[0067] Over a period of time, through The variables are averaged and the test statistic is calculated as:

[0068] (13)

[0069] in After averaging the pseudorange and carrier phase noise of L1 and L5 The residual error in is obtained by averaging the data for N seconds. ;

[0070] Run a recursive filter of the form , and change the Substitute the following formula:

[0071] (14)

[0072] in represents the pseudorange and carrier phase noise at two frequencies, represents the differential pseudorange, represents the differential carrier phase; is a composite fuzzy term in dual frequency carrier processing, ignoring , assuming that thermal noise and multipath noise are white Gaussian noise, in the In the iteration The variance of the noise is as follows:

[0073] (15)

[0074] The standard deviation of the error is given by the following equation:

[0075] (16)

[0076] in is the standard deviation describing the combination of thermal noise and multipath errors on L1 and L5.

[0077] Preferably, step S2 is as follows:

[0078] S21. Use historical observation data, including observation residual values, signal strength, elevation and azimuth angles, to construct a multipath error feature matrix, eliminate the influence of unit and scale differences between observation features, and normalize the observation features;

[0079] S22. Design a parallel time convolutional neural network model to model multipath errors offline and suppress multipath online.

[0080] S23. Use the stochastic gradient descent method to update the neural network parameters, and use the data-driven method to predict multipath and alleviate the multipath effect.

[0081] Preferably, the content of S21 is as follows:

[0082] To construct the multipath error feature matrix, the selected observation features are divided into three categories: the first category is closely related to the multipath error studied by previous researchers, such as satellite elevation and azimuth angles, signal strength, and observation residuals; the second category is GNSS time; and the third category is the quality characteristics of the observations. The observation features are normalized, and the normalized data set is as follows:

[0083] (17)

[0084] in Indicates the The input of samples is the selected observation data features; Indicates the The output of samples is the unmodeled multipath error.

[0085] Preferably, the contents of S22 and S23 are as follows:

[0086] The parallel time convolutional neural network consists of an input layer, 4 convolutional layers, a pooling layer, a flatten layer, a fully connected layer and an output layer; the size of each convolution kernel in the network is , the stride in both directions is set to 1; the number of convolution kernels in each convolution layer is set to 64, 64, 32 and 32 respectively; the convolution layer The convolution operation is as follows:

[0087] (18)

[0088] in Indicates the The convolution layer input feature matrix on each channel; Indicates the number of channels of the convolutional layer input; Indicates the The convolution layer corresponding to the input convolution kernels; represents the convolutional layer The bias of the convolution kernel; It is the convolutional layer The output feature matrix of the convolution kernel;

[0089] The Flatten layer is used to convert multidimensional data into one-dimensional data. The value of the loss function is obtained through each epoch of forward propagation. The loss function is defined as follows:

[0090] (19)

[0091] in To optimize the parameters; is the number of samples;

[0092] The parameter information is adjusted through the gradient of the loss function, and the weight is updated using the stochastic gradient descent method, as shown below:

[0093] (20)

[0094] After obtaining the gradient of the loss function, the parameters are updated and the network training enters the next epoch.

[0095] Therefore, the present invention adopts the above-mentioned method and device for airborne multi-source integrity monitoring of aviation navigation operation network, which has the following beneficial effects:

[0096] (1) The present invention helps to solve the aviation safety risks and efficiency degradation problems caused by frequent interference of aviation radio navigation signals; provides an air-ground coordinated integrity monitoring method for future aviation navigation positioning, while ensuring navigation accuracy and improving system reliability; establishes a new overall design architecture for airborne multi-source high-performance navigation receivers, which can fully utilize multi-source information to ensure efficient operation of aviation navigation throughout the entire flight phase;

[0097] (2) The present invention adopts the ionospheric gradient monitoring method of carrier phase differential to accurately detect ionospheric anomalies and reduce the low resolution problem of traditional land-based navigation monitoring. It optimizes the switching mechanism from DFree to IFree mode, effectively eliminates the influence of ionospheric gradient on the navigation system, and improves the stability and reliability of the navigation system under different ionospheric conditions. It combines the multivariate characteristics of the observation data to establish a feature matrix and improve the recognition and processing capabilities of multipath errors.

[0098] (3) The present invention uses a parallel time convolutional neural network to predict and mitigate the multipath effect, significantly improving the error processing effect of carrier phase differential positioning; through intelligent algorithms and machine learning technology, it realizes real-time processing and updating of navigation data, improving the real-time response capability of the system; it can effectively improve the economic benefits of aviation navigation, and by improving the accuracy and safety of aviation navigation positioning systems, it can reduce the costs of airlines, reduce the number of flight delays and cancellations, and improve the efficiency and operational benefits of flights.

[0099] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 This is a schematic diagram of the overall design architecture of an airborne multi-source high-performance navigation receiver according to an embodiment of the present invention;

[0101] Figure 2 This is a flow chart of ionospheric gradient monitoring based on carrier phase differential according to an embodiment of the present invention;

[0102] Figure 3 This is a schematic diagram of the structure of a convolutional neural network model for multipath mitigation suppression according to an embodiment of the present invention. DETAILED DESCRIPTION

[0103] The following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort shall fall within the scope of protection of the present invention.

[0104] See also Figure 1 , an airborne multi-source integrity monitoring device for aviation navigation operation network,

[0105] This diagram demonstrates the overall design and functional roles of the various modules of the onboard multi-source aviation navigation technology device. A development environment based on the ARM A76 + A55 processors was established, completing the overall development framework and basic dynamic library development. The diagram includes the navigation signal receiving modules and the pan-source navigation information processing module, along with interfaces such as ARINC 429 and RS422, illustrating how these components interact within the system. It also shows the internal system connections, such as the use of RJ45 connectors.

[0106] The GNSS module receives signals transmitted by the Global Positioning System (GPS) to determine the aircraft's position and time. The VDB module receives navigation correction information from the ground, enhancing GNSS accuracy. The Instrument Landing System (ILS) / VHF Omnidirectional Range (VOR) receiver provides additional navigation signals to aid in precise landing or heading positioning. The inertial navigation system (INS) is independent of external signals and calculates position and velocity by measuring the device's own acceleration, thereby enhancing navigation redundancy and independence. The dual-frequency, multi-mode receiver can simultaneously receive and process multiple navigation signals. The GBAS and SBAS modules provide additional navigation correction information to enhance position accuracy. The GNSS / VOR / DME / INS / LDACS data fusion processing module integrates information from multiple sources to achieve comprehensive and flexible navigation during different flight phases. In the communication reception design, ARINC429 and RS422 interfaces are used to implement data communication standards between avionics components for inter-device information transmission. The RJ45 interface is used for network connection and can be used for device configuration and maintenance communications. These modules work together to ensure that the aircraft can obtain accurate position and navigation information in various environments and conditions, thereby improving flight safety and efficiency.

[0107] See also Figure 2 A method for monitoring the airborne multi-source integrity of an aviation navigation operation network includes an ionospheric gradient monitoring algorithm for aviation navigation and a multipath suppression mitigation algorithm for aviation navigation.

[0108] Applicable to aviation navigation ionospheric gradient monitoring algorithm, the specific calculation steps are as follows:

[0109] Step 1: Ionospheric gradient estimation based on carrier phase differential

[0110] Consider that the ground station uploads raw observation data to the airborne terminal instead of differential corrections, and uses pure carrier phase for positioning on the airborne terminal. The observations broadcast by the ground station to the airborne terminal are as follows:

[0111]

[0112]

[0113] in, Represents a specific frequency Pseudorange measurement under Represents the carrier phase measurement at this frequency. represents the length of direct access (LOS), and are the satellite and receiver clock biases, respectively. and represent the ionospheric and tropospheric delays, respectively. Indicates that the ionospheric divergence is at a frequency The cumulative error of the filter under DFree and IFree smoothing is 0. and It is a term related to multipath error. and It indicates factors such as receiver thermal noise. Indicates the frequency The antenna group delay under and They represent the code deviation of the receiver and satellite RF front end at this frequency respectively. Indicates the frequency The change of the antenna phase center under . and Indicates the phase deviation between the receiver and the satellite RF front end, The combined effect of these parameters determines the positioning accuracy of aviation navigation.

[0114] Taking the difference in carrier phase measurement between the L1 and L5 frequencies yields:

[0115]

[0116] in represents the sum of carrier phase thermal noise and multipath error, and represent the differential phase deviations of the receiver and the satellite respectively. represents the combined form of the ambiguity after dual-frequency carrier differencing. Similar to the pseudorange case, the difference between the carrier phase measurements at the L1 and L5 frequencies can be used to derive an expression that separates the observables and isolates the noise term on the right side of the equation:

[0117]

[0118] (4)

[0119] Carrier phase ambiguity , Differential Phase Center Change , and the differential phase deviation are known quantities contained in the measurement equation. If these values ​​can be determined, they can be used to Calculation, thus estimating This estimate is affected only by thermal noise and multipath errors in the carrier phase, which are typically very small. Carrier tracking errors due to thermal noise are typically on the millimeter scale, while carrier phase multipath errors are limited to a physical quarter wavelength (approximately ±2.5 cm at the L1 frequency). In situations where all multipath signal powers are lower than the direct signal power, the impact of multipath is typically on the centimeter scale or less. Phase center variations typically fluctuate within a few centimeters or less. As long as carrier phase tracking is not interrupted, the carrier phase ambiguity remains constant.

[0120] The estimate will 、 、 and are considered as known quantities. However, they are usually not known quantities. One possible strategy is to estimate these quantities first and then use the differences in carrier phase observations to calculate Ionospheric studies often use carrier phase to code balance (CCL) for estimation. Some key concepts can be borrowed to derive ionospheric estimates based primarily on carrier measurements.

[0121] Consider simplifying the differential observation:

[0122] (5)

[0123] (6)

[0124] Note the error term above 、 are random and zero-mean, and represent the sum of thermal noise and multipath errors for pseudorange and carrier phase, respectively. Therefore, averaging over time will reduce the impact of these errors.

[0125] Combining the carrier difference with the pseudorange difference yields:

[0126]

[0127] (7)

[0128] ;

[0129] in, represents the antenna group delay variation;

[0130] Thus, the ionospheric term can be eliminated, and the remaining are noise and bias terms. It can be averaged over time using the above formula, so after a long period of averaging we can get:

[0131]

[0132] (8).

[0133] In a typical CCL, the ionospheric observations are calculated as follows:

[0134] (twenty one)

[0135] By rearranging the above formula, the estimated ionospheric delay is:

[0136]

[0137] (9)

[0138] The above formula shows that the carrier-code average ionospheric delay estimate will still contain the antenna group delay term and the differential code bias between the receiver and the satellite, but the carrier-related noise and multipath are reduced. will be projected into the clock error term, The estimated It can be used to make differential estimates of the ionospheric differences between the airborne terminal and the ground station:

[0139]

[0140] (10)

[0141] in, is the receiver differential airborne code deviation, the receiver differential ground code deviation is the noise term, is the ground-side antenna group delay characteristic, is the group delay characteristic of the airborne antenna, is the carrier phase differential ground-side ionospheric gradient estimation, The ionospheric gradient estimation of the airborne side at the L1 frequency point;

[0142] It should be noted that here, the satellite differential code bias has been eliminated. However, there is antenna group delay in both the airborne antenna and the ground reference station antenna. The two additional terms and should be universal across all satellites, and when the test statistic is projected into the position domain, these bias terms will be projected into the clock error state. Finally, the antenna group delay term will still be The noise term also includes and , the thermal noise and multipath effects of the ground station and airborne receiver are at the centimeter level. Finally, the noise term is related to The averaging time scale reduces the noise, but does not eliminate it completely. However, compared to the ionospheric estimates based on pseudoranges, and The proportion of will decrease. Both of the above methods will have a differential group delay term in the test statistic. To improve the accuracy of ionospheric gradient estimation, the ionospheric delay of the L1 frequency can be estimated as accurately as possible at the ground station and broadcast to the airborne terminal.

[0143] Since the receiver DCB term in (9) is common across all satellites, the DCB term can be jointly solved using weighted least squares or multi-epoch filters to improve the accuracy of ionospheric delay estimation and eliminate the impact of DCB.

[0144] The current GAST X proposal is to have the ground station estimate the ionospheric delay on the L1 frequency as accurately as possible and uplink this value to MT23. The ground station should understand the antenna group delay characteristics , and over time, the ionospheric delay can be accurately estimated. By providing this measurement to the airborne system, using VHF Data Broadcast (VDB), the airborne system can use equation (9) to estimate the ionospheric delay combination for each satellite observed by the ground reference receiver. Equation (9) is reconstructed as follows:

[0145] (11)

[0146] in, It represents the observed value of the wave phase at the L5 frequency. Represents the observed value of the carrier phase at L1 frequency.

[0147] Airborne equipment can calculate an estimate of the ionospheric delay seen by the ground station using the following expression:

[0148] (12)

[0149] The accuracy of this ionospheric estimate can be inferred from the ground uplink's Sigma_Iono_L1 parameter. Airborne systems can also use the CCL process described above to estimate their own ionospheric delay. However, this requires a long averaging time to converge to , and the air receiver does not understand characteristics of ground stations and therefore cannot mitigate this error source as effectively as ground stations.

[0150] Step 2: IGM detection statistical error characteristics based on carrier phase

[0151] Over a period of time, through The variables are averaged and the test statistic is calculated as:

[0152] (13)

[0153] in After averaging the pseudorange and carrier phase noise of L1 and L5, The residual error in is obtained by averaging the data for N seconds. Examining the other terms in (13) reveals other deviations, and will be projected into the clock state again or will be eliminated after the air and ground estimates are differencing. Therefore, the only term that will contribute to the residual is the one after averaging , antenna group delay variation , and carrier phase tracking noise The carrier phase tracking noise should be in the millimeter range and will be ignored in the rest of this analysis, including the effect of carrier phase tracking noise on Therefore, there are two terms that contribute to the error, one is the noise-like , and the other is like We will assume that the noise-like errors are effectively reduced by long-term averaging, while the slowly varying biases are not. We will assume that the DUFMAN model is sufficient to accurately describe the L1 and L5 Therefore, it is necessary to Describe the statistical characteristics.

[0154] calculate One way to do this is to run a recursive filter of the following form and substitute equation (6) into the following:

[0155] (14)

[0156] in Representing the pseudorange and carrier phase noise at two frequencies, these errors resemble noise and will be reduced by averaging. represents the differential pseudorange, Represents the differential carrier phase. is a complex ambiguity term in the dual frequency carrier process, which should be a constant and will not be reduced by averaging as long as carrier tracking is maintained. Therefore, only the noise term and will be reduced by averaging. Again, ignore , because it is very small, assuming that thermal noise and multipath noise are white Gaussian noise, in the In the iteration The variance of the noise is given by equation (15):

[0157] (15)

[0158] The standard deviation of the error is given by the following equation:

[0159] (16)

[0160] in is the standard deviation describing the combination of thermal noise and multipath errors on L1 and L5. Seconds averaging actually reduces the standard deviation of code and multipath errors by Unfortunately, multipath is not truly white Gaussian noise, so The factor is too optimistic. As mentioned earlier, the extent to which longer smoothing times will reduce multipath noise remains an open question. For the purposes of this study, we assume that the multipath errors are Gaussian white noise and that their associated standard deviations are, on average, Zoom.

[0161] Applicable to aviation navigation multipath suppression mitigation algorithm, the specific calculation steps are as follows:

[0162] Step 1: Establish multipath error characteristic matrix

[0163] To construct the multipath error feature matrix, the selected observation features can be divided into three categories: one is closely related to multipath error studied by previous researchers, such as satellite elevation and azimuth angles, signal strength, and observation residuals; another is GNSS time, which reflects the temporal characteristics of the error; and finally, the quality characteristics of the observations, such as the number of cycle slips, ambiguity resolution success rate, and ambiguity resolution status. To eliminate the influence of unit and scale differences between observation features and treat all dimensional features equally, the observation features need to be normalized. The normalized dataset is as follows:

[0164] (17)

[0165] in Indicates the The input of samples, that is, the selected observation data features, Indicates the The output of samples is the unmodeled multipath error.

[0166] Step 2: Design a parallel time convolutional neural network

[0167] A parallel time convolutional neural network architecture is designed to suppress multipath online. The network consists of an input layer, 4 convolutional layers, a pooling layer, a flatten layer, a fully connected layer and an output layer, as shown in Figure 3 shown.

[0168] The size of each convolution kernel in the network is , the stride in both directions is set to 1. The convolution layer adopts an effective zero padding strategy, that is, refusing to add zeros, resulting in the length and width of the output data being reduced by 2 after each convolution layer. The number of convolution kernels in each convolution layer is set to 64, 64, 32, and 32 respectively to extract multiple types of features. The convolution operation is as follows:

[0169] (18)

[0170] in Indicates the The convolution layer input feature matrix on each channel, Indicates the number of channels of the convolutional layer input, Indicates the The convolution layer corresponding to the input convolution kernels; represents the convolutional layer The bias of the convolution kernel; It is the convolutional layer The output feature matrix of the convolution kernel. The final Flatten layer is responsible for converting multidimensional data into one-dimensional data. Through each epoch of forward propagation, the value of the loss function can be obtained. The loss function is defined as follows:

[0171] (19)

[0172] in To optimize the parameters, is the number of samples. The parameter information is adjusted by the gradient of the loss function, and the weight is updated using the stochastic gradient descent method, that is:

[0173] (20)

[0174] After obtaining the gradient of the loss function, the parameters are updated and the network training enters the next epoch. Ultimately, the multipath error prediction is achieved using a data-driven approach, effectively alleviating and suppressing multipath.

[0175] Therefore, the present invention adopts the above-mentioned method and device for airborne multi-source integrity monitoring of aviation navigation operation network, and provides an improved navigation technology device and its air-ground coordinated integrity monitoring method to address the shortcomings of existing airborne multi-source aviation navigation operation network in ionospheric gradient monitoring and multipath effect mitigation. By constructing a high-performance navigation receiver architecture and fusing multi-source data, the present invention achieves accurate monitoring of ionospheric gradients and optimizes the switching mechanism between DFree and IFree modes. At the same time, a parallel time convolutional neural network model is proposed for intelligent identification and mitigation of multipath errors, thereby improving the integrity monitoring capability and system availability of the aviation navigation operation network.

[0176] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for airborne multi-source integrity monitoring of an aviation navigation operations network, comprising the following steps: S1. Propose an ionospheric gradient monitoring algorithm based on carrier phase difference, study the statistical error characteristics of IGM detection of carrier phase, establish the ionospheric gradient monitoring threshold, and implement the switching mechanism from DFree to IFree mode onboard; S11. Optimize multiple navigation sources according to different flight phases of route operation and design a flexible integrated navigation algorithm; S12, take the carrier phase observations at the L1 and L5 frequency points, and use the carrier phase difference to estimate the ionospheric gradient on the airborne side; S13, carrier phase difference and pseudo-range residual are combined to eliminate ionospheric delay term, and time average is used to eliminate ; S14, project the receiver differential code bias to the clock error term, eliminate the satellite differential code bias by differentiating it with the ground observation, and then use the carrier phase to code balance CCL to estimate the ionospheric delay in the air ; S15, using the carrier phase observation, by combining pseudorange and carrier phase Variable averaging, constructing the test statistic for ionospheric gradient monitoring, and calculating using recursive filters In the iteration Noise variance and error standard deviation, and calculate the monitoring threshold; S2. Propose an aviation navigation multipath mitigation and suppression algorithm. Combined with the multivariate feature extraction of observation data, the observation data feature matrix is ​​established. A parallel time convolutional neural network architecture is designed. The weights are updated through the stochastic gradient descent method to achieve the prediction and mitigation of multipath errors. S21. Use historical observation data, including observation residual values, signal strength, elevation and azimuth angles, to construct a multipath error feature matrix, eliminate the influence of unit and scale differences between observation features, and normalize the observation features; S22. Design a parallel time convolutional neural network model to model multipath errors offline and suppress multipath online. S23. Use the stochastic gradient descent method to update the neural network parameters, and use the data-driven method to predict multipath and alleviate the multipath effect.

2. The method for airborne multi-source integrity monitoring of an aviation navigation operation network according to claim 1, characterized in that: S11 content is as follows: The ground station uploads the original observation data to the airborne terminal, and uses the pure carrier phase for positioning on the airborne terminal. The observation quantities broadcast by the ground station to the airborne terminal are as follows: （1） （2） in, Represents a specific frequency Pseudorange measurement under ; Represents the carrier phase measurement at this frequency; Indicates the direct path LOS; and are the satellite and receiver clock biases, respectively; and represent the ionospheric and tropospheric delays, respectively; Indicates that the ionospheric divergence is at a frequency The cumulative error of the filter under DFree and IFree smoothing is 0; and is a term related to multipath error; and It represents the receiver thermal noise factor; Indicates the frequency Antenna group delay under ; and Respectively represent the code deviation of the receiver and satellite RF front end at this frequency; Indicates the frequency Changes in the antenna phase center under ; and Indicates the phase deviation between the receiver and the satellite RF front end; Represents the carrier phase ambiguity.

3. The method for airborne multi-source integrity monitoring of an aviation navigation operation network according to claim 2, characterized in that: S12 and S13 contain the following: Taking the difference in carrier phase measurements between the L1 and L5 frequencies, we obtain Equation (3): （3） in represents the sum of carrier phase thermal noise and multipath error; and represent the differential phase deviations of the receiver and the satellite respectively; Indicates the differential phase center change; represents the combined form of the ambiguity after dual-frequency carrier difference; the difference between the carrier phase measurements of L1 and L5 frequencies is used to derive an expression for the ionospheric gradient on the airborne side, as shown in Equation (4), which is used to separate the observable values ​​and isolate the noise term on the right side of the equation: （4） Carrier phase ambiguity , Differential Phase Center Change , and the differential phase deviation are included as known quantities in the measurement equation; The differential observation is simplified as follows: （5） （6） Error term 、 are random and zero-mean, and represent the sum of thermal noise and multipath error of pseudorange and carrier phase, respectively; Combining the carrier difference with the pseudorange difference yields: （7） ； in, represents the antenna group delay variation; Indicates the receiver differential code deviation; Indicates the satellite differential code deviation; The ionospheric term is eliminated and the average term By averaging the above formula over time, we can get: （8）。 4. The method for airborne multi-source integrity monitoring of an aviation navigation operation network according to claim 3, characterized in that: S14 content is as follows: According to the calculation method of ionospheric observation in CCL, the estimated ionospheric delay is: （9） Estimated Used to make differential estimates of the ionospheric differences between the airborne terminal and the ground station: （10） in, is the receiver differential airborne code deviation and the receiver differential ground code deviation, is the noise term, is the ground-side antenna group delay characteristic, is the group delay characteristic of the airborne antenna, is the carrier phase differential ground-side ionospheric gradient estimation, The ionospheric gradient estimation of the airborne side at the L1 frequency point; Since the receiver DCB term in the ionospheric delay is common across all satellites, the DCB term is solved jointly using weighted least squares or multi-epoch filters; Using VHF data broadcast VDB, the airborne system uses Equation (9) to estimate the ionospheric delay combination of each satellite observed by the ground reference receiver. Equation (9) is reconstructed as follows: （11） in, It represents the observed value of the wave phase at the L5 frequency. It represents the observed value of the wave phase at the L1 frequency; The airborne equipment calculates an estimate of the ionospheric delay seen by the ground station using the following expression: （12） The airborne system uses the CCL process described above to estimate the ionospheric delay in the air. .

5. The method for airborne multi-source integrity monitoring of an aviation navigation operation network according to claim 4, characterized in that: S15 content is as follows: Over a period of time, through The variables are averaged and the test statistic is calculated as: （13） in After averaging the pseudorange and carrier phase noise of L1 and L5, The residual error in is obtained by averaging the data for N seconds. ; Run a recursive filter of the form , and substitute equation (6) into the following equation: （14） in represents the pseudorange and carrier phase noise at two frequencies, represents the differential pseudorange, represents the differential carrier phase; is a composite fuzzy term in dual frequency carrier processing, ignoring , assuming that thermal noise and multipath noise are white Gaussian noise, in the In the iteration The variance of the noise is as follows (15): （15） The standard deviation of the error is given by the following equation: （16） in is the standard deviation describing the combination of thermal noise and multipath errors on L1 and L5.

6. The method for airborne multi-source integrity monitoring of an aviation navigation operation network according to claim 5, characterized in that: S21 content is as follows: To construct a multipath error feature matrix, the selected observation features are divided into three categories: the first category is closely related to the multipath error studied by previous researchers; the second category is GNSS time; and the third category is the quality characteristics of the observation values. The observation features are normalized, and the normalized data set is as follows: （17） in Indicates the The input of samples is the selected observation data features; Indicates the The output of samples is the unmodeled multipath error.

7. The method for airborne multi-source integrity monitoring of an aviation navigation operation network according to claim 6, characterized in that: The contents of S22 and S23 are as follows: The parallel time convolutional neural network consists of an input layer, 4 convolutional layers, a pooling layer, a flatten layer, a fully connected layer and an output layer; the size of each convolution kernel in the network is , the stride in both directions is set to 1; the number of convolution kernels in each convolution layer is set to 64, 64, 32 and 32 respectively; the convolution layer The convolution operation is as follows: （18） in Indicates the The convolution layer input feature matrix on each channel; Indicates the number of channels of the convolutional layer input; Indicates the The convolution layer corresponding to the input convolution kernels; represents the convolutional layer The bias of the convolution kernel; It is the convolutional layer The output feature matrix of the convolution kernel; The Flatten layer is used to convert multidimensional data into one-dimensional data. The value of the loss function is obtained through each epoch of forward propagation. The loss function is defined as follows: （19） in To optimize the parameters; m is the number of samples; The parameter information is adjusted through the gradient of the loss function, and the weight is updated using the stochastic gradient descent method, as shown below: （20） After obtaining the gradient of the loss function, the parameters are updated and the network training enters the next epoch.

Images