Cycle slip detection method of LSTM-GF combined adaptive threshold

Abstract

The invention relates to the technical field of satellite navigation signal processing, in particular to an LSTM-GF combined adaptive threshold cycle slip detection method, which comprises the following steps: firstly, constructing a GF combined sequence through GNSS carrier phase observation data, and carrying out first-order difference on the sequence to obtain a GF combined cycle slip detection threshold delta GF; then, a predicted TEC fluctuation intensity coefficient value TECVariation is obtained by using an LSTM (Long Short Term Memory) prediction model and inputting a historical TEC sequence value; and finally, constraining a TEC fluctuation intensity coefficient value by using an adjustment coefficient alpha, carrying out weighted average on the GF combination detection threshold value and the TEC fluctuation coefficient value to obtain an adaptive threshold value LGFt of the LSTM-GF combination, if delta GFt is greater than LGFt, determining that cycle slip occurs, and otherwise, determining that cycle slip does not exist. According to the invention, the accuracy of GNSS cycle slip detection of the GF combination under the condition that the ionosphere is active can be improved, and the false detection rate is reduced.

Description

Technical Field

[0001] This invention relates to the field of satellite navigation signal processing technology, and specifically to a cycle slip detection method using LSTM-GF combined adaptive threshold. Background Technology

[0002] With the gradual improvement of global GNSS systems, high-precision GNSS positioning technology has also developed and been applied rapidly. Carrier phase measurement technology plays a crucial role in high-precision GNSS positioning. In carrier phase measurement, the receiver needs to continuously observe satellites to obtain carrier phase observation values ​​for high-precision positioning. However, during the observation process, the complex observation environment, such as obstructions from tall buildings or trees, often causes the receiver to lose lock, resulting in degraded observation data quality. This leads to cycle slips in the received satellite signals, causing errors in subsequent integer ambiguity calculations and a significant decrease in positioning accuracy.

[0003] To address this technical challenge, many researchers have proposed various cycle slip detection methods, but certain limitations remain. The polynomial fitting method requires the absence of cycle slips within the fitting window for accurate detection; its detection capability is limited by the sampling rate and observation errors of the carrier phase observations, making it insensitive to small cycle slips. The higher-order difference method must consider errors caused by clock jumps and epoch intervals, amplifying both cycle slips and noise, limiting its applicability to large cycle slips. The ionospheric residual method is significantly affected by multipath effects and ionospheric transformation rates, enabling the detection of small cycle slips, but it suffers from multi-valuedness. The MW combination method is affected by pseudorange noise, exhibiting inaccuracies in detecting 1-2 small cycle slips and threshold diffusion after failed cycle slip detection. The GF combination method is sensitive to ionospheric abrupt changes; in active ionospheric environments, a fixed detection threshold can easily lead to false positives. The MW and GF combination detection method suffers from a simplistic threshold model, significantly impacting accuracy in complex observation environments and low data sampling rates, potentially causing missed or incorrect cycle slip detections. Summary of the Invention

[0004] The purpose of this invention is to provide a cycle slip detection method with LSTM-GF combined adaptive threshold, which aims to solve the problem of false cycle slip detection caused by the simplification of the cycle slip detection threshold in the GF combination method when the ionosphere is active, thereby improving the cycle slip detection effect and enhancing the positioning accuracy and reliability of carrier phase measurement technology in GNSS.

[0005] To achieve the above objectives, this invention provides a cycle slip detection method using LSTM-GF combined adaptive thresholding, comprising the following steps:

[0006] Step 1: Construct a GF combination sequence using carrier phase observation data, perform first-order difference processing on the sequence, and determine the GF combination detection threshold ΔGF;

[0007] Step 2: Extract historical TEC sequence values ​​from ionospheric GNSS-TEC data and input them into the LSTM prediction model to predict the TEC wave intensity coefficient value TEC_Variation. t ;

[0008] Step 3: Constrain the TEC fluctuation intensity coefficient value TEC_Variation using the adjustment coefficient α. t The adaptive threshold LGF for the LSTM-GF combination is obtained by weighting the GF combination detection threshold and the TEC fluctuation coefficient value. t ;

[0009] Step 4: Perform cycle slip determination: If |ΔGF t |>LGF t If the cycle slip is not detected, it is considered a cycle slip; otherwise, there is no cycle slip.

[0010] Optionally, the GF combination in step 1, i.e., the geometrically independent combination, is constructed using dual-frequency carrier data. The formula for calculating the GF combination observations is as follows:

[0011]

[0012] in, Let λ be the combined observation value of GF. L1 λ is the carrier wavelength at frequency L1. L2 The carrier wavelength at the L2 frequency point, This refers to the integer ambiguity term corresponding to the L1 frequency point. This refers to the integer ambiguity term corresponding to the L2 frequency point. This represents the ionospheric delay error corresponding to the L1 frequency point. This is the ratio of the squared frequencies of the L1 carrier and the L2 carrier;

[0013] The first-order difference is calculated for the GF combination observations at adjacent epochs t and t+1, using the following formula:

[0014]

[0015] Let λ be the first difference component of the GF combined observations at times t and t+1. L1 λ is the carrier wavelength at frequency L1. L2 The carrier wavelength at the L2 frequency point, and These represent the integer ambiguity variations of the L1 and L2 carriers, respectively. This represents the change in the ionosphere between epochs;

[0016] The formula for calculating the GF combined detection threshold ΔGF is as follows:

[0017] ΔGF=f×σ comb

[0018] Where f is the multiplication factor that determines the confidence level, f∈[3,5], σ comb This represents the standard deviation of carrier phase noise.

[0019] Optionally, in step 2, the LSTM unit of the LSTM prediction model is controlled by three gates: the forget gate, the input gate, and the output gate.

[0020] The forgetting gate controls which information in a memory unit should be retained or forgotten. The calculation formula is as follows:

[0021]

[0022] Where σ represents the Sigmoid activation function, which restricts the received values ​​to between 0 and 1; W f and b f Let h represent the weight matrix and bias vector of the forget gate, respectively. t-1 Let x represent the hidden state at the previous time step. t Indicates the current input;

[0023] The input gate determines the extent to which the current input information is written into the memory unit. The input gate calculation formula is as follows:

[0024] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0025] Where σ represents the Sigmoid activation function, which restricts the received values ​​to between 0 and 1; W i and b i Let h represent the weight matrix and bias vector of the input gate, respectively. t-1 Let x represent the hidden state at the previous time step. t Indicates the current input;

[0026] The output gate determines the output value at the current moment, and the calculation formula is as follows:

[0027]

[0028] in, The output gate value is represented by σ, which represents the Sigmoid activation function, and W... o and b oLet h represent the weight matrix and bias vector of the output gate, respectively. t-1 Let x represent the hidden state at the previous time step. t This indicates the current input.

[0029] Optionally, the formula for calculating the TEC volatility coefficient is as follows:

[0030]

[0031] Among them, TEC t and TEC t-1 Let t be the absolute change in TEC between time t and the previous time t-1, W be the window size, and max represent taking the maximum value.

[0032] Optionally, the LSTM-GF combined adaptive threshold calculation formula is as follows:

[0033] LGF t =ΔGF×(1+α×TEC_Variation t )

[0034] =f×σ comb ×(1+α×TEC_Variation t )

[0035] Where f is the multiplication factor used to determine the confidence level in step 1, and σ comb TEC_Variation represents the standard deviation of carrier phase noise, α represents the adjustment coefficient, and TEC_Variation represents the standard deviation of carrier phase noise. t The ionospheric TEC fluctuation coefficient is used to control the degree of influence of ionospheric disturbance on the threshold. The formula for calculating the adjustment coefficient α is as follows:

[0036] α t =α base ×(1+ROTI t )

[0037] Where, α base ROTI is a fixed base value. t The normalized real-time perturbation coefficient;

[0038] ROTI t The calculation formula is as follows:

[0039]

[0040] Where W is the sliding window length, Δt is the sampling interval, and TEC t and TEC t-1 Let be the absolute change of TEC between time t and the previous time t-1.

[0041] This invention provides a cycle slip detection method based on an adaptive threshold using LSTM-GF combined data. First, a GF combined sequence is constructed using GNSS carrier phase observation data, and the sequence is first-order differencing is performed to obtain the cycle slip detection threshold ΔGF for the GF combined sequence. Then, using an LSTM prediction model, the predicted TEC fluctuation intensity coefficient value TEC_Variation is obtained by inputting historical TEC sequence values. t Finally, the TEC fluctuation intensity coefficient value is constrained using the adjustment coefficient α. The adaptive threshold LGF for the LSTM-GF combination is obtained by weighted averaging of the GF combination detection threshold and the TEC fluctuation coefficient value. t If |ΔGF t |>LGF t If the condition is stable, a cycle slip is identified; otherwise, no cycle slip occurs. This invention can improve the accuracy of GNSS cycle slip detection using GF combinations and reduce the false detection rate under ionospheric conditions. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a flowchart illustrating a cycle slip detection method based on an LSTM-GF combined adaptive threshold according to the present invention.

[0044] Figure 2 This is a schematic diagram of the structure of the LSTM prediction model in an embodiment of the present invention. Detailed Implementation

[0045] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0046] This invention provides a cycle slip detection method based on LSTM-GF combined adaptive thresholding, comprising the following steps:

[0047] Step 1: Construct a GF combination sequence using carrier phase observation data, perform first-order difference processing on the sequence, and determine the GF combination detection threshold ΔGF;

[0048] Step 2: Extract historical TEC sequence values ​​from ionospheric GNSS-TEC data and input them into the LSTM prediction model to predict the TEC wave intensity coefficient value TEC_Variation. t ;

[0049] Step 3: Constrain the TEC fluctuation intensity coefficient value TEC_Variation using the adjustment coefficient α. t The adaptive threshold LGF for the LSTM-GF combination is obtained by weighting the GF combination detection threshold and the TEC fluctuation coefficient value. t ;

[0050] Step 4: Perform cycle slip determination: If |ΔGF t |>LGF t If the cycle slip is not detected, it is considered a cycle slip; otherwise, there is no cycle slip.

[0051] The following description, in conjunction with specific embodiments and execution steps, provides further details:

[0052] Please see Figure 1 In this embodiment, carrier phase observation data processing is performed based on data provided by satellite navigation systems including BeiDou Navigation Satellite System (BDS) and Global Positioning System (GPS), and the dataset required to construct the LSTM model is constructed using ionospheric TEC data provided by the International GNSS Service (IGS).

[0053] In step 1, after processing the carrier phase observation data, a GF combination sequence is constructed using the carrier phase observation data. The sequence is then subjected to first-order differential processing to eliminate errors such as satellite-to-ground distance, satellite clock error, receiver clock error, and tropospheric delay corresponding to the L1 and L2 frequency points, and to further reduce ionospheric delay errors. Finally, the GF combination detection threshold ΔGF is determined.

[0054] At time t, the formulas for calculating the carrier phase observations at frequencies L1 and L2 are as follows:

[0055]

[0056] In the above formula, The carrier phase observation value at frequency L1. For the carrier phase observation at frequency L2, λ L1 λ is the carrier wavelength at frequency L1. L2 For the L2 frequency carrier wavelength, ρ t,L1 ρ is the satellite-to-ground distance corresponding to the L1 frequency point. t,L2 The distance between the satellite and Earth corresponding to the L2 frequency point, c is the speed of light, and δt is the distance between the satellite and Earth. r,t,L1 δt represents the receiver clock bias corresponding to frequency L1. r,t,L2 δt represents the receiver clock bias corresponding to the L2 frequency point.s ,t,L1 δt represents the satellite clock bias corresponding to frequency L1. s,t,L2 This refers to the satellite clock bias corresponding to the L2 frequency point. This represents the ionospheric delay error corresponding to the L1 frequency point. This represents the ionospheric delay error corresponding to the L2 frequency point. This represents the tropospheric delay error corresponding to the L1 frequency point. This represents the tropospheric delay error corresponding to the L2 frequency point. This refers to the integer ambiguity term corresponding to the L1 frequency point. For the integer ambiguity term corresponding to the L2 frequency point, ε c,t,L1 For other errors corresponding to the L1 frequency point, ε c,t,L2 Other errors corresponding to the L2 frequency point.

[0057] The GF combination, also known as the geometrically independent combination, is constructed using dual-frequency carrier data. It eliminates satellite orbital errors, receiver and satellite clock errors, and tropospheric errors, but ionospheric delay errors still exist. The GF combination is constructed using the L1 and L2 dual-frequency carrier phase value data. The formula for calculating GF combination observations is as follows:

[0058]

[0059] In the above formula, Let λ be the combined observation value of GF. L1 λ is the carrier wavelength at frequency L1. L2 The carrier wavelength at the L2 frequency point, This refers to the integer ambiguity term corresponding to the L1 frequency point. This refers to the integer ambiguity term corresponding to the L2 frequency point. This represents the ionospheric delay error corresponding to the L1 frequency point. This is the ratio of the squared frequencies of the L1 carrier and the L2 carrier.

[0060] The first-order difference is performed on the GF combination observations at adjacent epoch times t, t+1, and the calculation formula is as follows:

[0061]

[0062] Let λ be the first difference component of the GF combined observations at times t and t+1. L1 λ is the carrier wavelength at frequency L1. L2 The carrier wavelength at the L2 frequency point, and These represent the integer ambiguity variations of the L1 and L2 carriers, respectively. This represents the change in the ionosphere between epochs.

[0063] When the ionosphere remains stable between adjacent epochs: The value is negligible. The residuals of the GF combination without cycle slips follow a Gaussian error:

[0064]

[0065] In the above formula, σ comb This represents the standard deviation of the carrier phase noise. At this point, the GF combination detection threshold ΔGF can be determined as follows:

[0066] ΔGF=f×σ comb

[0067] In the above formula, f is the multiplication factor that determines the confidence level, f∈[3,5], σ comb This represents the standard deviation of the carrier phase noise. The GF combination detection threshold is: when If the threshold is met, a cycle slip is detected. Otherwise, it is not. This threshold can lead to a higher false positive rate when the ionosphere is active.

[0068] Furthermore, in step 2, the dataset constructed using ionospheric TEC data is input into the LSTM prediction model, and the TEC wave intensity coefficient value TEC_Variation is obtained by predicting the data through the LSTM prediction model. t The LSTM prediction model structure is as follows: Figure 2 As shown, the LSTM model is a recurrent neural network model specifically designed to process sequential data. An LSTM unit is controlled by three gates: the forget gate, the input gate, and the output gate.

[0069] The forget gate controls which information in a memory cell should be retained or forgotten. The output of the forget gate. The cell controls the memory passed from the previous node by receiving the hidden state from the previous time step, and at the same time receives the current input to determine which memory needs to be discarded in the current cell state. The calculation formula is as follows:

[0070]

[0071] In the above formula, σ represents the Sigmoid activation function, which restricts the received values ​​to between 0 and 1. W f and b f Let h represent the weight matrix and bias vector of the forget gate, respectively. t-1 Let x represent the hidden state at the previous time step. t This indicates the current input.

[0072] The input gate determines the extent to which the current input information is written into the memory unit. The input gate receives the current input x. t Compared to the hidden state h in the previous time step t-1First, the write probability is generated, and the calculation formula is as follows:

[0073] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0074] In the above formula, σ represents the Sigmoid activation function, which restricts the received values ​​to between 0 and 1. W i and b i Let h represent the weight matrix and bias vector of the input gate, respectively. t-1 Let x represent the hidden state at the previous time step. t This indicates the current input.

[0075] Then, a new candidate value is generated and activated using tanh, calculated as follows:

[0076]

[0077] In the above formula, tanh represents the activation function. c and b c Let h represent the weight matrix and bias vector of the input gate, respectively. t-1 Let x represent the hidden state at the previous time step. t This indicates the current input.

[0078] Updating the cell state combines the results of the forgetting gate and the input gate to update the cell memory. The calculation formula is as follows:

[0079]

[0080] In the above formula, For the forget gate output, C t-1 This is the cellular memory from the previous moment. The candidate values ​​generated by the tanh activation function, i t The write probability generated for the Sigmoid activation function.

[0081] The output gate determines the output value at the current moment. The calculation formula is as follows:

[0082]

[0083] In the above formula, σ represents the Sigmoid activation function, and W o and b o Let h represent the weight matrix and bias vector of the input gate, respectively. t-1 Let x represent the hidden state at the previous time step. tThis represents the current input. The cell state is then processed through the tanh activation function and multiplied by the output gate value to obtain the final hidden state output. The calculation formula is as follows:

[0084]

[0085] In the above formula, C represents the value of the output gate. t Let be the cell state, and tanh be the activation function. The hidden state will be passed as output to the next time step.

[0086] Utilizing these gating structures and cell state functions, LSTM can effectively construct the input ionospheric TEC sequence values. Through continuous updating, iteration, and transmission within the network, the ionospheric TEC fluctuation coefficient value TEC_Variation is obtained. t To make a prediction, the TEC volatility coefficient is calculated using the following formula:

[0087]

[0088] In the above formula, TEC t and TEC t-1 The absolute change in TEC between time t and the previous time t-1, with W being the window size. By normalizing the changes in adjacent TEC, the TEC fluctuation coefficient effectively quantifies the relative intensity of ionospheric disturbances, providing a key input for LSTM-GF adaptive adjustment.

[0089] Figure 2 The LSTM ionospheric TEC fluctuation coefficient prediction model constructed is a time series prediction model. Mean Relative Accuracy (MRA) is used to evaluate the model's performance. MRA is commonly used to measure the overall average degree of the model's predictions relative to the actual values; essentially, it converts the mean absolute percentage error into accuracy. MRA intuitively reflects the proportion of prediction bias as a percentage. A larger MRA indicates a more accurate prediction of the ionospheric TEC fluctuation coefficient. The MRA calculation formula is as follows:

[0090]

[0091] In the above formula, n is the number of TEC sequence data points, and y i This represents the actual TEC sequence observations. This represents the predicted TEC fluctuation coefficient value from the model. MRA (Mechanical Analysis and Responsiveness) can provide a relatively comprehensive evaluation of the performance of the LSTM (Low-Speed ​​Matrix Electro-Ray) ionospheric TEC fluctuation coefficient prediction model.

[0092] Step 3: Use the adjustment coefficient α to constrain the TEC fluctuation intensity coefficient value TEC_Variation tThe adaptive threshold LGF for the LSTM-GF combination is obtained by weighting the GF combination detection threshold and the TEC fluctuation coefficient value. t LSTM-GF Adaptive Combined Threshold LGF t The calculation formula is as follows:

[0093] LGF t =ΔGF×(1+α×TEC_Variation t )

[0094] =f×σ comb ×(1+α×TEC_Variation t )

[0095] In the above formula, f is the multiplication factor for determining the confidence level, and σ comb TEC_Variation represents the standard deviation of carrier phase noise, α represents the adjustment coefficient, and TEC_Variation represents the standard deviation of carrier phase noise. t This represents the ionospheric TEC fluctuation coefficient. The influence of ionospheric disturbances on the threshold is controlled using an adjustment coefficient α. When the ionosphere is calm, α approaches 0, and the LGF... t =ΔGF, maintaining the original GF combination detection threshold; when ionospheric disturbances occur, α increases, LGF t =ΔGF×(1+α×TEC_Variation t The threshold is amplified to avoid false detections due to drastic fluctuations in TEC. The formula for calculating the adjustment coefficient is as follows:

[0096] α t =α base ×(1+ROTI t )

[0097] In the above formula, α base ROTI is a fixed base value. t This is the normalized real-time disturbance coefficient.

[0098] ROTI t The calculation formula is as follows:

[0099]

[0100] In the above formula, W is the sliding window length, Δt is the sampling interval, and TEC is the sampling interval. t and TEC t-1 Let be the absolute change of TEC between time t and the previous time t-1.

[0101] Step 4: Determine the cycle slip: If |ΔGF t |>LGF t If the cycle slip is not detected, it is considered a cycle slip; otherwise, there is no cycle slip.

[0102] Compared with the prior art, the beneficial technical effects of the present invention are as follows:

[0103] This invention comprehensively considers that in practical applications, the fixed cycle slip detection threshold ΔGF of the GF combined cycle slip detection method is prone to false detections during periods of ionospheric activity. Therefore, it proposes using an LSTM prediction model to measure the ionospheric TEC wave intensity coefficient TEC_Variation. t Predictions are made, and the fluctuation intensity coefficient is constrained using an adjustment coefficient. Then, the constrained fluctuation intensity coefficient is weighted and averaged with the original GF combined detection threshold to obtain the adaptive detection threshold LGF. t This can improve the accuracy of GNSS cycle slip detection by GF combination under active ionospheric conditions and reduce the false detection rate.

[0104] The above description discloses only one or more preferred embodiments of the present invention, and should not be construed as limiting the scope of the present invention. Those skilled in the art will understand that all or part of the processes of the above embodiments can be implemented, and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. A cycle slip detection method using LSTM-GF combined adaptive thresholding, characterized in that, Includes the following steps: Step 1: Construct a GF combination sequence using carrier phase observation data, perform first-order difference processing on the sequence, and determine the GF combination detection threshold ΔGF; Step 2: Extract historical TEC sequence values ​​from ionospheric GNSS-TEC data and input them into the LSTM prediction model to predict the TEC wave intensity coefficient value TEC_Variation. t ; Step 3: Constrain the TEC fluctuation intensity coefficient value TEC_Variation using the adjustment coefficient α. t The adaptive threshold LGF for the LSTM-GF combination is obtained by weighting the GF combination detection threshold and the TEC fluctuation coefficient value. t ; Step 4: Perform cycle slip determination: If |ΔGF t |>LGF t If the cycle slip is not detected, it is considered a cycle slip; otherwise, there is no cycle slip.

2. The cycle slip detection method with LSTM-GF combined adaptive threshold as described in claim 1, characterized in that, The GF combination in step 1 is the geometrically independent combination, which is constructed using dual-frequency carrier data. The formula for calculating the GF combination observations is as follows: in, Let λ be the combined observation value of GF. L1 λ is the carrier wavelength at frequency L1. L2 The carrier wavelength at the L2 frequency point, This refers to the integer ambiguity term corresponding to the L1 frequency point. This refers to the integer ambiguity term corresponding to the L2 frequency point. This represents the ionospheric delay error corresponding to the L1 frequency point. This is the ratio of the squared frequencies of the L1 carrier and the L2 carrier; The first-order difference is calculated for the GF combination observations at adjacent epochs t and t+1, using the following formula: Let be the first difference component of the GF combined observations at times t and t+1. and These represent the integer ambiguity variations of the L1 and L2 carriers, respectively. This represents the change in the ionosphere between epochs; The formula for calculating the GF combined detection threshold ΔGF is as follows: ΔGF=f×σ comb Where f is the multiplication factor that determines the confidence level, f∈[3,5], σ comb The standard deviation of the phase observation noise for the GF combined carrier is given.

3. The cycle slip detection method with LSTM-GF combined adaptive threshold as described in claim 2, characterized in that, In step 2, the LSTM unit of the LSTM prediction model is controlled by three gates: the forget gate, the input gate, and the output gate. The forgetting gate controls which information in a memory unit should be retained or forgotten. The calculation formula is as follows: Where σ represents the Sigmoid activation function, which restricts the received values ​​to between 0 and 1; W f and b f Let h represent the weight matrix and bias vector of the forget gate, respectively. t-1 Let x represent the hidden state at the previous time step. t Indicates the current input; The input gate determines the extent to which the current input information is written into the memory unit. The input gate calculation formula is as follows: i t =σ(W i ·[h t-1 ,x t ]+b i ) Where σ represents the Sigmoid activation function, which restricts the received values ​​to between 0 and 1; W i and b i Let h represent the weight matrix and bias vector of the input gate, respectively. t-1 Let x represent the hidden state at the previous time step. t Indicates the current input; The output gate determines the output value at the current moment, and the calculation formula is as follows: in, The output gate value is represented by σ, which represents the Sigmoid activation function, and W... o and b o Let h represent the weight matrix and bias vector of the output gate, respectively. t-1 Let x represent the hidden state at the previous time step. t This indicates the current input.

4. The cycle slip detection method with LSTM-GF combined adaptive threshold as described in claim 3, characterized in that, The formula for calculating the TEC volatility coefficient is as follows: Among them, TEC t and TEC t-1 Let t be the absolute change in TEC between time t and the previous time t-1, W be the window size, and max represent taking the maximum value.

5. The cycle slip detection method with LSTM-GF combined adaptive threshold as described in claim 4, characterized in that, The formula for calculating the adaptive threshold of the LSTM-GF combination is as follows: LGF t =ΔGF×(1+α×TEC_Variation t ) =f×σ comb ×(1+α×TEC_Variation t ) Where f is the multiplication factor used to determine the confidence level in step 1, and σ comb TEC_Variation represents the standard deviation of carrier phase noise, α represents the adjustment coefficient, and TEC_Variation represents the standard deviation of carrier phase noise. t The ionospheric TEC fluctuation coefficient is used to control the degree of influence of ionospheric disturbance on the threshold. The formula for calculating the adjustment coefficient α is as follows: a t =a base ×(1+ROTI t ) Where, α base ROTI is a fixed base value. t The normalized real-time perturbation coefficient; ROTI t The calculation formula is as follows: Where W is the sliding window length, Δt is the sampling interval, and TEC t and TEC t-1 Let be the absolute change of TEC between time t and the previous time t-1.

Images