GNSS-RTK multi-model cascade robust method
Through CEEMDAN and intelligent optimized wavelet packet decomposition technology, the problems of multipath error and random noise in GNSS-RTK measurement are solved, and the efficient filtering of the signal is realized, which improves the positioning accuracy and the accuracy of monitoring results.
Patent Information
- Application Number
- CN202410985902.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-12-03
AI Technical Summary
There are multipath errors and random noise errors in GNSS-RTK measurements, which affect the positioning accuracy and the accuracy of monitoring results. It is difficult for the prior art to effectively weaken these errors.
Adaptive noise-complete ensemble empirical modal decomposition (CEEMDAN) and intelligently optimized wavelet packet decomposition threshold noise reduction technology, the signal is decomposed through CEEMDAN and the wavelet packet threshold is optimized by using the marine predator algorithm to achieve high-efficiency filtering of multipath errors and random noise deal with.
It effectively improves the clarity and accuracy of the signal, automatically finds the optimal threshold, improves the efficiency and accuracy of noise reduction technology, processes signals in depth, and enhances the accuracy and reliability of information.
Smart Images

Figure CN119064963A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of satellite navigation and positioning, and in particular relates to a GNSS-RTK multi-model cascade anti-error method. Background Art
[0002] The Global Navigation Satellite System (GNSS) is widely used in navigation and positioning, large civil structure monitoring, surveying and mapping and other related technical fields due to its excellent positioning performance. It has the advantages of real-time positioning of monitoring points around the clock, high measurement accuracy, simultaneous monitoring of multiple measurement points, and good continuity. In the actual monitoring process, there are many measurement errors in GNSS measurement, such as satellite ephemeris error, satellite and receiver clock error, ionosphere error, troposphere error, multipath delay error, receiver noise, etc. GNSS-RTK is a real-time dynamic differential positioning technology using the carrier phase double difference model. In the case of a short baseline, the satellite ephemeris error, troposphere and current layer error, satellite and receiver species error, etc. can be basically eliminated by using the correlation of errors and the double difference model. However, errors with weak correlation cannot be eliminated, such as the multipath error caused by reflectors near the monitored object and the random noise error generated by the receiver itself, which become important factors affecting positioning accuracy and monitoring results. After the pseudo signal generated by the multipath effect is received by the receiver, it will directly interfere with the received satellite signal, thus affecting the positioning accuracy and the true value of the monitoring result. Studies have shown that the multipath error generated by the reflection source close to the monitored object is low-frequency, while the multipath error generated by the reflection source farther away is high-frequency; the random noise generated by the receiver itself is distributed in a wider frequency band and generally has low energy. The random noise and general signals can be distinguished by the statistical characteristics of the random signal. At present, the main methods for reducing multipath errors include measurement point location selection (selecting a relatively weak reflection environment), receiver hardware improvement (improvement of receiving antenna and internal phase-locked loop) and software post-processing (data processing and filtering methods). Among them, the software post-processing method can greatly reduce the impact of the multipath effect while saving costs, so it is more widely used in the multipath error correction of monitoring systems. In software post-processing, wavelet packet decomposition and noise reduction have a good separation effect on the signal mutation part and the noise component, and have a better ability to analyze white noise; empirical mode decomposition is a way to adaptively process non-stationary signals. It can decompose the signal into a series of characteristic signals with different components from high frequency to low frequency according to the signal characteristics without setting the number of decomposition layers in advance. Wavelet packet decomposition and empirical mode decomposition can analyze the signal components more carefully, and can filter out noise information while retaining the real information of the signal. Therefore, they have excellent performance in processing multipath errors and random noise. Summary of the invention
[0003] In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a GNSS-RTK multi-model cascade anti-error method.
[0004] To achieve the above object and other related objects, the present invention provides a GNSS-RTK multi-model cascade anti-error method, which includes the following steps:
[0005] Step 1: Import the observed GNSS-RTK monitoring result data x(t);
[0006] Step 2: Establish CEEMDAN parameter conditions, perform CEEMDAN decomposition on the monitoring results, and obtain a series of IMF components;
[0007] Step 3: Calculate the mutual correlation coefficient ρ between each IMF component and x(t), and divide the IMF according to the division criteria. Directly remove the low-frequency components that are less than the threshold; for the remaining parts, perform wavelet packet denoising on the noise-dominated part according to the cross-correlation extreme value judgment condition, and directly participate in the reconstruction of the signal-dominated part without processing;
[0008] Step 4: Decompose the noise-dominated IMF components by wavelet packets to obtain wavelet packet coefficients, pre-process the decomposed coefficients through the energy distribution of wavelet packet decomposition coefficients and determine the level to be denoised;
[0009] Step 5: According to the decomposed coefficients, the thresholds of each wavelet packet are established, and the parameters of the improved wavelet threshold function with parameters are optimized through the marine predator intelligent optimization algorithm, the optimal threshold function parameters are established, and the wavelet packet coefficients are thresholded. The denoised wavelet packet coefficients are inversely operated to obtain the denoised IMF components;
[0010] Step 6: Reconstruct the signal-dominated IMF component and the noise-dominated IMF component denoised by wavelet packets to obtain the denoised monitoring signal.
[0011] Preferably, CEEMDAN input parameters are established, CEEMDAN parameter conditions are established, and CEEMDAN decomposition is performed on the monitoring results to obtain a series of IMF components.
[0012] The CEEMDAN decomposition method is to add adaptive white noise to the original signal and calculate the unique residual component;
[0013] The IMF components are obtained by initializing the first-order mode and performing EMD decomposition on the noisy signal.
[0014] Preferably, the mutual correlation coefficient ρ between the original signal x(t) and each order IMF component is calculated, and the IMF is divided according to the division criteria. The low-frequency components less than the threshold are directly eliminated; the remaining parts are subjected to wavelet packet denoising based on the mutual correlation extreme value judgment conditions, and the signal-dominated parts are directly involved in the reconstruction without processing.
[0015] The principle of the dual-value division criterion of modal components is that when the receiver is close to the reflector, the multipath error appears in the low frequency band, while when it is far away, the multipath error appears in the high frequency band;
[0016] The correlation coefficient is used to obtain a global partition threshold δ, and the weakly correlated low-frequency IMF coefficients less than the threshold δ are eliminated to obtain the remaining components.
[0017] Preferably, the noise-dominated IMF components are subjected to wavelet packet decomposition to obtain wavelet packet coefficients.
[0018] Preferably, the wavelet packet decomposition coefficient energy distribution pre-processes the decomposed coefficients and establishes the level to be denoised.
[0019] Preferably, risk assessment is performed based on a series of processing of the decomposed coefficients to establish the thresholds of each wavelet packet.
[0020] Initializing the population number in the marine predator algorithm, the maximum number of iterations, and establishing the optimal threshold function parameters;
[0021] Preferably, threshold processing is performed on the wavelet packet coefficients to obtain a denoised signal, and the root mean square error of the wavelet packet is calculated as its fitness function.
[0022] The fitness value of the prey matrix in the initial state is calculated and optimized, and then the prey position is transmitted to the predator, and an elite matrix is constructed;
[0023] Preferably, the main loop part of the optimization algorithm updates the prey position, calculates the fitness value, updates the elite matrix, completes memory storage and solves the FADs effect, performs inverse operation on the denoised wavelet packet coefficients, and obtains the denoised IMF components through a three-stage search formula.
[0024] Preferably, the signal-dominated IMF component and the noise-dominated IMF component denoised by wavelet packets are reconstructed to obtain a noise-reduced monitoring signal.
[0025] As described above, the GNSS-RTK multi-model cascade anti-error method proposed in the present invention has the following beneficial effects:
[0026] (1) In signal processing, the present invention adopts the CEEMDAN decomposition method, which adds adaptive white noise to the original signal and calculates the unique residual component. This method can almost perfectly realize the decomposition and reconstruction of the signal to the greatest extent and solve the problem of false modes caused by additional modes in the EEMD algorithm.
[0027] (2) In terms of model denoising, the wavelet packet decomposition denoising used has a good separation effect on the signal mutation part and the noise component, and has a better ability to analyze white noise. Compared with wavelet analysis, wavelet packets have more detailed local analysis capabilities and can further subdivide the high-frequency part. Therefore, the signal can be decomposed into high-frequency and low-frequency components at multiple levels, reflecting its excellent application value.
[0028] (3) In terms of algorithm, based on the idea of cascaded robustness, the monitoring results are decomposed by CEEMDAN to obtain a series of IMF components. The noise-dominated IMF components are decomposed by wavelet packets to obtain wavelet packet coefficients. The decomposed coefficients are preprocessed through the energy distribution of wavelet packet decomposition coefficients and the level to be denoised is determined. The parameters of the improved parameterized wavelet threshold function are optimized using the marine predator intelligent optimization algorithm, the optimal threshold function parameters are determined, and the wavelet packet coefficients are thresholded. The denoised wavelet packet coefficients are inversely operated to obtain the denoised IMF components.
[0029] (4) The proposed GNSS-RTK multi-model cascade anti-error method effectively solves the multipath effect error and random error problems. Through the adaptive noise complete set empirical mode decomposition (CEEMDAN) and intelligent optimization wavelet packet decomposition threshold denoising technology, these noises and errors can be effectively filtered and processed, improving the clarity and accuracy of the signal. The difference from the traditional method is that the traditional denoising technology often requires manual setting of the threshold, while this method introduces an intelligent optimization algorithm that can automatically find the optimal threshold, greatly improving the efficiency and accuracy of the denoising technology. Therefore, through this cascade processing method, deeper signal processing can be achieved, and the accuracy and reliability of the information in the signal can be improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] In order to further illustrate the contents described in the present invention, the specific embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. It should be understood that these drawings are only used as typical examples and should not be regarded as limiting the scope of the present invention.
[0031] Figure 1 It is a flow chart of the GNSS-RTK multi-model cascade anti-error method provided by the present invention;
[0032] Figure 2 It is the overall flow chart of the multi-model cascade robustness algorithm;
[0033] Figure 3 It is the flow chart of the adaptive noise complete set empirical mode decomposition algorithm;
[0034] Figure 4 It is a denoising flow chart with intelligent optimization and weighted modulus average threshold function. DETAILED DESCRIPTION
[0035] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following specific embodiments are only used to explain this application.
[0036] The embodiment of the present invention provides a GNSS-RTK multi-model cascade anti-error method, the specific idea is to effectively filter these noises and errors through the adaptive noise complete set empirical mode decomposition (CEEMDAN) and intelligent optimization wavelet packet decomposition threshold denoising technology, thereby improving the clarity and accuracy of the signal. Specifically, it includes the following steps:
[0037] Step 1: Import the observed GNSS-RTK monitoring result data x(t);
[0038] Step 2: Establish CEEMDAN parameter conditions, perform CEEMDAN decomposition on the monitoring results, and obtain a series of IMF components. The decomposition process is as follows:
[0039] 1) Definition is the i-th order IMF component obtained by EMD decomposition; w j (t) is the jth time Gaussian white noise with mean 0 and variance 1 is added, j = 1, 2, 3, ..., N, N is the number of times white noise is added; ε0 is the standard deviation of the initial Gaussian white noise.
[0040] 2) Initialize the first-order mode. Adding the jth white noise to the monitoring data x(t) to obtain the noise-added signal can be expressed as x(t)+ε0w j (t), and perform EMD decomposition on it to obtain the first-order IMF component. The first-order CEEMDAN component is obtained by averaging N groups of first-order IMF components. The specific formula is as follows:
[0041]
[0042] Calculate the first residual component, the specific formula is as follows:
[0043] r1(t)=x(t)-IMF1(t)
[0044] 3) Add white noise ε1E1(w j (t)), then the specific formula for defining the second-order IMF component is as follows:
[0045]
[0046] Where ε1 is the standard deviation of the first residual component.
[0047] Calculate the second residual component, the specific formula is as follows:
[0048] r2(t)=r1(t)-IMF2(t)
[0049] 4) Similarly, the i-th residual signal is calculated as:
[0050] r i (t) = r i-1 (t)-IMF i (t)
[0051] The i+1th order CEEMDAN component is calculated as:
[0052]
[0053] 5) Repeat step 4) until it cannot be decomposed or the stopping condition is reached, and finally I IMF components and a unique residual component are obtained. The formula of the unique residual component is as follows:
[0054]
[0055] The CEEMDAN decomposition method is to add adaptive white noise to the original signal and calculate the unique residual component, such as Figure 3 It is the flow chart of the adaptive noise complete set empirical mode decomposition algorithm.
[0056] Step 3: Calculate the mutual correlation coefficient ρ between each IMF component and x(t), and divide the IMF according to the division criteria. The low-frequency components less than the threshold are directly eliminated; the remaining parts are subjected to wavelet packet denoising based on the cross-correlation extreme value judgment conditions, and the signal-dominated parts are directly involved in the reconstruction without processing.
[0057] Calculate the mutual correlation coefficient ρ between the original signal x(t) and each order IMF component i , the specific formula is as follows:
[0058]
[0059] The global partition threshold δ is obtained according to the correlation coefficient. The weakly correlated low-frequency IMF coefficients less than the threshold δ are eliminated (mainly multipath errors) to obtain the remaining components {IMF1, IMF2, ..., IMF n}.
[0060] where δ = max(ρ i ) / γ, γ is taken as 8.
[0061] The IMF component corresponding to the first local extreme point in the mutual correlation coefficient is used as the dividing point. The IMF component is the dividing line between the noise-dominated and signal-dominated analysis. The noise-dominated component is subjected to wavelet packet analysis and denoising; the signal-dominated component is not processed.
[0062] Step 4: Decompose the noise-dominated IMF components by wavelet packets to obtain wavelet packet coefficients, pre-process the decomposed coefficients through the energy distribution of wavelet packet decomposition coefficients and establish the level to be denoised.
[0063] The components divided by the correlation coefficient are decomposed by wavelet packets respectively, and the decomposition process is as follows:
[0064] 1) Set The i-th wavelet packet coefficient of the signal on the j-th layer. s-1} are assigned to the top wavelet coefficients respectively Initialize the number of decomposition layers It means that the integer M is the number of collection points.
[0065] 2) For the top wavelet coefficients Perform wavelet packet decomposition respectively. The specific formula of wavelet packet decomposition process is as follows:
[0066]
[0067] where i=1,2,...,2 j is the number of nodes in the jth layer; j = 1, 2, ..., J-1 is the number of wavelet packet decomposition layers; h0 and g0 are the low-pass and high-pass filter coefficients during wavelet packet decomposition, respectively.
[0068] 3) Calculate the energy distribution of wavelet packet coefficients in each layer. The energy of the nth wavelet packet in the jth layer is calculated as:
[0069]
[0070] Calculate the average energy of the j-th layer wavelet packet decomposition coefficients:
[0071]
[0072] 4) Calculate the energy of each layer of wavelet packet coefficients Less than average energy If num is greater than half of the number of wavelet packets in this layer for the first time, this layer will be used as the reconstruction and denoising level. Get each wavelet node on this scale And the wavelet packet coefficients of each node {w1,w2,...,w n}.
[0073] Step 5: According to the decomposed coefficients, establish the threshold of each wavelet packet. The specific steps are as follows:
[0074] 1) The wavelet packet node The coefficients {w1,w2,...,w n}Take the absolute value, then arrange them from small to large, and then square each coefficient to get a new sequence.
[0075]
[0076] 2) Perform risk assessment on f(k):
[0077]
[0078] 3) Calculate the minimum risk value min(Rish(k)) and the corresponding point is recorded as k min , then the threshold is estimated as:
[0079]
[0080] Where σ is the variance estimate of the noise factor. The calculation formula is:
[0081] Similarly, calculate each wavelet packet node The thresholds {λ1,λ2,...,λ i}.
[0082] The parameters of the improved wavelet threshold function with parameters are optimized by the marine predator intelligent optimization algorithm, the optimal threshold function parameters are determined, and the wavelet packet coefficients are thresholded. The denoised wavelet packet coefficients are inversely operated to obtain the denoised IMF components. Figure 4 It is a denoising flow chart with intelligent optimization and weighted modulus average threshold function.
[0083] The specific algorithm steps are as follows:
[0084] 1) Initialize the number of populations l, the maximum number of iterations Max_Iter, the lower boundary lb and upper boundary ub of the threshold function parameters (α, β), and the parameter dimension dim(α, β) in the marine predator algorithm (MPA). The threshold function formula is as follows:
[0085]
[0086] In the formula, α and β are both real numbers greater than 0.
[0087] 2) Use the initial prey matrix P as the parameter value of the threshold function to perform threshold processing on the wavelet packet coefficients to obtain the denoised signal. Calculate the denoised wavelet packet coefficients and the initial wavelet packet coefficients The root mean square error is used as the fitness function. The root mean square error is:
[0088]
[0089] 3) The fitness value of the prey matrix in the initial state is calculated through the fitness function and optimized. The prey position with the best fitness is used as the initial optimal solution of the algorithm. The prey position is then passed to the predator and the elite matrix E is constructed.
[0090] 4) Enter the main loop of the optimization algorithm, search the formula through three stages, update the prey position, calculate the fitness value, update the elite matrix, complete the memory storage and solve the FADs effect. The main loop of the algorithm is as follows:
[0091] ① Update the prey position through three stages of optimization.
[0092] Step 1: Initial iteration period
[0093]
[0094] STEP i =R B *(E i -R B *P i )(i=1,2,...,l)
[0095] P i =P i +0.5R*STEP i
[0096] In the formula, R B is a random number based on normal distribution, representing Brownian motion. R is a uniform random number vector in [0,1]. Iter is the current iteration number, and STEP is the step size of the prey's motion.
[0097] Step 2: Half of the group is used for exploration and the other half is used for development.
[0098] i: For the first half of the population:
[0099]
[0100] STEP i =R L *(E i -R L *P i )(i=1,2,...,l / 2)
[0101]
[0102] ii: For the second half of the population
[0103] STEP i =R B *(R B *E i -P i )(i=l / 2,...,l)
[0104]
[0105] Where R L is a random number vector based on Levy distribution, representing Levy motion. CF is an adaptive parameter that controls the predator movement step size. The expression is as follows:
[0106]
[0107] Step 3: Development stage
[0108]
[0109] STEP i =R L *(R L *E i -P i )(i=1,2,...,l)
[0110]
[0111] ② Calculate the fitness value of the prey position after the update and compare it with the fitness value of the predator position. If the fitness value of the prey position is better than the fitness value of the predator position, update the predator position to the prey position to complete the memory storage.
[0112] ③ Solve the FADs effect of prey position and escape from local optimum.
[0113]
[0114] Where, FADs = 0.2 is the probability that FADs affects the optimization process; U is a binary vector, the array includes 0 and 1. r is a uniform random number in [0, 1]; X min and X maxare the dimensional vectors of the lower and upper bounds of the threshold function parameters respectively; the subscripts r1 and r2 of the prey matrix P represent the random indicators of the prey matrix respectively.
[0115] ④ Determine whether the current number of iterations Iter is greater than the maximum number of iterations Max_Iter. If so, exit the loop and output the optimal threshold function parameters; if not, repeat the above steps ①-③.
[0116] 5) Substitute the optimal threshold parameter into the wavelet packet threshold function to obtain the denoised wavelet packet coefficients Reconstruct the denoised wavelet node coefficients into the original subband signal Finally, the denoised IMF components {IMF1',IMF2',...,IMF s ' -1}. The reconstruction formula is as follows:
[0117]
[0118] Where h1 and g1 are the low-pass and high-pass filter coefficients during reconstruction.
[0119] Step 6: Reconstruct the signal-dominated IMF component and the noise-dominated IMF component denoised by wavelet packets to obtain the denoised monitoring signal.
[0120] The component de-noised by wavelet packet and the component dominated by other signals {IMF k ,IMF k+1 ,...,IMF n}Reconstruct the signal to obtain the monitoring signal after noise reduction. The reconstruction formula is as follows:
[0121]
[0122] Calculate the noise reduction signal x(t) and the original signal The standard root mean square error between is used to evaluate the noise reduction effect. The standard root mean square error formula is as follows:
[0123]
[0124] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. By referring to the detailed description of the embodiments, any person skilled in the art may modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed by the present invention shall still be covered by the claims of the present invention.
Claims
1. A GNSS-RTK multi-model cascade anti-error method uses the cascaded adaptive noise complete ensemble empirical mode decomposition (CEEMDAN) improved intelligent optimized wavelet packet decomposition threshold denoising technology to filter multipath errors and random errors. The method is characterized in that it comprises the following steps: Step 1: Import the observed GNSS-RTK monitoring result data x(t); Step 2: Establish CEEMDAN parameter conditions, perform CEEMDAN decomposition on the monitoring results, and obtain a series of IMF components; Step 3: Calculate the mutual correlation coefficient ρ between each IMF component and x(t), and divide the IMF according to the division criteria. Directly remove the low-frequency components that are less than the threshold; for the remaining parts, perform wavelet packet denoising on the noise-dominated part according to the cross-correlation extreme value judgment condition, and directly participate in the reconstruction of the signal-dominated part without processing; Step 4: Decompose the noise-dominated IMF components by wavelet packets to obtain wavelet packet coefficients, pre-process the decomposed coefficients through the energy distribution of wavelet packet decomposition coefficients and determine the level to be denoised; Step 5: According to the decomposed coefficients, the thresholds of each wavelet packet are established, and the parameters of the improved wavelet threshold function with parameters are optimized through the marine predator intelligent optimization algorithm, the optimal threshold function parameters are established, and the wavelet packet coefficients are thresholded. The denoised wavelet packet coefficients are inversely operated to obtain the denoised IMF components; Step 6: Reconstruct the signal-dominated IMF component and the noise-dominated IMF component denoised by wavelet packets to obtain the denoised monitoring signal.
2. A GNSS-RTK multi-model cascade anti-error method according to claim 1, characterized in that: The step 2 specifically includes: establishing CEEMDAN input parameters, establishing CEEMDAN parameter conditions, and performing CEEMDAN decomposition on the monitoring results to obtain a series of IMF components. The CEEMDAN decomposition method is to add adaptive white noise to the original signal and calculate the unique residual component; The IMF components are obtained by initializing the first-order mode and performing EMD decomposition on the noisy signal.
3. The GNSS-RTK multi-model cascade anti-error method according to claim 2 is characterized in that: The step three is specifically as follows: calculate the mutual correlation coefficient ρ between the original signal x(t) and each order IMF component, and divide the IMF according to the division criteria. Directly remove the low-frequency components that are less than the threshold; for the remaining part, perform wavelet packet denoising on the noise-dominated part according to the mutual correlation extreme value judgment condition, and directly participate in the reconstruction of the signal-dominated part without processing. The principle of the dual-value division criterion of modal components is that when the receiver is close to the reflector, the multipath error appears in the low frequency band, while when it is far away, the multipath error appears in the high frequency band; The correlation coefficient is used to obtain a global partition threshold δ, and the weakly correlated low-frequency IMF coefficients less than the threshold δ are eliminated to obtain the remaining components.
4. The GNSS-RTK multi-model cascade anti-error method according to claim 3 is characterized in that: The step 4 is specifically as follows: The noise-dominated IMF components are subjected to wavelet packet decomposition to obtain wavelet packet coefficients; The energy distribution of the wavelet packet decomposition coefficients pre-processes the decomposed coefficients and establishes the level to be denoised.
5. The GNSS-RTK multi-model cascade anti-error method according to claim 3 is characterized in that: The step five is specifically as follows: The risk assessment is carried out according to a series of processing of the decomposed coefficients to establish the threshold of each wavelet packet; The initialization initializes the population number l in the marine predator algorithm, the maximum number of iterations, and establishes the optimal threshold function parameters; The wavelet packet coefficients are subjected to threshold processing to obtain a denoised signal, and the root mean square error of the wavelet packet is calculated as its fitness function; The fitness value of the prey matrix in the initial state is calculated and optimized, and then the prey position is transmitted to the predator, and the elite matrix E is constructed; The main loop of the optimization algorithm updates the prey position, calculates the fitness value, updates the elite matrix, completes memory storage and solves the FADs effect, performs inverse operation on the denoised wavelet packet coefficients, and obtains the denoised IMF components through a three-stage search formula.
6. The GNSS-RTK multi-model cascade anti-error method according to claim 5, characterized in that: The step six is specifically as follows: reconstructing the signal-dominated IMF component and the noise-dominated IMF component denoised by wavelet packets to obtain a noise-reduced monitoring signal.
Citation Information
Cited By
Multi-source index fused GNSS monitoring data denoising method for strip mine slope
CN119884614A
GNSS (Global Navigation Satellite System) monitoring data abnormal value detection method and device based on CEEMD (Continuous Empirical Mode Decomposition) and local abnormal factors, electronic equipment and storage medium
CN120408128A
