A multi-modal navigation signal optimization method and device

Abstract

The present invention claims protection for a navigation signal optimization method and device under multimodality, which performs spectrum clustering on the navigation signal to be optimized to obtain multiple navigation signal optimization spectrums to form a navigation signal optimization spectrum feature list; trains a navigation signal optimization model, performs feature recognition on multiple navigation signal optimization spectrums to obtain a semantic coding set of each spectrum; uses a mapping scheduling algorithm to compare the correlation of the semantic coding set of the candidate navigation signal between two spectra in the multiple navigation signal optimization spectrums to obtain a candidate travel path of the candidate navigation signal; splices and restores the candidate travel path to obtain a predicted travel path of the candidate navigation signal. The present invention improves the registration rate, reduces the loss rate caused by optimizer errors and occlusion overlap, uses splicing to optimize the path, uses environmental calculation and historical path information to predict and restore the navigation signal, improves the robustness of the path, and reduces the dependence on hardware.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer-aided navigation signal model analysis, and in particular, relates to a navigation signal optimization method and device under multi-modality. Background Art

[0002] The Global Position System (GPS) combines satellite and wireless technology to provide users with accurate location, speed and time information. Currently, GPS has been widely used in mobile devices on the market, such as mobile phones, personal digital assistants, car computers, etc. In addition to providing traditional positioning functions, such mobile devices equipped with GPS can also realize advanced functions such as route planning, voice navigation, and scenic spot search when combined with electronic maps and navigation software, which can really help users solve travel problems.

[0003] By adjusting the map display scale, the navigation device can only display a portion of the navigation route when executing the navigation function. However, if the user wants to view the entire navigation route during navigation, the map display scale needs to be manually adjusted. Even if the adjusted map can display the entire navigation route, the user still cannot understand the actual road conditions, such as where to turn or where there is a traffic jam. Therefore, it is necessary to provide a navigation route recovery method that can truly reflect the road conditions, so that the user can understand the actual road conditions at any time. Summary of the invention

[0004] In view of the above-mentioned defects, the purpose of the present invention is to provide a navigation signal optimization method and device under multimodality, aiming to propose a neural network algorithm to reduce the cost of data set production, improve the comparison accuracy, reduce the impact of occlusion and overlap on candidate movement and reduce the errors caused by optimizer missed detection and false detection, and adopt a non-linear model prediction method to reduce the algorithm's dependence on hardware.

[0005] According to a first aspect of the present invention, the present invention claims protection for a navigation signal optimization method under multi-modality, comprising:

[0006] Performing frequency spectrum clustering on the navigation signal to be optimized to obtain multiple navigation signal optimization frequency spectra, and forming a navigation signal optimization frequency spectrum feature list;

[0007] Training a navigation signal optimization model, performing feature recognition on the plurality of navigation signal optimization spectra, and obtaining a semantic coding set of each spectrum;

[0008] Using a mapping scheduling algorithm, a correlation comparison is performed on the semantic coding sets of the candidate navigation signals between two frequency spectra in the plurality of navigation signal optimization frequency spectra to obtain a candidate traveling path of the candidate navigation signal;

[0009] The candidate travel paths are spliced ​​and restored to obtain a predicted travel path of the candidate navigation signal.

[0010] Furthermore, the mapping scheduling algorithm adopts a multi-strategy comparison method, including a spectrum correlation strategy and a path transition strategy;

[0011] The spectrum correlation strategy is the characteristic correlation of the navigation signal moving from the m-1th spectrum to the mth spectrum through the navigation signal recognition model;

[0012] The path transition strategy is the correlation between the displacement of the navigation signal from the m-1th spectrum to the mth spectrum and the type of travel speed and the change characteristics;

[0013] The importance of the spectrum correlation strategy and the path transition strategy, wherein the importance of the spectrum correlation strategy is greater than the importance of the path transition strategy.

[0014] Furthermore, the mapping scheduling algorithm is used to compare the correlation of the semantic coding sets of the candidate navigation signals between two spectra in the multiple navigation signal optimization spectra to obtain the candidate travel paths of the candidate navigation signals, including:

[0015] Calculate the jth navigation signal of the m-1th spectrum and the kth navigation signal of the mth spectrum The feature correlation fs obtains the corresponding comparison m s , according to the path transition strategy, the corresponding comparison degree m is obtained v , the final comparison degree M is obtained by summing up the preset importance;

[0016] Combined with the missing spectrum comparison scenario when there is a retained march, the M of all navigation signals in the m-1th spectrum and the navigation signal in the mth spectrum is calculated to obtain the correlation matrix;

[0017] Using the mapping scheduling algorithm to perform mapping scheduling on the navigation signals in the m-1th spectrum and the mth spectrum;

[0018] The mapping scheduling is repeated until all adjacent spectra are compared, and the candidate travel paths of the candidate navigation signals are obtained.

[0019] Furthermore, the characteristic correlation of the navigation signal is calculated, including:

[0020] The distance between the navigation signal l in the m spectrum and the navigation signal k in the m-1 spectrum

[0021] The Euclidean distance is calculated and the calculation expression is as follows:

[0022]

[0023] in, are the coordinate positions of the navigation signal l in the spectrum m respectively; is the coordinate position of navigation signal k in the m-1 spectrum;

[0024] The feature correlation fs uses the dot product operation using the expression:

[0025]

[0026] in, is the data element description of the characteristics of the navigation signal s in the m-1th spectrum, is the data element description of the kth feature of the navigation signal in the mth spectrum, and the value range is optimized to obtain

[0027] Path transition strategy in inferring navigation signals from m-1 spectrum The mth spectrum appears at the point Distance At the same time, consider Changes No deviation occurs in the next spectrum, using the expression:

[0028]

[0029] where τ s ,τ v is a hyperparameter, is the angle between the navigation signal l in the m spectrum and the navigation signal k in the m-1 spectrum.

[0030] According to the importance w s ,w v Add up and get the total comparison degree M = w s m s +w v m v .

[0031] Furthermore, the candidate travel paths are spliced, including:

[0032] Filter the segment paths whose path heads appear in the middle of the map center of the non-candidate spectrum from the candidate travel paths to obtain the segment path feature list U sa = {sa m ,m=1,2,......};

[0033] judge sa m , if the path sa m If there is a lack of spectrum, the new path is disconnected at the lack of spectrum and added to the feature list U sa ;

[0034] in sa m Query the path that misses part of the spectrum and the path that terminates the path in the neighborhood of the head path element to get the feature list For paths where spectra are missing from the feature list, compare the missing parts with sa m Whether the head path element is consistent and the number of missing spectra and sa m Are the lengths the same? If they are the same, keep them in in, otherwise remove;

[0035] Then the feature list The path in the example is used to find the optimal splicing path based on the strategy of the most similar speed change and the shortest starting distance. m Splice it onto the optimal splicing path, and the new path obtained is recorded as the splicing path sa m ;

[0036] judge sa m The end point of the path. If the end point is close to the edge or the spectrum where the end point is located is the last spectrum of the signal, then sa m Path from feature list U sa Remove it from sa m In the next spectrum of the spectrum where the tail path element of is located, check whether there is a path where the head path element is located at the adjacent position of the tail path element. If there is a path, find the optimal splicing path based on the strategy of the most similar speed change and the shortest starting distance. m The fragment path of is spliced ​​with the optimal splicing path, and the spliced ​​path is recorded as the splicing path sa m ;

[0037] Repeat until U sa Remove U when it is empty or the feature list type does not change. sa Path in .

[0038] Furthermore, the restoration process includes:

[0039] Get the path with omissions in the middle of the splicing path. If the mth spectrum is omitted, calculate the navigation signal environment, and obtain the environment change vector and the environment type vector And the speed change vector of the navigation signal and speed type vector

[0040] According to the preset importance, the environment and speed are added to obtain the feature change vector before moving forward. Moving forward feature type vector

[0041] μ1, μ2 are weight coefficients;

[0042] The forward-travel feature is added as a path to the spliced ​​path, and the process is repeated until the entire path is complete to obtain a predicted travel path of the candidate navigation signal.

[0043] According to a second aspect of the present invention, the present invention claims protection for a navigation signal optimization device under multi-modality, comprising: a navigation signal spectrum clustering module, which performs spectrum clustering on the navigation signal to be optimized to obtain multiple navigation signal optimization spectrums to form a navigation signal optimization spectrum feature list;

[0044] A navigation signal feature optimization module is used to train a navigation signal optimization model, perform feature recognition on the plurality of navigation signal optimization spectra, and obtain a semantic coding set of each spectrum;

[0045] A feature similarity comparison module, which uses a mapping scheduling algorithm to compare the relevance of the semantic coding sets of the candidate navigation signals between two spectra in the plurality of navigation signal optimization spectra to obtain a candidate travel path of the candidate navigation signal;

[0046] The path optimization module performs splicing and restoration processing on the candidate travel paths to obtain a predicted travel path of the candidate navigation signal.

[0047] The scheme of the present invention trains a navigation signal candidate recognition model through a neural network algorithm, and uses the model to automatically optimize the navigation signal candidates in each spectrum in the signal sequence, and at the same time saves the feature description data element of each navigation signal candidate for mapping scheduling, uses multi-strategy comparison during mapping scheduling, improves the registration rate, uses a method of retaining the comparison of the progress missing spectrum to reduce the loss rate caused by optimizer errors and occlusion overlap, uses splicing processing to optimize the path, uses environmental calculation and historical path information to predict and restore the navigation signal, improves the robustness of the path and reduces the dependence on hardware. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flowchart of a navigation signal optimization method under multi-mode according to an embodiment of the present invention;

[0049] Figure 2 A second working flow chart of a navigation signal optimization method under multi-modality involved in an embodiment of the present invention;

[0050] Figure 3 It is a structural module diagram of a navigation signal optimization device under multi-mode involved in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the accompanying drawings to clearly and thoroughly describe the examples in the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] According to the first embodiment of the present invention, referring to the attached Figure 1 The present invention claims a method for optimizing navigation signals under multi-mode, comprising:

[0053] Performing frequency spectrum clustering on the navigation signal to be optimized to obtain multiple navigation signal optimization frequency spectra, and forming a navigation signal optimization frequency spectrum feature list;

[0054] Training a navigation signal optimization model, performing feature recognition on the plurality of navigation signal optimization spectra, and obtaining a semantic coding set of each spectrum;

[0055] Using a mapping scheduling algorithm, a correlation comparison is performed on the semantic coding sets of the candidate navigation signals between two frequency spectra in the plurality of navigation signal optimization frequency spectra to obtain a candidate traveling path of the candidate navigation signal;

[0056] The candidate travel paths are spliced ​​and restored to obtain a predicted travel path of the candidate navigation signal.

[0057] Furthermore, the mapping scheduling algorithm adopts a multi-strategy comparison method, including a spectrum correlation strategy and a path transition strategy;

[0058] The spectrum correlation strategy is the characteristic correlation of the navigation signal moving from the m-1th spectrum to the mth spectrum through the navigation signal recognition model;

[0059] The path transition strategy is the correlation between the displacement of the navigation signal from the m-1th spectrum to the mth spectrum and the type of travel speed and the change characteristics;

[0060] The importance of the spectrum correlation strategy and the path transition strategy, wherein the importance of the spectrum correlation strategy is greater than the importance of the path transition strategy.

[0061] Furthermore, the mapping scheduling algorithm is used to compare the correlation of the semantic coding sets of the candidate navigation signals between two spectra in the multiple navigation signal optimization spectra to obtain the candidate travel paths of the candidate navigation signals, including:

[0062] Calculate the jth navigation signal of the m-1th spectrum and the kth navigation signal of the mth spectrum The feature correlation fs obtains the corresponding comparison m s, according to the path transition strategy, the corresponding comparison degree m is obtained v , the final comparison degree M is obtained by summing up the preset importance;

[0063] Combined with the missing spectrum comparison scenario when there is a retained march, the M of all navigation signals in the m-1th spectrum and the navigation signal in the mth spectrum is calculated to obtain the correlation matrix;

[0064] Using the mapping scheduling algorithm to perform mapping scheduling on the navigation signals in the m-1th spectrum and the mth spectrum;

[0065] The mapping scheduling is repeated until all adjacent spectra are compared, and the candidate travel paths of the candidate navigation signals are obtained.

[0066] Furthermore, the characteristic correlation of the navigation signal is calculated, including:

[0067] The distance between the navigation signal l in the m spectrum and the navigation signal k in the m-1 spectrum

[0068] The Euclidean distance is calculated and the calculation expression is as follows:

[0069]

[0070] in, are the coordinate positions of the navigation signal l in the spectrum m respectively; is the coordinate position of navigation signal k in the m-1 spectrum;

[0071] The feature correlation fs uses the dot product operation using the expression:

[0072]

[0073] in, is the data element description of the characteristics of the navigation signal s in the m-1th spectrum, is the data element description of the kth feature of the navigation signal in the mth spectrum, and the value range is optimized to obtain

[0074] Path transition strategy in inferring navigation signals from m-1 spectrum The mth spectrum appears at the point Distance At the same time, consider Changes No deviation occurs in the next spectrum, using the expression:

[0075]

[0076] where τ s ,τ v is a hyperparameter, is the angle between the navigation signal l in the m spectrum and the navigation signal k in the m-1 spectrum;

[0077] According to the importance w s ,w v Add up and get the total comparison degree M = w s m s +w v m v .

[0078] The candidate travel paths are spliced, including:

[0079] Filter the segment paths whose path heads appear in the middle of the map center of the non-candidate spectrum from the candidate travel paths to obtain the segment path feature list U sa = {sa m ,m=1,2,......};

[0080] judge sa m , if the path sa m If there is a lack of spectrum, the new path is disconnected at the lack of spectrum and added to the feature list U sa ;

[0081] in sa m Query the path that misses part of the spectrum and the path that terminates the path in the neighborhood of the head path element to get the feature list For paths where spectra are missing from the feature list, compare the missing parts with sa m Whether the head path element is consistent and the number of missing spectra and sa m Are the lengths the same? If they are the same, keep them in in, otherwise remove;

[0082] Then the feature list The path in the example is used to find the optimal splicing path based on the strategy of the most similar speed change and the shortest starting distance. m Splice it onto the optimal splicing path, and the new path obtained is recorded as the splicing path sa m ;

[0083] judge sa m The end point of the path. If the end point is close to the edge or the spectrum where the end point is located is the last spectrum of the signal, then sa m Path from feature list U sa Remove it from sa m In the next spectrum of the spectrum where the tail path element of is located, check whether there is a path where the head path element is located at the adjacent position of the tail path element. If there is a path, find the optimal splicing path based on the strategy of the most similar speed change and the shortest starting distance. mThe fragment path of is spliced ​​with the optimal splicing path, and the spliced ​​path is recorded as the splicing path sa m ;

[0084] Repeat until U sa Remove U when it is empty or the feature list type does not change. sa Path in .

[0085] The above m It represents a superimposed iterative path that needs to be spliced, and therefore represents both the path before splicing and the path after splicing.

[0086] For further information, see Attachment Figure 2 , restore the candidate travel path, including:

[0087] Get the path with omissions in the middle of the splicing path. If the mth spectrum is omitted, calculate the navigation signal environment, and obtain the environment change vector and the environment type vector And the speed change vector of the navigation signal and speed type vector

[0088] According to the preset importance, the environment and speed are added to obtain the feature change vector before moving forward. Moving forward feature type vector

[0089] μ1, μ2 are weight coefficients;

[0090] The forward-travel feature is added as a path to the spliced ​​path, and the process is repeated until the entire path is complete to obtain a predicted travel path of the candidate navigation signal.

[0091] In this embodiment, when the mapping scheduling algorithm is used to perform mapping scheduling on the navigation signals in the m-1th spectrum and the mth spectrum, the following steps are included:

[0092] According to the correlation matrix, the Hungarian algorithm is used to find the best comparison object for each navigation signal. Its best comparison object The comparison degree is compared with the threshold T. When it is greater than the threshold T, Join If the path is smaller than the threshold T, it cannot be compared.

[0093] In case of non-matching:

[0094] Case 1: For the navigation signal in the m-1 spectrum There is no navigation signal in the m-th spectrum for comparison. At this time, due to the actual navigation signal situation, it can also be divided into two cases. In case one, the navigation signal deviates from the center of the map, resulting in inability to compare. At this time, it is judged whether the position of the navigation signal appears at the edge of the center of the map. If so, the navigation signal will no longer proceed. In case two, the optimizer error is undetected, and its position is fixed at the position where the m-1 spectrum is located and proceeds.

[0095] In case two, for the uncompared navigation signals in the m-th spectrum, they are set as new objects to proceed. If no navigation signal for comparison is found in the subsequent h-th spectrum, the object will no longer proceed.

[0096] Repeat the steps until the comparison steps for all navigation signals in the m-1 spectrum are completed.

[0097] Assume that the signal has a total of F spectra. After each spectrum is recognized by Fast-RCNN, a list of navigation signal features NS for each spectrum is obtained. m :

[0098]

[0099] Among them, m = 1, 2,......, F. Proceeding is the process of comparing the navigation signals in NS 1 with NS 2 , NS 2 with NS 3 ,......, NS F-1 with NS F in which s is the optimized navigation signal.

[0100] Let the retained proceeding feature list be

[0101]

[0102] It represents all the navigation signal feature lists that have not found comparison objects so far when comparing NS c-1 with NS c . Among them, c represents the current comparison of NS c-1 with NS c . It means that the a-th navigation signal in the x-th spectrum has not found any comparison objects so far, and c - h < x < c, a < s. When comparing NS c-1 with NS c , all the navigation signals in R c will be compared with NS c . At this time, due to the omission of spectrum comparison, it is necessary to consider how many spectra the navigation signal has actually proceeded during the comparison. The comparison degree m v should be:

[0103]

[0104] For example: Assume No with NS m ,NS m+1 ,......,NS m+4 If the navigation signal of any feature list in the list is successfully matched, it will be added to the reserved moving feature list R m+5 In NS m+4 With NS m+5 When comparing, R m+5 With NS m +x Therefore, when calculating the correlation during the comparison, m v Need to consider In fact, it travels 6 spectrum intervals, so its

[0105] According to the obtained navigation signal path, it can be used for navigation signal analysis. First, the navigation signal obtained by the navigation signal recognition model can obtain the number of navigation signals traveling in the center of the map during the signal time, which can assist the optimization personnel in counting. Secondly, the navigation signal's travel parameters can be calculated according to the path, and the navigation signal's travel category and travel speed can be determined. It is even possible to segment the optimized navigation signal and then perform morphological discrimination on it to reduce subjective errors in automatic recognition.

[0106] Embodiment 2:

[0107] Refer to the attached Figure 3 , a navigation signal optimization device under multi-mode, comprising:

[0108] The navigation signal spectrum clustering module performs spectrum clustering on the navigation signal to be optimized to obtain multiple navigation signal optimization spectrums to form a navigation signal optimization spectrum feature list;

[0109] A navigation signal feature optimization module is used to train a navigation signal optimization model, perform feature recognition on the plurality of navigation signal optimization spectra, and obtain a semantic coding set of each spectrum;

[0110] A feature similarity comparison module, which uses a mapping scheduling algorithm to compare the relevance of the semantic coding sets of the candidate navigation signals between two spectra in the multiple navigation signal optimization spectra to obtain a candidate travel path of the candidate navigation signal;

[0111] The path optimization module performs splicing and restoration processing on the candidate travel paths to obtain a predicted travel path of the candidate navigation signal.

[0112] Those skilled in the art will appreciate that the contents disclosed in this disclosure may be subject to various modifications and improvements. For example, the various devices or components described above may be implemented by hardware, or by software, firmware, or a combination of some or all of the three.

[0113] Flowcharts are used in this disclosure to illustrate the steps of the method according to the embodiments of the present disclosure. It should be understood that the preceding or following steps are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously. At the same time, other processing may also be added to these processes.

[0114] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of a software functional module. The present disclosure is not limited to any particular form of combination of hardware and software.

[0115] Unless otherwise defined, all terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the present disclosure belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or highly formal sense, unless explicitly defined as such herein.

[0116] The above is an explanation of the present disclosure and should not be considered as a limitation thereof. Although several exemplary embodiments of the present disclosure are described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present disclosure. Therefore, all such modifications are intended to be included within the scope of the present disclosure as defined in the claims. It should be understood that the above is an explanation of the present disclosure and should not be considered to be limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present disclosure is defined by the claims and their equivalents.

[0117] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0118] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A navigation signal optimization method under multi-mode, characterized in that: include: Performing frequency spectrum clustering on the navigation signal to be optimized to obtain multiple navigation signal optimization frequency spectra, and forming a navigation signal optimization frequency spectrum feature list; Training a navigation signal optimization model, performing feature recognition on the plurality of navigation signal optimization spectra, and obtaining a semantic coding set of each spectrum; Using a mapping scheduling algorithm, a correlation comparison is performed on the semantic coding sets of the candidate navigation signals between two frequency spectra in the plurality of navigation signal optimization frequency spectra to obtain a candidate traveling path of the candidate navigation signal; The mapping scheduling algorithm adopts a multi-strategy comparison method, including a spectrum correlation strategy and a path transition strategy; the path transition strategy is the correlation between the displacement of the navigation signal from the m-1th spectrum to the mth spectrum and the travel speed type and change characteristics; The using a mapping scheduling algorithm to compare the relevance of the semantic coding sets of the candidate navigation signals between two spectra in the multiple navigation signal optimization spectra to obtain the candidate travel paths of the candidate navigation signals includes: Calculate the jth navigation signal of the m-1th spectrum and the kth navigation signal of the mth spectrum The feature correlation Get the corresponding comparison , according to the path transition strategy, the corresponding comparison degree is obtained , the final comparison degree is obtained by summing up the preset importance ; Combined with the missing spectrum comparison scenario when there is a retained march, calculate the ratio of all navigation signals in the m-1th spectrum to the navigation signal in the mth spectrum. Get the correlation matrix; Using the mapping scheduling algorithm to perform mapping scheduling on the navigation signals in the m-1th spectrum and the mth spectrum; Repeat the mapping scheduling until all adjacent spectra are compared to obtain the candidate travel path of the candidate navigation signal; The candidate travel paths are spliced ​​and restored to obtain a predicted travel path of the candidate navigation signal.

2. The navigation signal optimization method under multi-mode according to claim 1, characterized in that: The spectrum correlation strategy is the characteristic correlation of the navigation signal moving from the m-1th spectrum to the mth spectrum through the navigation signal recognition model; The importance of the spectrum correlation strategy and the path transition strategy, wherein the importance of the spectrum correlation strategy is greater than the importance of the path transition strategy.

3. The navigation signal optimization method under multi-mode according to claim 2, characterized in that: Calculate the characteristic correlation of the navigation signal, including: Navigation signal in m-spectrum and navigation signals in the m-1 spectrum Distance The Euclidean distance is calculated and the calculation expression is as follows: ； in, are respectively the navigation signals in the m spectrum The coordinate position of , is the navigation signal in the m-1 spectrum The coordinate position of Feature Correlation Using the dot product operation, we use the expression: ； in, is the navigation signal in the m-1th spectrum The metadata description of the feature, is the navigation signal in the mth spectrum The data element description of the feature is obtained by optimizing the value range ; Path transition strategy in inferring navigation signals from m-1 spectrum The mth spectrum appears at the point Distance At the same time, consider Changes No deviation occurs in the next spectrum, using the expression: ； in is a hyperparameter, is the navigation signal in the m spectrum and navigation signals in the m-1 spectrum The angle value of .

4. The multi-modal navigation signal optimization method according to claim 3, characterized in that: The candidate travel paths are spliced, including: Filter the segment paths whose path heads appear in the middle of the map center of the non-candidate spectrum from the candidate travel paths to obtain a segment path feature list ; judge , if the path If there is a lack of spectrum, disconnect the missing spectrum to obtain a new path and add it to the feature list ; exist Query the path that misses part of the spectrum and the path that terminates the path in the neighborhood of the head path element to get the feature list For the paths with missing spectra in the feature list, compare the missing parts with Whether the first path element is consistent and the number of missing spectra and Are the lengths the same? If they are the same, keep them in in, otherwise remove; Then the feature list The path in the example is used to find the optimal splicing path based on the strategy of the most similar speed change and the shortest starting distance. Splice it onto the optimal splicing path, and the new path obtained is recorded as the splicing path ; judge The end point of the path. If the end point is close to the edge or the spectrum where the end point is located is the last spectrum of the signal, Path from feature list Remove it from In the next spectrum of the spectrum where the tail path element of is located, check whether there is a path where the head path element is at the adjacent position of the tail path element. If there is a path, find the optimal splicing path based on the strategy of the most similar speed change and the shortest starting distance. The fragment path of is spliced ​​with the optimal splicing path, and the spliced ​​path is recorded as the splicing path ; Repeat until Removed when empty or feature list type no longer changes Path in .

5. The multi-modal navigation signal optimization method according to claim 4, characterized in that: The recovery process includes: Get the path with omissions in the middle of the splicing path. If the mth spectrum is omitted, calculate the navigation signal environment, and obtain the environment change vector and the environment type vector And the speed change vector of the navigation signal and velocity type vector ; According to the preset importance, the environment and speed are added to obtain the feature change vector before moving forward. , the forward feature type vector ; is the weight coefficient; The forward-travel feature is added as a path to the spliced ​​path, and the process is repeated until the entire path is complete to obtain a predicted travel path of the candidate navigation signal.

6. A navigation signal optimization device under multi-mode, characterized in that: include: The navigation signal spectrum clustering module performs spectrum clustering on the navigation signal to be optimized to obtain multiple navigation signal optimization spectrums to form a navigation signal optimization spectrum feature list; A navigation signal feature optimization module is used to train a navigation signal optimization model, perform feature recognition on the plurality of navigation signal optimization spectra, and obtain a semantic coding set of each spectrum; A feature similarity comparison module, which uses a mapping scheduling algorithm to compare the relevance of the semantic coding sets of the candidate navigation signals between two spectra in the multiple navigation signal optimization spectra to obtain a candidate travel path of the candidate navigation signal; A path optimization module performs splicing and restoration processing on the candidate travel paths to obtain a predicted travel path of the candidate navigation signal; To implement the navigation signal optimization method under multimodality as described in any one of claims 1-5.

Images