A composite spatial positioning method for fiber grating sensor networks

Abstract

The present invention belongs to the field of fiber optic sensing technology, and specifically relates to a composite spatial positioning method for fiber optic Bragg grating sensor networks that combines a distributed individual differential evolution (DIDE) method with a four-point positioning (FPP) method (DIDE-FPP). The present invention provides a composite spatial positioning method for a fiber optic Bragg grating sensor network, which solves the three-dimensional spatial positioning problem of a large-capacity fiber optic Bragg grating sensor network by constructing a composite positioning algorithm. The DIDE-FPP method provided by the present invention includes: applying the DIDE method to search for peaks; and performing three-dimensional spatial positioning by the FPP method. The present invention solves the multi-peak optimization problem (MMOPS) existing in the process of tunnel fire positioning by a composite spatial positioning method, and realizes three-dimensional spatial positioning and achieves centimeter-level positioning accuracy by further analyzing and calculating the peak search results.

Description

Technical Field

[0001] The present invention belongs to the technical field of optical fiber sensing, and in particular relates to a composite spatial positioning method for a fiber grating sensor network that combines a distributed individual differential evolution (DIDE) method with a four-point positioning (FPP) method (DIDE-FPP). Background Art

[0002] Fiber optic sensing technology is a sensing technology that uses optical fiber as a medium and light waves as a carrier to sense and transmit external signals (to be measured). Due to the advantages of optical fiber, such as high sensitivity, light weight, high precision, bendability, passivity, anti-electromagnetic interference and strong multiplexing ability, it is widely used in military, aerospace and other fields. The use of online writing of ultra-weak fiber grating arrays can greatly improve the multiplexing ability and monitoring accuracy of the system. However, when ultra-weak grating arrays are applied to tunnel fire monitoring systems, it is difficult to accurately locate due to the multi-peak optimization problem (MMOPS) in the process of tunnel fire positioning. Therefore, solving the multi-peak optimization problem can improve the positioning accuracy of the entire tunnel fire monitoring system.

[0003] At present, the analysis and research on multi-peak optimization problems at home and abroad are constantly improving. Since R.Storn and KVPrice proposed differential evolution DE (differential evolution) in 1997, a new method that requires few control variables, is robust, easy to use, and very suitable for parallel computing, a large number of scholars have made improvements and upgrades based on DE. In 2004, R.Thomsen et al. expanded DE through a crowding scheme to enable it to track and maintain multiple optimal values. This algorithm is called the population-based differential evolution CDE (crowding differential evolution) method. A year later (2005), X.Li et al. also proposed the species-based differential evolution SDE (species-based differential evolution) method based on DE. It can simultaneously locate multiple global optimal values ​​by adaptively forming multiple species (or subpopulations) in the DE population in each iteration step. SDE is more computationally efficient than CDE, but this method still cannot cope with composite functions. In 2020, Zong-Gan Chen et al. proposed the DIDE algorithm based on their predecessors. The algorithm controls the evolution of the virtual population through the adaptive range adjustment (ARA) strategy, solves the diversity problem in the multi-modal optimization problem (MMOP) through the life cycle mechanism, and solves the accuracy problem in the multi-modal optimization problem (MMOP) through the elite learning mechanism (ELM). The four-point positioning (FPP) method is currently widely used in high-precision radio navigation positioning systems based on artificial earth satellites, such as the Global Positioning System (GPS). It can provide accurate geographic location, vehicle speed and precise time information anywhere in the world and in near-Earth space.

[0004] In view of the shortcomings of the above-mentioned method, the present invention uses a method combining DIDE and FPP to accurately identify and locate the ultra-weak grating array in three-dimensional space due to the change of the wavelength of the reflected wave caused by the change of the external temperature environment. The algorithm has a significant effect on the peak-finding and positioning of the dense wave peaks reflected by the high-density grating array. The specific content is as follows. Summary of the invention

[0005] The purpose of the present invention is to provide a composite spatial positioning method for a fiber Bragg grating sensor network, and to solve the three-dimensional spatial positioning problem of a large-capacity fiber Bragg grating sensor network by constructing a composite positioning algorithm.

[0006] In view of this, the present invention provides a composite spatial positioning algorithm based on DIDE-FPP. The DIDE-FPP composite spatial positioning method provided by the present invention includes a distributed individual differential evolution method (DIDE) and a four-point positioning method (FPP).

[0007] The DIDE method described in the present invention consists of four parts; the FPP method consists of two steps: distance calculation and coordinate solution.

[0008] Optionally, the DIDE method improves the performance of evolutionary algorithms in solving MMOPs through a new distributed multimodal individual (DIMP) framework;

[0009] Optionally, the DIDE method solves the problem that the mutation process in the DE algorithm cannot be applied to the multimodal distribution individual (DIMP) framework by creating a virtual population through an adaptive range adjustment (ARA) strategy;

[0010] Optionally, the DIDE method solves the peak diversity problem in MMOP through a life cycle mechanism;

[0011] Optionally, the DIDE method applies an elite learning mechanism (ELM) to improve the peak location accuracy, thereby solving the multi-peak optimization problem and improving the location accuracy of the entire tunnel fire monitoring system.

[0012] Optionally, the distance calculation step of the FPP method: In this step, the peak points found by the DIDE algorithm are grouped into four groups. Since the position of each grating sensor in the ultra-weak grating array is fixed after preparation, the distance of each point to the tunnel ground can be easily known. Assuming that the four points are A, B, C, and D, and their coordinates are (x1, y1, z1), (x2, y2, z2), (x3, y3, z3), and (x4, y4, z4), their distances to the tunnel ground are ρ1, ρ2, ρ3, and ρ4, respectively. According to the distance formula, the following equation can be obtained:

[0013]

[0014] All groups were subjected to the above operations;

[0015] Optionally, the FPP method coordinate solution step: In this step, the equation obtained in the distance calculation stage is simplified to obtain the following equation:

[0016]

[0017] Represent the equation in matrix form: Ac = b

[0018] in:

[0019]

[0020]

[0021]

[0022] According to c=A -1 b, we can get each set of virtual points for positioning, summarize them and draw a three-dimensional graph, thus completing the three-dimensional space positioning.

[0023] The present invention provides an operation process of the DIDE-FPP composite spatial positioning method applied to a fiber Bragg grating sensor network as described in the above technical solution, comprising the following steps:

[0024] The present invention uses MATLAB simulation software and takes the C band as the input signal (light source);

[0025] Add adjustable temperature changes;

[0026] The DIDE method was applied to find peaks;

[0027] Group the peak points obtained by peak searching;

[0028] Three-dimensional spatial positioning is performed using the FPP method.

[0029] The DIDE-FPP method provided by the present invention includes the DIDE method and the FPP technology. The DIDE-FPP provided by the present invention first uses the DIDE method to search for peaks, and then combines the FPP technology to further locate the peak points in three-dimensional space. The present invention not only solves the multi-peak optimization problem (MMOPS) existing in the tunnel fire location process, but also realizes three-dimensional space positioning and reaches centimeter-level positioning accuracy through further analysis and calculation of the peak search results. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0031] Figure 1 It is the flow chart of DIDE-FPP composite algorithm;

[0032] Figure 2 This is the DIDE algorithm flow chart;

[0033] Figure 3 Run a flowchart for the ARA strategy;

[0034] Figure 4Run flowcharts for lifecycle mechanisms;

[0035] Figure 5 Flowchart for the operation of the Elite Learning Mechanism ELM;

[0036] Figure 6 The middle is a comparison chart of PR results of DIDE-FPP and other four advanced algorithms;

[0037] Figure 7 The middle is a comparison chart of SR results of DIDE-FPP and other four advanced algorithms; DETAILED DESCRIPTION

[0038] The technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0039] The purpose of the present invention is to provide a composite spatial positioning method for a fiber Bragg grating sensor network, and to solve the multi-peak optimization problem in the peak-finding link of a large-capacity fiber Bragg grating sensor network by creating a composite spatial positioning method.

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] like Figure 1 As shown, the present invention provides a DIDE-FPP composite spatial positioning method, which specifically includes: a DIDE method and an FPP technology.

[0042] The DIDE-FPP method provided by the present invention consists of four parts: differential evolution, peak search optimization, peak grouping and spatial positioning, specifically including:

[0043] Step 101: Perform differential evolution (DE) on the waveform data reflected by the optical fiber array

[0044] In practical applications, step 101 specifically includes: firstly, treating each waveform data reflected by the optical fiber array as an independent individual, and creating their own virtual population for each independent individual, and then performing mutation operation to select the individual with the best performance to complete the differential evolution process;

[0045] Step 102: DIDE applies the ARA strategy, life cycle mechanism and ELM to optimize the results of differential evolution within the DIMP framework, completes the peak search process, and obtains the peak point;

[0046] Step 103: Regroup the peak points to prepare for coordinate calculation;

[0047] Step 104: Use the FPP method to locate the regrouped peak points in three-dimensional space.

[0048] like Figure 2 As shown, the present invention uses MATLAB simulation software, takes the C band as the input signal, adds an adjustable external temperature change, and applies the DIDE method to perform a peak search operation. The steps are as follows:

[0049] Step 1: Number the original signal into a specific individual X α , initialize X α and its corresponding R,cg,ht;

[0050] Step 2: For each individual X α Generate two virtual populations VX α,1 ,VX α,2 , perform mutation operation to generate the corresponding virtual individual V α ,

[0051] V α =X α +F×(VX α，1 -VX α，2 )

[0052]

[0053] Where F is the scale factor parameter, L is the upper bound of the search space, and U is the lower bound of the search space;

[0054] Step 3: Perform a crossover operation and select the information of each dimension according to the crossover rate CR. α and V α The value is paid to the offspring U α ,

[0055] where rnbr α Is individual X α A random integer in the range of {1,2,...,D}, where D is the dimension of the optimization problem;

[0056] Step 4: Make a selection by selecting α and X α The individual with better fitness is selected as the next generation of the evolutionary algorithm, and the cg parameters corresponding to each individual are updated to assist the ARA strategy;

[0057] Step 5: Figure 3 As shown, the adaptive range adjustment (ARA) strategy is implemented.

[0058] Where mcg = 10 × 2 [D / 10]+1 ;

[0059] Step 6: Figure 4 As shown, the execution life cycle mechanism,

[0060] Where mht is the maximum halving time and rank is the number of cycles;

[0061] Step 7: Figure 5 As shown, the elite learning mechanism (ELM) is executed. If the termination condition is met, the peak search result is output; otherwise, return to step 2 and continue the operation until the termination condition is met and the peak search result is output.

[0062] Among them B α =Gaussian(A α ,σ α ), σ α is the standard deviation of the corresponding individual, number is the stagnation counter, and dt is the falling threshold.

[0063] The present invention also provides an operation scheme of the four-point positioning (FPP) technology, which consists of two steps: distance calculation and coordinate solution. The four-point positioning (FPP) technology provided by the present invention is based on calculating the coordinates of the outlier points according to the distance formula, and realizes three-dimensional spatial positioning while constructing the coordinate interval.

[0064] In the present invention, the distance calculation step of the FPP method is as follows: In this step, the peak points found by the DIDE algorithm are grouped into four groups, because the position of each grating sensor of the ultra-weak grating array is fixed after preparation, and the distance of each point to the tunnel ground can be easily known. Assuming that the four points are A, B, C, and D, and the coordinates are (x1, y1, z1), (x2, y2, z2), (x3, y3, z3), and (x4, y4, z4), respectively, their distances to the tunnel ground are ρ1, ρ2, ρ3, and ρ4, respectively. According to the distance formula, the following equation can be obtained:

[0065]

[0066] All groups were subjected to the above operations;

[0067] In the present invention, the coordinate solution step of the FPP method is as follows: in this step, the equation obtained in the distance calculation stage is simplified to obtain the following equation:

[0068]

[0069] Represent the equation in matrix form: Ac = b

[0070] in:

[0071]

[0072]

[0073]

[0074] According to c=A -1 b, we can get each set of virtual points for positioning, summarize them and draw a three-dimensional graph, thus completing the three-dimensional space positioning.

[0075] In the actual test, in order to objectively evaluate the peak search performance, the peak ratio (PR) and success rate (SR) are used as evaluation indicators, and a comparative experiment is set to verify the denoising effect of the DIDE-FPP method.

[0076]

[0077] where PR represents the average percentage of global peaks found in multiple runs, NR is the number of runs, and NPF is α is the number of global peaks found in the run, TNP is the total number of global peaks in the optimization problem. SR represents the percentage of successful runs among multiple runs, where NSR is the number of successful runs among the total NR runs (a successful run is one in which all global peaks are found).

[0078] Comparative Example

[0079] To verify the peak finding performance of the DIDE-FPP algorithm, the CDE, SDE, NCDE, NSDE and DIDE-FPP peak finding methods are used respectively, and the peak finding effects on 20 functions in the CEC'2013 benchmark set are compared.

[0080] CDE extends the performance of differential evolution algorithms in tracking multiple peaks through a population scheme. SDE locates multiple global optima simultaneously by adaptively forming multiple species (or subpopulations) in the DE population. The NCDE (neighborhood based crowding differential evolution) method with neighborhood variation adds neighborhood variation technology to CDE. The NSDE (neighborhood based species differential evolution) method with neighborhood variation adds neighborhood variation technology to SDE. The DIDE-FPP composite method adds a distributed individual evolution scheme to DE and combines it with a four-point positioning technology.

[0081] In the actual test, the present invention sets the DIDE operating parameters as follows: the proportional factor F and the crossover rate CR are set to 0.3 and 0.9 respectively; in the life cycle mechanism, the maximum half time mht is 10; in the ELM, the drop threshold dt is 40, and σ β With σ min The 20 functions in Table 1 are used as input signals. Then DIDE-FPP, CDE, SDE, NCDE and NSDE are run 50 times on these 20 functions respectively. The 50 peak search results are averaged and sorted into PR and SR values ​​to complete the peak search performance comparison.

[0082] Table 1 20 functions in the CEC'2013 benchmark set

[0083] Function Name Local optimal solution Global optimal solution Data Dimensions 1 Five-Uneven-PeaK Trap 3 2 1 2 Equal Maxima 0 5 1 3 Uneven Decreasing Maxima 4 1 1 4 Himmelblau 0 4 2 5 Six-Hump Camel Back 2 2 2 6 Shubert with 2D many 18 2 7 Vincent with 2D 0 36 2 8 Shubert with 3D many 81 3 9 Vincent with 3D 0 216 3 10 Modified Rastrigin 0 12 2 11 Composition Function 1 with 2D many 6 2 12 Composition Function 2 with 2D many 8 2 13 Composition Function 3 with 2D many 6 2 14 Composition Function 3 with 3D many 6 3 15 Composition Function 4 with 3D many 8 3 16 Composition Function 3 with 5D many 6 5 17 Composition Function 4 with 5D many 8 5 18 Composition Function 3 with 10D many 6 10 19 Composition Function 4 with 10D many 8 10 20 Composition Function 4 with 20D many 8 20

[0084] In actual tests, the peak-finding effect, namely the peak ratio (PR), is obtained by applying different combination methods. Figure 6 The CDE, SDE, NCDE, NSDE and DIDE-FPP models can all perform low-dimensional peak-finding operations, but the DIDE-FPP model has a higher peak ratio under high-dimensional functions, that is, the peak-finding effect is better than other models.

[0085] The present invention obtains the peak search effect by applying different combination methods, that is, the success rate (SR) as follows Figure 7 As shown in the figure, the SR of DIDE-FPP on 20 functions is significantly higher than that of other models, that is, the comprehensive peak finding effect is better than that of other models.

[0086] In actual tests, the DIDE-FPP method has the best comprehensive running results on 20 functions in the CEC'2013 benchmark set, and obtains the highest peak ratio (PR) and the highest success rate (SR) among the five methods, proving that the DIDE-FPP model has the best peak-finding effect.

[0087] The present invention combines the distributed individual differential evolution method (DIDE) with the four-point positioning method (FPP) and proposes a DIDE-FPP method that can be used for peak finding in fiber Bragg grating sensor networks. This method can adaptively process the ultra-weak fiber Bragg grating reflection spectrum in the C band. Compared with models such as CDE and SDE, this method can perform peak finding more accurately and find more peak points.

[0088] The present invention uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples and comparative examples are only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A composite spatial positioning method for a fiber grating sensor network, characterized in that: The distributed differential evolution of individuals (DIDE) method is combined with the four-point positioning (FPP) method; The DIDE-FPP composite spatial positioning method specifically includes: Step 101: performing differential evolution (DE) on the waveform data reflected by the optical fiber array; Step 101 specifically includes: firstly, treating each waveform data reflected by the optical fiber array as an independent individual, and creating a virtual population for each independent individual, and then performing a mutation operation to select the individual with the best performance to complete the differential evolution process; Step 102: DIDE applies the adaptive range adjustment (ARA) strategy, life cycle mechanism and elite learning mechanism (ELM) to optimize the results of differential evolution within the multi-peak distributed individual (DIMP) framework, completes the peak search process, and obtains the peak point; Using MATLAB simulation software, taking the C band as the input signal, adding adjustable external temperature changes, and applying the DIDE method to perform peak search operations, the steps are as follows: Step 1: Number the original signal into a specific individual X α , initialize X α and its corresponding R,cg,ht; Step 2: For each individual X α Generate two virtual populations VX α,1 ,VX α,2 , perform mutation operation to generate the corresponding virtual individual V α , V α =X α +F×(VX α，1 -VX α，2 ) Where F is the scale factor parameter, L is the upper bound of the search space, and U is the lower bound of the search space; Step 3: Perform a crossover operation and select the information of each dimension according to the crossover rate CR. α and V α The value is paid to the offspring U α , where rnbr α Is individual X α A random integer in the range of {1,2,...,D}, where D is the dimension of the optimization problem; Step 4: Make a selection by selecting α and X α The individual with better fitness is selected as the next generation of the evolutionary algorithm, and the cg parameters corresponding to each individual are updated to assist the ARA strategy; Step 5: Execute the adaptive range adjustment (ARA) strategy; Step 6: Execute the life cycle mechanism; Step 7: Execute the elite learning mechanism (ELM). If the termination condition is met, the peak search result is output; otherwise, return to step 2 and continue the operation until the termination condition is met and the peak search result is output; Step 103: Regroup the peak points to prepare for coordinate calculation; Step 104: Using the FPP method to locate the regrouped peak points in three-dimensional space; The distance calculation step of the FPP method is as follows: In this step, the peak points found by the DIDE algorithm are grouped into four groups. Since the position of each grating sensor of the ultra-weak grating array is fixed after preparation, the distance of each point to the tunnel ground can be easily known. Assuming that the four points are A, B, C, and D, and the coordinates are (x1, y1, z1), (x2, y2, z2), (x3, y3, z3), and (x4, y4, z4), their distances to the tunnel ground are ρ1, ρ2, ρ3, and ρ4, respectively. According to the distance formula, the following equation can be obtained: All groups were subjected to the above operations; The coordinate solution step of the FPP method is as follows: In this step, the equation obtained in the distance calculation stage is simplified to obtain the following equation: Represent the equation in matrix form: Ac = b in: According to c=A -1 b, we can get each set of virtual points for positioning, summarize them and draw a three-dimensional graph, thus completing the three-dimensional space positioning.

Images