A multi-objective optimization method for sensor arrays for magnetic field signal denoising

Abstract

The present invention discloses a sensor array multi-objective optimization method for magnetic field signal denoising. The implementation steps include: establishing a magnetic field interference model under a constrained mode and performing transient analysis to obtain time series data of magnetic field signals collected by all sensors; setting an optimization index, using the optimization index as a multi-objective fitness function, optimizing the multi-objective fitness function, and obtaining an optimal auxiliary sensor layout scheme when the fitness function value is maximized; based on the optimal auxiliary sensor layout scheme, establishing a main sensor and an optimal auxiliary sensor array; in an environment affected by interference sources, processing the noise signal of the optimal auxiliary sensor array based on a trained deep learning network model to obtain a denoised magnetic field signal of the main sensor, and reconstructing the main position noise through sensor layout optimization and auxiliary sensor array noise to achieve magnetic field signal denoising when the object is disturbed, thereby improving the accuracy of magnetic field testing.

Description

Technical Field

[0001] The present invention relates to the field of magnetic field signal technology, and in particular to a sensor array multi-objective optimization method for magnetic field signal denoising. Background Art

[0002] Magnetic field measurement technology is a core tool in fields such as power equipment monitoring, biomedical imaging, and industrial nondestructive testing. Its accuracy directly impacts the accuracy of fault diagnosis and the reliability of signal analysis. However, in practical applications, the target magnetic field is often contaminated by complex environmental noise, such as switching surges in power systems, electromagnetic radiation from nearby equipment, and stray magnetic fields in biomedical settings. This noise often overlaps with the target signal in the frequency domain and exhibits nonstationary and nonlinear characteristics, posing significant challenges to traditional denoising methods.

[0003] Existing single-sensor filtering schemes (such as adaptive filtering and wavelet threshold denoising) can suppress some noise, but they struggle to distinguish signals in the same frequency band from interference components. This can easily cause signal distortion, especially when the noise intensity is comparable to the target signal. To address spatial noise correlation, fixed multi-sensor arrays (e.g., uniform or regularly distributed) are used for spatial filtering. However, their rigid layout leads to two key drawbacks: First, the sensor positions cannot adapt to changes in the spatial distribution of interference sources. When interference sources move or are added, the array's noise suppression effectiveness decreases significantly. Second, the fixed layout does not optimize signal correlation between sensors. Some sensors may collect low-correlation noise, which in turn introduces redundant computation and error propagation.

[0004] On the other hand, deep learning-based denoising methods (such as CNN and LSTM) demonstrate potential for handling nonlinear noise through end-to-end modeling, but their performance is highly dependent on the spatial representativeness of the input data. Existing technologies typically directly use fixed sensor arrays as network input without incorporating layout optimization, resulting in the network needing to additionally learn the spatially redundant characteristics of the noise, increasing model complexity and the risk of overfitting. Furthermore, traditional linear denoising assumptions (such as principal component analysis) struggle to capture the complex coupling relationship between interfering noise and the target magnetic field, further limiting denoising accuracy. Summary of the Invention

[0005] The technical solution of the present invention is to provide a new denoising architecture that integrates sensor dynamic optimization and deep learning, screens highly correlated auxiliary sensors through multi-objective indicators, and utilizes nonlinear mapping capabilities to decouple noise and signals, thereby improving the robustness of magnetic field measurements in complex electromagnetic environments and breaking through the technical bottlenecks of traditional methods in adaptive layout and nonlinear denoising.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a multi-objective optimization method for a sensor array for magnetic field signal denoising, comprising the following steps: S1: establishing a magnetic field interference model under a constrained mode and performing transient analysis to obtain time series data of magnetic field signals collected by all sensors; S2: setting an optimization index based on the magnetic field signal data, using the optimization index as a multi-objective fitness function, optimizing the multi-objective fitness function, and obtaining the optimal auxiliary sensor layout scheme when the fitness function value is maximized; S3: establishing a main sensor and an optimal auxiliary sensor array based on the optimal auxiliary sensor layout scheme; S4: in an environment affected by interference sources, processing the noise signal of the optimal auxiliary sensor array based on a trained deep learning network model to obtain the denoised magnetic field signal of the main sensor.

[0007] Furthermore, the magnetic field interference model under the constrained mode in step 1 is established by finite element analysis software, and the process is as follows: the finite element analysis software is used to build a magnetic field interference model, and the magnetic field interference model includes an interference source and a disturbed plane, and N+1 sensors for detecting magnetic field signals are set on the surface of the disturbed plane, of which one sensor is the main sensor located at the main position of the disturbed plane, and the remaining N sensors are selected auxiliary sensors that are evenly distributed radially on the disturbed plane.

[0008] Furthermore, the process of obtaining the magnetic field signal data collected by all sensors in step 1 is as follows: in the finite element analysis software, the material properties of the established magnetic field interference model are defined, and meshing is performed. Based on the selected unit type and mesh density, fixed constraints are imposed and the current load that varies with time is defined. Transient analysis is performed to obtain the magnetic field signal data in three directions collected by all sensors.

[0009] Furthermore, the optimization indicators in step 2 include optimization indicator 1 and optimization indicator 2, wherein optimization indicator 1 is the comprehensive dynamic time warping distance between the sensor magnetic field signal data of the selected candidate auxiliary sensor layout and the main sensor, and optimization indicator 2 is the comprehensive Chatterjee correlation coefficient between the sensor magnetic field signal data of the selected candidate auxiliary sensor layout and the main sensor; the optimal auxiliary sensor layout scheme refers to the optimal auxiliary sensor layout scheme selected from N candidate auxiliary sensor layouts; and the multi-objective fitness function is optimized using a genetic algorithm with adaptive dual structure encoding.

[0010] Furthermore, the process of optimizing the multi-objective fitness function using a genetic algorithm with adaptive dual structure coding is as follows: S21: Initialize the population size, the maximum number of iterations, the number of auxiliary sensors to be selected, the number of selected auxiliary sensors, the mutation selection probability, the crossover selection probability, the optimization index weight ratio, the number of elite individuals, and the stagnation threshold iteration number; S22: Construct individuals of the dual structure coding population, and the dual structure of the individual chromosome consists of two lines of position code and confirmation code, where the position code indicates that the auxiliary sensor can select a layout position, and the confirmation code indicates whether the position is selected; S23: Set the multi-objective fitness function; S24: Optimize the multi-objective fitness function to obtain the optimal auxiliary sensor layout plan when the fitness function value is maximized.

[0011] Furthermore, S23: setting a multi-objective fitness function specifically includes the following steps: S231: using optimization index 1 and optimization index 2 as a multi-objective fitness function, wherein optimization index 1 is negative and optimization index 2 is positive;

[0012]

[0013] S232: Normalize the calculated optimization index 1 and optimization index 2, and weight them according to the optimization index weight ratio to obtain the comprehensive fitness value of a single chromosome. The formula expression is:

[0014] Integrated_Fitness=[weight index_1 ,weight index_2 ]×[Norm Index_1 ,Norm Index_2 ] T ;

[0015] Among them, weight index_1 and weight index_2 They are respectively expressed as the weight ratio of optimization indicators, Norm Index_1 and Norm Index_2 They are respectively represented as the normalized results of optimization index 1 and optimization index 2 within the population.

[0016] Furthermore, S24: optimizing the multi-objective fitness function to obtain the optimal auxiliary sensor layout solution when the fitness function value is maximized. The specific process is as follows:

[0017] S241: Calculate the comprehensive fitness value of all chromosomes in the initial population, and after roulette wheel selection, partial matching crossover, inversion mutation, and using the elite preservation strategy, retain the chromosomes with the highest fitness value of elite individuals to generate a new population;

[0018] S242: Propose a nonlinear adaptive regulation mechanism by quantifying population distribution characteristics in real time;

[0019] By statistically analyzing the standard deviation and range distribution of population fitness, a composite diversity evaluation index is constructed, and the crossover probability P is realized using the Sigmoid function. c and mutation probability P m Smooth dynamic adjustment of , the formula is:

[0020]

[0021] Among them, the comprehensive fitness value of the contemporary population is set to have:

[0022]

[0023] P c is the crossover probability, P m is the mutation probability, D is the composite diversity index, e is the natural constant, P size represents the population size, f p represents the comprehensive fitness value of the pth chromosome in the population, f max Indicates the maximum comprehensive fitness in the population, f min It represents the minimum comprehensive fitness in the population, μ is the average fitness, σ is the standard deviation of fitness, and Δ is the range normalization factor;

[0024] S243: In S241, a nonlinear adaptive adjustment mechanism is added, and this step is repeated until the maximum number of iterations is reached or the chromosome with the maximum fitness in the population does not change after the stagnation threshold number of iterations, a large fitness chromosome is obtained, and the dual structure encoding is deconstructed to obtain the optimal auxiliary sensor layout.

[0025] Furthermore, the main sensor layout in the optimal auxiliary sensor layout scheme is close to the detected position. The optimal auxiliary sensor layout is obtained by comprehensive optimization using a multi-objective fitness function. The optimal auxiliary sensor layout changes with the location and type of the interference source; the interference source affects the environment when the current-carrying conductor is not energized and only the interference source generates a magnetic field.

[0026] Furthermore, during the training process of the deep learning network model, the noise signal corresponding to the optimal auxiliary sensor in the noise signal of the optimal auxiliary sensor array is taken as input, and the noise signal corresponding to the main sensor is taken as output.

[0027] Furthermore, the noise signal of the optimal auxiliary sensor array is processed based on the trained deep learning network model to obtain the denoised magnetic field signal of the main sensor. The process is as follows: the noise signal of the optimal auxiliary sensor array is reconstructed using the trained deep learning network model to output the noise signal at the main position; the noise signal at the main position is subtracted from the disturbed signal obtained by the main sensor to obtain the denoised magnetic field signal.

[0028] The present invention has the following beneficial effects:

[0029] (1) Multimodal optimization improves anti-interference capabilities: By integrating multi-objective optimization of dynamic time warping (DTW) distance and Chatterjee correlation coefficient, while taking into account the time domain alignment characteristics and nonlinear correlation of sensor signals, the robustness of auxiliary sensor layout is significantly improved. Compared with traditional single-index optimization methods, it can more accurately screen auxiliary nodes with the strongest correlation with the main sensor noise, laying a data foundation for subsequent denoising.

[0030] (2) High-precision noise separation driven by deep learning: A deep learning network is used to model the noise of the primary and secondary sensors, which can adaptively capture nonlinear interference patterns in complex environments, breaking through the limitations of traditional filtering algorithms that rely on prior noise models. The network trained with massive data can dynamically reconstruct the primary position noise, especially in transient or non-stationary interference scenarios, and significantly improves denoising accuracy.

[0031] (3) Full-process closed-loop optimization design: From finite element modeling, multi-objective optimization layout, to online noise elimination, a closed-loop system is formed, combining the dual advantages of physical mechanisms and data-driven design. The finite element model provides theoretical constraints for sensor layout, reducing experimental costs; data-driven optimization dynamically adapts to actual working conditions, ensuring the universality of the solution in complex electromagnetic environments.

[0032] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flow chart of a multi-objective optimization method for a sensor array for magnetic field signal denoising according to the present invention;

[0034] Figure 2 Flowchart for realizing the method of the present invention;

[0035] Figure 3 This is the overall technical roadmap for this method;

[0036] Figure 4 This is a physical structure simulation diagram of electromagnetic interference based on this method;

[0037] Figure 5Simulation diagram of main-auxiliary sensor layout;

[0038] Figure 6 This is the simulation result diagram under electromagnetic interference;

[0039] Figure 7 A diagram encoding the position of the auxiliary sensor;

[0040] Figure 8 The curve diagrams of various indicators in the optimization process of the standard GA algorithm and the new GA algorithm when the number of auxiliary sensors is 4;

[0041] Figure 9 This is a comprehensive comparison diagram of the optimization process of the standard GA algorithm and the new GA algorithm when the number of auxiliary sensors is 4;

[0042] Figure 10 This is the result diagram of the optimized auxiliary sensor layout;

[0043] Figure 11 The main sensor signal waveform and the reconstructed noise waveform when the current conductor is energized under electromagnetic interference;

[0044] Figure 12 A comparison diagram of the denoised signal waveform and the ideal waveform obtained by implementing the method of the present invention;

[0045] Figure 13 To achieve the denoising signal error comparison diagram obtained by the method of the present invention compared with the random layout of auxiliary sensors. DETAILED DESCRIPTION

[0046] See also Figure 1 The embodiment of the present invention provides a technical solution: a multi-objective optimization method for sensor array for magnetic field signal denoising, the method flow and overall technical route are as follows Figure 2 and Figure 3 As shown, the following steps are included: S1: establishing a magnetic field interference model under a constrained mode and performing transient analysis to obtain time series data of magnetic field signals collected by all sensors;

[0047] The magnetic field interference model under the constrained mode in step 1 is established using finite element analysis software. The process is as follows: use finite element analysis software to build a magnetic field interference model, which includes an interference source and a disturbed plane, and set N+1 sensors for detecting magnetic field signals on the surface of the disturbed plane, of which one sensor is the main sensor located at the main position of the disturbed plane, and the remaining N sensors are selected auxiliary sensors that are evenly distributed radially on the disturbed plane.

[0048] The process of obtaining the magnetic field signal data collected by all sensors in step 1 is as follows: In the finite element analysis software, define the material properties of the established magnetic field interference model, and perform meshing. Based on the selected unit type and mesh density, apply fixed constraints and define the current load that varies with time. Perform transient analysis to obtain the magnetic field signal data in three directions collected by all sensors.

[0049] Simulation: The premise for optimizing the auxiliary sensor layout is to obtain as much magnetic field distribution data as possible at the main position on the disturbed plane and its surrounding area under electromagnetic interference. Electromagnetic simulation analysis software, such as Ansys and COMSOL, is used to obtain magnetic field distribution data on the disturbed plane under different electromagnetic interference conditions. For different interference sources, perform the following steps:

[0050] (1) Importing the electromagnetic interference physical structure model, which mainly includes: interference source, disturbed plane, and main sensor and auxiliary sensor to be selected arranged on the plane;

[0051] (2) Define the material properties of the model and select the appropriate element type and mesh density;

[0052] (3) Define boundary conditions based on the interference source operating parameters;

[0053] (4) Select the solver and set the solution parameters;

[0054] (5) Perform simulation calculations;

[0055] (6) Obtain the magnetic field signal data collected by all sensors under the current interference source and export the data to Excel;

[0056] First, conduct a detailed survey and measurement of the interference source scene. Collect various dimensional information about the scene, including the length, width, and height of the affected surface to be detected, as well as the type and location of the interference source. Then, use professional 3D modeling software, such as SolidWorks, to convert the collected dimensional data into a visual 3D model. Finally, import the model into finite element analysis software and perform the above steps.

[0057] The physical structure simulation diagram for electromagnetic interference, the main-auxiliary sensor layout simulation diagram and the simulation result diagram under electromagnetic interference are as follows: Figure 4 、 Figure 5 as well as Figure 6 The interference source is a cylindrical energized spiral tube, to which a fixed constraint and a time-varying current load are applied. A main sensor is located at the center of the disturbed plane, and 42 auxiliary sensors are selected and evenly distributed radially on the disturbed plane. A transient analysis of the model is performed to obtain the time series of the X / Y / Z three-axis magnetic field signals of all sensors.

[0058] S2: Setting an optimization index based on the magnetic field signal data, using the optimization index as a multi-objective fitness function, optimizing the multi-objective fitness function, and obtaining the optimal auxiliary sensor layout solution when the fitness function value is maximized;

[0059] The optimization indicators in step 2 include optimization indicator 1 and optimization indicator 2, where optimization indicator 1 is the comprehensive dynamic time warping (DTW) distance between the sensor magnetic field signal data of the selected auxiliary sensor layout and the main sensor, and the smaller the optimization indicator, the better; optimization indicator 2 is the comprehensive Chatterjee correlation coefficient between the sensor magnetic field signal data of the selected auxiliary sensor layout and the main sensor, and the larger the better;

[0060] The specific process is:

[0061] The magnetic field signal data collected by the sensor is defined as a discrete time series with a length of len output through the analog-to-digital converter: S = {s 1 ,s 2 ,...,s len}.

[0062] Optimization indicator 1 is:

[0063]

[0064] Set up two sets of time series with time series lengths i and j respectively: and

[0065] Then, the dynamic time warping distance dtw(i,j) between the two is calculated by the following recursive formula:

[0066]

[0067] Optimization indicator 2 is:

[0068]

[0069] Among them, Index_1 represents optimization index 1, and Index_2 represents optimization index 2;

[0070] n sensor Indicates the number of selected auxiliary sensors, m axis Indicates the direction of magnetic field strength (1 represents the X axis, 2 represents the Y axis, and 3 represents the Z axis);

[0071] Indicates the m collected by the main sensor axis The length of the axis data, Indicates the m data collected by the nth selected auxiliary sensoraxis The length of the axis data;

[0072] Represented as time series data S a The i-th element in Similarly;

[0073] Define w as the Sakoe-Chiba band constraint window parameter, and the symmetric band region |ij|≤w;

[0074] l g Represents the index array S after sorting the time series S order In the formula, the rank corresponding to the g-th element of the time series S is the number of elements in the sequence S that are greater than this element.

[0075]

[0076] Among them, |() represents the indicator function, S order [h] represents the hth element of the time series S in the index array S order The corresponding size, S order [g]Same.

[0077] r g Yes g The difference is that when dealing with ties (elements are equal during the sorting process), there will be different handling methods. If the g-th element of the time series S is equal to the elements other than it, the following is adopted: randomly sort and select values ​​among these equal elements.

[0078] The optimal auxiliary sensor layout scheme refers to the optimal auxiliary sensor layout scheme selected from N candidate auxiliary sensor layouts; the multi-objective fitness function is optimized using a genetic algorithm with adaptive dual structure coding.

[0079] The optimal sensor layout scheme in this example refers to the four optimal auxiliary sensor layouts selected from 42 candidate auxiliary sensor positions. The optimization method used is a new adaptive dual-structure coding genetic algorithm;

[0080] The process of optimizing the multi-objective fitness function using the genetic algorithm with adaptive dual structure coding is as follows: S21: initializing the population size, the maximum number of iterations, the number of auxiliary sensors to be selected, the number of selected auxiliary sensors, the probability of mutation selection, the probability of crossover selection, the weight ratio of optimization indicators, the number of elite individuals, and the number of stagnation threshold iterations;

[0081] The following are some parameters of the genetic algorithm for the new adaptive dual structure encoding:

[0082]

[0083]

[0084] S22: Constructing a dual-structure coded population individual. The dual structure of individual chromosomes consists of two lines: position code and confirmation code. The position code indicates that the auxiliary sensor can select a layout position, and the confirmation code indicates whether the position is selected.

[0085] The position coding diagram of 42 auxiliary sensors to be selected is as follows: Figure 7 As shown, a dual structure encoding structure is:

[0086]

[0087] S23: Setting multi-objective fitness function;

[0088] S231: Optimization index 1 and optimization index 2 are used as a multi-objective fitness function, where optimization index 1 is negative and optimization index 2 is positive;

[0089]

[0090] S232: Normalize the calculated optimization index 1 and optimization index 2, and weight them according to the optimization index weight ratio to obtain the comprehensive fitness value of a single chromosome. The formula expression is:

[0091] Integrated_Fitness=[weight index_1 ,weight index_2 ]×[Norm Index_1 ,Norm Index_2 ] T ;

[0092] Among them, weight index_1 and weight index_2 They are respectively expressed as the weight ratio of optimization indicators, Norm Index_1 and Norm Index_2 They are respectively represented as the normalized results of optimization index 1 and optimization index 2 within the population.

[0093] S24: Optimizing the multi-objective fitness function to obtain an optimal auxiliary sensor layout solution when the fitness function value is maximized.

[0094] S241: Calculate the comprehensive fitness value of all chromosomes in the initial population, and after roulette wheel selection, partial matching crossover, inversion mutation, and using the elite preservation strategy, retain the chromosomes with the highest fitness value of elite individuals to generate a new population;

[0095] S242: Propose a nonlinear adaptive regulation mechanism by quantifying population distribution characteristics in real time;

[0096] By statistically analyzing the standard deviation and range distribution of population fitness, a composite diversity evaluation index is constructed, and the crossover probability P is realized using the Sigmoid function. c and mutation probability P m Smooth dynamic adjustment of , the formula is:

[0097]

[0098] Among them, the comprehensive fitness value of the contemporary population is set to have:

[0099]

[0100] P c is the crossover probability, P m is the mutation probability, D is the composite diversity index, e is the natural constant, P size represents the population size, f p represents the comprehensive fitness value of the pth chromosome in the population, f max Indicates the maximum comprehensive fitness in the population, f min It represents the minimum comprehensive fitness in the population, μ is the average fitness, σ is the standard deviation of fitness, and Δ is the range normalization factor;

[0101] S243: In S241, a nonlinear adaptive adjustment mechanism is added, and this step is repeated until the maximum number of iterations is reached or the chromosome with the maximum fitness in the population does not change after the stagnation threshold number of iterations, a large fitness chromosome is obtained, and the dual structure encoding is deconstructed to obtain the optimal auxiliary sensor layout.

[0102] The curves of various indicators in the optimization process of the standard GA algorithm and the new GA algorithm are as follows Figure 8 As shown, a represents the optimization process result of the standard GA algorithm, and b represents the optimization process result of the new GA algorithm. Figure 9 The figure shows a comprehensive comparison of the optimization process of the standard GA algorithm and the new GA algorithm. From the trend of the chart, it can be seen that the new GA algorithm reaches a higher performance value faster under the same number of iterations, while the standard GA algorithm improves more slowly. This shows that the adaptive mechanism can accelerate search efficiency and reduce resource consumption. The optimal sensor layout is finally obtained as follows Figure 10 .

[0103] S3: Based on the optimal auxiliary sensor layout plan, establish the main sensor and the optimal auxiliary sensor array;

[0104] S4: In an environment affected by interference sources, the noise signal of the optimal auxiliary sensor array is processed based on the trained deep learning network model to obtain the denoised magnetic field signal of the main sensor.

[0105] The main sensor layout in the optimal auxiliary sensor layout scheme is close to the detected position. The optimal auxiliary sensor layout is obtained by comprehensive optimization using a multi-objective fitness function. The optimal auxiliary sensor layout changes with the location and type of the interference source; the interference source affects the environment when the current-carrying conductor is not energized and only the interference source generates a magnetic field.

[0106] During the training process of the deep learning network model, the noise signal corresponding to the optimal auxiliary sensor in the noise signal of the optimal auxiliary sensor array is taken as input, and the noise signal corresponding to the main sensor is taken as output. The network is supervised training using a large amount of data, where the main sensor noise and the optimal auxiliary sensor noise are both time series data. The deep learning network input is the one-dimensional time series signal of the number of auxiliary sensors, and the output is the one-dimensional time series data of the single-channel main sensor.

[0107] The noise signal of the optimal auxiliary sensor array is processed based on the trained deep learning network model to obtain the denoised magnetic field signal of the main sensor. The process is as follows: the noise signal of the optimal auxiliary sensor array is reconstructed using the trained deep learning network model to output the noise signal at the main position; the noise signal at the main position is subtracted from the disturbed signal obtained by the main sensor to obtain the denoised magnetic field signal.

[0108] In an environment affected by interference sources and where a current-carrying conductor generates a magnetic field, the trained deep learning network model is used to reconstruct the noise at the main position from the auxiliary sensor array noise. According to Ampere's law, when a current-carrying conductor is energized, a magnetic field proportional to the current intensity is formed around it. The main sensor is placed above the current-carrying conductor. In an environment affected by interference sources, the signal collected by the main sensor is the superposition of the magnetic fields generated by the current-carrying conductor and the interference source, such as Figure 11 The red curve part in the middle, and the signal collected by the auxiliary sensor is the magnetic field of the interference source at its location. The trained deep learning network model is used to reconstruct the noise at the main location from the auxiliary sensor noise, as shown in Figure 11 The green curve part in .

[0109] The denoised magnetic field signal can be obtained by subtracting the noise signal at the main position reconstructed by the auxiliary sensor array from the disturbed signal obtained by the main sensor. Figure 12 As shown in the figure, the red line is the denoised signal, and the black line is the ideal signal. Figure 13A comparison of the denoised signal errors achieved by the proposed method compared to a random placement of auxiliary sensors is shown. The lower the RMSE (root mean square error) and MAE (mean absolute error), the better the denoising effect. This demonstrates that through a scientific sensor placement strategy, the proposed method effectively improves signal acquisition efficiency and reduces environmental noise interference, maintaining excellent performance even in complex scenarios.

[0110] An electronic device comprises: a processor; and a memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the sensor array multi-objective optimization method for magnetic field signal denoising as described above.

[0111] A computer-readable storage medium is used to store a program, which, when executed by a processor, implements the sensor array multi-objective optimization method for magnetic field signal denoising as described above.

[0112] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0113] The present invention is described with reference to flowcharts and / or block diagrams of systems, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0114] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0115] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0116] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0117] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A multi-objective optimization method for sensor arrays for magnetic field signal denoising, characterized in that: The following steps are involved: S1: Establish a magnetic field interference model under constrained mode and perform transient analysis to obtain the time series data of magnetic field signals collected by all sensors; S2: Setting an optimization index based on the magnetic field signal data, using the optimization index as a multi-objective fitness function, optimizing the multi-objective fitness function, and obtaining the optimal auxiliary sensor layout solution when the fitness function value is maximized; The optimization indicators include optimization indicator 1 and optimization indicator 2, where optimization indicator 1 is the comprehensive dynamic time warping distance between the sensor magnetic field signal data of the selected candidate auxiliary sensor layout and the main sensor, and optimization indicator 2 is the comprehensive Chatterjee correlation coefficient between the sensor magnetic field signal data of the selected candidate auxiliary sensor layout and the main sensor; The optimization index 1 and the optimization index 2 are used as the multi-objective fitness function, where the optimization index 1 is negative and the optimization index 2 is positive; ； ； in, Index_1 Indicates optimization index 1, Index_2 Indicates optimization index 2; n sensor Indicates the number of selected auxiliary sensors, m axis represents the direction of magnetic field strength, m axis When it is 1, it indicates the X axis. m axis When it is 2, it indicates the Y axis. m axis When it is 3, it indicates the Z axis; Indicates the data collected by the main sensor m axis The length of the axis data, Indicates the data collected by the nth selected auxiliary sensor m axis The length of the axis data; Represented as time series data The i-th element in Similarly; w is the Sakoe-Chiba band constraint window parameter, and the symmetrical band area |i j∣≤w; l g Indicates that in time series S Sorted index array S order In time series S No. g The rank corresponding to the elements, that is, the sequence S is greater than the number of elements in , r g yes l g a copy of; dtw(i,j) is the dynamic time warping distance; Normalize the calculated optimization index 1 and optimization index 2, and weight them according to the optimization index weight ratio to obtain the comprehensive fitness value of a single chromosome. The formula is: ； in, and They are respectively expressed as the weight ratio of optimization indicators, and They are respectively represented as the normalized results of optimization index 1 and optimization index 2 within the population; The optimal auxiliary sensor layout scheme refers to the optimal auxiliary sensor layout scheme selected from N candidate auxiliary sensor layouts; The adaptive dual structure coding genetic algorithm is used to optimize the multi-objective fitness function; S3: Based on the optimal auxiliary sensor layout plan, establish the main sensor and the optimal auxiliary sensor array; S4: In an environment affected by interference sources, the noise signal of the optimal auxiliary sensor array is processed based on the trained deep learning network model to obtain the denoised magnetic field signal of the main sensor.

2. A sensor array multi-objective optimization method for magnetic field signal denoising according to claim 1, characterized in that: The magnetic field interference model under the constrained mode in step 1 is established using finite element analysis software. The process is as follows: Finite element analysis software is used to build a magnetic field interference model. The magnetic field interference model includes the interference source and the disturbed plane, and N+1 sensors for detecting magnetic field signals are set on the surface of the disturbed plane, of which one sensor is the main sensor located at the main position of the disturbed plane, and the remaining N sensors are auxiliary sensors to be selected and evenly distributed radially on the disturbed plane.

3. The sensor array multi-objective optimization method for magnetic field signal denoising according to claim 2, characterized in that: The process of obtaining the magnetic field signal data collected by all sensors in step 1 is as follows: In the finite element analysis software, the material properties of the established magnetic field interference model are defined, and meshing is performed. Based on the selected unit type and mesh density, fixed constraints are imposed, and time-varying current loads are defined. Transient analysis is performed to obtain magnetic field signal data in three directions collected by all sensors.

4. The sensor array multi-objective optimization method for magnetic field signal denoising according to claim 1, characterized in that: The process of optimizing the multi-objective fitness function using the genetic algorithm with adaptive dual structure coding is as follows: S21: Initialize population size, maximum number of iterations, number of auxiliary sensors to be selected, number of selected auxiliary sensors, mutation selection probability, crossover selection probability, optimization index weight ratio, number of elite individuals, and stagnation threshold iteration number; S22: Constructing a dual-structure coded population individual. The dual structure of individual chromosomes consists of two lines: position code and confirmation code. The position code indicates that the auxiliary sensor can select a layout position, and the confirmation code indicates whether the position is selected. S23: Setting multi-objective fitness function; S24: Optimizing the multi-objective fitness function to obtain an optimal auxiliary sensor layout solution when the fitness function value is maximized.

5. The sensor array multi-objective optimization method for magnetic field signal denoising according to claim 4, characterized in that: S24: Optimize the multi-objective fitness function to obtain the optimal auxiliary sensor layout solution when the fitness function value is maximized. The specific process is as follows: S241: Calculate the comprehensive fitness value of all chromosomes in the initial population, and after roulette wheel selection, partial matching crossover, inversion mutation, and using the elite preservation strategy, retain the chromosomes with the highest fitness value of elite individuals to generate a new population; S242: Propose a nonlinear adaptive regulation mechanism by quantifying population distribution characteristics in real time; By statistically analyzing the standard deviation and range distribution of population fitness, a composite diversity evaluation index is constructed, and the Sigmoid function is used to realize the crossover probability. P c and mutation probability P m Smooth dynamic adjustment of , the formula is: ； Among them, the comprehensive fitness value of the contemporary population is set to ,have: ； ； P c is the crossover probability, P m is the mutation probability, D is a composite diversity index, e is a natural constant, P size represents the population size, f p Indicates the first p The comprehensive fitness value of chromosomes, f max represents the maximum comprehensive fitness in the population, f min represents the minimum comprehensive fitness in the population, is the average fitness, is the standard deviation of fitness, is the range normalization factor; S243: In S241, a nonlinear adaptive adjustment mechanism is added, and this step is repeated until the maximum number of iterations is reached or the chromosome with the maximum fitness in the population does not change after the stagnation threshold number of iterations, a large fitness chromosome is obtained, and the dual structure encoding is deconstructed to obtain the optimal auxiliary sensor layout.

6. The sensor array multi-objective optimization method for magnetic field signal denoising according to claim 1, characterized in that: The optimal auxiliary sensor layout scheme places the main sensor close to the detected location. The optimal auxiliary sensor layout is obtained by comprehensive optimization using a multi-objective fitness function. The optimal auxiliary sensor layout changes with the location and type of the interference source. The interference source affects the environment when the current-carrying conductor is not energized and only the interference source generates a magnetic field.

7. The sensor array multi-objective optimization method for magnetic field signal denoising according to claim 1, characterized in that: During the training process of the deep learning network model, the noise signal corresponding to the optimal auxiliary sensor in the noise signal of the optimal auxiliary sensor array is taken as input, and the noise signal corresponding to the main sensor is taken as output.

8. The sensor array multi-objective optimization method for magnetic field signal denoising according to claim 7, characterized in that: The noise signal of the optimal auxiliary sensor array is processed based on the trained deep learning network model to obtain the denoised magnetic field signal of the main sensor. The process is as follows: The noise signal of the optimal auxiliary sensor array is reconstructed using the trained deep learning network model to output the noise signal at the main position; The noise signal at the main position is subtracted from the disturbed signal obtained by the main sensor to obtain a denoised magnetic field signal.

Images