A Multi-Parameter Extraction Method for Superparamagnetic Effect Based on Improved Particle Swarm Optimization Algorithm
By improving the particle swarm optimization algorithm and combining the Cole-Cole magnetization model and Logistic chaotic mapping, the problem of extracting multiple parameters of the superparamagnetic effect was solved, and the magnetization parameters were accurately extracted, thus improving the accuracy of data processing.
Patent Information
- Application Number
- CN202310526439.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-11
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-05-11
AI Technical Summary
Existing technologies cannot effectively explain and extract the effects of superparamagnetism, leading to errors in the processing of magnetic environment data.
An improved particle swarm optimization algorithm is adopted, which introduces the Cole-Cole magnetic susceptibility model and the nonlinear changing inertial weight of the Logistic chaotic mapping, and combines it with an adaptive particle velocity and position update strategy to achieve accurate extraction of multiple parameters of the superparamagnetic effect.
It improves the accuracy of magnetization parameters, avoids immature convergence and local minima, and achieves accurate extraction of multiple parameters of superparamagnetic effect.
Smart Images

Figure CN116562018B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method for extracting multiple parameters of superparamagnetic effect based on an improved particle swarm optimization algorithm. Background Technology
[0002] The Time-Domain Electromagnetic Method (AEM) has advantages such as speed, efficiency, and wide detection range, and has been widely used in various fields such as mineral resource exploration, environmental monitoring, and basic geological surveys (Lei, 2006; Zhang, 2005). During the transient electromagnetic method, the on-time phase of the emitted current magnetizes underground magnetic particles; after the current is turned off, the magnetic particles lose their magnetization, and the receiving system observes a decaying signal with an approximately -1 power rate. This process is called the superparamagnetic effect. Domestic and international experimental results show that the superparamagnetic effect exists not only in various rocks and minerals (Buselli, 1982), but also in magnetite and hematite particles in the 0.1-15 μm radius range (Bournas, 2017; Dabas and Skinner, 2002). The superparamagnetic effect is an important characterization of information about underground magnetic media, but traditional parameter extraction methods, when processing measured data of the magnetic environment, may lead to incorrect interpretation of the data due to neglecting this effect. Magnetization parameters have a significant impact on the superparamagnetic effect. Therefore, it is crucial to accurately extract the magnetization parameters to achieve accurate interpretation of data containing the superparamagnetic effect.
[0003] Chinese patent CN113887106A discloses a three-dimensional numerical simulation method for induction-magnetization effects based on the Chikazumi model. It introduces the Chikazumi complex magnetic susceptibility model into the frequency domain Maxwell equations, proposes a matrix operation form for complex permeability, optimizes the calculation method of the governing equations, and re-establishes a scientific mapping matrix between permeability and magnetic field response based on the positional relationship of the magnetic field on the grid after differential discretization. It also defines a special boundary setting method for magnetic anomalies and uses a sine / cosine transform numerical filtering algorithm to transform the frequency domain results to the time domain. This method can achieve three-dimensional numerical simulation of induction-magnetization effects in the time domain, but it cannot accurately extract magnetization parameters.
[0004] Chinese patent CN112666618A discloses a method for extracting geometric and physical property multi-feature parameters for multiphase media. Based on a generalized equivalent polarization model, the underground medium is defined as consisting of surrounding rock, strongly polarized medium, and weakly polarized medium. The time-domain magnetic field response under trapezoidal wave excitation in two-phase and multiphase media is calculated, and the objective function is formed by combining it with magnetic field data collected by a superconducting quantum sensor. The conductivity of the surrounding rock is calculated based on early measured magnetic field data as a constraint condition. For cases where the extracted parameters are multi-dimensional and have few constraints, dimensions are set to correspond to two-phase and multiphase media respectively, and the quantum particle swarm optimization algorithm is used to extract the parameters separately. This method effectively achieves global extraction of multi-dimensional parameters, but it cannot explain the influence of superparamagnetic effects.
[0005] While the methods for explaining and extracting feature parameters of the superparamagnetic effect as described above have been published, domestic and international patents have not yet addressed the use of particle swarm optimization algorithms to simultaneously extract multiple superparamagnetic parameters. Therefore, this invention employs forward numerical simulation of a layered superparamagnetic model based on the Cole-Cole magnetic susceptibility model, simultaneously achieving global extraction of multiple parameters of the superparamagnetic effect. A nonlinearly varying inertial weight based on a Logistic chaotic mapping is introduced to ensure a larger particle search space. An adaptive particle velocity and position update strategy is adopted to enhance the algorithm's global search capability in the early stages of evolution and demonstrate strong local exploration capability in the later stages, thereby improving the accuracy of simultaneous extraction of multiple parameters. Summary of the Invention
[0006] The technical problem this invention aims to solve is to provide a multi-parameter extraction method for superparamagnetic effects based on an improved particle swarm optimization algorithm. By introducing the Cole-Cole magnetic susceptibility model, numerical simulation of layered superparamagnetic effects is achieved, and the influence of magnetization parameters on the superparamagnetic effect is analyzed. By employing a nonlinearly changing inertia weight based on a Logistic chaotic mapping, the accuracy of magnetization parameter extraction is ensured, improving phenomena such as premature convergence or local minima that occur during the search process. An adaptive particle velocity and position update strategy is proposed by defining the adaptiveness judgment condition of particles, ultimately achieving accurate extraction of multiple parameters of the superparamagnetic effect. This invention is implemented as follows: A multi-parameter extraction method for superparamagnetic effects based on an improved particle swarm optimization algorithm:
[0007] Includes the following steps:
[0008] 1) The formula for the superparamagnetic response generated by layered magnetized media is derived using Maxwell's equations in the frequency domain;
[0009] 2) To simulate and analyze the superparamagnetic effect, a Cole-Cole magnetic susceptibility model with a log-normal distribution in the relaxation time of magnetic particles is introduced. Using magnetic vector potential iteration, the magnetic field strength H generated by the induction-magnetization effect under the layered model is derived.z Magnetic induction intensity B z ;
[0010] 3) For the extraction of multiple parameters of superparamagnetic effect, first set the spatial dimension D of the particle swarm, input relevant parameters such as population size M, number of iterations K, etc., initialize the position X, velocity V, individual optimal value Pbest, population optimal value Gbest of each particle, and construct the objective function.
[0011] 4) To ensure a larger particle search space, a nonlinear change inertial weight based on Logistic chaotic mapping is used, and the inertial weight is updated according to the number of iterations;
[0012] 5) The adaptive particle velocity and position update strategy uses a linear combination formula of individual optimal solution and group optimal solution to replace the corresponding optimal solution, and achieves multi-parameter extraction of superparamagnetic response through multiple iterations;
[0013] In step 1), the magnetic vector potential in the layered superparamagnetic response satisfies:
[0014]
[0015] The subscript i represents the stratigraphic level, and k is the wave number, which can be written as: When the frequency is less than 10 5 At Hz, the displacement current can be ignored, and the wave number is k. 2 =iμσω;
[0016] Using the orthogonality of Hankel integrals, the vector potential expression can be derived as follows:
[0017]
[0018] Where m is the magnetic moment at z = -h (z ≤ 0), and the magnetic moment of each layer is m i for The wave number of each layer is k i 2 =iμ i (ω)σ i ω, the permeability μ of each layer i (ω) can be expressed as:
[0019]
[0020] function for:
[0021]
[0022] In step 2), the Cole-Cole magnetic susceptibility model is:
[0023]
[0024] Where χ0 is the zero-frequency magnetic susceptibility of the material, χ ∞ Let ω be the magnetic susceptibility at infinite frequency, iω be the complex frequency variable, and τ be the magnetic susceptibility at infinite frequency. c The magnetic susceptibility time constant is α Controlling the relaxation time distribution;
[0025] The magnetic field strength H at the air receiver point can be obtained through iteration. z for
[0026]
[0027] The magnetic susceptibility of each layer can be approximated as:
[0028]
[0029] Where z i =h + d1 + d2 + ... + d i The magnetic flux density can be derived as follows:
[0030]
[0031] In step 3), the position X, velocity V, individual optimal value Pbest, and population optimal value Gbest of the i-th particle can be expressed as:
[0032] X i =(x i1 ,x i2 ,…,x iD )i=1,2,…,M (9)
[0033] V i =(v i1 ,v i2 ,…,v iD )i=1,2,…,M (10)
[0034] Pbest=(p i1 ,p i2 ,…,p iD )i=1,2,…,M (11)
[0035] Gbest=(p g1 ,p g2 ,…,p gD g = 1, 2, ..., M (12)
[0036] Next, we set the objective function, which is to configure the fitness function Y(x) of the improved particle swarm optimization algorithm.
[0037]
[0038] Among them, Bm This is the actual measured electromagnetic response result, B t The result is the induction-magnetization response calculated using a forward modeling algorithm. This transforms the magnetization parameter extraction problem into a minimum optimization problem. Furthermore, the fitting function can be constructed as a multi-parameter function to achieve simultaneous extraction.
[0039] In step 4), the linear change inertia weight based on Logistic regression is defined as follows:
[0040]
[0041] Among them, w min =0.4, w max =0.9; r(k) is a random number generated iteratively, specifically calculated as follows:
[0042] r(k+1)=4r(k)(1-r(k))r(0)=rand (15)
[0043] Where r(0)≠{0,0.25,0.75,1};
[0044] In step 5), the adaptive decision condition for the particle is:
[0045]
[0046] Where, f(x) i (k) represents the fitness value of the current particle, and N is the number of particles in the population. By calculating and comparing the average fitness of the particles in the current population and the adaptive decision criterion for each particle, a better strategy for updating the particle's position and velocity can be selected. The adaptive particle velocity and position update is defined as follows:
[0047]
[0048] The two formulas of the adaptive update strategy take effect in the early and late stages of the search, respectively, allowing the algorithm to have a wider search range during iteration and increasing the probability of finding the optimal solution. By setting the learning factors c1 = c2 = 2, r1, r2 = rand, and the threshold parameter δ = 1, multiple iterations can achieve multi-parameter extraction of the superparamagnetic effect.
[0049] Compared with the prior art, the beneficial effects of this invention are as follows: it can overcome the data misinterpretation caused by neglecting the superparamagnetic effect when processing measured data of magnetic environment in traditional parameter extraction methods, and realize the accurate extraction of magnetization parameters, laying the foundation for data processing of superparamagnetic effect. Attached Figure Description
[0050] Figure 1This is a schematic diagram of a multi-parameter extraction method for superparamagnetic effect based on an improved particle swarm optimization algorithm;
[0051] Figure 2 This is a schematic diagram illustrating the effect of zero-frequency magnetic susceptibility on the superparamagnetic effect;
[0052] Figure 3 This is a schematic diagram illustrating the effect of electrical conductivity on the superparamagnetic effect;
[0053] Figure 4 This is a schematic diagram illustrating the effect of fractional order on the superparamagnetic effect;
[0054] Figure 5 This is a graph showing the relationship between inertia weight and the number of iterations;
[0055] Figure 6 This is a fitness curve of the Cole-Cole magnetic susceptibility model extracted from multiple parameters; Detailed Implementation
[0056] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0057] Example
[0058] See Figure 1 A method for extracting multiple parameters of superparamagnetic effect based on an improved particle swarm optimization algorithm, comprising:
[0059] 1) Set the radius of the transmitting coil to 10m, the transmitting current to 300A, the magnetization layer to be the first layer with a thickness of 60m, and the magnetic susceptibility to be 0 at infinite frequency. The magnetic susceptibility of the other two layers is 0, and the height of the transmitting coil is 30m.
[0060] 2) Through forward modeling, the influence of magnetization parameters on the superparamagnetic effect was analyzed, and the zero-frequency magnetic susceptibility, fractional order, and conductivity of the magnetization layer were determined as relevant parameters. A Cole-Cole layered magnetic susceptibility model was set up, with the zero-frequency magnetic susceptibility χ0 = 0.03, the fractional order α = 0.23, and the conductivity σ = 0.01.
[0061] 3) Set the basic parameters of the particle swarm algorithm: set the spatial dimension of the particle swarm D=3, the population size M=30, the number of iterations K=1000, the learning factors c1=c2=2, r1,r2=rand, and the threshold parameter δ=1.
[0062] 4) Algorithm initialization: Set particle position boundaries to [0.01, 0.1; 0.1, 1; 0.005, 0.05] and velocity boundaries to [-0.001, 0.001; -0.01, 0.01; -0.0005, 0.0005]. Initialize the position X, velocity V, individual optimal value Pbest, and population optimal value Gbest for each particle.
[0063] 5) Calculate the fitness of the particles, compare whether the individual and group fitness are better than the historical values. If they are not better, recalculate the particle fitness and compare them again. If they are better, update the individual optimal value and the global optimal value.
[0064] 6) The definition (16) is the adaptive degree judgment condition of the particle. The latest position and velocity are obtained by using the velocity and position update strategy given by equation (17).
[0065] 7) When the maximum number of iterations (1000) is reached, the calculation is complete, and the extracted magnetic susceptibility, fractional order, and conductivity are displayed.
[0066] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A superparamagnetic effect multi-parameter extraction method based on an improved particle swarm algorithm, characterized in that, Includes the following steps: 1) The formula for the induction-magnetization response generated by the layered magnetization medium is derived using Maxwell's equations in the frequency domain; 2) To realize the simulation and analysis of superparamagnetic effect, the Cole-Cole magnetic susceptibility model is introduced, in which the relaxation time of magnetic particles is logarithmically normally distributed. The magnetic field intensity , magnetic induction intensity generated by superparamagnetic effect under the layered model is derived by using magnetic vector iteration. 3) For the extraction of multiple parameters of superparamagnetic effect, first set the spatial dimension d of the particle swarm, input the relevant parameters, initialize the position x, velocity v, individual optimal value pbest, population optimal value gbest of each particle, and construct the objective function. 4) To ensure a larger particle search space, a nonlinear change inertial weight based on Logistic chaotic mapping is used, and the inertial weight is updated according to the number of iterations; 5) The adaptive particle velocity and position update strategy uses a linear combination formula of individual optimal solution and group optimal solution to replace the corresponding optimal solution, and achieves multi-parameter extraction of superparamagnetic response through multiple iterations; In step 1), the vector potential in the layered superparamagnetic response satisfies: (1) subscript denotes the layer, is the wave number, written as ; when the frequency is less than , the displacement current is neglected, in which case ; Using the orthogonality of Hankel integrals, the vector potential expression is derived as follows: (2) wherein is magnetic moment, the magnetic moment of each layer is the wave number of each layer is the magnetic permeability of each layer is represented as: (3) function for: (4) In step 2), the Cole-Cole magnetic susceptibility model is: (5) wherein is the zero frequency magnetic susceptibility of the material, is the magnetic susceptibility at infinite frequency, is the complex frequency variable, is the magnetic susceptibility time constant, controls the relaxation time distribution; By iteration the magnetic field strength at the aerial reception point is found is: (6) The magnetic susceptibility of each layer is approximately: (7) wherein The magnetic induction B is derived as (8) In step 3), the position x, velocity v, individual optimal value pbest, and population optimal value gbest of the i-th particle are represented as follows: (9) (10) (11) (12) Then, the objective function is set, that is, the fitness function of the configuration improved particle swarm algorithm : (13) wherein, is the actual measured electromagnetic response result, is the superparamagnetic response result calculated by the forward algorithm; that is, the magnetization parameter extraction problem is converted into a minimum optimization problem; and the fitting function is constructed as a multi-parameter function to achieve simultaneous extraction; In step 4), the linear change inertia weight based on Logistic regression is defined as follows: (14) wherein , ; is an iteratively generated random number, calculated as: (15) wherein ; In step 5), the adaptive decision condition for the particle is: (16) in, Let M be the fitness value of the current particle, and M be the number of particles in the population. By calculating and comparing the average fitness of the particles in the current population and the adaptive decision criteria for each particle, a better strategy for updating the particle's position and velocity can be selected. The adaptive particle velocity and position update is defined as follows: (17) Among them, learning factor , Threshold parameter ; The two formulas of the adaptive update strategy take effect in the early and late stages of the search, respectively, allowing the algorithm to have a wider search range during the iteration process and increasing the probability of finding the optimal solution; multiple iterations are used to extract multiple parameters of the superparamagnetic effect.
Citation Information
Patent Citations
Induction-magnetization effect three-dimensional numerical simulation method based on Chikazumi model
CN113887106A
Geometrical physical property multi-feature parameter extraction method for multi-phase medium
CN112666618A
Motor cascade active disturbance rejection control parameter optimization method based on improved particle swarm
CN112671291A
