Method for measuring year of implement based on residual magnetism detection technology
Through electromagnetic wave processing and neural network analysis based on residual magnetic detection technology, the problem of insufficient measurement accuracy of complex magnetic signals and multi-material instruments in the existing technology is solved, and high-precision instrument date is realized, which is suitable for complex and multi-layer structures. Cultural relics.
Patent Information
- Application Number
- CN202411232756.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-12-03
AI Technical Summary
When the existing instrument year measurement methods face complex magnetic signals and multi-material instruments, the measurement results are not accurate enough and rely on a single physical characteristic or material, making dating accuracy difficult to guarantee.
Using a method based on residual magnetic detection technology, an electromagnetic wave generator is used to generate electromagnetic waves for specific frequency to detect the instrument, and the electromagnetic wave data is processed through a neural network model, resonance phenomena are judged, and the age information of the instrument is determined in combination with a mapping table. The method includes wavelet decomposition, filtering, standardization processing, time-domain and frequency-domain feature extraction, analysis of electromagnetic signals using twin LSTM network models, and calculation of resonance frequency to infer historical information of objects.
It improves measurement sensitivity and accuracy, reduces manual analysis errors, and achieves accurate chronology of complex objects, without loss and non-invasiveness.
Smart Images

Figure CN119064381A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of cultural relics identification, and in particular to a method for determining the age of artifacts based on residual magnetism detection technology. Background Art
[0002] After an object undergoes drastic physical or chemical changes, its internal magnetic material will produce residual magnetism. This phenomenon is due to the rearrangement of magnetic domains in the body under the action of an external magnetic field. However, when the external force disappears, these magnetic domains do not completely return to their original disordered state, but retain a certain amount of residual magnetism. This process is called remanent magnetization.
[0003] As time goes by, the residual magnetic signal of an object will gradually decay. This phenomenon of change over time is called magnetic relaxation. Magnetic relaxation is a complex dynamic process, which is affected by many factors such as temperature, environmental conditions and the microstructure of the material itself. By accurately measuring and analyzing the attenuation curve of the residual magnetic signal of an object, the time point when the object was last subjected to drastic physical or chemical changes can be obtained. This technology is similar to radiocarbon dating, but it uses changes in magnetic properties rather than the decay of radioactive isotopes. Therefore, through the study of the residual magnetic signal of an object, a non-destructive and non-invasive dating method can be achieved, which is particularly suitable for the dating of craft products, archaeological relics and geological samples.
[0004] However, most current methods for dating artifacts use average values for inference, and lack the ability to effectively interpret complex magnetic signals. Since residual magnetic signals are affected by multiple factors such as the environment, material properties, and temperature, inferring only by average values will lead to large errors in the dating results, and it is impossible to effectively distinguish the magnetic characteristics of different layers of artifacts, especially in artifacts with multi-layer structures or sites with drastic environmental changes, the dating accuracy is difficult to guarantee. In addition, current dating methods rely on a single physical property or a single material for determination. For example, radiocarbon dating can only be used for carbon-containing organic matter, and magnetic dating methods can only process information on one type of material when determining artifacts made of multiple heterogeneous materials. When faced with complex, multi-material artifacts, the determination results are not accurate enough. Therefore, there is an urgent need for an artifact dating method based on residual magnetic detection technology to solve such problems. Summary of the invention
[0005] To this end, the present invention provides a method for determining the age of an artifact based on residual magnetic detection technology, so as to overcome the problem that the prior art lacks effective interpretation capabilities when faced with complex magnetic signals. Since the residual magnetic signal is affected by multiple factors such as the environment, material properties, temperature, etc., inferring only by the average value will lead to large errors in the dating results, and it is impossible to effectively distinguish the magnetic characteristics of different levels of the artifact. The determination method relies on a single physical property or a single material for determination. When faced with complex, multi-material artifacts, the determination results are not accurate enough.
[0006] To achieve the above object, the present invention provides a method for determining the age of an object based on residual magnetism detection technology, comprising the following steps:
[0007] Step S1, using an electromagnetic wave generator to generate and emit electromagnetic waves of a specific frequency to detect the object;
[0008] Step S2, receiving electromagnetic wave data reflected from the object and recording its intensity and phase change;
[0009] Step S3, constructing a neural network model, using the model to process the received electromagnetic wave data, and comparing it with the standard data to determine whether there is a resonance phenomenon;
[0010] Step S4, presetting a mapping table, the mapping table contains age information corresponding to electromagnetic waves of different frequencies, and after confirming the resonant frequency, searching the corresponding age of the artifact through the mapping table;
[0011] Step S5, converting the matching result into age information.
[0012] Further, in step S3, the electromagnetic wave data includes:
[0013] Resonance contrast signal 1, i.e., the electrical signal converted from the electromagnetic wave data of the detected object;
[0014] Resonance contrast signal 2, i.e., the detected electromagnetic signal;
[0015] The data format includes time domain data and frequency domain data. The time domain data directly records the intensity of electromagnetic waves that changes with time, and the frequency domain data is obtained by Fourier transform of the time domain data.
[0016] Furthermore, in step S3, the received electromagnetic wave data is processed in the following manner:
[0017] Perform the following operations on electromagnetic wave data:
[0018] Wavelet decomposition, obtain wavelet coefficients of different scales, threshold the wavelet coefficients to remove the noise;
[0019] Wavelet reconstruction is used to obtain the denoised signal, and low-pass, high-pass and band-pass filters are used for filtering;
[0020] Standardization is performed to process the signal amplitude of different sample data to the interval [-1,1]. where μ is the mean, σ is the standard deviation, and X and X ′ Represent the signal amplitude before and after processing respectively.
[0021] Furthermore, in step S3, the method of processing the received electromagnetic wave data also includes:
[0022] Perform time domain feature extraction, including peak value, mean μ, variance σ 2 ;
[0023] Perform frequency domain feature extraction, including main frequency, spectrum energy E, E = ∑|X(k)| 2 , where X(k) represents the signal spectrum and k represents the spectrum length;
[0024] The extracted frequency domain features also include spectral entropy, which is a measure of uncertainty in the energy distribution of the spectrum, denoted as H, H = -∑p(k)log2p(k), where Represents the energy distribution probability of the spectrum;
[0025] Use short-time Fourier transform (STFT) to convert from the time domain to the frequency domain. The STFT calculation formula is: X(n,ω) = ∑x(m)·w(nm)·e -jωm , where x(m) is the signal, w(nm) is the window function, n is the time index, ω is the frequency, and e -jωm It is a negative exponent, X(n,ω), which means that the frequency domain features include the energy distribution and spectrum centroid of electromagnetic wave data.
[0026] Furthermore, in step S3, the method of processing the received electromagnetic wave data also includes:
[0027] The extracted features are organized into a neural network input format, and the time series data is input as a one-dimensional vector sequence. Each sequence corresponds to a feature vector sequence of an electromagnetic signal, denoted as x1, x2, …, x T , where x t is the feature vector at time t, and T is the sequence length.
[0028] Furthermore, in step S3, the neural network model is constructed as follows:
[0029] The twin LSTM is used to construct a neural network model. The neural network model consists of two identical LSTM networks. The two networks share the same weights and parameters. Each LSTM network receives the feature vector sequence of the electromagnetic signal as input and outputs a vector of fixed length, namely the embedding vector.
[0030] Furthermore, in step S3, the neural network model calculation process includes:
[0031] Two LSTM networks process different input sequences x respectively. (a) and x (b) , and calculate the final hidden state of different sequences and The L1 norm difference between them generates a similarity score y.
[0032] Furthermore, in step S3, the neural network model calculation process is specifically as follows:
[0033] Will and Input into two LSTM networks respectively;
[0034] The output of each LSTM network is a sequence of hidden states. For the first LSTM network, the hidden state is For the second LSTM network, the hidden state is
[0035] Calculate the L1 norm, or Manhattan distance, between the last hidden states of two LSTM networks: in, Represents the L1 norm between vectors, calculated as: in and are the components of the hidden state vector;
[0036] The distance is converted into a similarity score y through an exponential function, namely:
[0037] Furthermore, the method of determining whether there is a resonance phenomenon in step S3 is:
[0038] An alternating current with a frequency of ω is input into the circular coil, and the alternating current forms an alternating magnetic field, which is used to detect cultural relics containing magnetic minerals and having thermal remanent magnetism; when nuclear magnetic resonance is generated inside the artifact under the alternating magnetic field, the atomic nuclei of the magnetic minerals in the artifact will form a macroscopic magnetic moment; the magnetic moment produces a precession motion in the alternating magnetic field, and its precession frequency is ω i , the resonant frequency is also ω i ;
[0039] When an artifact containing magnetic minerals and having thermal residual magnetism is detected by an alternating current with a frequency of ω, if nuclear magnetic resonance is generated inside the artifact, its resonance frequency is ω i ,ω i represents the resonance frequency of the ith state or the ith particle;
[0040] According to the Langevin equation, the motion of the magnetic moment μ in the external magnetic field B is: Where γ is the cyclotron ratio, M represents the magnetic moment of the object, and B represents the external magnetic field. represents the rate of change of magnetic moment with time;
[0041] Determine the paleomagnetic field strength B i , Among them B i represents the magnetic field strength of the ith state or the ith particle, and the measured resonant frequency ω i and the known cyclotron ratio γ, calculate the paleo-geomagnetic field strength B at the time of production i ;
[0042] By adding B i Map to the mapping table to find the corresponding time coordinates and determine the formation age of the cultural relics;
[0043] The mapping data comes from the prior information of the remanent magnetic information of known objects of different types and ages.
[0044] Furthermore, in step S4, the measured paleomagnetic field intensity is matched with a preset age mapping table.
[0045] Compared with the prior art, the present invention has the following beneficial effects:
[0046] The present invention determines the age of artifacts based on residual magnetism detection technology. Specifically, the residual magnetism detection technology is mainly based on the principle of magneto-electric conversion. The magnetic direction and strength of an object are detected and analyzed by electromagnetic waves. Since the residual magnetism signal is extremely weak and complex, it is necessary to use high-precision electromagnetic detection equipment to perform detailed electromagnetic detection and comparison in different frequency ranges. In order to improve the accuracy of the detection, a neural network model is used to compare and analyze the standard electromagnetic wave and the electromagnetic wave of the object to be measured. When a resonance phenomenon occurs, the information of the electromagnetic frequency can be determined, thereby inferring the historical information of the object. For example, when a material with thermal residual magnetism is placed in an alternating magnetic field within a certain frequency or frequency range, the alternating magnetic field will produce an induction phenomenon with the thermal residual magnetism of the material. This induction phenomenon is similar to the nuclear magnetic resonance effect. The user can set the target frequency of the electromagnetic wave by inputting instructions. The target frequency corresponds to specific age information in a preset mapping table. Subsequently, the electromagnetic wave generator in the electromagnetic detector generates electromagnetic waves of the target frequency according to the user's instructions, and emits these electromagnetic waves through the detection component. If the emitted electromagnetic waves resonate with the thermal residual magnetism electromagnetic waves of the object to be identified, the age of the object can be quickly and conveniently determined.
[0047] The present invention uses a highly sensitive magnetic measuring device to capture weak magnetic signals on the surface and inside of an object, greatly improving the measurement sensitivity. A mathematical model is then established to fit the magnetic relaxation process, the initial value of the residual magnetic signal and its decay rate are deduced, and the year when the object enters a stable state is determined in combination with known parameters.
[0048] The present invention, through the refined modeling of the magnetic relaxation process, can achieve accurate estimation of the time point when an object last experienced drastic physical and chemical changes. Through the established mathematical model, this process is fitted and analyzed, and the initial value of the residual magnetic signal and its decay rate are accurately calculated. The measurement method based on magnetic properties adopted by the present invention is similar to the radiocarbon dating method, and utilizes the unique properties of magnetic materials to achieve non-destructive and non-invasive dating of cultural relics.
[0049] The present invention combines the neural network model with the frequency domain analysis technology, greatly improving the accuracy of signal processing and analysis. It uses an electromagnetic wave generator and a detection component to perform detailed electromagnetic detection of objects within different frequency ranges. By comparing and analyzing the standard electromagnetic waves with the electromagnetic waves of the object being measured through the neural network, it can quickly determine the electromagnetic frequency information when resonance occurs, thereby inferring the historical information of the object; it greatly reduces the errors in manual analysis and improves the measurement efficiency through automated processing. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Figure 1 It is a schematic flow chart of the method for determining the age of an object based on the residual magnetism detection technology of the present invention.
[0051] Figure 2 Schematic diagram of the model structure of the neural network model of the present invention. DETAILED DESCRIPTION
[0052] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0053] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0054] The present invention is further described in detail below in conjunction with the accompanying drawings:
[0055] Example 1
[0056] See also Figure 1 and Figure 2 The present invention provides a method for determining the age of an artifact based on residual magnetism detection technology, comprising:
[0057] Step S1, using an electromagnetic wave generator to generate and emit electromagnetic waves of a specific frequency to detect the object;
[0058] Step S2, receiving electromagnetic wave data reflected from the object and recording its intensity and phase change;
[0059] Step S3, constructing a neural network model, using the model to process the received electromagnetic wave data, and comparing it with the standard data to determine whether there is a resonance phenomenon;
[0060] In step S3, the electromagnetic wave data includes:
[0061] Resonance contrast signal 1, i.e., the electrical signal converted from the electromagnetic wave data of the detected object;
[0062] Resonance contrast signal 2, i.e., the detected electromagnetic signal;
[0063] The data format includes time domain data and frequency domain data. The time domain data directly records the intensity of electromagnetic waves that changes with time. The frequency domain data is obtained by Fourier transform of the time domain data and displays the intensity of each frequency component in the form of a spectrum diagram.
[0064] In step S3, the received electromagnetic wave data is processed in the following manner:
[0065] Perform the following operations on electromagnetic wave data:
[0066] Wavelet decomposition, obtain wavelet coefficients of different scales, threshold the wavelet coefficients to remove the noise;
[0067] Wavelet reconstruction is used to obtain the denoised signal, and low-pass, high-pass and band-pass filters are used for filtering;
[0068] Standardization is performed so that the signal amplitude of different sample data is processed to the interval [-1,1]. where μ is the mean, σ is the standard deviation, and X and X ′ Respectively represent the signal amplitude before and after processing;
[0069] In step S3, the method of processing the received electromagnetic wave data also includes:
[0070] Perform time domain feature extraction, including peak value, mean μ, variance σ 2 ;
[0071] Perform frequency domain feature extraction, including main frequency, spectrum energy E, E = ∑|X(k)| 2 , where X(k) represents the signal spectrum and k represents the spectrum length;
[0072] The extracted frequency domain features also include spectral entropy, which is a measure of uncertainty in the energy distribution of the spectrum, denoted as H, H = -∑p(k)log2p(k), where Represents the energy distribution probability of the spectrum;
[0073] Use short-time Fourier transform (STFT) to convert from the time domain to the frequency domain. The STFT calculation formula is: X(n,ω) = ∑x(m)·w(nm)·e -jωm , where x(m) is the signal, w(nm) is the window function, n is the time index, ω is the frequency, and e -jωm It is a negative index, X(n,ω), indicating that the frequency domain features include the energy distribution and spectrum centroid of electromagnetic wave data;
[0074] In step S3, the method of processing the received electromagnetic wave data also includes:
[0075] The extracted features are organized into a neural network input format, and the time series data is input as a one-dimensional vector sequence. Each sequence corresponds to a feature vector sequence of an electromagnetic signal, denoted as x1, x2, …, x T , where x t is the feature vector at time t, T is the sequence length;
[0076] In step S3, the neural network model is constructed as follows:
[0077] The twin LSTM is used to construct a neural network model. The neural network model consists of two identical LSTM networks. The two networks share the same weights and parameters. Each LSTM network receives the feature vector sequence of the electromagnetic signal as input and outputs a vector of fixed length, namely the embedding vector, denoted as h. The calculation process of the LSTM network is expressed as:
[0078] Input Gate i t Expressed as: i t =σ(W i *x t +U i *h t-1 +b i ),
[0079] Forget Gate t Expressed as: f t =σ(W f *x t +U f *h t-1 +b f ),
[0080] Output gate o t Expressed as: o t =σ(W o *x t +U o *h t-1 +b o ),
[0081] The cell state update is represented as: c t =f t ⊙c t-1 +i t ⊙tanh(W c *x t +U c *h t-1 +b c ),
[0082] The hidden state update is represented as: c t =f t ⊙c t-1 +i t ⊙tanh(W c x t +U c h t-1 +b c ),
[0083] Among them, σ is the sigmoid activation function, ⊙ represents the element-wise product, W and U are the weight matrices of the corresponding gates, and b is the bias vector of the corresponding gate.
[0084] In step S3, the neural network model calculation process includes:
[0085] Two LSTM networks process different input sequences x respectively. (a) and x (b) , and calculate the final hidden state of different sequences and The L1 norm difference between them generates a similarity score y;
[0086] In step S3, the neural network model calculation process is specifically as follows:
[0087] Will and Input into two LSTM networks respectively;
[0088] The output of each LSTM network is a sequence of hidden states. For the first LSTM network, the hidden state is For the second LSTM network, the hidden state is
[0089] Calculate the L1 norm, or Manhattan distance, between the last hidden states of two LSTM networks: in, Represents the L1 norm between vectors, calculated as: in and are the components of the hidden state vector;
[0090] The distance is converted into a similarity score y through an exponential function, namely:
[0091] The detailed LSTM network calculation steps include:
[0092] The formula for calculating the similarity d between the embedding vectors output by two LSTM networks is: Among them, h1 and h2 are the embedding vectors output by two LSTM networks, and T represents the transposition operation. The TripletLoss loss function is used to force the model to pair similar waveforms while trying to separate dissimilar models, which greatly improves the robustness and stability of the model for mismatched electromagnetic waves and untrained data.
[0093]
[0094] Among them, L represents the value of the loss function, Y represents the label variable, Y = 1 means that the two samples are similar, Y = 0 means that the two samples are not similar, D W Represents the Euclidean distance between sample features and is defined as: Among them, X1 and X2 are the feature vectors of the two samples respectively, P is the dimension of the feature, and m represents margin, which is used to define the minimum allowed distance between non-matching samples. When the distance between dissimilar samples is less than this value, the model will increase the loss, thereby encouraging the model to learn a larger degree of separation;
[0095] When Y = 1 (similar samples), the first term of the formula disappears, and the loss function only has the second term, which is expressed as: It means that when samples are similar, if their Euclidean distance D in the feature space is W Beyond the boundary m, the loss will increase, prompting the model to pull similar samples closer in the feature space;
[0096] When Y = 0 (the samples are not similar), the second term of the formula disappears, and the loss function only has the first term, which is expressed as: It means that when samples are not similar, the loss is related to their Euclidean distance D W Proportional to the distance between samples, the model is prompted to push dissimilar samples further away in the feature space. By minimizing the loss function, similar samples are brought closer and dissimilar samples are separated, thus enhancing the model's ability to distinguish.
[0097] in, Dw is defined as the Euclidean distance between the outputs of the twin network, representing the Euclidean distance (two-norm) between the features X1 and X2 of two samples. P represents the feature dimension of the sample, Y represents the label of whether the two samples match, Y=1 represents that the two samples are similar or matched, Y=0 represents mismatch, and m, i.e., margin, is the set threshold;
[0098] Therefore, for the twin neural network, when the two input electromagnetic waves are resonant, the Euclidean distance between them is small, and the loss is also small; when they are not resonant, the Euclidean distance is large, and the loss is also large. In simple terms, it is to minimize the distance between data of the same class and maximize the distance between different classes. In the above loss function:
[0099] When Y = 1 (i.e., the samples are similar), the loss function only has to express, That is, when the samples are similar, if the Euclidean distance in the feature space is large, it means that the current model is not good, so the loss is increased;
[0100] When Y = 0 (i.e., the samples are not similar), the loss function is expressed as, That is, when the samples are not similar, if the Euclidean distance in the feature space is small, the loss value will become larger, which meets the requirements of electromagnetic wave resonance detection;
[0101] A threshold margin is set here, which means that only dissimilar features with a Euclidean distance between 0 and margin are considered. When the distance exceeds margin, its loss is regarded as 0 (that is, dissimilar features are far apart, and their loss should be very low; while similar features are far apart, their loss is increased, thereby continuously updating the matching degree of paired samples);
[0102] Based on a large number of items with known year information, each item i has a known year information y i , for each item i, a resonant electromagnetic wave signal REW is generated i and non-resonant electromagnetic wave signal NREW i .
[0103] Resonance electromagnetic wave signal indicates REW i =f(y i ), non-resonant electromagnetic wave signal represents NREW i =g(y i ), where f and g are functions for generating resonant and non-resonant electromagnetic wave signals;
[0104] Construct training data pairs: positive example pair representation (REW i ,REW j ), negative example pair represents (REW i ,NREW k );
[0105] Perform model training. Based on the training data generated in the previous step, optimize the model by minimizing the similarity loss of similar signal pairs and the dissimilarity loss of dissimilar signal pairs. Specifically, use the gradient descent optimization algorithm to train the model and gradually update the weights and parameters of the model to minimize the loss function.
[0106] After training, the twin LSTM model is used to calculate the similarity of new electromagnetic signal pairs. The specific process is to input the feature vector sequences of the two electromagnetic signals into the trained twin network, obtain two embedding vectors respectively, use the embedding vectors to calculate the similarity between them, and judge whether the two signals are similar signals according to the preset similarity threshold;
[0107] Specifically, the signal processing and analysis accuracy is improved by combining neural network models and frequency domain analysis, electromagnetic detection of objects is performed within different frequency ranges, and a comparison is made between standard electromagnetic waves and the electromagnetic waves of the object being measured through a neural network. When resonance occurs, the electromagnetic frequency information is quickly determined, thereby inferring the historical information of the object, greatly reducing errors in manual analysis.
[0108] The method for determining whether there is a resonance phenomenon in step S3 is:
[0109] An alternating current with a frequency of ω is input into the circular coil, and the alternating current forms an alternating magnetic field, which is used to detect cultural relics containing magnetic minerals and having thermal remanent magnetism; when nuclear magnetic resonance is generated inside the artifact under the alternating magnetic field, the atomic nuclei of the magnetic minerals in the artifact will form a macroscopic magnetic moment; the magnetic moment produces a precession motion in the alternating magnetic field, and its precession frequency is ω i , the resonant frequency is also ω i ;
[0110] When an artifact containing magnetic minerals and having thermal residual magnetism is detected by an alternating current with a frequency of ω, if nuclear magnetic resonance is generated inside the artifact, its resonance frequency is ω i ,ω i represents the resonance frequency of the ith state or the ith particle;
[0111] According to the Langevin equation, the motion of the magnetic moment μ in the external magnetic field B is: Where γ is the cyclotron ratio, M represents the magnetic moment of the object, and B represents the external magnetic field. represents the rate of change of magnetic moment with time;
[0112] Determine the paleomagnetic field strength B i , Among them B i represents the magnetic field strength of the ith state or the ith particle, and the measured resonant frequency ω i and the known cyclotron ratio γ, calculate the paleo-geomagnetic field strength B at the time of production i ;
[0113] By adding B i Map to the mapping table to find the corresponding time coordinates and determine the formation age of the cultural relics;
[0114] The mapping data is derived from the prior information of the remanent magnetic information of known objects of different types and ages;
[0115] Specifically, by modeling the magnetic relaxation process, the time point when the object last underwent drastic physical and chemical changes can be accurately estimated, and the initial value of the residual magnetic signal and its decay rate can be calculated. The measurement method based on magnetic properties adopted in the present invention is similar to the radiocarbon dating method, and utilizes the unique properties of magnetic materials to achieve non-destructive and non-invasive dating of cultural relics.
[0116] In step S3, the electromagnetic wave resonance detection method for the actual age information is:
[0117] The user sets a year number through the host computer software interface, which generates electromagnetic waves of a specific frequency to scan the object being detected;
[0118] The electromagnetic detection device (such as an electromagnetic plate or an electromagnetic carbon rod) generates electromagnetic waves of a certain frequency according to the set parameters, which are used for resonance comparison with the electromagnetic waves of the object being detected;
[0119] Since the magnetization intensity of the thermal residual magnetism of cultural relics is generally small, the electromagnetic detection of this method only needs to emit electromagnetic waves of lower frequency to identify the production age, and an electromagnetic generating device capable of emitting extremely low frequency, ultra-low frequency, extra-low frequency or very low frequency electromagnetic waves is used, where the frequency range is defined as:
[0120] Extremely Low Frequency (ELF): 1Hz to 30Hz;
[0121] Super Low Frequency (SLF): 30Hz to 300Hz;
[0122] Ultra Low Frequency (ULF): 300Hz to 3kHz;
[0123] Very Low Frequency (VLF): 3kHz to 30kHz;
[0124] Perform electromagnetic wave frequency mapping. There is a preset mapping relationship between the electromagnetic wave frequency of objects of different materials and the production years. The specific mapping relationship is:
[0125] When the frequency range is 1Hz≤f≤310Hz, it corresponds to artifacts before the Shang Dynasty (about 1600 BC);
[0126] When the frequency range is 2700Hz≤f≤2800Hz, it corresponds to modern times (approximately 1949 to present);
[0127] The host computer software is responsible for frequency adjustment, confirmation, and triggering of electromagnetic wave scanning. Its main functions include:
[0128] Frequency adjustment button:
[0129] Used to increase or decrease the set frequency value f, increase frequency: f = f + Δf, decrease frequency: f = f - Δf;
[0130] Frequency confirmation button:
[0131] Confirm the currently set frequency value f c and prepare for scanning;
[0132] Clear button:
[0133] Reset the currently set frequency value f r ;
[0134] Touchpad peripheral trigger scan:
[0135] The user taps the touchpad once, triggering a complete electromagnetic wave emission and detection;
[0136] Electromagnetic waves are transmitted through an electromagnetic transmitting device (such as an electromagnetic coil plate or an electromagnetic transmitting rod) at a set frequency f s Electromagnetic waves are used to detect the object being detected;
[0137] The intensity and direction of the reflected wave generated by the detected object and the preset electromagnetic wave information are input into the neural network model, and the neural network model calculates and fits the resonance and non-resonance results.
[0138] Step S4, presetting a mapping table, the mapping table contains age information corresponding to electromagnetic waves of different frequencies, and after confirming the resonant frequency, searching the corresponding age of the artifact through the mapping table;
[0139] In step S4, the measured paleo-geomagnetic field intensity is matched with a preset age mapping table;
[0140] Step S5, converting the matching result into age information;
[0141] Specifically, the advantage of the present invention is that it can maintain the integrity of the object being measured, and is particularly suitable for precious cultural relics and samples that cannot or should not be destroyed.
[0142] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
Claims
1. A method for determining the age of an object based on residual magnetism detection technology, characterized in that: include: Step S1, using an electromagnetic wave generator to generate and emit electromagnetic waves of a specific frequency to detect the object; Step S2, receiving electromagnetic wave data reflected from the object and recording its intensity and phase change; Step S3, constructing a neural network model, using the model to process the received electromagnetic wave data, and comparing it with the standard data to determine whether there is a resonance phenomenon; Step S4, presetting a mapping table, the mapping table contains age information corresponding to electromagnetic waves of different frequencies, and after confirming the resonant frequency, searching the corresponding age of the artifact through the mapping table; Step S5, converting the matching result into age information.
2. The method for determining the age of an object based on residual magnetism detection technology according to claim 1, characterized in that: In step S3, the electromagnetic wave data includes: Resonance contrast signal 1, i.e., the electrical signal converted from the electromagnetic wave data of the detected object; Resonance contrast signal 2, i.e., the detected electromagnetic signal; The data format includes time domain data and frequency domain data. The time domain data directly records the intensity of electromagnetic waves that changes with time, and the frequency domain data is obtained by Fourier transform of the time domain data.
3. The method for determining the age of an object based on residual magnetism detection technology according to claim 2, characterized in that: In step S3, the received electromagnetic wave data is processed in the following manner: Perform the following operations on electromagnetic wave data: Wavelet decomposition, obtain wavelet coefficients of different scales, threshold the wavelet coefficients to remove the noise; Wavelet reconstruction is used to obtain the denoised signal, and low-pass, high-pass and band-pass filters are used for filtering; Standardization is performed so that the signal amplitude of different sample data is processed to the interval [-1,1]. where μ is the mean, σ is the standard deviation, and X and X ′ Represent the signal amplitude before and after processing respectively.
4. The method for determining the age of an object based on residual magnetism detection technology according to claim 3 is characterized in that: In step S3, the method of processing the received electromagnetic wave data also includes: Perform time domain feature extraction, including peak value, mean μ, variance σ 2 ; Perform frequency domain feature extraction, including main frequency, spectrum energy E, E = ∑|X(k)| 2 , where X(k) represents the signal spectrum and k represents the spectrum length; The extracted frequency domain features also include spectral entropy, which is a measure of uncertainty in the energy distribution of the spectrum, denoted as H, H = -∑p(k)log2 p(k), where Represents the energy distribution probability of the spectrum; Use short-time Fourier transform (STFT) to convert from the time domain to the frequency domain. The STFT calculation formula is: X(n,ω) = ∑x(m)·w(nm)·e -jωm , where x(m) is the signal, w(nm) is the window function, n is the time index, ω is the frequency, and e -jωm It is a negative exponent, X(n,ω), which means that the frequency domain features include the energy distribution and spectrum centroid of electromagnetic wave data.
5. The method for determining the age of an object based on residual magnetism detection technology according to claim 4 is characterized in that: In step S3, the method of processing the received electromagnetic wave data also includes: The extracted features are organized into a neural network input format, and the time series data is input as a one-dimensional vector sequence. Each sequence corresponds to a feature vector sequence of an electromagnetic signal, denoted as x1, x2, …, x T , where x t is the feature vector at time t, and T is the sequence length.
6. The method for determining the age of an object based on residual magnetism detection technology according to claim 5, characterized in that: In step S3, the neural network model is constructed as follows: The twin LSTM is used to construct a neural network model. The neural network model consists of two identical LSTM networks. The two networks share the same weights and parameters. Each LSTM network receives the feature vector sequence of the electromagnetic signal as input and outputs a vector of fixed length, namely the embedding vector.
7. The method for determining the age of an object based on residual magnetism detection technology according to claim 6, characterized in that: In step S3, the neural network model calculation process includes: Two LSTM networks process different input sequences x respectively. (a) and x (b) , and calculate the final hidden state of different sequences and The L1 norm difference between them generates a similarity score y.
8. The method for determining the age of an object based on residual magnetism detection technology according to claim 7, characterized in that: In step S3, the neural network model calculation process is specifically as follows: Will and Input into two LSTM networks respectively; The output of each LSTM network is a sequence of hidden states. For the first LSTM network, the hidden state is For the second LSTM network, the hidden state is Calculate the L1 norm, or Manhattan distance, between the last hidden states of two LSTM networks: in, Represents the L1 norm between vectors, calculated as: in and are the components of the hidden state vector; The distance is converted into a similarity score y through an exponential function, namely:
9. The method for determining the age of an object based on residual magnetism detection technology according to claim 8, characterized in that: The method for determining whether there is a resonance phenomenon in step S3 is: An alternating current with a frequency of ω is input into the toroidal coil, and the alternating current forms an alternating magnetic field. When nuclear magnetic resonance is generated inside the object under the alternating magnetic field, the magnetic moment produces a precession motion in the alternating magnetic field, and its precession frequency is ω. i , the resonant frequency is also ω i ; When an object containing magnetic minerals and having thermal residual magnetism is detected by an alternating current with a frequency of ω, if nuclear magnetic resonance is generated inside the object, its resonance frequency is ω i ,ω i represents the resonance frequency of the ith state or the ith particle; According to the Langevin equation, the motion of the magnetic moment μ in the external magnetic field B is: Where γ is the cyclotron ratio, M represents the magnetic moment of the object, and B represents the external magnetic field. represents the rate of change of magnetic moment with time; Determine the paleomagnetic field strength B i , Among them B i represents the magnetic field strength of the ith state or the ith particle, and the measured resonant frequency ω i and the known cyclotron ratio γ, calculate the paleo-geomagnetic field strength B at the time of production i ; By adding B i Map to the mapping table to find the corresponding time coordinates and determine the age of the artifact.
10. The method for determining the age of an object based on residual magnetism detection technology according to claim 9, characterized in that: In step S4, the measured paleomagnetic field intensity is matched with a preset age mapping table.
Citation Information
Patent Citations
Transform domain filtering narrowband interference suppression method and device
CN114200486A
Method for identifying production time of cultural relics and electromagnetic detector
CN114895361A
Blast furnace abnormal furnace condition detection method for extracting stationary features based on twin neural network
CN116842402A
Rock destruction observing method and rock bed slope degradation monitor
JP2003161782A
Method and apparatus for PCB treatment
JP2005168573A
Cited By
Wireless power transmission system coil parameter prediction method, device and medium
CN120068593A
