A method and system for analyzing a polymorphic non-metallic substance

Abstract

The application discloses a kind of multi-form nonmetallic substance analysis method and system, belong to terahertz substance measuring instrument technical field.The application is output by microwave excitation signal generation unit local oscillator signal and radio frequency signal;Terahertz signal generation array is handled to local oscillator signal and radio frequency signal, generates terahertz signal;Terahertz signal is focused to terahertz signal detection array by optical array with reflected signal and transmission signal;Computer is acquired by multichannel data acquisition unit and is analyzed with transmission signal and reflected signal of the sample to be measured.Can realize the organic fusion of substance three-dimensional structure and component analysis, satisfy the test demand of multi-form nonmetallic substance performance evaluation;Solved the problem that " cannot satisfy the test demand of multi-form nonmetallic substance performance evaluation " in prior art.

Description

Technical Field

[0001] This invention relates to the field of terahertz material measurement instruments, and in particular to a method and system for analyzing multi-morphological non-metallic substances. Background Technology

[0002] The statements in this section merely refer to the background art related to this invention and do not necessarily constitute prior art.

[0003] Non-metallic substances typically include various forms such as solid, liquid, gas, powder, and thin film. The three-dimensional structure and composition of a substance are key parameters for measuring its properties.

[0004] Three-dimensional structural imaging instruments and material composition analysis instruments are typically two relatively independent types of measuring instruments, each with its own focus, forming a complementary and interdependent relationship. Terahertz technology, with its fingerprint characteristics, low-energy safety, and strong penetration, has become an important supplementary means in the field of non-metallic material measurement.

[0005] However, existing terahertz material measurement instruments have not yet achieved the organic integration of three-dimensional structure and composition analysis of materials, and cannot meet the testing needs for performance evaluation of non-metallic materials in multiple forms such as solid, liquid, and gas. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method and system for analyzing multimorphic nonmetallic substances, achieving an organic integration of the three-dimensional structure and composition of multimorphic nonmetallic substances.

[0007] In a first aspect, the present invention provides a method for analyzing multimorphic nonmetallic substances;

[0008] A method for analyzing multi-morphological nonmetallic substances, comprising:

[0009] The transmission and reflection signals of the sample to be tested are acquired, and the transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum of the sample to be tested are determined based on the transmission and reflection signals; wherein, the transmission and reflection signals are acquired by a multi-morphology non-metallic material analysis system;

[0010] The thickness information of the sample is obtained by performing an inverse Fourier transform on the reflected terahertz characteristic spectrum of the sample. Based on the thickness information of the sample, the transmitted terahertz characteristic spectrum and the reflected terahertz characteristic spectrum are filtered in the time domain.

[0011] The composition of the sample under test is obtained based on the time-domain filtered transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum.

[0012] Based on the time-domain filtered transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum, a two-dimensional image matrix of the sample under test is obtained; the two-dimensional image matrices of different frequencies are input into a preset convolutional neural network for processing to achieve cross-scale three-dimensional structural tomography and obtain three-dimensional structural images of the sample under test.

[0013] Furthermore, the inverse Fourier transform of the reflected terahertz characteristic spectrum of the sample to be tested to obtain the thickness information of the sample includes:

[0014] The inverse Fourier transform of the reflected terahertz spectrum is performed on the reflected terahertz characteristic spectrum amplitude and the reflected terahertz characteristic spectrum phase of the sample to obtain the terahertz characteristic spectral lines in the time domain.

[0015] The first and second reflection peaks in the terahertz characteristic spectral lines are extracted, and the thickness information of the sample to be tested is determined based on the first and second reflection peaks.

[0016] Furthermore, obtaining the composition of the sample to be tested based on the time-domain filtered transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum includes:

[0017] Wavelet decomposition is performed on the transmitted terahertz characteristic spectrum and the reflected terahertz characteristic spectrum. Based on the wavelet coefficients after wavelet decomposition, the transmitted terahertz characteristic spectrum and the reflected terahertz characteristic spectrum are reconstructed.

[0018] Characteristic peaks were determined in the reconstructed transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum;

[0019] If a characteristic peak is present, the sample to be tested is a single-component sample. The category of the sample to be tested is determined based on the position of the characteristic peak.

[0020] If no characteristic peaks are found, the characteristic parameters of the sample to be tested are calculated based on the reconstructed transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum.

[0021] Based on the morphology of the sample to be tested, the feature parameters of the sample are selected and input into the trained machine learning network for processing to determine the category of the sample.

[0022] Furthermore, determining the transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum of the sample under test based on the transmission signal and the reflection signal includes:

[0023] Based on the transmission signal of the sample to be tested, and using the transmission reference signal, the transmission terahertz characteristic spectrum of the sample to be tested is determined; wherein, the transmission reference signal is the transmission signal to air.

[0024] Based on the reflection signal of the sample under test, and using the reflection reference signal, the reflection terahertz characteristic spectrum of the sample under test is determined; wherein, the reflection reference signal is the reflection terahertz signal of the strongly reflective metal calibration component.

[0025] Furthermore, the transmitted signal includes a transmitted signal amplitude matrix and a transmitted signal phase matrix, and the reflected signal includes a reflected signal phase matrix and a reflected signal amplitude matrix.

[0026] Secondly, the present invention provides a multi-morphological non-metallic material analysis system;

[0027] A multi-morphological non-metallic material analysis system includes a microwave excitation signal generation unit, a switching matrix, a power divider amplifier module, a terahertz signal generation array, a first terahertz signal detection array, a second terahertz signal detection array, a multi-channel data acquisition unit, a first optical array, a second optical array, and a computer.

[0028] The microwave excitation signal generating unit outputs a local oscillator signal and a radio frequency (RF) signal. The local oscillator signal is processed by a power divider and amplification module and then transmitted to the switching matrix. The switching matrix transmits the RF signal and the local oscillator signal to the terahertz signal generating array, and the local oscillator signal is transmitted to the terahertz signal detection array. Under the excitation of the local oscillator signal and the RF signal, the terahertz signal generating array generates a terahertz signal and a reference intermediate frequency (IF) signal. The terahertz signal is transmitted through a first optical array to the sample under test, and a transmitted IF signal is generated by the sample under test. This transmitted IF signal is then focused onto the first terahertz signal detection array by a second optical array. The terahertz signal is then reflected by the first optical array and focused onto the second terahertz signal detection array.

[0029] The multi-channel data acquisition unit acquires the reference intermediate frequency signal, the transmitted intermediate frequency signal, and the reflected intermediate frequency signal via the switching matrix, performs mixing processing on each signal, acquires the transmitted signal and the reflected signal, and transmits them to the computer.

[0030] The computer acquires the transmission and reflection signals of the sample to be tested and performs the steps of the above-described multimorphic nonmetallic material analysis method.

[0031] Furthermore, it also includes a precision control module, which is disposed between the first optical array and the second optical array;

[0032] The sample to be tested is placed in the precision control module.

[0033] Furthermore, the terahertz signal generating array includes a power divider, a harmonic signal generating unit, and a silicon lens;

[0034] The local oscillator signal and the radio frequency signal are evenly split by the power divider and then transmitted to multiple harmonic signal generating units. Under the excitation of the local oscillator signal and the radio frequency signal, the harmonic signal generating units generate terahertz sub-signals and reference intermediate frequency signals.

[0035] Terahertz sub-signals are spatially power-combined using silicon lenses to obtain terahertz signals.

[0036] Furthermore, the first terahertz signal detection array and the second terahertz signal detection array have the same composition. The first terahertz signal detection array includes a silicon lens, an on-chip antenna, a harmonic frequency multiplication amplification link, a power amplifier, and a power divider.

[0037] The local oscillator signal is divided equally by the power divider and then transmitted to the power amplifier. The power amplifier transmits the local oscillator signal to the harmonic frequency doubling amplification link. The first terahertz signal is transmitted to the harmonic frequency doubling amplification link via a silicon lens and an on-chip antenna, and is mixed with the local oscillator signal processed by the harmonic frequency doubling detection link to realize the detection of the terahertz signal.

[0038] Furthermore, the first optical array includes a first parabolic mirror, a second parabolic mirror, a beam splitter, and a third parabolic mirror, and the second optical array includes a fourth parabolic mirror and a fifth parabolic mirror;

[0039] The terahertz signal generated by the terahertz signal generating array is incident on the second parabolic mirror. The second parabolic mirror collimates the terahertz signal and then incident it on the beam splitter. The beam splitter divides the terahertz signal into two beams. One beam of terahertz signal is reflected to the first parabolic mirror and focused by the first parabolic mirror onto the silicon lens in the second terahertz signal detection array. The other beam of terahertz signal is transmitted to the third parabolic mirror and focused by the third parabolic mirror onto the surface of the sample to be tested. It is then transmitted through the sample to the fourth parabolic mirror, collimated by the fourth parabolic mirror, and focused by the fifth parabolic mirror onto the silicon lens in the first terahertz signal detection array.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] The technical solution provided by this invention employs a spatially synthesized terahertz signal generating array, which can effectively improve signal power and signal-to-noise ratio; the use of a terahertz signal generating and detecting array enables miniaturization and high integration of the entire instrument; the detection resolution of terahertz characteristic spectral lines spans the GHz, MHz, kHz, or Hz levels, filling the gap in refined terahertz fingerprint spectra; the reflection adopts a transceiver separation technology, which can significantly improve the dynamic range of the test; the integrated transmission and reflection testing technology meets the testing needs of various sample forms such as solid, liquid, gas, powder, and thin film; and it realizes cross-scale three-dimensional structural tomography imaging with imaging resolution spanning the centimeter-millimeter-micrometer levels. Attached Figure Description

[0042] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0043] Figure 1 A schematic flowchart of the multimorphic nonmetallic material analysis method provided in the embodiments of the present invention;

[0044] Figure 2 This is a schematic diagram of the architecture of the multimorphic nonmetallic material analysis system provided in an embodiment of the present invention;

[0045] Figure 3 A schematic diagram of the architecture of a terahertz signal generating array provided in an embodiment of the present invention;

[0046] Figure 4 This is a schematic diagram of the architecture of the i-th harmonic frequency multiplication amplification link provided in an embodiment of the present invention;

[0047] Figure 5 This is a schematic diagram of the architecture of the frequency multiplier amplification link provided in an embodiment of the present invention;

[0048] Figure 6 A schematic diagram of the architecture of a terahertz signal detection array provided in an embodiment of the present invention;

[0049] Figure 7 This is a schematic diagram of the architecture of the multi-channel data acquisition unit and the microwave excitation signal generation unit provided in an embodiment of the present invention. Detailed Implementation

[0050] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0051] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0052] Example 1

[0053] Existing terahertz material measurement instruments have not yet achieved the organic integration of three-dimensional structure and composition analysis of materials, and cannot meet the testing requirements for performance evaluation of non-metallic materials in multiple forms such as solid, liquid, and gas. Therefore, this invention provides a method for analyzing non-metallic materials in multiple forms.

[0054] Next, combined Figure 1 This embodiment discloses a method for analyzing multi-morphological non-metallic substances, which includes the following steps:

[0055] S1. Acquire the transmission reference signal, reflection reference signal, transmission signal and reflection signal of the sample under test.

[0056] Specifically, the signal is acquired using a multi-morphology non-metallic material analysis system, and the specific process is as follows:

[0057] First, the system test parameters of the multimorphic nonmetallic material analysis system are set by a computer equipped with measurement and control software. The system test parameters include the start and end frequencies, frequency resolution, radio frequency local oscillator signal power, intermediate frequency bandwidth, and shaft motion speed, which are the initial parameters of the system.

[0058] Then, strong reflection signals are acquired, the strong reflection metal calibration piece is placed on the precision control module, and the rotation angle is set. 0=0 o (That is, ensure that the strong reflective metal calibration component is orthogonal and perpendicular to the terahertz signal transmission direction), set the position of the metal calibration component (X0, Y0, Z0), and after the metal calibration component reaches the set position, collect the reflected terahertz signal as the reflected reference signal. λ1 represents the amplitude in dB; λ2 represents the phase in o.

[0059] Then, acquire the transmission signal into the air, keeping the set position parameters unchanged, remove the strongly reflective metal calibration piece, and acquire the transmitted terahertz signal as the transmission reference signal. μ1 represents the amplitude in dB; μ2 represents the phase in 0°.

[0060] Finally, the transmission and reflection signals of the sample under test are collected. The sample is placed on the test fixture, and the test area (X0, Y0, Z0, ...) is set. 0)-(X) e Y e , Z0, e (), where Z0 is a fixed value that remains constant; under the control of the computer's built-in analysis software, firstly in 0=0 o Under these conditions, the sample to be tested can be placed in the region (X0, Y0) - (X... e Y e Acquisition of amplitude and phase data of transmitted and reflected signals within the system; 0 moved to At point 1, the sample to be tested is located in the region (X0, Y0) - (X... e Y e Acquisition of amplitude and phase data of transmitted and reflected signals within the internal transmission signal; until... 0=360 o -Φ oAt the end of the test, Φo represents the set angular step interval, ultimately resulting in (360 / Φ) M×N×P three-dimensional matrices ψ1, ψ2, ψ3, ψ4. ψ1 represents the reflected signal amplitude matrix, ψ2 represents the reflected signal phase matrix, ψ3 represents the transmitted signal amplitude matrix, and ψ4 represents the transmitted signal phase matrix. Here, M represents the sample length sampling value, N represents the sample width sampling value, and P represents the start and end frequency sampling values.

[0061] S2. Based on the transmission and reflection signals, determine the transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum of the sample under test. Represented as follows:

[0062] The amplitude of the reflected terahertz characteristic spectrum of the sample under test is: Ж1=ψ1 / λ1, and the phase is: Ж2=ψ2 / λ2;

[0063] The amplitude of the transmission terahertz characteristic spectrum of the sample under test is: Ж3=ψ3 / μ1, and the phase is: Ж4=ψ4 / μ2.

[0064] S3. Perform an inverse Fourier transform on the reflected terahertz characteristic spectrum of the sample to obtain the thickness information of the sample; including the following steps:

[0065] S301. Perform an inverse Fourier transform on the reflected terahertz spectrum based on the amplitude Ж1 and phase Ж2 of the reflected terahertz characteristic spectrum of the sample to be tested, and obtain the terahertz characteristic spectral lines in the time domain.

[0066] S 302. Extract the first and second reflection peaks from the terahertz characteristic spectral lines, and determine the thickness information of the sample to be tested based on the first and second reflection peaks.

[0067] Specifically, by performing an inverse Fourier transform on the reflected terahertz characteristic spectrum based on Ж1 and Ж2, the time-domain terahertz characteristic spectral lines are obtained. The first reflection peak τ1 and the second reflection peak τ2 are extracted. The difference between the two peaks is then obtained. Based on the difference between the two peaks, the thickness information of the sample under test can be obtained, expressed as:

[0068] The thickness information of the sample to be tested = Ⅰτ1-τ2Ⅰ×3×10 8 m / s.

[0069] S4. Based on the thickness information of the sample to be tested, perform time-domain filtering on the transmitted terahertz characteristic spectrum and the reflected terahertz characteristic spectrum.

[0070] Specifically, the bandwidth of the time-domain filtering of the transmitted terahertz characteristic spectrum and the reflected terahertz characteristic spectrum is 2-3 times the estimated thickness of the sample to be tested, thereby obtaining the filtered and denoised terahertz characteristic spectrum.

[0071] Setting the bandwidth of the time-domain filtering to 2-3 times the estimated thickness of the sample can effectively improve the signal-to-noise ratio and enhance the filtering and noise reduction effect.

[0072] S5. Based on the time-domain filtered transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum, obtain the composition of the sample to be tested. This includes the following steps:

[0073] S501. Perform wavelet decomposition on the transmitted terahertz characteristic spectrum and the reflected terahertz characteristic spectrum, and reconstruct the transmitted terahertz characteristic spectrum and the reflected terahertz characteristic spectrum based on the wavelet coefficients after wavelet decomposition.

[0074] S502. Determine the characteristic peaks of the reconstructed transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum.

[0075] S503. If a characteristic peak exists, the sample to be tested is a single-component sample. The sample category is determined based on the position of the characteristic peak. If no characteristic peak exists, the characteristic parameters of the sample to be tested are calculated based on the reconstructed transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum.

[0076] S504. Based on the morphology of the sample to be tested, select the feature parameters of the sample to be tested and input them into the trained machine learning network for processing to determine the category of the sample to be tested.

[0077] For example, the specific process is as follows:

[0078] (1) Principal component extraction: Wavelet transform algorithm is used to decompose Ж1, Ж2, Ж3 and Ж4. Based on the decomposed wavelet coefficients, the wavelet coefficients with a proportion ≥95% are extracted as the principal components of the original matrix, and the reconstruction of Ж1, Ж2, Ж3 and Ж4 is achieved. 11 Ж 22 Ж 33 and Ж 44 .

[0079] (2) Characteristic peak identification: Use the peak finding function to find the terahertz characteristic peak. The basic operation is to solve Ж 11 Ж 22 Ж 33 and Ж 44 The difference between adjacent amplitude data corresponding to a frequency is recorded as a characteristic peak when the sign of the difference changes and the absolute value of the difference is ≤10dB; if and only if Ж 11 Ж 22 Ж 33 and Ж 44 The frequency point is confirmed as a characteristic peak only when the difference between the amplitude and phase of the reflection characteristic peak and the amplitude and phase of the transmission characteristic peak is ≤100MHz; otherwise, there is no characteristic peak.

[0080] (3) When the sample to be tested has a characteristic peak, it indicates that the sample is likely a single-component sample. The sample category can be identified by comparing the position of the characteristic peak.

[0081] (4) When the sample to be tested does not have a characteristic peak, according to Ж 11 Ж 22 Ж 33 and Ж 44 The characteristic parameters of the sample to be tested, including dielectric constant, absorption coefficient, refractive index, loss tangent, etc., are calculated and denoted as characteristic parameter matrix Й.

[0082] When the sample to be tested is in the form of solid sheet, liquid, gas, powder, film, etc., the dielectric constant, absorption coefficient, refractive index, and loss tangent are selected as the input parameters of the trained machine learning network.

[0083] When the sample to be tested is a solid powder sample, the dielectric constant and refractive index are selected as the input parameters of the trained machine learning network.

[0084] In this embodiment, convolutional neural networks and extreme learning machines are selected as training models for component analysis, respectively. The specific processing flow of inputting input parameters into convolutional neural networks or extreme learning machines and outputting classification results and content is the existing technology of analysis using convolutional neural networks or extreme learning machines. This embodiment does not improve on this and will not be described in detail here.

[0085] The decision process after a convolutional neural network or extreme learning machine outputs a classification result is as follows:

[0086] When both networks determine that the component of the sample to be tested is A, the sample to be tested is determined to be A, and the output result is that the probability of a single sample being A is 100%.

[0087] When the convolutional neural network determines that the sample to be tested is component A, and the extreme learning machine determines that the sample to be tested is component B, then the probability of the sample being component A is 80%, the probability of component B is 50%, and the output result is a mixed sample with a probability of 80% for component A and a content of A1, and a probability of 50% for component B and a content of B1.

[0088] S6. Based on the time-domain filtered transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum, obtain the two-dimensional image matrix of the sample under test; input the two-dimensional image matrices of different frequencies into a preset convolutional neural network for processing to achieve cross-scale three-dimensional structural tomography and obtain three-dimensional structural images of the sample under test. The specific process is as follows:

[0089] (1) Principal component extraction: Wavelet transform algorithm is used to decompose Ж1, Ж2, Ж3 and Ж4. Based on the decomposed wavelet coefficients, the wavelet coefficients with a proportion ≥95% are extracted as the principal components of the original matrix, and the reconstruction of Ж1, Ж2, Ж3 and Ж4 is achieved. 11 Ж 22 Ж 33 and Ж 44 ;

[0090] (2) Based on Ж1, Ж2, Ж3 and Ж4, a two-dimensional image matrix of the sample to be tested can be obtained. According to the test requirements, different two-dimensional images are extracted and three-dimensional structural tomography of the sample to be tested is realized using a convolutional neural network. For example, to achieve centimeter-level resolution, test data in the 0.11THz-0.17THz frequency band can be selected; to achieve millimeter-level resolution, test data in the 0.11THz-0.5THz frequency band can be selected; to achieve micrometer-level resolution, test data in the 0.11THz-3THz frequency band can be selected. Thus, cross-scale three-dimensional structural tomography is realized, and the imaging resolution spans the centimeter-millimeter-micrometer scale.

[0091] The use of existing convolutional neural networks for three-dimensional structural tomography is a method for three-dimensional structural tomography of scenes in the prior art. This embodiment does not improve upon it and will not be described in detail here.

[0092] Example 2

[0093] Combination Figures 2-7 This embodiment discloses a multi-morphological non-metallic material analysis system, including a microwave excitation signal generation unit, a power divider amplification module, a switching matrix, a terahertz signal generation array, a first terahertz signal detection array, a second terahertz signal detection array, a multi-channel data acquisition unit, a first optical array, a second optical array, a precision control module, and a computer.

[0094] The precision control module is positioned between the first and second optical arrays, and the sample to be tested is placed within the precision control module. The precision control module's function is to realize the XYZ-axis rotation of the sample to be tested. Rotation in four directions, Indicating the rotation angle, the computer's role is to use built-in software to program the hardware used in the entire system and facilitate communication between hardware components; at the same time, it achieves the organic integration of the analysis of the three-dimensional structure and composition of matter. The microwave excitation signal generation unit generates and outputs a local oscillator signal and a radio frequency (RF) signal. The two excitation signals undergo time-base processing to ensure they are in phase and frequency. The local oscillator signal is processed by a power divider and amplified module before being transmitted to a switching matrix. The switching matrix transmits the RF signal and the local oscillator signal to a terahertz signal generation array, and then to a terahertz signal detection array. The terahertz signal generation array, excited by the local oscillator and RF signals, generates a terahertz signal and a reference intermediate frequency (IF) signal. The terahertz signal is transmitted through a first optical array to the sample under test, generating a transmitted IF signal, which is then focused onto the first terahertz signal detection array by a second optical array. The terahertz signal generates a reflected IF signal through the first optical array and is focused onto the second terahertz signal detection array. The multi-channel data acquisition unit acquires the reference IF signal, the transmitted IF signal, and the reflected IF signal via the switching matrix and performs mixing processing on each signal to obtain the transmitted and reflected signals, which are then transmitted to a computer. The computer acquires the transmitted and reflected signals of the sample under test through the multi-channel data acquisition unit and executes the following steps:

[0095] The transmission and reflection signals of the sample to be tested are acquired, and the transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum of the sample to be tested are determined based on the transmission and reflection signals; wherein, the transmission and reflection signals are acquired by a multi-morphology non-metallic material analysis system;

[0096] The thickness information of the sample is obtained by performing an inverse Fourier transform on the reflected terahertz characteristic spectrum of the sample. Based on the thickness information of the sample, the transmitted terahertz characteristic spectrum and the reflected terahertz characteristic spectrum are filtered in the time domain.

[0097] The composition of the sample under test is obtained based on the time-domain filtered transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum.

[0098] Based on the time-domain filtered transmission terahertz characteristic spectrum and reflection terahertz characteristic spectrum, a two-dimensional image matrix of the sample under test is obtained; the two-dimensional image matrices of different frequencies are input into a preset convolutional neural network for processing to achieve cross-scale three-dimensional structural tomography and obtain three-dimensional structural images of the sample under test.

[0099] Furthermore, the microwave excitation signal generation unit includes a 20-60GHz excitation source, a first 50-110GHz excitation source, a second 50-110GHz excitation source, a third 50-110GHz excitation source, and a 110GHz directional coupler. The first 20-60GHz excitation source and the first 50-110GHz excitation source are connected. The first 50-110GHz excitation source is connected to the second 50-110GHz excitation source and the third 50-110GHz excitation source, respectively. The second 50-110GHz excitation source is connected to the 110GHz directional coupler. The multi-channel data acquisition unit includes a frequency reference unit, a 20-60GHz synthesizer module, a fourth 50-110GHz excitation source, a power divider module, a multi-channel mixer module, a multi-channel intermediate frequency conditioning module, and a multi-channel high-speed acquisition module. The excitation unit is connected to the 20-60GHz synthesizer module and the 20-60GHz excitation source, respectively. The 20-60GHz synthesizer module is connected to the fourth 50-110GHz excitation source, the fourth 50-110GHz excitation source is connected to the power divider module, the power divider module is connected to the multi-channel mixer module, the multi-channel mixer module is connected to the multi-channel intermediate frequency conditioning module, the multi-channel intermediate frequency conditioning module is connected to the multi-channel high-speed acquisition module, and the multi-channel high-speed acquisition module is connected to the computer.

[0100] The frequency reference unit simultaneously transmits a frequency reference signal to the 20-60GHz synthesizer module and the 20-60GHz excitation source. At this time, radio frequency signals are generated by the 20-60GHz excitation source, the first 50-110GHz excitation source, and the third 50-110GHz excitation source. A local oscillator signal is generated by the 20-60GHz excitation source, the first 50-110GHz excitation source, the second 50-110GHz excitation source, and the 110GHz directional coupler. The local oscillator signal undergoes power distribution and amplification by the power divider module, splitting into two local oscillator signals. Simultaneously, a second local oscillator signal is generated by the 20-60GHz synthesizer module, the fourth 50-110GHz excitation source, and the power divider module and transmitted to the multi-channel mixer module.

[0101] Furthermore, the terahertz signal generating array employs a spatially synthesized terahertz signal radiation mode, significantly improving the terahertz signal radiation power. The terahertz signal generating array includes a power divider, one harmonic signal generating unit, and a silicon lens. One local oscillator signal and one radio frequency signal are transmitted to the power divider via a switching matrix. After being split into multiple paths by the power divider, they are transmitted to the one harmonic signal generating unit. A 50GHz-110GHz power amplifier drives a harmonic multiplier in the harmonic frequency multiplication amplification link to simultaneously generate the first to first harmonics, thus producing terahertz sub-signals in the 0.05THz-1×0.11THz frequency band. The frequency resolution can be on the order of GHz, MHz, kHz, or Hz. The generated terahertz sub-signals are spatially synthesized using an on-chip antenna and a silicon lens to acquire the terahertz signal and improve the signal generation power.

[0102] The harmonic frequency multiplier amplification link includes a harmonic frequency multiplier, an amplifier, a unidirectional coupler, and an amplifier connected in sequence, which sequentially amplify, couple, and process the terahertz signal generated by the harmonic frequency multiplier.

[0103] Each harmonic signal generation unit consists of a 4×4 array, in which one harmonic frequency multiplier amplification link is mixed with the frequency multiplier amplification link driven by the local oscillator signal to obtain a reference intermediate frequency signal, which is then transmitted to the multi-channel intermediate frequency conditioning module via a switching matrix.

[0104] Compared with traditional centralized array structures, this transmitter array (terahertz signal generating array) has good frequency / phase consistency, increases array scalability, and can be expanded by simply stacking more similar units to increase the size of the transmitter array and achieve the purpose of increasing the terahertz signal radiation power.

[0105] Furthermore, the first optical array includes a first parabolic mirror, a second parabolic mirror, a beam splitter, and a third parabolic mirror, and the second optical array includes a fourth parabolic mirror and a fifth parabolic mirror. The terahertz signal generated by the terahertz signal generating array is incident on the second parabolic mirror. The second parabolic mirror collimates the terahertz signal and then incident it on the beam splitter. The beam splitter divides the terahertz signal into two beams. One terahertz signal (the first terahertz signal) is reflected back to the first parabolic mirror and focused by the first parabolic mirror onto the silicon lens in the second terahertz signal detection array. The other terahertz signal (the second terahertz signal) is transmitted to the third parabolic mirror and focused by the third parabolic mirror onto the surface of the sample to be tested. The sample to be tested transmits the signal to the fourth parabolic mirror, which collimates it and then focuses it through the fifth parabolic mirror onto the silicon lens in the first terahertz signal detection array.

[0106] Furthermore, the first terahertz signal detection array and the second terahertz signal detection array have the same composition, and the first terahertz signal detection array will be used as an example for explanation.

[0107] The first terahertz signal detection array includes a silicon lens, an on-chip antenna, a mixer, a harmonic frequency doubling amplification link, a power amplifier, and a power divider. Another local oscillator signal is evenly split by the power divider and transmitted to the power amplifier. The power amplifier transmits the local oscillator signal to the harmonic frequency doubling amplification link. The first terahertz signal is mixed with the local oscillator signal processed by the harmonic frequency doubling detection link through the silicon lens and on-chip antenna, and then mixed by the mixer to obtain the transmitted intermediate frequency signal, thereby realizing the detection of the terahertz signal.

[0108] Similarly, the second terahertz signal is mixed with the local oscillator signal, which has been processed by the harmonic frequency multiplication amplification link, through a silicon lens and on-chip antenna, and then mixed by a mixer to obtain the reflected intermediate frequency signal, so as to realize the detection of terahertz signal.

[0109] The reference intermediate frequency (IF) signal, transmitted IF signal, and reflected IF signal are transmitted to a multi-channel mixing module via a switching matrix. The multi-channel mixing module mixes the reference IF signal and the second local oscillator signal with the transmitted IF signal and the reflected IF signal, respectively, to obtain a 7.6MHz reflected IF signal and a 7.6MHz transmitted IF signal, which are then output to a multi-channel IF conditioning module. The multi-channel IF conditioning module filters and amplifies the IF signals before they are digitized by a multi-channel high-speed acquisition module and transmitted to a computer for further processing.

[0110] In this embodiment, the microwave excitation signal generation unit is a microwave signal generator, and the precision control module is a four-axis control module composed of optical devices such as translation stage, lifting stage and rotary stage in the prior art.

[0111] The descriptions of each embodiment in the above embodiments have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0112] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for analyzing multi-morphological nonmetallic substances, characterized in that, include: The transmission and reflection signals of the sample to be tested are acquired. Based on the transmission and reflection signals, the transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum of the sample to be tested are determined. The transmission and reflection signals are acquired by a multi-morphology non-metallic material analysis system. The transmission signal includes the transmission signal amplitude matrix ψ3 and the transmission signal phase matrix ψ4. Based on the amplitude μ1 and phase μ2 of the transmission reference signal, the amplitude Ж3=ψ3 / μ1 and the phase Ж4=ψ4 / μ2 of the transmission terahertz characteristic spectrum of the sample to be tested are determined. The transmission reference signal is a transmission signal to air. The reflection signal includes the reflection signal amplitude matrix ψ1 and the reflection signal phase matrix ψ2. Based on the amplitude λ1 and phase λ2 of the reflection reference signal, the amplitude Ж1=ψ1 / λ1 and the phase Ж2=ψ2 / λ2 of the reflection terahertz characteristic spectrum of the sample to be tested are determined. The reflection reference signal is the reflected terahertz signal of a strongly reflective metal calibration component. The thickness information of the sample is obtained by performing an inverse Fourier transform on the reflected terahertz characteristic spectrum of the sample: Based on the amplitude and phase of the reflected terahertz characteristic spectrum of the sample, an inverse Fourier transform is performed on the reflected terahertz spectrum to obtain the time-domain terahertz characteristic spectral lines. The first and second reflection peaks in the terahertz characteristic spectral lines are extracted, and the thickness information of the sample is determined based on the first and second reflection peaks. Time-domain filtering is then performed on the transmitted and reflected terahertz characteristic spectra based on the thickness information of the sample. The composition of the sample to be tested is obtained based on the time-domain filtered transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum. Wavelet decomposition is performed on the time-domain filtered transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum. Based on the wavelet coefficients after wavelet decomposition, the transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum are reconstructed. Characteristic peaks are judged on the reconstructed transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum. If characteristic peaks exist, the sample to be tested is a single-component sample. The category of the sample to be tested is determined based on the position of the characteristic peaks. If no characteristic peaks exist, the characteristic parameters of the sample to be tested are calculated based on the reconstructed transmission terahertz characteristic spectrum and the reflection terahertz characteristic spectrum. Based on the morphology of the sample to be tested, the characteristic parameters of the sample to be tested are selected and input into the trained machine learning network for processing to determine the category of the sample to be tested. Two-dimensional image matrices of the sample under test are obtained based on Ж1, Ж2, Ж3, and Ж4. The two-dimensional image matrices of different frequencies are input into a preset convolutional neural network for processing to achieve cross-scale three-dimensional structural tomography and obtain three-dimensional structural images of the sample under test.

2. A multi-morphological non-metallic substance analysis system, characterized in that, It includes a microwave excitation signal generation unit, a switching matrix, a power divider amplifier module, a terahertz signal generation array, a first terahertz signal detection array, a second terahertz signal detection array, a multi-channel data acquisition unit, a first optical array, a second optical array, and a computer; The microwave excitation signal generating unit outputs a local oscillator signal and a radio frequency (RF) signal. The local oscillator signal is processed by a power divider and amplification module and then transmitted to the switching matrix. The switching matrix transmits the RF signal and the local oscillator signal to the terahertz signal generating array and the local oscillator signal to the terahertz signal detection array. Under the excitation of the local oscillator signal and the RF signal, the terahertz signal generates a terahertz signal and a reference intermediate frequency (IF) signal. The terahertz signal is transmitted to the sample under test through a first optical array, and the sample under test generates a transmitted IF signal, which is then focused onto the first terahertz signal detection array through a second optical array. The terahertz signal is reflected by the first optical array to generate an intermediate frequency signal and then focused onto the second terahertz signal detection array. The multi-channel data acquisition unit acquires the reference intermediate frequency signal, the transmitted intermediate frequency signal, and the reflected intermediate frequency signal via the switching matrix, performs mixing processing on each signal, acquires the transmitted signal and the reflected signal, and transmits them to the computer. The computer acquires the transmission and reflection signals of the sample to be tested and performs the steps of the multimorphic nonmetallic material analysis method as described in claim 1.

3. The multi-morphological nonmetallic substance analysis system as described in claim 2, characterized in that, It also includes a precision control module, which is disposed between the first optical array and the second optical array; The sample to be tested is placed in the precision control module.

4. The multi-morphological nonmetallic substance analysis system as described in claim 2, characterized in that, The terahertz signal generating array includes a power divider, a harmonic signal generating unit, and a silicon lens; The local oscillator signal and the radio frequency signal are evenly split by the power divider and then transmitted to multiple harmonic signal generating units. Under the excitation of the local oscillator signal and the radio frequency signal, the harmonic signal generating units generate terahertz sub-signals and reference intermediate frequency signals. Terahertz sub-signals are spatially power-combined using silicon lenses to obtain terahertz signals.

5. The multi-morphological nonmetallic substance analysis system as described in claim 2, characterized in that, The first terahertz signal detection array and the second terahertz signal detection array have the same composition. The first terahertz signal detection array includes a silicon lens, an on-chip antenna, a harmonic frequency multiplication amplification link, a power amplifier, and a power divider. The local oscillator signal is divided equally by the power divider and then transmitted to the power amplifier. The power amplifier transmits the local oscillator signal to the harmonic frequency doubling amplification link. The first terahertz signal is transmitted to the harmonic frequency doubling amplification link via a silicon lens and an on-chip antenna, and is mixed with the local oscillator signal processed by the harmonic frequency doubling detection link to realize the detection of the terahertz signal.

6. The multi-morphological nonmetallic substance analysis system as described in claim 2, characterized in that, The first optical array includes a first parabolic mirror, a second parabolic mirror, a beam splitter, and a third parabolic mirror; the second optical array includes a fourth parabolic mirror and a fifth parabolic mirror. The terahertz signal generated by the terahertz signal generating array is incident on the second parabolic mirror. The second parabolic mirror collimates the terahertz signal and then incident it on the beam splitter. The beam splitter divides the terahertz signal into two beams. One beam of terahertz signal is reflected to the first parabolic mirror and focused by the first parabolic mirror onto the silicon lens in the second terahertz signal detection array. The other beam of terahertz signal is transmitted to the third parabolic mirror and focused by the third parabolic mirror onto the surface of the sample to be tested. It is then transmitted through the sample to the fourth parabolic mirror, collimated by the fourth parabolic mirror, and focused by the fifth parabolic mirror onto the silicon lens in the first terahertz signal detection array.

Images