Extremely low frequency signal receiver based on man-machine interaction algorithm and signal processing method thereof

Abstract

The invention discloses an extremely-low-frequency signal receiver based on a human-computer interaction algorithm and a signal processing method thereof. The receiver comprises a multi-rate FIR band-pass filter, a combined filter, a self-adaptive intensifier, a correlation detection filter and a threshold decoder which are electrically connected together in sequence, the combined filter comprises a self-adaptive wave trap and an FIR Wiener filter which are connected in parallel. According to the receiver and the signal processing method thereof disclosed by the invention, real-time reliable detection and communication decoding capability of extremely-low-frequency weak signals can be realized, and the engineering application range is effectively expanded.

Description

Technical Field

[0001] The present invention belongs to the field of extremely low frequency signal reception, and in particular relates to an extremely low frequency signal receiver based on a human-computer interaction algorithm and a signal processing method thereof in the field, and focuses on receiving and processing extremely low frequency signals within the frequency range of 3 Hz to 30 Hz. Background Art

[0002] With its good penetration, long-wave propagation characteristics and anti-interference ability, extremely low frequency has demonstrated unique application value in submarine communications, geological exploration, pipeline detection, tunnel communications and other fields.

[0003] With the rapid development of electronic technology and the growing demand for high-precision, long-distance communications, extremely low frequency communications have been continuously optimized in design. The use of high-sensitivity receiving antennas combined with low-noise analog front ends and high-performance digital processing technology can achieve efficient signal capture, precise analysis and reliable transmission, further promoting scientific and technological progress and development in related fields.

[0004] Extremely low frequency (ELF) signals usually refer to signals with a frequency range of 3Hz to 30Hz. Such signals have the following significant characteristics:

[0005] Good penetration: Extremely low frequency signals have good penetration into metal, seawater, soil and other media, which makes them have unique application value in some specific scenarios.

[0006] Long wave propagation: Due to the low frequency, the wavelength of the extremely low frequency signal is longer, which can form a waveguide effect between the earth's surface and the ionosphere, thereby achieving long-distance propagation.

[0007] Strong anti-interference ability: Compared with high-frequency signals, ultra-low frequency signals are less affected by electromagnetic interference, so they have better stability in some scenarios with complex electromagnetic environments.

[0008] Technologies related to extremely low frequency communications are currently mainly used in the following areas:

[0009] (1) Communications:

[0010] Submarine communication: Extremely low frequency signals are widely used for communication between submarines and surface ships or land command posts because they can penetrate seawater. This communication method overcomes the difficulty of high-frequency radio waves to penetrate seawater, and enables submarines to communicate reliably over long distances underwater.

[0011] Long-distance communication: Since the wavelength of ELF signals is extremely long and the attenuation is extremely low, attenuating only 1 to 2 dB per thousand kilometers, a single transmitter has the potential to communicate around the world. Therefore, in some special occasions, ELF signal receivers are used to achieve long-distance communication across the ocean.

[0012] Electromagnetic wave engineering communication while drilling: Extremely low frequency electromagnetic wave communication while drilling is a technology that uses electromagnetic waves to transmit information between the underground and the ground during the drilling process. This technology has the advantages of fast transmission speed, stable signal, and little influence by drilling fluid. It has been widely used in drilling operations in the oil, natural gas and other industries.

[0013] (2) Geological exploration and monitoring:

[0014] Geophysical exploration: Extremely low frequency signal receivers play an important role in geological exploration. By receiving extremely low frequency electromagnetic signals naturally generated by underground rocks, ore bodies, etc., it is possible to infer underground geological structures and search for mineral resources.

[0015] Earthquake monitoring: ELF signal receivers can also be used in earthquake monitoring networks. By receiving the ELF signals generated by seismic waves propagating inside the earth, they provide data support for earthquake early warning and scientific research.

[0016] (3) Industrial monitoring:

[0017] Pipeline detection: In the field of pipeline detection, the extremely low frequency signal receiver can receive the extremely low frequency magnetic field signal emitted by the mobile robot in the pipeline to achieve the robot's tracking and positioning. This method has been widely used in non-destructive detection of pipeline defects and other aspects.

[0018] Other industrial applications: ELF signal receivers can also be used to monitor the operating status of industrial equipment. By receiving the ELF electromagnetic signals generated by the equipment, potential faults can be discovered in a timely manner to ensure production safety.

[0019] In electromagnetic wave communication while drilling, the wireless communication signal from the borehole to the ground is transmitted to the ground through the formation. The received signal will inevitably be affected by various noise interferences. As the transmission distance of wireless communication increases, the ground received signal becomes weaker and weaker, and the signal-to-noise ratio becomes worse and worse, which eventually makes it difficult to detect and decode. How to receive on the ground and better recover the original signal from the signal "contaminated" by noise is one of the key technologies to improve the communication capabilities in this field.

[0020] In the field of EM (electromagnetic wave) communication while drilling, the signal strength of extremely low frequency signals will be significantly attenuated during underground transmission due to changes in formation resistivity, electromagnetic shielding effects of drill pipes and other metal structures, and interference from other electromagnetic waves underground, and a large amount of noise may be mixed in. This will lead to a decrease in the signal-to-noise ratio of the signal received by the receiver, increasing the difficulty of signal processing. In order to improve the anti-interference ability of the signal, complex signal processing technologies such as filtering, demodulation, synchronization, etc. are usually required, but these technologies will also increase the complexity and cost of the system. In addition, existing extremely low frequency receivers while drilling may have certain limitations in data processing capabilities. With the continuous development of drilling technology, the requirements for data processing speed, accuracy and real-time performance are becoming higher and higher. However, the receiver may not be able to process a large amount of data in a timely and accurate manner, resulting in data loss or delay, affecting the efficiency and accuracy of drilling operations. During the underground transmission of extremely low frequency signals, their waveform and phase may be distorted by various factors. This makes the receiver face certain challenges in identifying and parsing signals. In order to accurately extract useful information, advanced signal recognition and parsing algorithms are required, but these algorithms may have high computational complexity and difficulty in implementation. Summary of the invention

[0021] The technical problem to be solved by the present invention is to provide an extremely low frequency signal receiver and a signal processing method thereof based on a human-computer interaction algorithm, which is based on extremely low frequency electromagnetic wave transmission communication while drilling and can also be applied to other fields related to extremely low frequency communication and monitoring.

[0022] The present invention adopts the following technical solution:

[0023] A very low frequency signal receiver based on a human-computer interaction algorithm, the improvement of which is that it comprises a multi-rate FIR bandpass filter, a combination filter, an adaptive enhancer, a correlation detection filter and a threshold decoder which are electrically connected in sequence, wherein the combination filter comprises an adaptive notch filter and a FIR Wiener filter connected in parallel.

[0024] Furthermore, the receiving antenna adopts a magnetic loop antenna.

[0025] A signal processing method uses the above-mentioned extremely low frequency signal receiver, and its improvement lies in that: the ELF-EM signal X(n) collected by the receiver is passed through a multi-rate FIR bandpass filter to obtain a bandpass signal X1(n); if the downhole measurement while drilling transmission excitation device is not working, a register is used to store noise data for a period of time, and a time-frequency analysis is performed on it to find out what type of noise it is; if there is in-band interference, the frequency, phase and amplitude information of the interference are required; after the above steps, the signal X1(n) is sent to a combined filter to obtain a filtered signal X2(n); X2(n) is then passed through a correlation detection filter to obtain X3(n); X3(n) is passed through a threshold decoder to obtain decoded data X4(n); and finally, an engineering value Y(n) is output through adaptive judgment.

[0026] Furthermore, the filtering algorithm of the multi-rate FIR bandpass filter is:

[0027] (1) Design the shaping filter F(z):

[0028] The upper and lower cutoff frequencies of the passband of F(z) are and The upper and lower cut-off frequencies of the stop band are and Its transfer function is:

[0029]

[0030] (2) Design the periodic filter F(z L ):

[0031] The mathematical expression of the upper sampler is:

[0032]

[0033] Cascading the two, we get a periodic filter F(z L ):

[0034]

[0035] (3) Design the low-pass filter I(z):

[0036] The transition band frequency range of the low-pass filter is: [ω p2 ,f s / L-ω s2 ];

[0037] (4) The periodic filter F(z L ) is cascaded with the low-pass filter I(z) to obtain the multi-rate FIR filter H IFIR (z).

[0038] Furthermore, the time-frequency analysis algorithm is:

[0039] Assume the sinusoidal signal is:

[0040]

[0041] Its discretized expression is:

[0042]

[0043] If we take the positive frequency part of the Fourier transform, we have:

[0044]

[0045] In the above formula, Assume that the frequency f0 of the sinusoidal signal is an integer multiple of Δf, where Δf is the frequency interval between the spectral lines in the spectrum, called the resolution, and f s / N, f0=lΔf, which means f0 coincides with a certain line on the spectrum after FFT. Let f0 coincide with the kth line, that is, k=l, so we have From K=0 and formula (4-5), we get:

[0046]

[0047] Further export:

[0048]

[0049] A and θ are the amplitude and phase angle of the desired signal respectively;

[0050] If f0 is not on a certain spectral line or between two spectral lines, a correction method is needed to analyze and calculate its true value:

[0051] With a single frequency signal Assume the angular frequency ω m =2πf m , and the frequency f m =αΔf, where α is not an integer. In the case of discretization, t=nΔt, then:

[0052]

[0053] In the above formula, A is a complex number, which is expressed as A=A m e jθ , θ is the initial phase angle of the signal; Δf is the frequency resolution, Δt is the time resolution, and ΔfΔt=(f s / N)(1 / f s )=1 / N;

[0054] x m The DFT of (n) is X m (k), derive Xm (k) is:

[0055]

[0056] Frequency correction:

[0057] Formula (4-9) represents the DFT value of a single-frequency signal after adding a rectangular window, and X m (k) modulus |X m (k)|is:

[0058]

[0059] k is the index value of the spectrum line, and the signal frequency f m =aΔf is between two spectral lines, let the index number of the spectral line with the maximum value in the spectrum be k, and the index number of the second largest spectral line be k-1 or k+1;

[0060] Signal frequency f m Between the spectrum lines k and k-1 or k+1, the frequencies corresponding to the spectrum lines k, k-1 and k+1 are kΔf, (k-1)Δf and (k+1)Δf, and their corresponding amplitudes are y k ,y k-1 and k+1 , frequency f m Also written as: f m =αΔf=(k+Δk)Δf, 0<|Δk|<1, and α=k+Δk, when f m Between the spectrum lines k and k+1, Δk>0;

[0061] The index numbers of the largest spectral line and the second largest spectral line are k and k-1 respectively. The estimated value of the abscissa of the center of gravity of the main lobe is obtained by the energy center of gravity method:

[0062]

[0063] The index numbers of the largest spectral line and the second largest spectral line are k and k+1 respectively, and the coordinates of the center of gravity of the main lobe are obtained as follows:

[0064]

[0065] Since x0 = k + Δk, the frequency correction formula is:

[0066]

[0067] The calibration frequency is:

[0068] f m =(k+Δk)·f s / N (4-12)

[0069] Amplitude correction:

[0070] When N>>1, and α=k+Δk, formula (4-10) is further calculated as:

[0071]

[0072] Obtain The above results are processed by single-sided spectrum to obtain the amplitude correction of the real signal:

[0073]

[0074] Furthermore, the adaptive notch algorithm is an adaptive algorithm with a variable step size:

[0075]

[0076] In the above formula: W(n) is the filter weight coefficient, α is the coefficient of the control function shape, m is the adjustment factor, and b(n) is the absolute value of the error change.

[0077] Furthermore, the adaptive enhancement algorithm is an adaptive algorithm with a variable step size:

[0078] The mathematical equation between expectation and input is:

[0079] d(n)=h T x(n)+c(n) (4-15)

[0080] The objective function of the Leaky-LMS algorithm is:

[0081] J(n)=e 2 +γW T (n)W(n) (4-16)

[0082] The error function is:

[0083] e(n)=d(n)-x T (n)W(n) (4-17)

[0084] γ is an embedding factor greater than 0, where γW T (n)W(n) is a sparse matrix;

[0085]

[0086] In the above formula: μ is the step size factor;

[0087] The condition for the algorithm to converge in formula (4-25) is that the value range of the step size factor μ is:

[0088]

[0089] The beneficial effects of the present invention are:

[0090] The receiver and signal processing method disclosed in the present invention can realize real-time reliable detection and communication decoding capabilities of extremely low frequency weak signals, effectively expanding the scope of engineering applications.

[0091] (1) Enhanced weak signal detection: The ELF receiver uses advanced signal processing technology and algorithms to reliably detect ELF weak signals with large dynamic range, short duration, and almost drowned by narrowband noise in real time. This is of great significance for obtaining useful signals in complex environments.

[0092] (2) Improve detection accuracy: By optimizing the receiver’s hardware design and signal processing flow, the detection accuracy of extremely low frequency signals can be improved and false alarms and missed alarms can be reduced.

[0093] The receiver and the signal processing method thereof disclosed in the present invention can effectively expand the application field.

[0094] Communications field: Extremely low frequency receivers have potential applications in submarine communications, long-distance communications and other fields, and can improve the reliability and concealment of communications.

[0095] Environmental monitoring: In the fields of earthquake monitoring, ocean exploration, etc., extremely low frequency reception can be used to detect geophysical signals and provide important data for scientific research.

[0096] Healthcare: In the biomedical field, ELF signals may be related to certain physiological activities of the human body, so ELF receivers also have potential applications in medical monitoring and diagnosis.

[0097] The receiver and signal processing method disclosed in the present invention can improve user experience.

[0098] In human-computer interaction scenarios, the ELF receiver can be used as one of the signal receiving devices, working in conjunction with other sensors and interactive devices to enhance the user's interaction experience with the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] Figure 1 It is a specific architecture flow chart of the human-computer interaction detection filtering algorithm;

[0100] Figure 2 It is the structure diagram of FIR digital filter;

[0101] Figure 3 It is the amplitude-frequency response diagram of the direct bandpass filter;

[0102] Figure 4 It is a diagram of the structure of a multi-rate interpolating FIR bandpass filter;

[0103] Figure 5 is the amplitude-frequency response diagram of the shaping filter F(z);

[0104] Figure 6 is F(z L )’s amplitude-frequency response diagram;

[0105] Figure 7 is the amplitude-frequency response diagram of the low-pass filter I(z);

[0106] Figure 8 is the multi-rate filter H IFIR (z) amplitude-frequency response diagram;

[0107] Fig. 9 It is the amplitude-frequency response diagram of the multi-rate FIR filter;

[0108] Fig.10 is the spectrum of the rectangular window signal;

[0109] Fig.11 It is the block diagram of the adaptive notch filter;

[0110] Fig.12 It is a diagram of the adaptive signal enhancer of the Leaky-LMS algorithm. DETAILED DESCRIPTION

[0111] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0112] Embodiment 1, this embodiment discloses an extremely low frequency signal receiver based on a human-computer interaction algorithm, which further improves the terminal receiver's detection and communication capabilities for weak signals in order to solve the problem that the extremely low frequency receiving terminal has poor weak signal detection capabilities. It includes a multi-rate FIR bandpass filter, a combined filter, an adaptive enhancer, a correlation detection filter, and a threshold decoder electrically connected in sequence, wherein the combined filter includes an adaptive notch filter and an FIR Wiener filter in parallel. The receiving antenna uses a magnetic loop antenna.

[0113] 1. System design and implementation:

[0114] (1) Antenna selection and installation:

[0115] I. Choose a magnetic ring antenna that is sensitive to magnetic fields as the receiving antenna, such as the HZ-10 antenna produced by Rohde & Schwarz, which has an operating bandwidth of 5Hz to 10MHz and high sensitivity. Other forms of magnetic antennas such as high-sensitivity flux negative feedback antennas can also be selected. In addition, electric field antenna reception can also be selected, such as the electric field antenna of the silver chloride electrode to measure the potential difference to receive signals.

[0116] II. Install the antenna, making sure that the antenna is positioned to best receive ELF signals.

[0117] (2) Low noise analog front-end design:

[0118] IDesign and implement low noise analog front-end circuits for antenna impedance matching, signal amplification and filtering.

[0119] II. Place the analog front-end circuit at the bottom of the antenna and seal it with a shielding box to reduce external interference.

[0120] III. Use a multi-core cable to connect the analog front end to the digital receiver to complete signal transmission and power supply.

[0121] (3) Digital receiver development:

[0122] I. Develop a data acquisition and transmission module based on USB 2.0 / 3.0 bus to achieve efficient data acquisition and transmission.

[0123] II. Integrate high-precision GPS synchronization module to ensure the accuracy and synchronization of data collection.

[0124] III. Use FPGA to complete timing control to ensure the stable operation of the entire system.

[0125] 2. Human-computer interaction interface development

[0126] (1) Demand analysis:

[0127] Clarify the functional requirements of the human-computer interaction interface, such as signal display, parameter setting, data processing, etc.

[0128] (2) Interface design:

[0129] Design an intuitive and easy-to-use graphical user interface (GUI), including signal waveform display area, parameter setting area, data processing result display area, etc.

[0130] Use appropriate programming languages ​​and tools for interface development, such as Python, C++, C#, etc.

[0131] (3) Functional implementation:

[0132] Realize the signal receiving and display functions, and display the extremely low frequency signal collected by the receiver on the interface in real time.

[0133] Implement parameter setting function, allowing users to adjust various parameters of the receiver as needed.

[0134] Realize the data processing function, filter, analyze, store and other operations on the received signals, and display the processing results on the interface.

[0135] 3. Receiver operation and data processing:

[0136] (1) Power on and initialization:

[0137] Turn on the receiver power and perform system initialization, including antenna calibration, parameter setting, etc.

[0138] (2) Signal reception:

[0139] I. Turn on the receiver and start receiving ELF signals.

[0140] II. Monitor signal quality in real time to ensure that the received signal is stable and reliable.

[0141] (3) Data processing and analysis:

[0142] I. Preprocess the received signal, such as filtering, denoising, etc.

[0143] II. Conduct in-depth data analysis, such as spectrum analysis, signal recognition, etc.

[0144] III. Store the analysis results locally or in a remote server for subsequent review and use.

[0145] (4) Human-computer interaction feedback:

[0146] I. The analysis results can be displayed to the user through the human-computer interaction interface, including signal waveform, spectrum diagram, analysis results, etc.

[0147] II. Adjust receiver parameters or perform further data processing based on feedback from on-site communication decoding.

[0148] In the extremely low frequency communication system, various types of noise generated by external electromagnetic interference such as industrial frequency and lightning have caused great interference to the recognition of effective signals received by the terminal and communication decoding. The present invention first uses part of the noise data collected in the time domain as prior information based on the characteristics of the extremely low frequency electromagnetic wave signal, and then accurately analyzes the noise through a time-frequency analysis algorithm. Based on the analysis results, different filtering methods are used to process the interference signal and random noise respectively.

[0149] The present embodiment also discloses a signal processing method, using the above-mentioned extremely low frequency signal receiver, the ELF-EM signal X(n) collected by the receiver passes through a multi-rate FIR bandpass filter to obtain a bandpass signal X1(n), if the downhole measurement while drilling transmission excitation device is not working, the noise data for a period of time is stored in a register, and an accurate time-frequency analysis is performed on it to find out what type of noise it is (interference or random noise or both), if there is in-band interference, the frequency, phase and amplitude information of the interference are required, after the above steps, the signal X1(n) is sent to a combined filter, and different filtering methods are selected according to the type of noise, if the noise contains two types, adaptive notch filtering is first performed, followed by FIR Wiener filtering and adaptive enhanced series filter group to obtain a filtered signal X2(n), X2(n) is then passed through a correlation detection filter to obtain X3(n), X3(n) is passed through a threshold decoder to obtain decoded data X4(n), and finally an engineering value Y(n) is output through adaptive judgment. If the decoding bit error rate is too high and does not meet the engineering requirements, the human-computer interaction phase is entered, and then the noise data is accurately analyzed in time and frequency, and then filtered again according to the method described until the engineering requirements are met. The key to this combined algorithm is the accurate time and frequency analysis of the noise data during the human-computer interaction process, and then different algorithms are used to process it step by step according to the characteristics of the noise.

[0150] Usually, the interference signal and random noise of the ELF-EM transmission channel have statistical characteristics of time-varying non-stationarity. The traditional method uses Fourier transform and window function to suppress out-of-band interference and random noise, but the window function cannot solve the in-band interference and random noise. Therefore, this paper first uses part of the noise data collected in the time domain as prior information based on the characteristics of ELF-EM measurement in drilling engineering, and then accurately analyzes the noise through the above-mentioned time-frequency analysis algorithm. Then, based on the analysis results, different filtering methods are used to solve the interference signal and random noise. The specific architecture flow of the human-computer interaction detection (HCID) algorithm is as follows Figure 1 As shown. At the engineering site, it is easy to determine whether the instrument is in working condition. When the instrument transmitter is not working, the data collected on site is the actual noise data (the present invention collects 2 minutes of noise data for preliminary analysis. The time-frequency analysis correction algorithm can adopt ratio correction, phase correction and other methods). If the current noise interference is different from the previous time interference distribution characteristics, the adaptive notch filter in the HCID combined algorithm cannot filter out the interference noise, which will inevitably lead to an increase in the terminal bit error rate and inability to decode normally. At this time, the on-site engineer can call the time-frequency analysis algorithm software to perform spectral analysis on the engineering site noise, and then feed the analysis results back to the HCID combined filtering algorithm for filtering until the interference noise characteristics of the engineering site change, and repeat the above process for filtering and decoding.

[0151] The filtering algorithm involved in the present invention is:

[0152] (1) Multi-rate FIR filtering algorithm:

[0153] Usually, a narrowband filter can be designed first to filter out the out-of-band noise. In actual engineering, narrowband filters are often designed using FIR digital bandpass filters. This is because FIR digital filters have linear phases and are relatively easy to implement in both software and hardware. Figure 2 As shown, the FIR digital filter is used To describe, h(k) is the filter coefficient, N is the filter order, z -1 is the unit delay. The time domain relationship is: y(n) and x(n) are the input and output sequences respectively.

[0154] Under the communication condition of extremely low frequency transmission, the performance of its FIR bandpass filter is simulated as follows Figure 3 As shown, the sampling rate fs=1000Hz and the filter order is 512.

[0155] Multi-rate interpolating FIR bandpass filter H IFIR (Z) is the periodic FIR filter F(z L ) and a low-pass filter I(z), which is constructed as follows Figure 4 The design steps are as follows:

[0156] ① Design the shaping filter F(z):

[0157] The upper and lower cutoff frequencies of the passband of F(z) are and The upper and lower cut-off frequencies of the stop band are and (The sparse factor L needs to be calculated and analyzed). Its transfer function is:

[0158]

[0159] At this time, the amplitude-frequency response of the shaping filter F(z) is as follows: Figure 5 shown.

[0160] ② Design periodic filter F(z L ):

[0161] The mathematical expression of the upper sampler is:

[0162]

[0163] Cascading the two, we get a periodic filter F(z L ):

[0164]

[0165] F(zL ) has an amplitude-frequency response such as Figure 6 shown.

[0166] ③ Design low-pass filter I(z):

[0167] The transition band frequency range of the low-pass filter is: [ω p2 ,f s / L-ω s2 ], its amplitude-frequency response is as follows Figure 7 shown.

[0168] ④The periodic filter F(z L ) is cascaded with the low-pass filter I(z) to obtain a multi-rate FIR filter H IFIR (z), its amplitude-frequency response is as follows Figure 8 shown.

[0169] The amplitude-frequency characteristics of the multi-rate filter analyzed by the above theory are as follows Fig. 9 As shown, the filter occupies 256-order resources, which saves more resources than the direct FIR filter and has a stronger ability to suppress out-of-band noise, 40 to 50 dB more than the direct type.

[0170] (2) Time-frequency analysis algorithm

[0171] For in-band interference noise and random noise, various traditional window function filters based on Fourier transform cannot fundamentally solve the problem. Therefore, the present invention first performs accurate time-frequency analysis on a section of interference data collected on-site, and then adopts different algorithms for interference noise and random noise to solve the in-band signal "pollution" problem, ultimately improving the terminal signal-to-noise ratio and increasing the transmission distance of the ELF-EM signal in the formation.

[0172] Assume the sinusoidal signal is:

[0173]

[0174] Its discretized expression is:

[0175]

[0176] If we take the positive frequency part of the Fourier transform, we have:

[0177]

[0178] In the above formula, Assume that the frequency f0 of the sinusoidal signal is an integer multiple of Δf, where Δf is the frequency interval between the spectral lines in the spectrum, called the resolution, and f s / N, f0=lΔf, which means f0 coincides with a certain line on the spectrum after FFT. Let f0 coincide with the kth line, that is, k=l, so we have From K=0 and formula (4-5), we get:

[0179]

[0180] Further export:

[0181]

[0182] A and θ are the amplitude and phase angle of the desired signal respectively;

[0183] The above analysis is for special cases. If f0 is not on a certain spectral line or between two spectral lines, a correction method is needed to analyze and calculate its true value:

[0184] With a single frequency signal Assume the angular frequency ω m =2πf m , and the frequency f m =αΔf, where α is not an integer. In the case of discretization, t=nΔt, then:

[0185]

[0186] In the above formula, A is a complex number, which is expressed as A=A m e jθ , θ is the initial phase angle of the signal; Δf is the frequency resolution, Δt is the time resolution, and ΔfΔt=(f s / N)(1 / f s )=1 / N;

[0187] x m The DFT of (n) is X m (k), derive X m (k) is:

[0188]

[0189] Frequency correction:

[0190] Formula (4-9) represents the DFT value of a single-frequency signal after adding a rectangular window, and X m (k) modulus value |X m (k)|is:

[0191]

[0192] The spectrum of formula (4-10) is as follows Fig.10 As shown, k is the index value of the spectrum line, and the signal frequency f m =aΔf is between two spectral lines. Let the index of the spectral line with the maximum value in the spectrum be k, and the index of the second largest spectral line be k-1 or k+1, respectively. Fig.10 as shown in (a) and (b).

[0193] Signal frequency f m Between the spectrum lines k and k-1 or k+1, the frequencies corresponding to the spectrum lines k, k-1 and k+1 are kΔf, (k-1)Δf and (k+1)Δf, and their corresponding amplitudes are y k ,y k-1 and k+1 , frequency f m Also written as: f m =αΔf=(k+Δk)Δf, 0<|Δk|<1, and α=k+Δk, when f m Between the spectrum lines k and k+1, Δk>0;

[0194] The index numbers of the largest spectral line and the second largest spectral line are k and k-1 respectively. The estimated value of the abscissa of the main lobe center of gravity is obtained by the energy center of gravity method:

[0195]

[0196] The index numbers of the largest spectral line and the second largest spectral line are k and k+1 respectively, and the coordinates of the center of gravity of the main lobe are obtained as follows:

[0197]

[0198] Since x0 = k + Δk, the frequency correction formula is:

[0199]

[0200] The calibration frequency is:

[0201] f m =(k+Δk)·f s / N (4-31)

[0202] Amplitude correction:

[0203] When N>>1, and α=k+Δk, formula (4-10) is further calculated as:

[0204]

[0205] Obtain

[0206] In actual communication applications, the processed signal is usually a real signal. The above formula is the calculation result of the complex signal. The above result is processed by single-sided spectrum to obtain the amplitude correction of the real signal:

[0207]

[0208] (3) Adaptive notch algorithm:

[0209] In order to achieve a good filtering effect, the traditional notch filter design needs to set a very high order to filter out interference signals as much as possible, which is unrealistic in engineering. The adaptive notch filter can adaptively track and filter out unwanted interference noise based on the least mean square algorithm and least squares algorithm based on statistical principles, effectively avoiding the shortcomings of traditional notch filters in this regard. The basic structure of the adaptive notch filter is as follows: Fig.11 In the figure a, X(n) is a signal with specific frequency interference, that is, the original input signal, Y(n) is the interference signal of adaptive iterative approximation, and e(n) is the signal after eliminating the interference. If X(n) contains multiple signals with specific frequency interference, its structure is as follows Fig.11 Figure b in .

[0210] Commonly used adaptive algorithms include LMS algorithm, NLMS algorithm, Leaky-LMS algorithm and RLS algorithm. The present invention adopts an adaptive interference notch filter algorithm based on variable step size LMS to further filter out specific frequency interference in the band. The variable step size LMS algorithm is an improved algorithm of the traditional LMS, which solves the contradiction between convergence speed and stability performance. The derivation process of this algorithm is described below.

[0211] The fixed step size LMS algorithm recursive formula is as follows:

[0212]

[0213] e(n)=X(n)-d(n)W(n) (4-35)

[0214] W(n+1)=W(n)+2μ(n)e(n)X(n) (4-36)

[0215] Where: X(n) is the input signal; W(n) is the weight coefficient; d(n) is the reference signal; e(n) is the error; μ(n) is the iteration step size.

[0216] The purpose of adaptive notch filtering is to filter out interference signals as much as possible, but in the fixed step size algorithm, since μ(n) is not 0, iteration will continue. The step size satisfies the following conditions:

[0217] μ(n)=μ, 0<μ<1 / λ max (4-37)

[0218] In the above formula, λ max is the maximum eigenvalue of Signal_in.

[0219] The variable step-size adaptive algorithm (VSS-LMS) proposed in the present invention uses the error change rate to control the change of the step-size parameter μ(n).

[0220]

[0221] In the above formula: W(n) is the filter weight coefficient, α is the coefficient of the control function shape, m is the adjustment factor, and b(n) is the absolute value of the error change.

[0222] (4) Adaptive enhancement algorithm:

[0223] The Leaky-LMS algorithm is an improved version of the traditional LMS algorithm. This algorithm is mainly suitable for broadband noise with strong randomness. Compared with the traditional LMS algorithm, this algorithm has improvements in convergence, stability and transient characteristics.

[0224] The mathematical equation between expectation and input is:

[0225] d(n)=h T x(n)+c(n) (4-39)

[0226] The objective function of the Leaky-LMS algorithm is:

[0227] J(n)=e 2 +γW T (n)W(n) (4-40)

[0228] The error function is:

[0229] e(n)=d(n)-x T (n)W(n) (4-41)

[0230] The Leaky-LMS algorithm adds a sparse matrix to the cost function of the LMS algorithm, where γ is an embedding factor greater than 0. T (n)W(n) is a sparse matrix;

[0231]

[0232] In the above formula: μ is the step size factor;

[0233] The condition for the algorithm to converge in formula (4-25) is that the value range of the step size factor μ is:

[0234]

[0235] The leakage factor γ can be analyzed through measured data, and needs to be considered in terms of convergence speed and convergence stability.

[0236] Leaky-LMS algorithm stability analysis:

[0237] For broadband noise, the instantaneous MSE (mean square error) is:

[0238] ε(n)=εmin +E{V T (n)ΔV} (4-44)

[0239] ε min =E{d 2 (n)}-P T W * (4-45)

[0240] The weight bias vector is:

[0241] L=W * -W L =[l1,l2,…,l N ] T (4-46)

[0242] W L =(R+rI -1 ) -1 P (4-47)

[0243] I is the identity matrix, W L is the steady-state weight coefficient, W * is the instantaneous weight coefficient.

[0244] Combining equations (4-30) and (4-27) we can obtain:

[0245] ε(n)=ε min +E{L T (n)ΔL} (4-48)

[0246] Find the range of step size μ:

[0247]

[0248] In the above formula: is the variance of the input signal x(n). In order to ensure the stable convergence of the Leaky-LMS algorithm, the value of μ must be within its constraint range.

[0249] Adaptive signal enhancer based on Leaky-LMS algorithm Fig.12 shown.

[0250] In summary, the solution of the present invention can be applied to deep earth communication, submarine communication, EM communication while drilling, resource exploration and other fields, including applications in the defense field and the civilian field, and realizes the following technical protection points:

[0251] This solution uses highly sensitive electric antenna and magnetic antenna multi-channel digital receiving technology;

[0252] This scheme adopts the high-performance filtering algorithm of HCID combined filtering (such as Figure 1 The data processing system shown);

[0253] The solution uses a low-noise analog front-end design for antenna impedance matching, signal amplification and filtering. The analog front-end circuit is placed at the bottom of the receiving antenna and sealed with a shielded iron box to reduce external interference.

[0254] The solution develops a data acquisition and transmission module based on USB 2.0 / 3.0 bus to achieve efficient data acquisition and transmission, and integrates a high-precision GPS synchronization module to ensure the accuracy and synchronization of data acquisition. FPGA is used to complete timing control to ensure the stable operation of the entire system;

[0255] The scheme designs a human-computer interaction feedback interface to display the analysis results to the operator or user through the human-computer interaction interface, including signal waveform, spectrum diagram, analysis results, etc. The receiver parameters can also be adjusted and further data processing can be performed based on the feedback of the on-site communication situation.

Claims

1. An extremely low frequency signal receiver based on a human-computer interaction algorithm, characterized in that: The invention comprises a multi-rate FIR bandpass filter, a combination filter, an adaptive enhancer, a correlation detection filter and a threshold decoder which are electrically connected in sequence. The combination filter comprises an adaptive notch filter and a FIR Wiener filter which are connected in parallel.

2. The extremely low frequency signal receiver based on the human-computer interaction algorithm according to claim 1, characterized in that: The receiving antenna adopts a magnetic loop antenna.

3. A signal processing method, using the extremely low frequency signal receiver according to claim 1, characterized in that: The ELF-EM signal X(n) collected by the receiver is passed through a multi-rate FIR bandpass filter to obtain a bandpass signal X1(n). If the downhole measurement while drilling transmission excitation device is not working, the noise data for a period of time is stored in a register, and time-frequency analysis is performed on it to find out what type of noise it is. If there is in-band interference, the frequency, phase and amplitude information of the interference are required. After the above steps, the signal X1(n) is sent to the combined filter to obtain the filtered signal X2(n), and X2(n) is then passed through the correlation detection filter to obtain X3(n). X3(n) is passed through the threshold decoder to obtain the decoded data X4(n), and finally the engineering value Y(n) is output through adaptive judgment.

4. The signal processing method according to claim 3, characterized in that: The filtering algorithm of the multi-rate FIR bandpass filter is: (1) Design the shaping filter F(z): The upper and lower cutoff frequencies of the passband of F(z) are and The upper and lower cut-off frequencies of the stop band are and Its transfer function is: (2) Design the periodic filter F(z L ): The mathematical expression of the upper sampler is: Cascading the two, we get a periodic filter F(z L ): (3) Design the low-pass filter I(z): The transition band frequency range of the low-pass filter is: [ω p2 ,f s / L-ω s2 ]; (4) The periodic filter F(z L ) is cascaded with the low-pass filter I(z) to obtain a multi-rate FIR filter H IFIR (z).

5. The signal processing method according to claim 3, characterized in that: The time-frequency analysis algorithm is: Assume the sinusoidal signal is: Its discretized expression is: If we take the positive frequency part of the Fourier transform, we have: In the above formula, Assume that the frequency f0 of the sinusoidal signal is an integer multiple of Δf, where Δf is the frequency interval between the spectral lines in the spectrum, called the resolution, and f s / N, f0=lΔf, which means f0 coincides with a certain line on the spectrum after FFT. Let f0 coincide with the kth line, that is, k=l, so we have From K=0 and formula (4-5), we get: Further export: A and θ are the amplitude and phase angle of the desired signal respectively; If f0 is not on a certain spectral line or between two spectral lines, a correction method is needed to analyze and calculate its true value: With a single frequency signal Assume the angular frequency ω m =2πf m , and the frequency f m =αΔf, where α is not an integer. In the case of discretization, t=nΔt, then: In the above formula, A is a complex number, which is expressed as A=A m e jθ , θ is the initial phase angle of the signal; Δf is the frequency resolution, Δt is the time resolution, and ΔfΔt=(f s / N)(1 / f s )=1 / N; x m The DFT of (n) is X m (k), derive X m (k) is: Frequency correction: Formula (4-9) represents the DFT value of a single-frequency signal after adding a rectangular window, and X m (k) modulus |X m (k)|is: k is the index value of the spectrum line, and the signal frequency f m =aΔf is between two spectral lines, let the index number of the spectral line with the maximum value in the spectrum be k, and the index number of the second largest spectral line be k-1 or k+1; Signal frequency f m Between the spectrum lines k and k-1 or k+1, the frequencies corresponding to the spectrum lines k, k-1 and k+1 are kΔf, (k-1)Δf and (k+1)Δf, and their corresponding amplitudes are y k ,y k-1 and k+1 , frequency f m Also written as: f m =αΔf=(k+Δk)Δf, 0<|Δk|<1, and α=k+Δk, when f m Between the spectrum lines k and k+1, Δk>0; The index numbers of the largest spectral line and the second largest spectral line are k and k-1 respectively. The estimated value of the abscissa of the center of gravity of the main lobe is obtained by the energy center of gravity method: The index numbers of the largest spectral line and the second largest spectral line are k and k+1 respectively, and the coordinates of the center of gravity of the main lobe are obtained as follows: Since x0 = k + Δk, the frequency correction formula is: The calibration frequency is: f m =(k+Δk)·f s / N (4-12) Amplitude correction: When N>>1, and α=k+Δk, formula (4-10) is further calculated as: Obtain The above results are processed by single-sided spectrum to obtain the amplitude correction of the real signal:

6. The signal processing method according to claim 3, characterized in that: The adaptive notch filter algorithm is an adaptive algorithm with variable step size: In the above formula: W(n) is the filter weight coefficient, α is the coefficient of the control function shape, m is the adjustment factor, and b(n) is the absolute value of the error change.

7. The signal processing method according to claim 3, characterized in that: The adaptive enhancement algorithm is an adaptive algorithm with variable step size: The mathematical equation between expectation and input is: d(n)=h T x(n)+c(n) (4-15) The objective function of the Leaky-LMS algorithm is: J(n)=e 2 +γW T (n)W(n) (4-16) The error function is: e(n)=d(n)-x T (n)W(n) (4-17) γ is an embedding factor greater than 0, where γW T (n)W(n) is a sparse matrix; In the above formula: μ is the step size factor; The condition for the algorithm to converge in formula (4-25) is that the value range of the step size factor μ is:

Images