Data transmission method for underwater measurement systems based on power line carrier communication

Abstract

This invention provides a data transmission method for an underwater measurement system based on power line carrier communication, belonging to the field of data transmission technology. The method includes: sampling an orthogonal frequency division multiplexing (OFDM) carrier signal coupled from a power line to obtain a discrete received signal; calculating local energy estimation and impulse noise probability estimation; constructing an adaptive time-varying blanking threshold using a preset nonlinear mapping function, automatically increasing the threshold when the impulse noise probability increases and automatically decreasing it when it decreases; setting the sampling points in the discrete received signal whose amplitude exceeds the adaptive time-varying blanking threshold to zero, obtaining a blanked signal; using the blanked signal as the input signal to cancel power frequency harmonic interference, obtaining a purified signal; and performing OFDM demodulation on the purified signal to recover the underwater measurement data. This invention effectively solves the problem of low communication reliability in ultra-deep underwater construction caused by the time-varying and non-stationary nature of noise in traditional methods.

Description

Technical Field

[0001] This invention relates to the field of data transmission technology, and specifically to a data transmission method for an underwater measurement system based on power line carrier communication. Background Technology

[0002] In underwater engineering projects such as the construction of ultra-deep cutoff walls, equipment such as borehole depth measurement devices, concrete surface liquid level monitoring robots, and drilling measurement-while-drilling devices utilize power line carrier communication technology to simultaneously achieve power supply and data transmission using existing armored power cables. To suppress strong pulse noise and power frequency harmonic interference present in the field environment, the receiving end is typically equipped with a pulse blanking module and a notch filter to preprocess the coupled carrier signal before sending it to the demodulator to recover the underwater measurement data.

[0003] However, the noise characteristics at construction sites exhibit significant time-varying and non-stationary properties. On one hand, the strong impulse noise generated by the frequent start-up and shutdown of large construction equipment fluctuates drastically in probability and energy level with different construction procedures. Traditional fixed-threshold blanking methods cannot adaptively adjust to changes in noise intensity, easily leading to missed or over-blanking, resulting in signal distortion. On the other hand, power frequency harmonic interference generated by nonlinear loads in the power grid also drifts in frequency and amplitude due to load fluctuations, making it difficult for fixed-parameter notch filters to effectively track and suppress it. The coupling and superposition of these two types of interference in the time and frequency domains severely degrades the signal-to-noise ratio of the communication link, making it difficult for existing receiver preprocessing methods to guarantee the reliability of measurement data transmission in ultra-deep underwater environments. Summary of the Invention

[0004] To address the problems in related technologies, this invention provides a data transmission method for an underwater measurement system based on power line carrier communication.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] This invention provides a data transmission method for an underwater measurement system based on power line carrier communication, comprising the following steps:

[0007] Step S1: Receive the orthogonal frequency division multiplexing carrier signal coupled from the power line, and sample the carrier signal to obtain a discrete received signal;

[0008] Step S2: In the time domain, calculate the local energy estimate and the impulse noise occurrence probability estimate based on the local statistics of the discrete received signal;

[0009] Step S3: Based on the local energy estimation and the impulse noise occurrence probability estimation, an adaptive time-varying blanking threshold is constructed using a preset nonlinear mapping function, so that the threshold automatically increases when the impulse noise occurrence probability increases and automatically decreases when it decreases.

[0010] Step S4: Set the sampling points in the discrete received signal whose amplitude exceeds the adaptive time-varying blanking threshold to zero to obtain the blanked signal;

[0011] Step S5: Using the blanking-processed signal as the input signal, the power frequency harmonic interference is canceled by an adaptive notch filter based on the normalized least mean square algorithm to obtain a purified signal; wherein, the reference signal of the adaptive notch filter is composed of orthogonal sine components corresponding to the frequencies of each power frequency harmonic.

[0012] Step S6: Perform orthogonal frequency division multiplexing demodulation on the purified signal to recover the underwater measurement data.

[0013] Optionally, in step S1, the received carrier signal is composed of a useful orthogonal frequency division multiplexing signal, strong impulse noise, power frequency harmonic interference, and colored background noise superimposed, and the strong impulse noise follows a Middleton Class A statistical model; the carrier signal is converted from analog to digital at the sampling frequency to obtain the discrete received signal, and the index range of the discrete received signal is from the first sampling point to the last sampling point within an orthogonal frequency division multiplexing symbol period.

[0014] Optionally, in step S2, the local energy estimate is the root mean square value of the discrete received signal within an orthogonal frequency division multiplexing symbol frame, which is calculated by taking the square root of the average of the sum of squares of the amplitudes of each sampling point within the frame.

[0015] Optionally, in step S2, the probability estimate of the impulse noise occurrence is determined as follows:

[0016] The amplitude of each sampling point within the frame is determined, and the number of sampling points whose amplitude exceeds three times the root mean square value is counted. This number is then divided by the total number of sampling points in the frame, and the resulting ratio is a dimensionless value between 0 and 1.

[0017] Optionally, in step S3, the adaptive time-varying blanking threshold is calculated as follows:

[0018] The base threshold coefficient is multiplied by the root mean square value, and then multiplied by an adjustment factor consisting of a weighted sum of the adaptive adjustment depth coefficient and the arctangent function value; wherein the input of the arctangent function is the probability sensitivity coefficient multiplied by the impulse noise occurrence probability estimate and then divided by the probability normalization benchmark value.

[0019] Optionally, in step S4, setting the sampling points in the discrete received signal whose amplitude exceeds the adaptive time-varying blanking threshold to zero specifically involves:

[0020] For each sampling point, if its amplitude is less than or equal to the adaptive time-varying blanking threshold, the value of the sampling point remains unchanged; if its amplitude is greater than the adaptive time-varying blanking threshold, the value of the sampling point is set to zero; thus, the blanking-processed signal is obtained.

[0021] Optionally, in step S5, the reference signal of the adaptive notch filter is composed of orthogonal sinusoidal components corresponding to the frequencies of each power frequency harmonic, specifically including:

[0022] For each harmonic from the fundamental frequency to the highest harmonic order that needs to be suppressed, a cosine reference signal and a sine reference signal are constructed respectively. The frequencies of the cosine reference signal and the sine reference signal are the frequencies of that harmonic order, and their phases differ by ninety degrees.

[0023] Optionally, in step S5, the cancellation process of the adaptive notch filter based on the normalized least mean square algorithm further includes:

[0024] Using the blanking-processed signal as the desired input, the difference between the blanking-processed signal and the sum of the products of the cosine weights and cosine reference signals corresponding to all harmonics, and the products of the sine weights and sine reference signals, is calculated as the error signal.

[0025] The weight coefficients of the next sampling point are updated using the normalized least mean square algorithm. The update amount of each weight coefficient is the step size factor divided by the sum of the norm square of the reference signal vector corresponding to the harmonic and the regularization constant, and then multiplied by the error signal and the corresponding reference signal value.

[0026] After iterating through all sampling points within the frame, the final error signal sequence is output as the purified signal.

[0027] Optionally, in step S6, performing orthogonal frequency division multiplexing demodulation on the purified signal includes:

[0028] After removing the cyclic prefix, the purified signal is subjected to a fast Fourier transform to convert the time-domain signal into a frequency-domain signal. The value at each frequency point is obtained by multiplying each sampling point in the time domain with the complex exponential basis function of the corresponding frequency and then summing the results. Then, the frequency-domain signal is subjected to channel equalization, demapping and decoding to recover the underwater measurement data.

[0029] Optionally, the method is applied to the construction environment of ultra-deep cutoff walls, with a construction depth exceeding 200 meters; and the value range of the basic threshold coefficient used in step S3 to construct the adaptive time-varying blanking threshold is 2.5 to 4.0, and the value range of the step size factor used in the normalized least mean square algorithm in step S5 is 0.01 to 0.5.

[0030] Beneficial effects:

[0031] 1. First, the method of the present invention has strong adaptability and can match the time-varying noise environment of the construction site in real time. Specifically, the present invention dynamically constructs a blanking threshold linked to the probability of impulse noise occurrence by calculating the local statistics of the received signal in real time. This allows the threshold to be automatically adjusted according to the noise intensity changes caused by the start and stop of construction equipment, thereby avoiding the problems of missed blanking or over-blanking that occur with traditional fixed thresholds when the noise environment changes.

[0032] Second, the method of this invention possesses frequency tracking capability in suppressing power frequency harmonic interference. Specifically, this invention employs an adaptive notch filter based on the normalized least mean square algorithm, utilizing orthogonal sinusoidal reference signals to cancel power frequency harmonics. This algorithm, through error signal-driven iterative updates of weight coefficients, can automatically track minute drifts in the fundamental frequency of the power grid, eliminating the need for pre-setting precise interference frequencies. This solves the problem of reduced suppression capability of traditional fixed-frequency notch filters in non-stationary harmonic environments.

[0033] Third, in the method of this invention, the cascaded processing order can ensure the robustness of signal processing. Specifically, the method of this invention first performs pulse blanking and then performs power frequency harmonic cancellation. This order utilizes the time-domain sparsity of pulse noise to remove high-energy pulse spikes first, which can provide a relatively stable input environment for the subsequent adaptive notch filter. This can avoid the impact of strong pulse noise on the convergence process of the least mean square algorithm and prevent the weight coefficients from diverging.

[0034] Fourth, the method of this invention provides controllable signal distortion and maintains stable operation even under extreme conditions. Specifically, when constructing the adaptive blanking threshold, this invention uses the arctangent function to perform a nonlinear mapping on the probability of impulse noise occurrence. This function tends to saturate when the impulse probability is high, thus providing an upper bound for the threshold's rise. This avoids the situation where a large number of normal signal samples are erroneously set to zero due to the threshold increasing infinitely under extreme conditions of extremely dense impulses, ensuring the stability of the system and the basic recoverability of the signal.

[0035] Fifth, the algorithm implementation of the method of the present invention mainly uses scalar multiplication and addition operations, with controllable complexity, and can be implemented in real time on the digital signal processor or field-programmable gate array at the receiving end. At the same time, the method of the present invention directly utilizes existing power line carrier transmission lines and devices, without the need for hardware modification of existing power lines, transmitting equipment or construction equipment, and has good engineering deployability and promotion prospects.

[0036] 2. Other beneficial effects or advantages of the present invention will be described in detail in the specific embodiments. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] in:

[0039] Figure 1 This is a flowchart illustrating the steps of a data transmission method for an underwater measurement system based on power line carrier communication, provided by an exemplary embodiment of the present invention. Detailed Implementation

[0040] To facilitate a clearer and more accurate understanding of the technical solutions of this invention by those skilled in the art, the existing related technologies and their technical problems will be described in more detail below.

[0041] Currently, when constructing ultra-deep anti-seepage walls with depths exceeding 200 meters, various intelligent devices used to monitor construction quality, such as automatic depth measuring devices for measuring trench depth, ultrasonic robots for monitoring the concrete pouring liquid level, and drilling measurement-while-drilling systems for real-time tracking of drill bit trajectories, are all placed in the mud and concrete environment hundreds of meters underground.

[0042] To power these devices and retrieve data, robust armored power cables are typically used. Data signals containing information such as depth and liquid level are modulated onto a high-frequency carrier wave using power line carrier communication technology. This signal is then transmitted to the ground along with DC or AC power supply current through the existing power line. The ground receiving end uses a coupler to separate the weak carrier signal from the high-voltage current.

[0043] The separated signal is not pure; it is mixed with harsh noise generated by various electrical devices on site. In order to restore the data, the receiving end usually sets up two key preprocessing stages before sending the signal to the demodulator. One is a pulse blanking module (generally, a fixed amplitude threshold is set to remove or zero out instantaneous high-energy spikes exceeding the threshold), and the other is a power frequency notch filter (for continuous interference from the power grid at 50Hz and its integer multiples, multiple narrowband filters of fixed frequencies are set to filter out the energy of these specific frequencies).

[0044] The aforementioned traditional fixed-parameter processing methods are effective in stable, ideal power environments. However, in heavy industrial construction sites such as ultra-deep cutoff walls, the actual situation is much more complex, making traditional methods almost impossible to operate normally. They face the following problems:

[0045] First, a fixed threshold cannot cope with time-varying strong impulse noise.

[0046] The noise environment on site is highly variable. For example, a 90-kilowatt high-power motor that powers the hydraulic grab bucket starts frequently during trench excavation. The instantaneous current at startup can reach 5 to 7 times the rated current. This sudden current will couple out a pulse peak of several volts, lasting from a few microseconds to a few milliseconds, through the power line in the carrier communication frequency band.

[0047] Meanwhile, when a twin-wheel milling machine is operating, its cutting wheels continuously impact the hard rock, causing the load to fluctuate wildly, and the inverter arm of the frequency converter to switch on and off at high speed. This continuously generates a series of pulse groups with varying amplitudes and random intervals. When the motor starts and the milling operation is performed simultaneously, the receiver sees not just occasional single high pulses, but a superposition of pulse groups and background noise.

[0048] If a high threshold (e.g., 1V) is set, it cannot effectively eliminate densely occurring 0.6V and 0.8V pulse clusters when the motor is not running and the background noise is low, leading to continuous decoding errors. If a low threshold (e.g., 0.2V) is set, although weak pulses can be eliminated, when a high-power device starts up and generates a normal 1.5V signal, the signal peak value will far exceed the threshold and be incorrectly set to zero, resulting in the complete loss of data blocks. This drastic noise power fluctuation determined by the construction process is the fundamental reason why a fixed threshold cannot be adapted.

[0049] Second, fixed notch filters have difficulty tracking harmonic interference caused by frequency drift.

[0050] Interference from the power frequency and its harmonics is not as constant as one might imagine. To drive a mud pump hundreds of meters deep, a high-power frequency converter outputs a pulse-width modulation (PWM) voltage with varying frequency. The harmonic interference generated by this voltage on the power line is not only high in energy, but its frequency also changes continuously with the motor speed. For example, a nominal 250Hz 5th harmonic may actually drift between 248Hz and 252Hz due to variations in the pump's load.

[0051] A notch filter with a fixed center frequency of 250Hz will significantly reduce its ability to suppress actual interference signals at 248.5Hz. The sideband energy of the interference signal bypasses the filter and enters the demodulator, increasing the noise floor and leading to a higher data error rate. Blindly widening the blocking bandwidth of the notch filter to cover frequency drift will filter out even more useful signal components, causing additional loss in signal-to-noise ratio.

[0052] In summary, the traditional combination of fixed threshold blanking and fixed frequency notch filtering is insufficient to guarantee reliable transmission of underwater measurement data in environments with strong electromagnetic interference, such as ultra-deep cutoff walls, which are subject to time-varying impulse noise and frequency drift harmonic interference. Therefore, a new method that can adaptively sense noise characteristics and dynamically adjust processing strategies is urgently needed.

[0053] In view of this, the present invention provides a novel solution: a data transmission method for underwater measurement systems based on power line carrier communication. The technical concept of this invention lies in abandoning the traditional passive approach of fixed thresholds and fixed filtering. First, in the time domain, by analyzing the local statistics of the received signal, an adaptive blanking threshold linked to the probability of impulse noise occurrence is dynamically constructed to achieve precise removal of transient strong interference. Subsequently, in the residual signal, an adaptive notch filter based on the normalized least mean square algorithm is applied to the power frequency harmonics, using orthogonal reference signals to track and eliminate narrowband interference caused by frequency drift in real time. That is, a coarse screening is performed in the time domain, followed by a finer refinement in the time domain. The sparsity of impulse noise and the quasi-periodicity of power frequency harmonics are utilized to ensure the robustness and self-consistency of the method under extreme operating conditions.

[0054] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0055] like Figure 1 As shown in the figure, this embodiment provides a data transmission method for an underwater measurement system based on power line carrier communication, comprising the following steps:

[0056] Step S1: Receive the orthogonal frequency division multiplexing carrier signal coupled from the power line, and sample the carrier signal to obtain a discrete received signal;

[0057] Step S2: In the time domain, calculate the local energy estimate and impulse noise probability estimate based on the local statistics of the discrete received signal;

[0058] Step S3: Based on local energy estimation and impulse noise occurrence probability estimation, an adaptive time-varying blanking threshold is constructed using a preset nonlinear mapping function, so that the threshold automatically increases when the impulse noise occurrence probability increases and automatically decreases when it decreases.

[0059] Step S4: Set the sampling points in the discrete received signal whose amplitude exceeds the adaptive time-varying blanking threshold to zero to obtain the blanked signal;

[0060] Step S5: Using the blanking-processed signal as the input signal, the power frequency harmonic interference is canceled by an adaptive notch filter based on the normalized least mean square algorithm to obtain a purified signal; wherein, the reference signal of the adaptive notch filter is composed of orthogonal sine components corresponding to the frequencies of each power frequency harmonic.

[0061] Step S6: Perform orthogonal frequency division multiplexing demodulation on the purified signal to recover the underwater measurement data.

[0062] Through this implementation, firstly, the method of the present invention can achieve adaptive suppression of time-varying impulse noise. Specifically, step S2 of the present invention extracts two key local statistics from the received signal in the time domain: local energy estimation and impulse noise occurrence probability estimation. These two statistics can reflect the intensity benchmark of the current channel noise environment and the sparsity of impulse interference in real time. Simultaneously, step S3 uses the above two statistics as input variables and dynamically calculates the adaptive time-varying blanking threshold through a preset nonlinear mapping function. Thus, when the probability of impulse noise increases due to frequent startup of high-power equipment at the construction site, the threshold automatically increases to prevent dense, medium-amplitude impulses from being misjudged as valid signals and missed (to avoid undersuppression); when construction enters a stable period and the impulse probability decreases, the threshold automatically decreases to avoid erroneously zeroing the sampling points of normal signals (to avoid oversuppression). Combined with step S4, the entire receiving preprocessing process can autonomously and dynamically find the optimal balance between maximizing the elimination of impulse interference and minimizing damage to valid signals according to the real-time changes in the noise environment. This effectively avoids the oversuppression or undersuppression problems caused by fixed thresholds in existing related technologies.

[0063] Secondly, the method of this invention can continuously cancel frequency-drifting power frequency harmonic interference without pre-setting precise values ​​for the interference frequency. Specifically, step S5 of this invention uses an adaptive notch filter based on the normalized least mean square algorithm. This differs from traditional fixed-frequency notch filters. The normalized least mean square algorithm is a closed-loop error-driven iterative algorithm, and its reference signal is limited to orthogonal sinusoidal components corresponding to each power frequency harmonic frequency. This set of orthogonal reference signals can provide the adaptive filter with an internal reference synchronized with the power frequency fundamental frequency and its harmonics. Thus, when the frequency of the actual harmonic interference drifts slightly due to fluctuations in the power grid load, the adaptive notch filter will automatically adjust its internal weighting coefficients to track this frequency change by continuously comparing the residual error between the input signal and the "weighted sum of reference signals," ensuring that the notch notch is always aligned with the true interference frequency. Based on this, the method of this invention does not require the fixed and precise prior assumptions about the power frequency and its harmonic frequencies as in traditional methods, and can still achieve robust suppression of non-stationary harmonic interference.

[0064] Third, the method of this invention can effectively guarantee the convergence stability of the adaptive filter. Specifically, in this invention, step S4 is executed first, followed by step S5; that is, pulse blanking is performed first, followed by harmonic cancellation. This allows the adaptive mechanism constructed in steps S2 and S3 to remove the most prominent pulse noise in the received signal in the time domain, leaving only relatively stable background noise and amplified residual interference in the signal fed into the subsequent adaptive notch filter. This provides a relatively clean convergence environment for the normalized least mean square algorithm, ensuring the robustness of the adaptive filter weight coefficient update and avoiding the risk of algorithm divergence. Finally, step S6 receives the purified signal that has undergone the above two-stage preprocessing and has had its main interference energy removed. This allows the subsequent orthogonal frequency division multiplexing demodulation and decoding modules to operate under high signal-to-noise ratio conditions, ultimately achieving the goal of recovering reliable underwater measurement data.

[0065] Understandably, if the execution order of steps S4 and S5 is reversed, the strong impulse noise, due to its suddenness and high energy, will severely impact the weight coefficient update process of the normalized least mean square algorithm, which is very likely to cause the algorithm to diverge and make the adaptive notch filter completely fail.

[0066] In one embodiment of the present invention, in step S1, the received carrier signal is composed of a useful orthogonal frequency division multiplexing signal, strong impulse noise, power frequency harmonic interference and colored background noise superimposed, and the strong impulse noise follows a Middleton Class A statistical model; the carrier signal is converted from analog to digital at the sampling frequency to obtain a discrete received signal, and the index range of the discrete received signal is from the first sampling point to the last sampling point within an orthogonal frequency division multiplexing symbol period.

[0067] For example, the received carrier signal It can be represented as:

[0068]

[0069] In the formula, For useful orthogonal frequency division multiplexed signals, This represents strong impulse noise, which follows a Middleton type A statistical model. This indicates power frequency harmonic interference. Indicates colored background noise;

[0070] right With sampling frequency Perform analog-to-digital conversion to obtain discrete received signals. ,in, For the sampling point index, and , This represents the number of sampling points within a single orthogonal frequency division multiplexing symbol period.

[0071] In this embodiment... Defined as a linear superposition of four components, interference in the received signal can be physically classified into two independent components with different statistical characteristics: strong impulse noise (i.e., According to the Middleton Class A model, it exhibits a sparse distribution in the time domain, with amplitudes significantly higher than background noise. It is precisely because of its sparseness and unusually high amplitude that it can be identified and eliminated through an amplitude threshold. This is in contrast to power frequency harmonic interference (i.e.,...). As a product of nonlinear loads in the power grid, it manifests in the frequency domain as narrowband energy concentrated at integer multiples of the power frequency, overlapping with broadband orthogonal frequency division multiplexing (OFDM) signals in the spectrum. Because of its narrow band and known or estimable frequency, it can be canceled by constructing orthogonal reference signals at the corresponding frequencies. In this way, various noises coupled and superimposed in the actual physical environment can be modeled at the receiver as identifiable and separable independent additive components, facilitating subsequent processing.

[0072] In one embodiment of the present invention, in step S2, the local energy is estimated as the root mean square value of the discrete received signal within an orthogonal frequency division multiplexing symbol frame. This root mean square value is calculated by taking the square root of the average of the sum of the squares of the amplitudes of each sampling point within the frame.

[0073] For example, the local energy estimate is the root mean square value of the discrete received signal within one orthogonal frequency division multiplexing symbol frame. Its calculation formula can be expressed as:

[0074]

[0075] In the formula, Indicates the first Sample values ​​of a discrete received signal.

[0076] In this implementation, for this formula, firstly, the squaring operation converts the signal's amplitude value into an equivalent quantity of its instantaneous power. Since power is proportional to energy, averaging the power means that the metric directly reflects the total energy level carried by the signal within a symbol period, rather than merely reflecting the surface characteristics of amplitude fluctuations. Then, the square root of the average power is taken, resulting in a calculated value... Regression to the original sampled value The same amplitude units. Therefore, if subsequent steps require... A threshold is set as the benchmark for comparison with the sampled value amplitude. Since both values ​​are within the same dimensional system, the proportional relationship can be directly constructed and the amplitude compared without unit conversion, reducing the complexity of system implementation. Finally, the root mean square (RMS) value is calculated based on the sum of squares of all sample points within the entire frame, making it a holistic statistic. Compared to instantaneous absolute values ​​or peak values, the RMS value does not fluctuate drastically due to a few outliers, and can stably characterize the fundamental energy level of the signal in that frame. This provides a reliable reference baseline for subsequently using this value as the basis for adaptive thresholding.

[0077] It is understandable that an orthogonal frequency division multiplexing (OFDM) symbol is the smallest unit of time for data transmission. By strictly aligning the energy estimation window to the duration of each symbol, it is possible to achieve... This reflects precisely the energy level resulting from the superposition of channel noise and signal during the symbol transmission, rather than a mixed statistical value across symbol boundaries. This ensures that when sudden changes in construction conditions cause noise energy to fluctuate drastically over several symbol periods, the calculated values ​​are accurate frame-by-frame. It can keep up with this change in a timely manner and provide an energy benchmark that accurately reflects the current symbol transmission environment for subsequent processing. If a sliding window across symbols or an energy accumulation method without fixed boundaries is used, it may introduce the inertial tail of historical noise information, which will weaken its sensitivity to instantaneous operating conditions.

[0078] In one embodiment of the present invention, in step S2, the probability estimation of impulse noise is determined by: judging the amplitude of each sampling point in the frame, counting the number of sampling points whose amplitude exceeds three times the root mean square value, and dividing the number by the total number of sampling points in the frame, the resulting ratio is a dimensionless value between 0 and 1.

[0079] For example, impulse noise occurrence probability estimation It can be determined by the following formula:

[0080]

[0081] In the formula, This is an indicator function that takes the value 1 when the condition inside the parentheses is true, and 0 otherwise; It is a dimensionless proportional value, and its range is [value missing]. .

[0082] In this embodiment, the discrimination criterion for this formula is: In other words, it is the result of the aforementioned calculations. It serves as a dynamic benchmark, rather than using a preset absolute amplitude value as the decision threshold. Simultaneously, it will... Defined as the proportion of samples identified as impulse noise pollution out of N sampling points in the entire OFDM symbol frame, this allows the "occurrence density" of impulse noise to be quantified into a continuous scalar index (impulse noise at a construction site manifests as a series of discrete interference events in the time domain. By counting these discrete events within a frame and dividing by the total number of samples, it can be transformed into a continuous scalar with values ​​between [0,1]. This allows the severity of impulse noise to be elevated from a qualitative concept of scattered events to a quantitative parameter that can be directly used by subsequent modules).

[0083] In this implementation, the reason for setting a factor of 3 is that, for Gaussian background noise and typical orthogonal frequency division multiplexing (OFDM) signals, the probability of their instantaneous amplitude exceeding three times the root mean square (RMS) value is extremely low. Therefore, when the amplitude of a sampling point exceeds... At this point, from a statistical perspective, it can be determined with high confidence that the sampling point is not a product of the superposition of normal signal and background noise, but rather contaminated by external impulse noise. Thus, it is possible to identify impulse-contaminated samples based solely on the statistical characteristics of the received signal, without needing to know the absolute amplitude range of the impulse noise beforehand. Of course, it is understandable that in practical implementation... The coefficient multiple can be selected and adjusted according to the actual situation, and the present invention does not impose specific limitations on it.

[0084] Furthermore, due to It is calculated frame by frame and updated in real time, therefore the judgment criterion It also fluctuates dynamically. When the overall construction noise increases, leading to an increase in the total power of the received signal, When the threshold is increased synchronously, the reference value is automatically raised; conversely, it is automatically lowered. In this way, the threshold for judging impulse noise always maintains a reasonable proportional relationship with the basic energy level of the current channel environment, and will not cause systematic misjudgment due to the overall fluctuation of background noise.

[0085] In one embodiment of the present invention, in step S3 of the present invention, the adaptive time-varying blanking threshold is calculated as follows: the basic threshold coefficient is multiplied by the root mean square value, and then multiplied by an adjustment factor consisting of a weighted sum of the adaptive adjustment depth coefficient and the arctangent function value; wherein, the input of the arctangent function is the probability sensitivity coefficient multiplied by the impulse noise occurrence probability estimate and then divided by the probability normalization benchmark value.

[0086] For example, the constructed adaptive time-varying hidden line elimination threshold It can be calculated using the following formula:

[0087]

[0088] in, Based on the threshold coefficient, To adaptively adjust the depth coefficient, It is the arctangent function. This is the probability sensitivity coefficient. This is the baseline value for probability normalization.

[0089] In this implementation, for this formula, firstly, the basic terms... A benchmark for the ratio of the threshold to local energy can be established to ensure the essential amplitude attribute of the threshold. Specifically, with As the base value for threshold construction, the threshold It is anchored to the current baseline energy level of the received signal, so that when background noise or changes in the construction electromagnetic environment cause a gradual change in the overall amplitude of the received signal, This will scale synchronously, ensuring that the decision benchmark and the signal environment maintain a reasonable proportional relationship. Simultaneously, under conditions where there is no impulse noise or its occurrence probability is extremely low, the amplitude of the superimposed normal signal and background noise is mainly distributed within... Within the range of the root mean square value. Multiply This allows us to construct a normalized safety range upper limit that is higher than the base energy level; anything exceeding this upper limit is preliminarily identified as a pulse anomaly.

[0090] Secondly, the adjustment factor term The pulse generation density can be nonlinearly mapped to a dynamic rise in the threshold; this is the core of achieving time-varying characteristics. Specifically, for its input variables... , In other words, It is the calculated probability estimate of impulse noise occurrence, obtained through... Adjust sensitivity, by After normalization, the product ( This allows the probability estimate to be transformed into a proportional quantity that fits the dynamic range of the arctangent function input. Simultaneously, for the arctangent function... In other words, when the probability of pulse occurrence At lower levels, The adjustment factor term can be approximated as At this point, the threshold rises moderately at an approximately linear rate with increasing pulse probability, enabling sensitive response to the initial change of pulse activity from zero to its occurrence. As the pulse occurrence probability continues to increase, the slope of the arctangent function gradually decreases and eventually approaches zero, with the function value approaching the asymptote π / 2. Thus, when construction enters its most severe condition and pulses are extremely dense, the threshold will not increase with... It does not expand infinitely with increasing size, but rather converges to a certain value. This avoids the risk of excessive blanking caused by raising the threshold indefinitely, which could lead to a large number of normal signal samples being incorrectly zeroed out, and ensures that the system can still maintain basic signal reception capability in extreme noise environments.

[0091] In addition, coefficient Located before the output of the arctangent function, it is used to control the overall dynamic range of the adjustment factor. The larger the value, the greater the rise in the threshold between the two extreme states of no pulse and full pulse. Used to set the baseline height. The two controls are independently adjustable to control the fluctuation range, allowing for flexible adaptation to the dynamic range of noise in different construction scenarios.

[0092] Overall, this implementation multiplies the base term by the adjustment factor to obtain the final result. It is determined by the current signal energy level and current pulse generation density The jointly driven adaptive threshold allows the threshold to scale synchronously with the overall fluctuations of the background energy (through...). Furthermore, it can independently perform limited secondary adjustments (through adjustment factors) above the base level based on the sparsity of the impulse interference itself. These two adaptive mechanisms operate independently and simultaneously, enabling the threshold to possess a more refined and stable adaptive adjustment capability than single-variable control strategies when facing complex and multi-dimensional noise environment evolution caused by changes in construction procedures.

[0093] In one embodiment of the present invention, in step S4, sampling points in the discrete received signal whose amplitude exceeds the adaptive time-varying blanking threshold are set to zero. Specifically, for each sampling point, if its amplitude is less than or equal to the adaptive time-varying blanking threshold, the value of the sampling point remains unchanged; if its amplitude is greater than the adaptive time-varying blanking threshold, the value of the sampling point is set to zero; thereby obtaining the blanking processed signal.

[0094] For example, setting the sampling points in the discrete received signal whose amplitude exceeds the adaptive time-varying blanking threshold to zero can be done as follows: for each sampling point Signal after blanking It is given by the following formula:

[0095]

[0096] In the formula, The amplitude of the discrete received signal.

[0097] In this implementation, the formula is a point-by-point binary determination and processing process; that is, for each sampling point within a frame, the system calculates its instantaneous amplitude. With the construction of adaptive threshold A one-to-one comparison is performed, and either the sample is retained or zeroed out based on the comparison result. In other words, the smallest processing unit for blanking is a single sample point, not a subcarrier group or a symbol block. This ensures that even if impulse noise contaminates only a few sample points within an orthogonal frequency division multiplexing (OFDM) symbol, the system can accurately identify and remove them individually without affecting other uncontaminated sample points within the same symbol. Simultaneously, sample points with amplitudes not exceeding a threshold are retained in their original value. This ensures that the useful signal and background noise information carried by uncontaminated sample points are fully transmitted to subsequent processing stages, avoiding additional signal distortion introduced by the blanking operation. Sample points identified as contaminated by impulse noise are assigned a value of zero, effectively erasing them from the valid data. This prevents the sample from contributing any energy to subsequent frequency domain transformations or adaptive filter weight updates, thus ensuring that the adaptive notch filter is not interfered with by these transient high-energy samples (impulse noise-contaminated sample points) during iteration, effectively reducing the risk of algorithm divergence.

[0098] In one embodiment of the present invention, in step S5, the reference signal of the adaptive notch filter is composed of orthogonal sinusoidal components corresponding to the frequencies of each power frequency harmonic. Specifically, it includes: for each harmonic from the fundamental wave to the highest harmonic order to be suppressed, constructing a cosine reference signal and a sine reference signal respectively, wherein the frequencies of the cosine reference signal and the sine reference signal are the frequencies of the harmonic order, and their phases differ by ninety degrees.

[0099] For example, the reference signal of the adaptive notch filter is composed of orthogonal sinusoidal components corresponding to the frequencies of each power frequency harmonic, specifically including:

[0100] For the Second harmonic , Construct a cosine reference signal for the highest harmonic order that needs to be suppressed. and sinusoidal reference signal They are respectively:

[0101]

[0102] In the formula, The fundamental frequency of the power grid. Sampling frequency, This is the index for the sampling points.

[0103] In this embodiment, a cosine reference signal is constructed. and a sine reference signal The two have the same frequency but different phases. The two orthogonal reference signals form a pair of orthogonal bases. Thus, when actual power frequency harmonic interference reaches the receiver, its phase is unknown and random (depending on various factors such as grid load characteristics and transmission line impedance distribution). If only a single-phase reference signal is constructed, the adaptive filter cannot completely cancel the interference by adjusting a single weight coefficient when the phase of the actual interference is not orthogonal to the reference signal. However, the two orthogonal reference signals can form a complete set of bases (according to the principle of linear superposition of trigonometric functions, any sine wave with the same frequency, amplitude, and phase as the reference signal can be uniquely decomposed into a weighted sum of a pair of cosine and sine components). Based on this, by adjusting the cosine and sine weight coefficients respectively, the adaptive notch filter can synthesize a cancellation signal that is completely in phase, frequency, and amplitude of the interference, thereby achieving accurate reconstruction and complete cancellation of any initial-phase harmonic interference.

[0104] Meanwhile, the frequency of the reference signal is determined by parameters. (Power grid fundamental frequency) and The sampling frequency is determined and directly generated through calculation (not extracted through a phase-locked loop or frequency estimator). This effectively ensures the stability of the frequency reference (the fundamental frequency of the power grid). These are inherent parameters of the power system; although they exhibit minor fluctuations, they are statistically highly stable. The nominal value, for example, 50Hz, is used as the frequency reference to construct the reference signal. This ensures that the frequency pointed to by the reference signal is always near the theoretical position where the power frequency harmonics appear, providing a reliable initial frequency anchor point for the adaptive notch filter. Moreover, it can effectively avoid frequency pulling (in a strong noise environment, if the interference frequency is estimated in real time from the received signal, the presence of strong impulse noise and broadband signals can easily cause the frequency estimator to produce a large estimation deviation, which in turn causes the reference frequency of the notch filter to deviate from the real interference frequency. However, this implementation adopts an open-loop method that directly calculates based on the nominal frequency and the sampling frequency. The frequency of its reference signal is predetermined and is not affected by the instantaneous state of the received signal. Therefore, it can effectively avoid the frequency pulling problem caused by noise and ensure that the reference signal remains stable even in extreme interference environments).

[0105] In addition, the construction traversal of reference information That is, for each integer multiple of frequency from the fundamental frequency to the Kth harmonic, an independent pair of orthogonal reference signals is constructed. In this way, each pair of orthogonal reference signals corresponds to only one specific harmonic frequency. The adaptive notch filter selectively cancels the frequency component using its own independent weighting coefficients. This one-to-one correspondence ensures that the suppression processes of each harmonic are independent and do not interfere with each other. Compared to a scheme that uses a broadband filter to uniformly filter out all harmonics, this implementation can achieve precise linear suppression of power frequency harmonics without affecting the useful signal frequency band between harmonic frequencies, thus preserving the spectral integrity of the useful orthogonal frequency division multiplexed signal to the greatest extent.

[0106] In one embodiment of the present invention, step S5 of the present invention further includes the following cancellation processing of the adaptive notch filter based on the normalized least mean square algorithm:

[0107] Using the blanking-processed signal as the desired input, the difference between the blanking-processed signal and the sum of the products of the cosine weights and cosine reference signals corresponding to all harmonics, and the products of the sine weights and sine reference signals, is calculated as the error signal.

[0108] The weight coefficients of the next sampling point are updated using the normalized least mean square algorithm. The update amount of each weight coefficient is the step size factor divided by the sum of the norm square of the reference signal vector corresponding to the harmonic and the regularization constant, and then multiplied by the error signal and the corresponding reference signal value.

[0109] After iterating through all sampling points within the frame, the final error signal sequence is output as the purified signal.

[0110] For example, the signal after blanking processing can be... As the desired input, calculate the error signal. for:

[0111]

[0112] In the formula, and For the current sampling point corresponding to the first Cosine and sine weights of subharmonics;

[0113] The weight coefficients of the next sampling point are updated using the normalized least mean square algorithm. The update formula is as follows:

[0114]

[0115] In the formula, and For the updated cosine and sine weights, It is the step size factor, and , Indicates by and The vector formed , ;

[0116] After iterating through all sampling points within the frame, the final error signal sequence is used as the purified signal. Output.

[0117] In this embodiment, for the error signal In terms of the calculation formula, the summation term ( Using the weighting coefficients of each harmonic, the two orthogonal reference signals are weighted and summed to mathematically provide a real-time estimate of all power frequency harmonic interference components in the received signal. The blanking-processed signal... Subtracting this estimate allows for the complete "stripping" of all estimated harmonic interference components from the received signal, leaving only the error signal. It retains useful orthogonal frequency division multiplexed signals, residual interference that has not been completely canceled, and background noise.

[0118] At the same time, (instead of the original received signal) As the desired input, after pulse blanking processing In the middle, high-energy pulse spikes have been removed. As the input to an adaptive notch filter, it can avoid the influence of impulse noise on the error signal. This generates transient shocks, which in turn ensures the stability of the subsequent weight coefficient update process.

[0119] For the update formula, using cosine weights For example, the update item ( This is the product of the error signal and the corresponding reference signal. According to the basic principle of the least mean square algorithm, this product is statistically proportional to the negative gradient direction of the mean square error with respect to the weight coefficients. Adding this term to the current weight coefficients is equivalent to iteratively updating the weight coefficients along the steepest descent direction that reduces the mean square error. The absolute value of the update term reflects the degree of deviation between the current weight coefficients and their optimal solution (the greater the deviation, the lower the error). Residual in The stronger the correlation, the larger the update step size, accelerating the approach to the optimal solution. (Normalized denominator term) The instantaneous power of the reference signal vector is denoted as . In the general normalized least mean square algorithm, dividing by the input signal power eliminates the influence of input signal amplitude fluctuations on the convergence speed. In the specific application scenario of this invention, since the constructed two orthogonal reference signals satisfy , Therefore, it is always true. It is always equal to 1. Thus, the normalization of the denominator is equivalent to removing fluctuations in the amplitude of the reference signal (which is itself a constant value in this scheme), making the effective step size solely determined by... This approach effectively simplifies the analysis of convergence behavior and avoids the transient instability problem caused by sudden changes in input signal power in the general normalized least mean square algorithm (because the normalization denominator in this implementation is constant).

[0120] Furthermore, the weighting coefficients are updated at each sampling point, rather than in batches at the end of the frame. This allows the adaptive notch filter to track amplitude variations and frequency drifts of power frequency harmonic interference on a sample-by-sample basis. When the characteristics of the harmonic interference change slowly within an orthogonal frequency division multiplexing (OFDM) symbol period, the weighting coefficients can follow this change in real time, maintaining effective cancellation. The final purified signal is directly derived from the error signal at each sampling point. Instead of subtracting the estimated harmonic interference from the original signal, the error signal at each moment in the point-by-point iteration process is itself the result of "the current input signal minus the current optimal harmonic estimate", thus achieving the cancellation purpose. Using the error signal sequence as the output avoids additional post-processing steps and ensures the immediacy of the output (once the sampling point is processed, it can enter the subsequent orthogonal frequency division multiplexing demodulation process without introducing frame-level processing delay).

[0121] In one embodiment of the present invention, step S6 of the present invention, performing orthogonal frequency division multiplexing demodulation on the purified signal includes: after removing the cyclic prefix, performing a fast Fourier transform on the purified signal to convert the time-domain signal into a frequency-domain signal, wherein the value at each frequency point is obtained by multiplying each sampling point in the time domain with the complex exponential basis function of the corresponding frequency and then accumulating the results; then performing channel equalization, demapping and decoding on the frequency-domain signal to recover the underwater measurement data.

[0122] For example, for purified signals Orthogonal frequency division multiplexing demodulation can include:

[0123] After removing the cyclic prefix, a fast Fourier transform is performed to obtain the frequency domain signal. for:

[0124]

[0125] In the formula, For the number of points in the Fast Fourier Transform, Subcarrier sequence number, The imaginary unit, It is an exponential function;

[0126] After that Channel equalization, demapping, and decoding are performed to recover underwater measurement data.

[0127] In this implementation, the core principle of this formula lies in its ability to convert time-domain discrete sequences... With a set of complex exponential basis functions The inner product operation is performed to extract the complex amplitude of the signal at each subcarrier frequency. The complex exponential basis function is used. For different subcarrier sequence numbers In length of They are mutually orthogonal within the interval. This orthogonality is the theoretical basis of orthogonal frequency division multiplexing—when The useful signal components are synthesized by the transmitter through inverse Fourier transform. When the subcarriers are strictly orthogonal, the receiver performs a forward transform using this formula, which theoretically allows for crosstalk-free synthesis of the independent modulation symbols carried on each subcarrier. They were separated one by one.

[0128] In this implementation method, firstly... The input to this formula is the purified signal after two stages of processing: pulse blanking in step S4 and power frequency harmonic cancellation in step S5. Since the high-energy spikes caused by pulse noise have been removed and the narrowband interference caused by power frequency harmonics has been cancelled, the signal is then fed into the Fast Fourier Transform. The interference energy contained within has been significantly reduced. Based on this, the linear transformation achieved by this formula can diffuse the energy of residual interference to each subcarrier, preventing it from causing concentrated damage to specific subcarriers, thus ensuring that the demodulated frequency domain symbols on each subcarrier are more efficient. The signal-to-interference-plus-noise ratio was improved overall.

[0129] Secondly, because the cyclic prefix (a guard interval added by the transmitter before each orthogonal frequency division multiplexing symbol, usually copied from the end of the symbol) is removed before performing the Fast Fourier Transform, impedance discontinuities and branch lines in ultra-deep underwater power cable channels can cause reflections and multipath propagation effects. The reflected wave from the previous symbol may arrive late and overlap with the current symbol, causing inter-symbol interference. The cyclic prefix provides a guard interval greater than the maximum multipath delay. When removing the cyclic prefix, the receiver discards these samples contaminated by the delayed wave from the previous symbol, ensuring that the remaining valid symbol samples no longer contain trailing interference from the previous symbol, thus guaranteeing the independence of the current symbol. Simultaneously, after removing the cyclic prefix, the remaining... In a sample, the multipath effect only causes the signal to exhibit a product of phase and amplitude changes on each subcarrier. In this case, by using the Fast Fourier Transform, the linear convolutional channel model can be equivalent to a single-tap multiplicative channel on each subcarrier, effectively avoiding the loss of orthogonality between subcarriers caused by multipath.

[0130] In one embodiment of the present invention, the method is applied to the construction environment of ultra-deep cutoff walls, with a construction depth exceeding 200 meters; and the basic threshold coefficient used in step S3 when constructing the adaptive time-varying blanking threshold is... The value ranges from 2.5 to 4.0, and the step size factor used in the normalized least mean square algorithm in step S5 is... The value range is from 0.01 to 0.5.

[0131] In this embodiment, it should first be noted that in ultra-deep construction environments exceeding 200 meters, the power line carrier communication link between the underwater measuring equipment and the ground receiving end has significantly different physical characteristics compared to conventional shallow construction. Specifically, the substantial increase in power cable length leads to increased signal transmission attenuation and more complex line impedance characteristics; simultaneously, ultra-deep construction inevitably utilizes higher-power trenching equipment, resulting in electromagnetic interference intensity and complexity far exceeding that of ordinary construction. Fixed-threshold blanking and fixed-frequency notch filtering schemes, which may be barely usable in conventional deep construction, completely fail in ultra-deep environments exceeding 200 meters due to the contradiction between noise intensity and communication distance. The adaptive two-level processing method proposed in this invention can effectively adapt to ultra-deep construction environments exceeding 200 meters.

[0132] Secondly, regarding the basic threshold coefficient In other words, it acts on the basic terms of the adaptive hidden surface elimination threshold. This determines the threshold relative to the root mean square (RMS) value of the signal under pulse-free or low-pulse probability conditions. The lower limit is set at 2.5 to allow sufficient margin for natural amplitude fluctuations after the superposition of the normal signal and background noise (for the superposition of an orthogonal frequency division multiplexed signal following a typical distribution with background noise, the instantaneous amplitude typically does not exceed 2.5 times the RMS value for more than 95% of the time). This can keep the probability of normal signal samples being misidentified as impulse noise at a low level. At the same time, limiting its upper limit to 4.0 is to avoid a decrease in the detection sensitivity for medium-amplitude impulse noise due to an excessively high threshold.

[0133] For step size factor Specifically, it is used for updating the weight coefficients in the normalized least mean square algorithm. In the normalized least mean square algorithm, The value of directly determines the trade-off between convergence speed and steady-state error. The larger the value, the larger the single-step correction, and the faster the weight coefficients approach the optimal solution, but the more drastic the steady-state fluctuations after convergence. The smaller the value, the smoother the convergence process and the smaller the steady-state error, but the slower the response to the time-varying characteristics of harmonic interference. The lower limit is set at 0.01, which ensures that the adaptive notch filter has an acceptable convergence speed in the slow time-varying electromagnetic environment such as the construction of ultra-deep cutoff walls, and will not fail to complete effective cancellation within the finite orthogonal frequency division multiplexing symbol period due to an excessively small step size. Setting the upper limit to 0.5 avoids overshoot and oscillation during convergence due to excessively large step sizes, prevents the weight coefficients from fluctuating around the optimal solution and failing to stabilize, and ensures a purified output signal. Stability.

[0134] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A data transmission method for an underwater measurement system based on power line carrier communication, characterized in that, Includes the following steps: Step S1: Receive the orthogonal frequency division multiplexing carrier signal coupled from the power line, and sample the carrier signal to obtain a discrete received signal; Step S2: In the time domain, calculate the local energy estimate and the impulse noise occurrence probability estimate based on the local statistics of the discrete received signal; Step S3: Based on the local energy estimation and the impulse noise occurrence probability estimation, an adaptive time-varying blanking threshold is constructed using a preset nonlinear mapping function, so that the threshold automatically increases when the impulse noise occurrence probability increases and automatically decreases when it decreases. Step S4: Set the sampling points in the discrete received signal whose amplitude exceeds the adaptive time-varying blanking threshold to zero to obtain the blanked signal; Step S5: Using the blanking-processed signal as the input signal, the power frequency harmonic interference is canceled by an adaptive notch filter based on the normalized least mean square algorithm to obtain a purified signal; wherein, the reference signal of the adaptive notch filter is composed of orthogonal sine components corresponding to the frequencies of each power frequency harmonic. Step S6: Perform orthogonal frequency division multiplexing demodulation on the purified signal to recover the underwater measurement data.

2. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 1, characterized in that, In step S1, the received carrier signal is composed of a useful orthogonal frequency division multiplexing signal, strong impulse noise, power frequency harmonic interference, and colored background noise, and the strong impulse noise follows a Middleton Class A statistical model; the carrier signal is converted from analog to digital at the sampling frequency to obtain the discrete received signal, and the index range of the discrete received signal is from the first sampling point to the last sampling point within an orthogonal frequency division multiplexing symbol period.

3. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 2, characterized in that, In step S2, the local energy estimate is the root mean square value of the discrete received signal within an orthogonal frequency division multiplexing symbol frame. This root mean square value is calculated by taking the square root of the average of the sum of the squares of the amplitudes of each sampling point within the frame.

4. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 3, characterized in that, In step S2, the probability estimate of the impulse noise occurrence is determined as follows: The amplitude of each sampling point within the frame is determined, and the number of sampling points whose amplitude exceeds three times the root mean square value is counted. This number is then divided by the total number of sampling points in the frame, and the resulting ratio is a dimensionless value between 0 and 1.

5. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 4, characterized in that, In step S3, the adaptive time-varying blanking threshold is calculated as follows: The base threshold coefficient is multiplied by the root mean square value, and then multiplied by an adjustment factor consisting of a weighted sum of the adaptive adjustment depth coefficient and the arctangent function value; wherein the input of the arctangent function is the probability sensitivity coefficient multiplied by the impulse noise occurrence probability estimate and then divided by the probability normalization benchmark value.

6. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 5, characterized in that, In step S4, setting the sampling points in the discrete received signal whose amplitude exceeds the adaptive time-varying blanking threshold to zero specifically involves: For each sampling point, if its amplitude is less than or equal to the adaptive time-varying blanking threshold, the value of the sampling point remains unchanged; if its amplitude is greater than the adaptive time-varying blanking threshold, the value of the sampling point is set to zero; thus, the blanking-processed signal is obtained.

7. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 1, characterized in that, In step S5, the reference signal of the adaptive notch filter is composed of orthogonal sinusoidal components corresponding to the frequencies of each power frequency harmonic, specifically including: For each harmonic from the fundamental frequency to the highest harmonic order that needs to be suppressed, a cosine reference signal and a sine reference signal are constructed respectively. The frequencies of the cosine reference signal and the sine reference signal are the frequencies of that harmonic order, and their phases differ by ninety degrees.

8. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 7, characterized in that, In step S5, the cancellation process of the adaptive notch filter based on the normalized least mean square algorithm further includes: Using the blanking-processed signal as the desired input, the difference between the blanking-processed signal and the sum of the products of the cosine weights and cosine reference signals corresponding to all harmonics, and the products of the sine weights and sine reference signals, is calculated as the error signal. The weight coefficients of the next sampling point are updated using the normalized least mean square algorithm. The update amount of each weight coefficient is the step size factor divided by the sum of the norm square of the reference signal vector corresponding to the harmonic and the regularization constant, and then multiplied by the error signal and the corresponding reference signal value. After iterating through all sampling points within the frame, the final error signal sequence is output as the purified signal.

9. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 1, characterized in that, In step S6, the orthogonal frequency division multiplexing demodulation of the purified signal includes: After removing the cyclic prefix, the purified signal is subjected to a fast Fourier transform to convert the time-domain signal into a frequency-domain signal. The value at each frequency point is obtained by multiplying each sampling point in the time domain with the complex exponential basis function of the corresponding frequency and then summing the results. Then, the frequency-domain signal is subjected to channel equalization, demapping and decoding to recover the underwater measurement data.

10. The data transmission method for an underwater measurement system based on power line carrier communication according to claim 1, characterized in that, The method is applied to the construction environment of ultra-deep cutoff walls with a construction depth of more than 200 meters; and the value range of the basic threshold coefficient used in step S3 to construct the adaptive time-varying blanking threshold is 2.5 to 4.0, and the value range of the step size factor used in the normalized least mean square algorithm in step S5 is 0.01 to 0.5.

Images