CN116776083B - Signal acquisition noise reduction method for multi-beam submarine topography measurement - Google Patents
Signal acquisition noise reduction method for multi-beam submarine topography measurement Download PDFInfo
- Publication number
- CN116776083B CN116776083B CN202310723169.9A CN202310723169A CN116776083B CN 116776083 B CN116776083 B CN 116776083B CN 202310723169 A CN202310723169 A CN 202310723169A CN 116776083 B CN116776083 B CN 116776083B
- Authority
- CN
- China
- Prior art keywords
- column
- detection signals
- signal
- detection
- curve
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 75
- 238000005259 measurement Methods 0.000 title claims abstract description 55
- 238000012876 topography Methods 0.000 title claims abstract description 44
- 230000009467 reduction Effects 0.000 title claims abstract description 15
- 238000001514 detection method Methods 0.000 claims abstract description 298
- 239000011159 matrix material Substances 0.000 claims abstract description 64
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 33
- 230000009466 transformation Effects 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims description 23
- 238000001228 spectrum Methods 0.000 claims description 8
- 230000011218 segmentation Effects 0.000 claims description 7
- 239000000523 sample Substances 0.000 claims description 6
- 230000003287 optical effect Effects 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 18
- 238000012545 processing Methods 0.000 abstract description 3
- 238000010276 construction Methods 0.000 description 14
- 238000010586 diagram Methods 0.000 description 8
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 230000000737 periodic effect Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Abstract
The invention relates to the technical field of data processing, and provides a signal acquisition noise reduction method for multi-beam submarine topography measurement, which comprises the following steps: collecting a detection matrix of multi-beam submarine topography measurement to obtain detection signals of each column; the method comprises the steps of carrying out Fourier transformation on a detection matrix to obtain high-frequency measuring points, and obtaining an effective point group and a prediction curve of each column of detection signals according to each column of detection signals, the high-frequency measuring points and a sine function; according to each column of detection signals, a local mean function and a prediction curve, an extension signal curve of each column of detection signals is obtained, EMD decomposition is carried out on the basis of the extension signal curve to obtain a plurality of IMF components, and the component removal quantity of each column of detection signals is obtained according to the IMF components and high-frequency measurement points to obtain a denoising curve and a denoising detection matrix of each column of detection signals; and according to the denoising detection matrix, completing multi-beam submarine topography measurement. The invention aims to solve the problem that signal denoising results are inaccurate due to signal distortion caused by end-point effect.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a signal acquisition noise reduction method for multi-beam submarine topography measurement.
Background
An automatic positioning method for the pile leg of the offshore construction platform based on a multi-beam sounding system (MBES) is a method for realizing the accurate positioning of the pile leg of the offshore construction platform by utilizing a multi-beam sounding technology; in offshore construction, accuracy of pile leg positioning is critical to stability and safety of a platform, and the multi-beam sounding system can provide high-resolution submarine topography data, so that high-precision pile leg positioning is facilitated, and however, the high-resolution submarine topography data has higher precision requirements on preprocessing of measurement signals.
In the process of carrying out high-precision measurement on the submarine topography through MBES equipment, a measurement signal of the device is interfered by submarine environment factors, such as water flow, ocean current and other environment factors, or other objects such as rocks, and the interference can cause random or periodic interference, so that the high-precision measurement calculation of the submarine topography is inaccurate, a noise signal with complex components is difficult to accurately remove by a conventional data denoising algorithm, and a certain information damage problem always exists; in the prior art, the signal components are decomposed through an EMD decomposition algorithm, and then component signals in noise concentration are smoothed, so that loss of effective information can be effectively avoided; however, due to the limitation of the offshore collection environment and equipment, insufficient setting of sampling frequency and sampling window parameters, an endpoint effect is often caused to the signal, the signal decomposition process is distorted at two ends, a good decomposition effect cannot be obtained, and the signal denoising result is inaccurate.
Disclosure of Invention
The invention provides a signal acquisition noise reduction method for multi-beam submarine topography measurement, which aims to solve the problem of inaccurate signal denoising results caused by signal distortion due to the existing end-point effect, and adopts the following technical scheme:
one embodiment of the present invention provides a signal acquisition noise reduction method for multi-beam seafloor topography measurement, the method comprising the steps of:
collecting a detection matrix of multi-beam submarine topography measurement, and obtaining each column of detection signals;
The method comprises the steps of carrying out Fourier transformation on a detection matrix to obtain a high-frequency information duty ratio and high-frequency measuring points of each column of detection signals, constructing a first objective function of each column of detection signals according to each column of detection signals and the high-frequency measuring points of each column of detection signals by combining a sine function, and obtaining an effective point group and a prediction curve of each column of detection signals by converging the first objective function;
According to each column of detection signals, a local mean function and a prediction curve, an extension signal curve of each column of detection signals is obtained, EMD decomposition is carried out on the basis of the extension signal curve to obtain a plurality of IMF components of each column of detection signals, a second objective function is constructed according to the IMF components and high-frequency measuring points, and a denoising curve and a denoising detection matrix of each column of detection signals are obtained by converging the second objective function;
And according to the denoising detection matrix, completing multi-beam submarine topography measurement.
Optionally, the specific method for acquiring the high-frequency information duty ratio and the high-frequency measurement point of each column of the detection signal includes:
Converting the detection matrix into a frequency spectrum through Fourier transformation, and carrying out OTSU (optical time series unit) Otsu (on-line per se) Otsu threshold segmentation through frequency information and amplitude information to obtain segmentation thresholds of high-frequency information and low-frequency information, so as to obtain the high-frequency information and the low-frequency information;
calculating the duty ratio of the high-frequency information in the frequency spectrum, and recording the duty ratio as the duty ratio of the high-frequency information;
And (3) carrying out inverse Fourier transform on the high-frequency information to a detection matrix, wherein the detection matrix only retains corresponding measuring points of the high-frequency information, and marking the high-frequency measuring points in each column of detection signals according to the positions of the corresponding measuring points of the high-frequency information in the detection matrix to obtain the high-frequency measuring points of each column of detection signals.
Optionally, the constructing the first objective function of each column of the detection signals includes the following specific methods:
taking any column of detection signals as target detection signals, wherein the expression of a first target function of the target detection signals is as follows:
Wherein L represents the abscissa length of the target detection signal, x i represents the abscissa value of the i-th measurement point in the target detection signal, F (x) represents the target detection signal, P (x) represents the local mean function of the target detection signal, sin ωx represents a sine function with amplitude of 1, circular frequency of ω and initial phase of 0, N H represents the number of high-frequency measurement points in the target detection signal, (sin ωx=0) ∈h represents the number of high-frequency measurement points at zero point of the sine function, G (sinωx＝0)∈H represents the number of high-frequency measurement points at zero point of the sine function, and|represents the absolute value;
A first objective function is constructed for each column of probe signals.
Optionally, the method for obtaining the valid point group and the prediction curve of each column of the detection signals includes the following specific steps:
Taking any column of detection signals as target detection signals, converging a first target function of the target detection signals to obtain the optimal circular frequency of the target detection signals, and taking all zero points of a sine function of the optimal circular frequency as an effective point group to obtain an effective point group of the target detection signals;
Fitting all the effective points in the effective point group by combining with corresponding data of the effective points in the target detection signal, extending the fitted curve at the end points of two sides, and marking the extended fitted curve as a predicted curve of the target detection signal;
and acquiring an effective point group and a prediction curve of each column of detection signals.
Optionally, the method for obtaining the optimal circle frequency of the target detection signal includes the following specific steps:
And iterating the circular frequency in the first objective function of the target detection signal by taking any column of detection signals as target detection signals, and when the output result of the first objective function is minimum in the iteration process, considering that the first objective function of the target detection signal is converged at the time when the circular frequency iterates, and recording the circular frequency at the time as the optimal circular frequency of the target detection signal.
Optionally, the method for obtaining the extension signal curve of each column of the detection signals includes the following specific steps:
Taking any column of detection signals as target detection signals, calculating the difference value of the target detection signals and the local mean function of the target detection signals on each data point, and recording the difference value as an error value of each data point;
Acquiring the mean square error of all error values, superposing the mean square error of the extension parts in the prediction curve, sequentially superposing the mean square error of adjacent measuring points in a positive-negative order according to the sequence of the extension parts from front to back, respectively adding the superposed extension parts to the outer sides of the left end point and the right end point of the target detection signal, and marking the added curve as an extension signal curve of the target detection signal;
An extended signal profile is obtained for each column of probe signals.
Optionally, the EMD decomposition is performed based on the extension signal curve to obtain a plurality of IMF components of each column of the detection signal, including the specific method that:
And carrying out EMD decomposition on each column of detection signals based on the extension signal curve to obtain a plurality of IMF components of each column of detection signals, wherein in the EMD decomposition process, a local mean function and an upper envelope function are obtained based on the extension signal curve, and then the local mean function and the upper envelope function of the original signal length are intercepted for decomposition.
Optionally, the constructing the second objective function according to the IMF component and the high-frequency measurement point includes the following specific methods:
The expression of the second objective function is:
wherein ρ represents the high frequency information duty ratio of the detection matrix, M represents the column number of the detection matrix, F c (x) represents the detection signal of the c-th column, Representing the constant integral of the detection signal of column c between 0 and L on the abscissa, L representing the length of the abscissa of the detection signal of each column, k c representing the iterative k-value of the detection signal of column c, f c,j (x) representing the function curve of the j-th IMF component of the detection signal of column c,The function curve representing the jth IMF component of the detection signal of column c is integrated at a constant value between the abscissa 0 and L, || represents the absolute value.
Optionally, the method for obtaining the denoising curve and the denoising detection matrix of each column of detection signals includes the following specific steps:
and (3) acquiring the component removal quantity of each column of detection signals by converging the second objective function, respectively removing the first K IMF components of each column of detection signals, wherein K represents the component removal quantity corresponding to each column of detection signals, recombining the residual IMF component signals, recording the obtained curve as a denoising curve of each column of detection signals, and obtaining a denoised detection matrix according to the denoising curve of each column of detection signals by extracting the data value at the measuring point, and recording the denoised detection matrix as a denoising detection matrix.
Optionally, the method for obtaining the component removal number of each column of the detection signal includes the following specific steps:
and obtaining the minimum value of the output value of the second objective function in the k-value iteration process of the detection signals in different columns, wherein the k-value of each column of the iteration signal corresponding to the minimum value is used as the component removal quantity of each column of the detection signals.
The beneficial effects of the invention are as follows: according to the invention, noise reduction processing is carried out on detection signals corresponding to the multi-beam submarine topography detection matrix, and the conventional smoothing algorithm is insufficient, so that the conventional EMD decomposition noise removal algorithm is optimized, and the problem of end effect in the signal decomposition process is solved; firstly, obtaining a detection matrix frequency spectrum through Fourier transformation, removing low-frequency information, restoring high-frequency information in the detection matrix, marking, then encircling a detection signal by utilizing a sine function, constructing a first objective function to obtain an intersection point of the sine function and the detection signal, namely, the minimum noise interference at a zero point, further obtaining an effective point, fitting to obtain a prediction curve and an endpoint extension length, and adding an error value to compensate the prediction loss to obtain an extension signal curve; the extension segments are involved in EMD decomposition together, so that the end effect can be generated in advance, the transition of the actual end point of the original signal is smooth, the influence of the end effect is avoided, and an accurate EMD decomposition result is obtained; then, each column of signal curves is subjected to self-adaptive noise component signal screening by using an information quantity statistics method, and a denoised detection signal is obtained; the application performance of the EMD algorithm in ocean exploration data is greatly improved, the error amount in the denoising process is reduced, more effective information is reserved as much as possible in a self-adaptive component screening mode, the problem of massive damage of information in the denoising process is restrained, and the method is helpful for the accuracy of the submarine topography measurement result.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a schematic flow chart of a signal acquisition noise reduction method for multi-beam seafloor topography measurement according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a multi-beam array design;
Fig. 3 is a schematic diagram of a multi-beam sound source;
FIG. 4 is a graph of a signal curve fitted to the detected data;
FIG. 5 is a schematic diagram of the end point distortion of the EMD decomposition process;
fig. 6 is a simplified model diagram of an offshore construction platform.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, a flowchart of a signal acquisition noise reduction method for multi-beam seafloor topography measurement according to an embodiment of the present invention is shown, the method includes the steps of:
and S001, collecting a detection matrix of multi-beam submarine topography measurement to obtain detection signals of each column.
The purpose of the embodiment is to denoise a detection matrix generated by multi-beam submarine topography measurement, so as to improve the precision of submarine topography measurement, so that the detection matrix of multi-beam submarine topography measurement needs to be acquired first; multi-beam submarine topography is a common marine topography measurement technique, and uses an array formed by a plurality of sound sources and receivers to measure the seabed at the same time; referring to fig. 2, a schematic diagram of a multi-beam array design is shown; referring to fig. 3, a schematic diagram of a multi-beam sound source is shown; the specific process of multi-beam submarine topography measurement is as follows:
Multi-beam array design: designing an array comprising a plurality of sound sources and receivers, and determining the distance and direction between each sound source and receiver; the sound wave emitted by each sound source in the array can be received by a receiver, and the propagation time and path of the sound wave are calculated;
Transmitting sound waves: exciting each sound source in turn to enable the sound source to emit a sound wave; parameters such as frequency, amplitude, phase and the like of the sound wave need to be controlled during emission so as to ensure that enough energy is possessed and mutual interference cannot be generated;
Receiving sound waves: when the sound wave reaches the receivers, the receivers convert the sound wave into electric signals and amplify the electric signals, and the electric signals are converted into water depth signal data according to the information such as the arrival time and the sound pressure of the sound wave, so that the detection data of the positions of the corresponding measuring points of each receiver are obtained.
Further, in multi-beam undersea terrain measurement, the accuracy and quality of signal acquisition have a very large impact on the final terrain data; the multi-beam signal records the detection data of the measuring points in the beam range in the form of a detection matrix, the detection matrix is decomposed according to columns or rows to obtain detection data of each column or each row, the detection data of each column or each row is fitted to obtain a plurality of two-dimensional signal curves, the two-dimensional signal curves are recorded as detection signals of each column or each row, and a signal curve obtained by fitting the detection data is shown in fig. 4; in this embodiment, the detection signals obtained from the detection data of each column are taken as an example to describe, and denoising of the detection matrix is completed by subsequently denoising the detection signals of each column.
Thus, a detection matrix and detection signals of each column are obtained.
And step S002, obtaining high-frequency measuring points by carrying out Fourier transformation on the detection matrix, and obtaining effective point groups and prediction curves of each column of detection signals according to each column of detection signals, the high-frequency measuring points and the sine function.
It should be noted that, in the process of performing EMD decomposition on the target detection signal by using any column of detection signals as the target detection signal, the local mean function and the envelope estimation function need to be obtained according to local extremum points, and the endpoint data of the target detection signal may be abnormal or missing due to reasons such as sampling frequency, and is not necessarily an extremum point; when calculating the local average value and the local envelope value at the end points, the end points are regarded as extreme points by default to calculate, so that distortion occurs when calculating the local average value function and the envelope estimation function at the end points of the signal curve, please refer to fig. 5, which shows a schematic diagram of the signal curve end point distortion in the EMD decomposition process, in which the solid line part in fig. 5 is a true local average value curve and an envelope estimation curve, the true meaning indicates the local average value curve and the envelope estimation curve obtained by the signal curve without distortion, and the dotted line part is the envelope estimation curve and the local average value curve, which are distortion curves caused by the end point effect.
The multi-beam submarine topography measurement is mainly used for detecting submarine topography of a predicted sea area, building an offshore construction platform in the predicted sea area and determining the pile point position of the offshore construction platform; the pile point position of the offshore construction platform is at the position of the detection plane close to the corner, and the end point effect leads to EMD decomposition distortion at the end point, so that the denoising effect of the detection signal is affected, and the positioning of the pile point and the submarine topography evaluation are greatly negatively affected; existing methods for solving the end-point effect include a continuation method, a zero padding method, a smoothing method and the like, but still introduce additional errors and noise or cause information blurring and loss; the extension method has obvious effect on solving the end effect of the periodic signal, the target detection signal can be regarded as water depth data (topographic data) which continuously changes along with the movement of the measuring point, the target detection signal is not the periodic signal, but the topographic change is necessarily low-frequency information for the signal detection frequency, and the predictability of the low-frequency signal is good, so the extension method can be applied; in particular, the extension method is adopted in the embodiment, rather than only extending an extreme point at two ends of the signal curve, because the target detection signal cannot be directly predicted when the noise signal exists, the high-frequency signal in the target detection signal needs to be filtered as noise, and a signal segment with a certain length is extended on the basis of the low-frequency signal, so that the end point effect in the EMD decomposition process occurs in advance and the offset is smaller, and the end point of the EMD decomposition process is well and smoothly transited to the end point of the target detection signal, and the core problem to be solved is that the effective point is selected and the optimal extension length is determined to be obtained, and the effective point is obtained by the low-frequency signal.
It should be further noted that, when the detection signals of each column are processed separately, the global high-frequency noise point in the detection matrix may not be found, so that the high-frequency noise point is detected by performing fourier transform on the detection matrix, and the high-frequency detection point in each column of detection signals is obtained by marking in the detection signal curve of each column according to the matrix position relationship, and then the selection of the low-frequency information, that is, the effective point, is performed according to the detection signals and the high-frequency detection points.
Specifically, firstly, a detection matrix is converted into a frequency spectrum through fourier transformation, the OTSU oxford threshold segmentation is performed through frequency information and amplitude information, segmentation thresholds of high-frequency information and low-frequency information are obtained, the high-frequency information and the low-frequency information are obtained, and the OTSU oxford threshold segmentation method is a known technology and is not repeated in the embodiment; calculating the duty ratio of the high-frequency information in the frequency spectrum, namely the duty ratio of the high-frequency information, which is denoted as rho, and obtaining the high-frequency information and calculating the duty ratio of the high-frequency information in the frequency spectrum are known techniques, and the embodiment is not repeated; the high-frequency information is inversely transformed into a detection matrix through Fourier transform, at the moment, the detection matrix only keeps corresponding measuring points of the high-frequency information, and according to the positions of the corresponding measuring points of the high-frequency information in the detection matrix, the high-frequency measuring points of each column of detection signals are marked, so that the high-frequency measuring points of each column of detection signals are obtained; since the topography change is continuously changed in a low-frequency trend, a large noise may exist at the high-frequency measuring points in the detected signal, and for any column of detected signals, the set formed by the marked high-frequency measuring points is denoted as H.
It should be further noted that, the effective points are signal measurement points that can be referred to when predicting and extending the signal through the low-frequency trend, the curve fitted by the effective points is a prediction curve, the effective points need to be locally subjected to the noise interference on the original signal curve, that is, the target detection signal curve, to be relatively minimum, and the quasi-periodic distribution is performed on the original signal curve, and the marked high-frequency measurement points obviously do not have the condition of being selected as the effective points; in order to ensure that the selection of the effective points has good periodicity, the embodiment utilizes the sine function to superimpose the target detection signals, namely, the sine function is enabled to surround along the base line of the target detection signal curve, then the effective points are taken at the zero point, the fitting loss of the effective points is minimized through iterating the circle frequency of the sine function, and the high-frequency measuring points with larger noise are avoided as far as possible to construct the first target function, so that the final effective point group is obtained.
Specifically, any one row of detection signals is taken as a target detection signal, and the expression of a first target function of the target detection signal is as follows by utilizing a sine function and high-frequency measuring points of the target detection signal:
Wherein L represents the abscissa length of the target detection signal, x i represents the abscissa value of the ith measurement point in the target detection signal, F (x) represents the target detection signal, P (x) represents the local mean function of the target detection signal, sin ωx represents a sine function with amplitude of 1, circular frequency of ω and initial phase of 0, and it is noted that the circular frequency controls the period of the sine function The larger the circle frequency is, the smaller the period is; n H represents the number of high-frequency measurement points in the target detection signal, (sin ωx=0) ∈h represents that the zero point of the sine function is a high-frequency measurement point, G (sinωx＝0)∈H represents that the zero point of the sine function is the number of high-frequency measurement points, and i represents that the absolute value is calculated; by superposing the local mean function and the sine function, the target detection signal is kept unchanged only at the zero point of the sine function, and by iterating the circular frequency omega, in the iteration processThe closer to 0, the closer to the fluctuation value of the target detection signal is the sine function surrounding the local mean function, the sine function under the round frequency of the iteration can cover the high-frequency information, namely noise information, to the greatest extent, namely, the noise is concentrated on the non-zero part of the sine function, so that the zero points of all the sine functions are used as effective points to fit and obtain a prediction curve with the minimum local noise interference, and the prediction curve predicts the low-frequency terrain change signal in the target detection signal; meanwhile, the comparison of the high-frequency measuring point and the zero point of the sine function is introduced as a punishment item, namely, the zero point of the sine function is required to avoid the high-frequency measuring point as much as possible, and thenThe smaller the closer to 0, the fewer the sine function zero points are indicated for the high frequency points, the smaller the penalty.
Further, iterating the circle frequency ω in the first objective function, calculating an initial value of the iterative process by using 1 in this embodiment, and setting 1 in this embodiment for each iteration increment value; when the output result of the first objective function in the iteration process is minimum, the first objective function regarded as the objective detection signal is iterated to be converged at the round frequency, the round frequency at the moment is recorded as the optimal round frequency of the objective detection signal, all zero points of the sine function of the optimal round frequency are used as the effective point groups, and then the effective point groups of the objective detection signal are obtained; fitting all the effective points in the effective point group by combining the corresponding data of the effective points in the target detection signal, extending the curve obtained by fitting at the end points of two sides through quadratic linear interpolation, and setting the extension length of two sides as a sine period in the embodimentLength, wherein omega' is the optimal circle frequency of the target detection signal, and the extended fitting curve is recorded as the prediction curve of the target detection signal; and acquiring an effective point group and a prediction curve of each column of detection signals according to the method.
So far, by acquiring high-frequency measuring points and combining a sine function, the effective point group and the prediction curve of each column of detection signals are acquired.
Step S003, an extension signal curve of each column of detection signals is obtained according to each column of detection signals, a local mean function and a prediction curve, EMD decomposition is carried out on the basis of the extension signal curve to obtain a plurality of IMF components, and the component removal quantity of each column of detection signals is obtained according to the IMF components and high-frequency measuring points to obtain a denoising curve and a denoising detection matrix of each column of detection signals.
It should be noted that, because the prediction curve is obtained by prediction according to the effective point group, the prediction curve belongs to the trend item and has prediction error; therefore, the noise value is required to be added to the extension part in the prediction curve, so that the prediction error is eliminated, the offset of the end effect when the EMD calculates the local mean function and the upper and lower envelope functions is reduced by reducing the prediction error, the offset of the extension section is smoothly transited to the actual end point of the original signal, and the influence on the subsequent signal decomposition is almost eliminated.
Specifically, taking any column of detection signals as target detection signals, calculating the difference value of the target detection signals and the local mean function of the target detection signals on each data point, and recording the difference value as the error value of each data point, namely calculating the value of the target detection signals on the same abscissa minus the value of the local mean function to obtain the error value; obtaining the mean square error of all error values, if the noise in the target detection signal accords with Gaussian distribution, regarding the error value as an estimated noise value, then the mean square error of the error value represents a noise fluctuation interval of the target detection signal, overlapping the mean square error of an extension part in a prediction curve, namely overlapping the mean square error of each abscissa of the extension part on the basis of the original numerical value, wherein in order to keep the endpoints, the overlapping value of the two endpoints of the prediction curve is 0, then the numerical values of the adjacent measurement points, namely the adjacent abscissas, are overlapped with the mean square error in sequence from front to back according to the extension part, namely the numerical value of one measurement point is added with the mean square error, the numerical value of the next measurement point is subtracted from the mean square error, the overlapped extension part is respectively added to the outer sides of the left endpoint and the right endpoint of the target detection signal, namely through the overlapped extension part of the prediction curve, and the added curve is recorded as an extension signal curve of the target detection signal; an extension signal curve of each column of detection signals is obtained according to the method.
It should be further noted that when calculating the local mean function and the upper and lower envelope functions of the target detection signal, the extension segments are calculated together, in the subsequent EMD decomposition process, the local mean function and the upper and lower envelope functions of the original signal length are intercepted, and the deviation caused by the end point effect is completely placed in the extension segments of the end points at the two sides to be directly discarded, so that the end point effect problem can be effectively solved, and a more accurate EMD decomposition signal is obtained; the EMD decomposition process is as follows:
Subtracting the local mean function from the input target detection signal, and obtaining the first order difference between the upper envelope mean value and the lower envelope mean value of the obtained output signal, and repeating the operation if the first order difference is not 0 until the first order difference between the upper envelope mean value and the lower envelope mean value is 0, namely the upper envelope and the lower envelope are symmetrical, and the average value of the upper envelope and the lower envelope is a straight line; then a first component signal is obtained, the first component signal is subtracted from the input signal to obtain a residual signal, the steps are repeated on the residual signal to decompose the residual signal until all the component signals are obtained, and the stopping condition is that the last component signal is less than two extreme points, and then the decomposition is stopped;
After all component signals are obtained, the embodiment analyzes IMF components distributed in noise concentration in each column of detection signals by constructing a second objective function, removes k IMF components in front of each column of detection signals, k represents the number of component removal, and recombines the remaining IMF components to obtain a denoising curve of each column of detection signals; in the existing method, k=3 is usually adopted for calculation, however, information loss is easy to be caused by setting an empirical value, each column of detection signals only comprises partial information of the detection matrix, and damage of the detection matrix is overlarge due to the fact that excessive component signals are lost, so that the k value is iterated through constructing a second objective function, and accurate component removal quantity is obtained.
In particular, EMD decomposition is firstly carried out on each column of detection signals based on an extension signal curve to obtain a plurality of IMF components of each column of detection signals, wherein in the EMD decomposition process, a local mean function and an upper envelope function and a lower envelope function are obtained based on the extension signal curve, and then the local mean function and the upper envelope function of the length of an original signal are intercepted to carry out subsequent decomposition; in this embodiment, the iteration is performed on the k value, where the k value is the component removal number, the iteration initial value is calculated by using 1, the increment value of each iteration is also 1, the maximum value of the iteration is set to 3 in this embodiment, that is, the iteration process of k is k=1, k=2, and k=3, and then the expression of the second objective function is:
wherein ρ represents the high frequency information duty ratio of the detection matrix, M represents the column number of the detection matrix, F c (x) represents the detection signal of the c-th column, Representing the constant integral of the detection signal of the c-th column between 0 and L on the abscissa, wherein L represents the length of the abscissa of the detection signal of each column, and the constant integral between 0 and L is the area surrounded by the detection signal and the abscissa; k c denotes the iterative k-value of the detection signal of column c, f c,j (x) denotes the function curve of the j-th IMF component of the detection signal of column c,A constant integral of a function curve representing the jth IMF component of the detection signal of the c-th column between the abscissa 0 and L, || represents an absolute value; because the IMF components can be overlapped and recombined into the detection signal, the ratio between the sum of the fixed integral of the first k c IMF components and the fixed integral of the detection signal of the c-th column can be expressed, namely the information duty ratio of the information included in the first k c IMF components in the detection signal, the ratio obtained by accumulating all the column detection signals is accumulated, and the ratio is compared with the high-frequency information duty ratio, the second objective function is more 0, so that the probability of the noise concentrated distribution in the first k IMF components corresponding to each column detection signal at the moment is more, namely the high-frequency information is concentrated and distributed in the first k IMF components corresponding to each column detection signal; it should be noted that, the iterative k values of each column of the detection signals are respectively subjected to iterative calculation, and the k value of each column of the iterative signal corresponding to the minimum value is used as the component removal quantity of each column of the detection signals by obtaining the minimum value of the output value of the second objective function in the k value iterative process of different columns of the detection signals; for example, there are 9 columns in the detection matrix, and the k-value iteration process of each column of detection signals is 3 times, so that 9×3=27 output values of the second objective function can be obtained in total.
Further, the first K IMF components of each column of the detection signals are removed according to the component removal number of each column of the detection signals, where K is the component removal number corresponding to each column of the detection signals, the remaining IMF component signals are recombined, the obtained curve is recorded as a denoising curve of each column of the detection signals, and the denoised detection matrix is recorded as a denoising detection matrix by extracting the data value at the measuring point according to the denoising curve of each column of the detection signals.
Thus, a denoising curve of each column of detection signals is obtained, and a denoising detection matrix is obtained.
And S004, completing multi-beam submarine topography measurement according to the denoising detection matrix.
After the denoising detection matrix is obtained, obtaining a submarine topography map of the multi-beam topography measurement sea area according to the denoising detection matrix, and completing multi-beam submarine topography measurement according to the detection matrix as the prior art, which is not described in detail in the embodiment; taking the average water depth in the range of the areas of the pile legs and the preset pile point position as the flatness of the preset pile point position; judging whether the construction conditions are met according to flatness standard regulations of different types of construction platforms, and further completing construction of the offshore construction platform, wherein pile point flatness and construction platform construction are known techniques, and are not points of the invention, and the embodiment is not repeated; referring to fig. 6, a simplified model diagram of an offshore construction platform is shown.
Therefore, the detection matrix and the denoising of the detection signals for multi-beam submarine topography measurement are completed, so that the end point effect cannot cause signal distortion to cause inaccurate signal denoising results, the submarine topography measurement results are affected, and the submarine topography measurement accuracy is improved.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (6)
1. A method of signal acquisition noise reduction for multi-beam seafloor topography measurements, the method comprising the steps of:
collecting a detection matrix of multi-beam submarine topography measurement, and obtaining each column of detection signals;
The method comprises the steps of carrying out Fourier transformation on a detection matrix to obtain a high-frequency information duty ratio and high-frequency measuring points of each column of detection signals, constructing a first objective function of each column of detection signals according to each column of detection signals and the high-frequency measuring points of each column of detection signals by combining a sine function, and obtaining an effective point group and a prediction curve of each column of detection signals by converging the first objective function;
According to each column of detection signals, a local mean function and a prediction curve, an extension signal curve of each column of detection signals is obtained, EMD decomposition is carried out on the basis of the extension signal curve to obtain a plurality of IMF components of each column of detection signals, a second objective function is constructed according to the IMF components and high-frequency measuring points, and a denoising curve and a denoising detection matrix of each column of detection signals are obtained by converging the second objective function;
Completing multi-beam submarine topography measurement according to the denoising detection matrix;
The method for constructing the first objective function of each column of detection signals comprises the following specific steps:
taking any column of detection signals as target detection signals, wherein the expression of a first target function of the target detection signals is as follows:
Wherein, Representing the length of the abscissa of the object detection signal,Representing the first of the object detection signalsThe abscissa value of each measuring point,Which is indicative of the detection signal of the object,Representing the local mean function of the object detection signal,Indicating an amplitude of 1 and a circular frequencyAnd the initial phase is a sine function of 0,Indicating the number of high frequency points in the target probe signal,The zero point representing the sine function is a high-frequency measuring pointThe zero point representing the sine function is the number of high frequency points,Representing absolute value;
Constructing a first objective function of each column of detection signals;
the method for acquiring the effective point group and the prediction curve of each column of detection signals comprises the following specific steps:
Taking any column of detection signals as target detection signals, converging a first target function of the target detection signals to obtain the optimal circular frequency of the target detection signals, and taking all zero points of a sine function of the optimal circular frequency as an effective point group to obtain an effective point group of the target detection signals;
Fitting all the effective points in the effective point group by combining with corresponding data of the effective points in the target detection signal, extending the fitted curve at the end points of two sides, and marking the extended fitted curve as a predicted curve of the target detection signal;
Acquiring an effective point group and a prediction curve of each column of detection signals;
the specific method for acquiring the extension signal curve of each column of detection signals comprises the following steps:
Taking any column of detection signals as target detection signals, calculating the difference value of the target detection signals and the local mean function of the target detection signals on each data point, and recording the difference value as an error value of each data point;
Acquiring the mean square error of all error values, superposing the mean square error of the extension parts in the prediction curve, sequentially superposing the mean square error of adjacent measuring points in a positive-negative order according to the sequence of the extension parts from front to back, respectively adding the superposed extension parts to the outer sides of the left end point and the right end point of the target detection signal, and marking the added curve as an extension signal curve of the target detection signal;
Acquiring an extension signal curve of each column of detection signals;
The method for constructing the second objective function according to the IMF component and the high-frequency measuring point comprises the following specific steps:
The expression of the second objective function is:
Wherein, Representing the high frequency information duty cycle of the detection matrix,Representing the number of columns of the detection matrix,Represent the firstThe detection signal of the column,Represent the firstThe column detection signal is integrated at a constant value between 0 and L on the abscissa,Representing the abscissa length of the probe signal of each column,Represent the firstIteration of column detection signalsThe value of the sum of the values,Represent the firstThe first of the column's detection signalsA function curve of the individual IMF components,Represent the firstThe first of the column's detection signalsThe functional curves of the individual IMF components are integrated at a constant value between 0 and L on the abscissa,Representing absolute values.
2. The method for noise reduction in signal acquisition for multi-beam submarine topography measurement according to claim 1, wherein the method for obtaining the high-frequency information duty ratio and the high-frequency measurement point of each column of the detection signals comprises the following specific steps:
Converting the detection matrix into a frequency spectrum through Fourier transformation, and carrying out OTSU (optical time series unit) Otsu (on-line per se) Otsu threshold segmentation through frequency information and amplitude information to obtain segmentation thresholds of high-frequency information and low-frequency information, so as to obtain the high-frequency information and the low-frequency information;
calculating the duty ratio of the high-frequency information in the frequency spectrum, and recording the duty ratio as the duty ratio of the high-frequency information;
And (3) carrying out inverse Fourier transform on the high-frequency information to a detection matrix, wherein the detection matrix only retains corresponding measuring points of the high-frequency information, and marking the high-frequency measuring points in each column of detection signals according to the positions of the corresponding measuring points of the high-frequency information in the detection matrix to obtain the high-frequency measuring points of each column of detection signals.
3. The method for noise reduction in signal acquisition for multi-beam undersea topography measurement according to claim 1, wherein the obtaining the optimal circular frequency of the target detection signal comprises the following specific steps:
And iterating the circular frequency in the first objective function of the target detection signal by taking any column of detection signals as target detection signals, and when the output result of the first objective function is minimum in the iteration process, considering that the first objective function of the target detection signal is converged at the time when the circular frequency iterates, and recording the circular frequency at the time as the optimal circular frequency of the target detection signal.
4. The method for signal acquisition and noise reduction for multi-beam seafloor topography measurement according to claim 1, wherein the EMD decomposition based on the extended signal curve obtains a plurality of IMF components of each column of the probe signal, comprising the specific steps of:
And carrying out EMD decomposition on each column of detection signals based on the extension signal curve to obtain a plurality of IMF components of each column of detection signals, wherein in the EMD decomposition process, a local mean function and an upper envelope function are obtained based on the extension signal curve, and then the local mean function and the upper envelope function of the original signal length are intercepted for decomposition.
5. The method for noise reduction in signal acquisition for multi-beam submarine topography measurement according to claim 1, wherein the obtaining a denoising curve and a denoising detection matrix of each column of detection signals comprises the following specific steps:
By converging the second objective function, the component removal quantity of each column of detection signals is obtained, and the front of each column of detection signals is respectively calculated Removing the IMF component, whereinAnd (3) representing the removal quantity of the components corresponding to each column of detection signals, recombining the residual IMF component signals, marking the obtained curve as a denoising curve of each column of detection signals, and obtaining a denoised detection matrix according to the denoising curve of each column of detection signals by extracting the data value at the measuring point, and marking the denoised detection matrix as a denoised detection matrix.
6. The method for noise reduction in signal acquisition for multi-beam seafloor topography measurement of claim 5, wherein said obtaining the component removal number of each column of probe signals comprises the specific steps of:
By acquisition of different columns of detection signals Minimum value of output value of second objective function in value iteration process, and each row of iterative signals corresponding to minimum valueThe value is taken as the component removal number of each column of the detection signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310723169.9A CN116776083B (en) | 2023-06-19 | Signal acquisition noise reduction method for multi-beam submarine topography measurement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310723169.9A CN116776083B (en) | 2023-06-19 | Signal acquisition noise reduction method for multi-beam submarine topography measurement |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116776083A CN116776083A (en) | 2023-09-19 |
| CN116776083B true CN116776083B (en) | 2024-07-16 |
Family
ID=
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111985426A (en) * | 2020-08-27 | 2020-11-24 | 南京信息工程大学 | Sea clutter hybrid denoising algorithm based on variational modal decomposition |
| WO2023004688A1 (en) * | 2021-07-29 | 2023-02-02 | 南京浙溧智能制造研究院有限公司 | Non-contact respiration monitoring method based on doppler radar |
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111985426A (en) * | 2020-08-27 | 2020-11-24 | 南京信息工程大学 | Sea clutter hybrid denoising algorithm based on variational modal decomposition |
| WO2023004688A1 (en) * | 2021-07-29 | 2023-02-02 | 南京浙溧智能制造研究院有限公司 | Non-contact respiration monitoring method based on doppler radar |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112526451B (en) | Compressed beam forming and system based on microphone array imaging | |
| US5581490A (en) | Contact management model assessment system for contact tracking in the presence of model uncertainty and noise | |
| CN110673108B (en) | Airborne marine laser radar signal processing method based on iteration Klett | |
| CN113625337B (en) | Ultra-shallow water high-precision seismic data rapid imaging method | |
| CN112703421B (en) | Echo signal processing method, device, system and storage medium | |
| CN113093280B (en) | Equal-floating correction method for virtual reflection travel-time cable based on coherent function control | |
| CN113608195B (en) | Laser radar full waveform data decomposition sounding method and device and electronic equipment | |
| CN110673109A (en) | Full waveform data decomposition method for satellite-borne large-light-spot laser radar | |
| KR102497000B1 (en) | Ultrasonic non-destructive inspection method and system using deep learning and auto-encoder-based predictive model learning method used therein | |
| CN117111048B (en) | Improved ITD-based multi-beam submarine topography measurement echo noise reduction method | |
| CN108646249B (en) | Parameterized leakage target detection method suitable for partial uniform reverberation background | |
| Bitenc et al. | Evaluation of wavelet denoising methods for small-scale joint roughness estimation using terrestrial laser scanning | |
| CN110490800B (en) | Shallow sea terrain reconstruction method | |
| CN115859116A (en) | Marine environment field reconstruction method based on radial basis function regression interpolation method | |
| CN116908853A (en) | High coherence point selection method, device and equipment | |
| CN110133680B (en) | Airborne laser sounding received waveform effective signal initial value determination method and system | |
| CN116776083B (en) | Signal acquisition noise reduction method for multi-beam submarine topography measurement | |
| CN113866736A (en) | Laser radar echo signal Gaussian inflection point selective decomposition method | |
| CN111445404B (en) | Phase deblurring method based on double-frequency and probability model | |
| CN116776083A (en) | Signal acquisition noise reduction method for multi-beam submarine topography measurement | |
| Hofton et al. | Algorithm Theoretical Basis Document (ATBD) | |
| CN116738375A (en) | Induced heave error detection and elimination method and system based on single-strip sounding data | |
| Beaudoin et al. | Automated correction of refraction residuals | |
| Lu et al. | Automatic outlier detection in multibeam bathymetric data using robust LTS estimation | |
| CN115170428A (en) | Noise reduction method for acoustic wave remote detection imaging graph |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Country or region after: China Address after: 310000 Gudun Road 997, Sandun Town, Xihu District, Hangzhou City, Zhejiang Province Applicant after: Zhejiang East China Geotechnical Survey and Design Institute Co.,Ltd. Applicant after: 715 Research Institute of China Shipbuilding Corp. Address before: 310000 Gudun Road 997, Sandun Town, Xihu District, Hangzhou City, Zhejiang Province Applicant before: ZHEJIANG HUADONG CONSTRUCTION ENGINEERING Co.,Ltd. Country or region before: China Applicant before: 715 Research Institute of China Shipbuilding Corp. |
|
| GR01 | Patent grant |