CN114898278B - Non-contact rockfall protection dynamic response signal automatic identification and feedback method - Google Patents

Non-contact rockfall protection dynamic response signal automatic identification and feedback method Download PDF

Info

Publication number
CN114898278B
CN114898278B CN202210653638.XA CN202210653638A CN114898278B CN 114898278 B CN114898278 B CN 114898278B CN 202210653638 A CN202210653638 A CN 202210653638A CN 114898278 B CN114898278 B CN 114898278B
Authority
CN
China
Prior art keywords
wavelet
signal
rockfall
gaussian
differential
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
Application number
CN202210653638.XA
Other languages
Chinese (zh)
Other versions
CN114898278A (en
Inventor
郭立平
余志祥
田永丁
廖林绪
张丽君
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southwest Jiaotong University
Original Assignee
Southwest Jiaotong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southwest Jiaotong University filed Critical Southwest Jiaotong University
Priority to CN202210653638.XA priority Critical patent/CN114898278B/en
Publication of CN114898278A publication Critical patent/CN114898278A/en
Application granted granted Critical
Publication of CN114898278B publication Critical patent/CN114898278B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising
    • G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
    • Y02A10/23—Dune restoration or creation; Cliff stabilisation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Complex Calculations (AREA)

Abstract

The invention relates to the technical field of side slope rockfall disaster protection, in particular to a non-contact rockfall protection dynamic response signal automatic identification and feedback method, which comprises the following steps: step (1): arranging a camera and recording a high-speed video in an impact process; step (2): calculating and calibrating a pixel-distance scale factor; and (3): automatically identifying the falling rock impact motion track; and (4): analyzing a time domain noise-containing signal frequency domain and energy distribution thereof; and (5): selecting a Gaussian function and a Gaussian wavelet family; and (6): determining a signal pseudo frequency and selecting a scale parameter; and (7): performing convolution operation on the time domain noise-containing signal and Gaussian wavelets; and (8): and (4) performing signal feedback of falling rock movement displacement, speed and acceleration and self-consistency check. The invention realizes the automatic identification and real-time feedback of the non-contact rockfall protection dynamic response signal.

Description

Non-contact rockfall protection dynamic response signal automatic identification and feedback method
Technical Field
The invention relates to the technical field of side slope rockfall disaster protection, in particular to a non-contact rockfall protection dynamic response signal automatic identification and feedback method.
Background
The flexible protective net is a strong nonlinear structure commonly used for protecting collapsed falling rocks, and is easy to encounter continuous impact action of the falling rocks during working. Under certain impact energy, multiple collision and rebound processes occur between the falling rocks and the intercepting net piece, and the falling rocks and the intercepting net piece are finally stabilized on the flexible net surface. When the rockfall interacts with the protective net, information such as deformation, protective energy level and impact load of the protective net can be indirectly obtained by monitoring the rockfall motion state in real time, and key data support is provided for impact resistance evaluation and engineering design of the protective net system. However, the physical characteristics and structural response of the protective net under the action of continuous impact continuously change, and the rockfall motion signal is a nonlinear and unstable transient pulse signal. When the traditional contact type wired equipment is used for measuring the dynamic response signal of the rockfall protection, equipment damage and signal instability are easy to occur. Particularly, under the condition of noise interference, it is very difficult to realize automatic feedback and information feedback of the rockfall motion data signal.
Disclosure of Invention
The invention provides a non-contact rockfall protection dynamic response signal automatic identification and feedback method based on wavelet transformation, which can overcome some or some defects in the prior art.
The non-contact rockfall protection dynamic response signal automatic identification and feedback method comprises the following steps:
step (1): arranging a camera and recording a high-speed video in an impact process;
step (2): calculating and calibrating a pixel-distance scale factor;
and (3): automatically identifying the falling rock impact motion track;
and (4): analyzing a time domain noise-containing signal frequency domain and energy distribution thereof;
and (5): selecting a Gaussian function and a Gaussian wavelet family;
and (6): determining a signal pseudo frequency and selecting a scale parameter;
and (7): performing convolution operation on the time domain noisy signal and the Gaussian wavelet;
and (8): and (3) carrying out signal feedback and self-consistency check on the displacement, the speed and the acceleration of the rockfall.
Preferably, in the step (1), 2 high-speed cameras are erected according to the installation positions of the flexible protective nets, and the camera positions and the recording targets meet the requirements of different distances and different angles; setting a camera frame rate, and recording the whole impact process; the minimum frame rate of the camera should meet the requirement of the sampling theorem, and the maximum frame rate should ensure that the recording time can be longer than or equal to the duration of the whole impact process.
Preferably, in the step (2), different target objects in the high-speed video picture are selected, and the actual characteristic length of each target object is measured respectively; the number of pixels occupied by a selected target object in a high-speed video picture is recorded, and a pixel-distance scaling factor is calculated according to the one-to-one correspondence between the video picture pixels and the characteristic lengths.
Preferably, the step (a)(3) Marking a rock falling outline in a video picture, and selecting feature points captured by rock falling motion by adjusting picture contrast and brightness; ensuring that the rockfall characteristic points are not completely shielded in the whole impact process; the automatic capture of the rockfall characteristic points and the automatic recognition of the rockfall motion tracks are realized by combining the motion analysis technology; converting the capture pixel points of the rockfall motion process into rockfall displacement time-domain signals f through the pixel-distance scale factors calculated in the step (2) M [t n ]Where N =1,2, \8230;, N.
Preferably, in the step (4), the falling rock displacement time-domain signal f obtained in the step (3) is subjected to M [t n ]Performing discrete Fourier transform to obtain Fourier transform result F (omega) of rockfall motion signal k )：
Figure GDA0003947768790000021
Wherein ω is k Is the frequency value of the signal; according to the discrete Fourier transform result of the rockfall displacement time-domain signal, the frequency domain energy E of the signal can be further obtained freq ：
Figure GDA0003947768790000031
Recording the signal frequency band with the energy ratio of 0-90% and the energy ratio of 0-99%, and recording the frequency of the energy band as omega 90 And omega 99 。
Preferably, in step (5), the gaussian function is one of the most important functions in the application of probability statistics and random signal analysis, and the expression of the gaussian function g (t) is as follows:
Figure GDA0003947768790000032
wherein C and alpha are constants; fourier transform of Gaussian function
Figure GDA0003947768790000033
Still gaussian function:
Figure GDA0003947768790000034
the Gaussian function has smooth infinite derivative, the n-order derivative of the Gaussian function has the characteristics of wavelet function oscillation and limited energy, and n Gaussian mother wavelets can be generated by performing n-order differential operation on the Gaussian function:
Figure GDA0003947768790000035
for wavelet function g n (t) a series of wavelet functions can be obtained by performing expansion and translation transformation, and the functions are called as a wavelet function family:
Figure GDA0003947768790000036
wherein g is u,s (t) is called a family of Gaussian wavelet functions, s is called a scale parameter, and u is called a translation parameter.
Preferably, in step (6), the wavelet scale parameter is not equivalent to a fourier frequency that is more easily understood in engineering; the signal frequency domain characteristic corresponding to the wavelet scale parameter s adopts a pseudo frequency f s Expressed, the corresponding relationship between the scale parameter and the pseudo frequency can be converted by the following formula:
Figure GDA0003947768790000037
wherein Δ t =1/f samp Is the sampling time interval of the time domain signal, f samp To the sampling frequency, f c ＝ω c The/2 pi is the central frequency of the wavelet, for the time domain impact signal, the frequency component greater than zero is more concerned, and the mother wavelet function g adopted in the wavelet transformation is selected n (t) after, mother wavelet time center t c Angular frequency center omega c Can be calculated from the following formula:
Figure GDA0003947768790000041
preferably, in step (7), performing convolution operation on the rockfall displacement signal obtained in step (2) and the gaussian wavelet selected in step (5) to obtain a gaussian wavelet transform result GWT (u, s) of the time-domain noisy signal:
Figure GDA0003947768790000042
wherein
Figure GDA0003947768790000043
Figure GDA0003947768790000044
Is the complex conjugate of g (t); depending on the differential nature of the convolution operation, the gaussian wavelet transform can eventually be written in the form:
Figure GDA0003947768790000045
the above formula shows that the expression of selecting Gaussian wavelet to perform wavelet transformation explicitly includes the differential operation d n /du n And convolution operation f (u) g s (u); because the integral of the Gaussian function in the real number domain is not zero, the convolution operation is explained as the function f in the kernel function g s (u) a weighted average smoothing process, wherein the Gaussian wavelets of different orders are correspondingly calculated according to the differential of different orders of the original noisy signal; through wavelet transformation, smoothing of the noisy vibration signal f and its n-order differentiation on the scale s are simultaneously achieved.
Preferably, in the step (8), for the actual time-domain vibration signal, the wavelet transformation differential result needs to ensure the accuracy of the amplitude of the differential result in addition to the accuracy of the curve shape and the positions of peaks and valleys; from step (7), it is found that the amplitude of the wavelet differential result is s of the wavelet scale parameter n Related to, areThe normalization processing of the family of time wavelet functions also affects the magnitude result of the differential operation; in order to eliminate the influence of the Gaussian wavelet transform process on the amplitude of the differential result, an amplitude parameter A is introduced mp The wavelet differential operation of a noisy signal is defined as:
Figure GDA0003947768790000051
the numerical differentiation process amplifies the influence of noise interference in the signals, and the numerical integration process plays a role in inhibiting the noise interference in the original signals in a reverse view; therefore, the accuracy of the wavelet differential result can be measured by performing reverse integration on the wavelet differential result and comparing the coincidence degree of the wavelet differential result and the original signal, and the higher the coincidence degree of the integration result and the original signal is, the more accurate and reliable the result after the differentiation can be considered; because the two time sequences before and after the wavelet transformation have the same length, the overlapping degree between the noisy signal and the integral result of the approximate differential of the wavelet is evaluated by adopting the Euclidean distance, and the Euclidean distance is calculated as follows:
Figure GDA0003947768790000052
wherein j represents equidistant sampling of wavelet differential results of the original time domain signals, m is the number of discrete points of the sampled signals, and n is the wavelet differential order;
Figure GDA0003947768790000053
for wavelet transformation of the result of an n-th order differentiation
Figure GDA0003947768790000054
Integral of the value of (t) k To find the product node, A k For the multiplication factor, also called the accompanying node t k The right of (1); when n =0, the number of the bits is set to n =0,
Figure GDA0003947768790000055
for noisy signals
Figure GDA0003947768790000056
And a gaussian function g 0 (t) convolution smoothing results;
Figure GDA0003947768790000057
amplitude parameter A mp The value of (d) can be obtained by iterative solution, when the Euclidean distance between the first integral of the n-order wavelet differential and the n-1 order wavelet differential reaches the minimum value ED min When the coincidence degree of the curves is the highest, the amplitude parameter obtains the optimal value; finally, feeding back a rockfall impact displacement signal through convolution operation of a noise-containing signal and a Gaussian function; feeding back a rockfall impact speed signal through convolution operation of a noisy signal and a first-order Gaussian wavelet; and feeding back a rockfall impact acceleration signal through convolution operation of the noise-containing signal and a second-order Gaussian wavelet.
The signal processing method based on Gaussian wavelet transform can be applied to rockfall motion analysis. The method can realize automatic identification and automatic capture of impact falling rocks in the high-speed video. The characteristic that the Gaussian function has infinite-order smooth derivative is utilized, the Gaussian wavelet is adopted to carry out wavelet transformation on the rockfall catching signal, the approximate differential calculation of any order of noisy signals can be realized, and the derivative order depends on the property of the wavelet function. The wavelet method effectively solves the problem of high sensitivity of the differential information feedback process of the rockfall capture signal to noise interference. The Gaussian wavelet transform is equivalent to the process of smoothing and deriving signal convolution, the differential result of the noise-containing rockfall displacement signal can be fed back through one-time convolution operation, the rockfall speed and acceleration time course close to the real rockfall speed and acceleration time course are obtained, and automatic identification and real-time feedback of the non-contact rockfall protection dynamic response signal are achieved.
Drawings
Fig. 1 is a flowchart of a non-contact rockfall protection dynamic response signal automatic identification and feedback method based on wavelet transform in embodiment 1;
fig. 2 is a schematic diagram of a high-speed camera arrangement and a rockfall motion capture in embodiment 1;
FIG. 3 is a schematic diagram of Gaussian function, first order Gaussian wavelet and second order Gaussian wavelet in example 1;
FIG. 4 is a schematic diagram of the time-domain signal of the falling rock motion capture and its frequency-domain distribution in example 1;
FIG. 5 is a flow chart of iteration of the Gaussian wavelet transform magnitude parameter in example 1;
fig. 6 is a graph of the feedback result of the signals of the displacement, the speed and the acceleration of the rockfall in the embodiment 1.
The method comprises the following steps of 1, rockfall, 2, a rockfall impact plane, 3, a scale, 4, a first-speed camera, 5, a second-speed camera, 6, a Gaussian function, 7, a first-order Gaussian wavelet, 8, a second-order Gaussian wavelet, 9, rockfall displacement capture signals, 10, rockfall displacement signal frequency domain distribution, 11, rockfall displacement signal frequency domain energy distribution, 12, rockfall displacement fed back by a wavelet method, 13, rockfall speed fed back by the wavelet method, and 14, rockfall acceleration fed back by the wavelet method.
Detailed Description
For a further understanding of the invention, reference should be made to the following detailed description taken in conjunction with the accompanying drawings and examples. It is to be understood that the examples are illustrative of the invention and not restrictive.
Example 1
As shown in fig. 1 to 6, the embodiment provides a non-contact rockfall protection dynamic response signal automatic identification and feedback method based on wavelet transform:
firstly, the following steps: high-speed video recording of camera position arrangement and impact process
Two high-speed cameras are erected according to the impact motion track range of the falling rocks 1, wherein the recording direction of the first high-speed camera 4 is vertical to the impact plane 2 of the falling rocks, and the distance is d 1 (ii) a The recording direction of the second high-speed camera 5 is parallel to the rockfall impact plane 2, and the distance is d 2 . The frame rates of the two cameras are both f sample And recording the whole impact process.
II, secondly, the method comprises the following steps: pixel-to-distance scale factor calculation and calibration
Selecting reference targets in a high-speed video picture as a falling rock 1 and a scale 3 in an impact plane respectively, and measuring the reference targets respectivelyDiameter l of rock 1 1 The length of the scale 3 is l 2 . Recording the number of pixels occupied by the diameter of the falling rock 1 in the high-speed video picture as m 1 The number of pixels occupied by the length of the scale 3 is m 2 . And testing different reference targets, and calculating a pixel-distance scale factor calculation value lambda according to the corresponding relation between the video picture pixel and the characteristic length. When the following expression is satisfied, it is considered that the calibration value [ λ ] of the pixel scale factor is obtained]。
Figure GDA0003947768790000071
Thirdly, the method comprises the following steps: automatic identification of falling rock impact motion track
Marking a rock falling outline in a video picture, and selecting feature points captured by rock falling motion by adjusting picture contrast and brightness. The rockfall characteristic points are ensured not to be completely shielded in the whole impact process. And the automatic capture of the rockfall characteristic points and the automatic identification of the rockfall motion trail are realized by combining the motion analysis technology. Calibration value lambda by pixel-distance scale factor]Converting capture pixel points in the falling rock motion process into falling rock displacement time domain signals f M [t n ]Where N is the number of recorded displacement data points, N =1,2, \ 8230;, N.
Fourthly, the method comprises the following steps: time domain noisy signal frequency domain analysis and energy distribution thereof
Signals 9 (f) are captured for rockfall displacement M [t n ]) Performing discrete Fourier transform to obtain frequency domain distribution 10 of falling rock displacement signal F (omega) k )：
Figure GDA0003947768790000072
Wherein omega k Is the frequency value of the signal. According to the discrete Fourier transform result of the rockfall displacement time-domain signal, the frequency-domain energy distribution 11 of the rockfall displacement signal can be further obtained to be E freq ：
Figure GDA0003947768790000073
Recording the signal frequency band with the energy ratio of 0-90% and the energy ratio of 0-99%, and recording the frequency of the energy band as omega 90 And omega 99 。
Fifthly: gaussian function and Gaussian wavelet family selection
Gaussian wavelets are a family of functions generated from the derivative of the infinite order of gaussian function 6, and the general expression of gaussian function g (t) can be written as:
Figure GDA0003947768790000081
wherein C and alpha are constants. Fourier transform of Gaussian function
Figure GDA0003947768790000082
Still gaussian function 6:
Figure GDA0003947768790000083
the gaussian function 6 has a smooth infinite derivative, and the nth derivative (n is a positive integer) of the gaussian function 6 has the characteristics of wavelet function oscillation and energy limitation, and n gaussian mother wavelets can be generated by performing nth differential operation on the gaussian function 6:
Figure GDA0003947768790000084
for wavelet function g n (t) a series of wavelet functions can be obtained by performing expansion and translation transformation, and the functions are called as a wavelet function family:
Figure GDA0003947768790000085
wherein g is u,s (t) is called a family of Gaussian wavelet functions, s is called a scale parameter, and u is called a translation parameter.
When the smooth result of the noisy time domain impact signal is solved, selecting a Gaussian function 6; when solving the first derivative (speed) of the noisy time domain impact signal, selecting a first-order Gaussian wavelet 7; when solving for the second derivative (acceleration) of the noisy time domain impulse signal, a second order gaussian wavelet 8 is selected.
Sixthly, the method comprises the following steps: signal pseudo-frequency determination and scale parameter selection
The wavelet scale parameters are not equivalent to the more readily understood fourier frequencies in engineering. The signal frequency domain characteristics corresponding to the wavelet scale parameters s can adopt pseudo frequencies f s It is shown that the correspondence of the scale parameter to the pseudo-frequency can be scaled by:
Figure GDA0003947768790000091
wherein Δ t =1/f samp Is the sampling time interval of the time domain signal, f samp To sample frequency, f c ＝ω c The/2 pi is the central frequency of the wavelet, for the time domain impact signal, the frequency component greater than zero is more concerned, and the mother wavelet function g adopted in the wavelet transformation is selected n (t) after, mother wavelet time center t c Angular frequency center omega c Can be calculated from the following formula:
Figure GDA0003947768790000092
seventhly, the method comprises the following steps: convolution operation of time domain noisy signal and Gaussian wavelet
Carrying out convolution operation on the rockfall displacement signal and the Gaussian wavelet to obtain a Gaussian wavelet transform result GWT (u, s) of the time domain noisy signal:
Figure GDA0003947768790000093
wherein
Figure GDA0003947768790000094
Figure GDA0003947768790000095
Is the complex conjugate of g (t). Depending on the differential nature of the convolution operation, the gaussian wavelet transform can ultimately be written in the form:
Figure GDA0003947768790000096
the expression of Gaussian wavelet transform includes d n ／du n And convolution operation f (u) g s (u) of the formula (I). Because the integral of the Gaussian function in the real number domain is not zero, the convolution operation can be interpreted as that the function f is in the kernel function g s (u) in the weighted average smoothing process, the gaussian wavelets with different orders are correspondingly calculated according to the differential of different orders of the original noise-containing signal. By means of wavelet transformation, smoothing of noisy vibration signal f and its n-order differentiation on scale s can be achieved simultaneously.
Eighthly: and (4) performing signal feedback of falling rock movement displacement, speed and acceleration and self-consistency check.
For an actual time domain impact signal, besides the accuracy of the curve shape and the positions of the wave crest and the wave trough, the wavelet transformation differential result also needs to ensure the accuracy of the amplitude of the differential result. Amplitude of wavelet differential result and s of wavelet scale parameter n Correlation, and the normalization processing of the wavelet function family also affects the amplitude result of the differential operation. In order to eliminate the influence of the gaussian wavelet transform process on the amplitude of the differentiation result, the wavelet differentiation operation of the noisy signal is defined as:
Figure GDA0003947768790000101
wherein A is mp Is an amplitude parameter. The numerical differentiation process amplifies the influence of noise interference in the signal, and the numerical integration process plays a role in inhibiting the noise interference in the original signal in the opposite direction. Therefore, the method can be used for measuring the accuracy and the product of the wavelet differential result by performing inverse integration on the wavelet differential result and comparing the coincidence degree with the original signalThe higher the degree of coincidence between the differentiation result and the original signal, the more accurate and reliable the differentiation result can be considered. Because the two time sequences before and after the wavelet transformation have the same length, the overlapping degree between the noisy signal and the integral result of the approximate differential of the wavelet can be conveniently evaluated by adopting the Euclidean distance which is calculated as follows
Figure GDA0003947768790000102
Wherein j represents equidistant sampling of the wavelet differential result of the original time domain signal, m is the number of discrete points of the sampled signal, and n is the wavelet differential order.
Figure GDA0003947768790000103
For wavelet transformation of the result of an n-th order differentiation
Figure GDA0003947768790000104
Integral of the value of (t) k To form an integral node, A k For the multiplication factor, also called the accompanying node t k The right of (c). In particular, when n =0,
Figure GDA0003947768790000105
for noisy signals
Figure GDA0003947768790000106
And a gaussian function g 0 And (t) convolution smoothing the result.
Figure GDA0003947768790000107
Amplitude parameter A mp Can be obtained by iterative solution, as shown in the flow chart 5, when the Euclidean distance between the first integral of the n-order wavelet differential and the n-1-order wavelet differential reaches the minimum value ED min I.e. the amplitude parameter A when the coincidence of the curves is highest mp An optimum value is obtained. Finally, the rockfall displacement 12 fed back by the wavelet method, the rockfall speed 13 fed back by the wavelet method and the wavelet are obtained respectivelyThe method feeds back the falling rock acceleration 14.
In the embodiment, a high-speed video capturing signal of the falling rocks is convoluted with a Gaussian wavelet family by utilizing the property that a Gaussian function has infinite-order smooth conductivity, the approximate differentiation process of wavelet transformation and the principle of resisting noise interference are explained, and a non-contact automatic falling rocks speed and acceleration information identification and feedback program based on the Gaussian wavelet transformation theory is established. And the matching degree between the wavelet transformation processing signal and the real signal is controlled by introducing an amplitude parameter. The wavelet scale parameters are calibrated according to the main frequency band of the time domain captured signal energy distribution. The method effectively solves the problem of high sensitivity of the differential information feedback process of the rockfall capturing signal to noise interference, improves the signal-to-noise ratio level of the impact signal, and avoids the difficult problems of instability and vulnerability of the traditional contact wired equipment for monitoring and recording the high-speed rockfall protection dynamic response signal.
The present invention and its embodiments have been described above schematically, and the description is not intended to be limiting, and what is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. Therefore, if the person skilled in the art receives the teaching, without departing from the spirit of the invention, the person skilled in the art shall not inventively design the similar structural modes and embodiments to the technical solution, but shall fall within the scope of the invention.

Claims (6)

1. The non-contact rockfall protection dynamic response signal automatic identification and feedback method is characterized in that: the method comprises the following steps:
step (1): arranging a camera and recording a high-speed video in an impact process;
step (2): calculating and calibrating a pixel-distance scale factor;
in the step (2), different target objects in the high-speed video picture are selected, and the actual characteristic length of each target object is measured respectively; recording the number of pixels occupied by a selected target object in a high-speed video picture, and calculating a pixel-distance scaling factor according to the one-to-one correspondence between the video picture pixels and the characteristic lengths;
and (3): automatically identifying the falling rock impact motion track;
and (4): analyzing a rockfall displacement time domain signal frequency domain and energy distribution thereof;
and (5): selecting a Gaussian function and a Gaussian wavelet family;
in step (5), the gaussian function is one of the most important functions in the application of probability statistics and random signal analysis, and the expression of the gaussian function g (t) is as follows:
Figure FDA0003947768780000011
wherein C and alpha are constants; fourier transform of Gaussian function
Figure FDA0003947768780000012
Still gaussian function:
Figure FDA0003947768780000013
the Gaussian function has smooth infinite derivative, the n-order derivative of the Gaussian function has the characteristics of wavelet function oscillation and limited energy, and n Gaussian mother wavelets are generated by the n-order differential operation of the Gaussian function:
Figure FDA0003947768780000014
for wavelet function g n (t) performing expansion and translation transformation to obtain a series of wavelet functions, which are called wavelet function families:
Figure FDA0003947768780000015
wherein g is u,s (t) is called Gaussian wavelet function family, s is called scale parameter, and u is called translation parameter;
and (6): determining a signal pseudo frequency and selecting a scale parameter;
in the step (6), the wavelet scale parameters are not equal to Fourier frequency which is easier to understand in engineering; the signal frequency domain characteristic corresponding to the wavelet scale parameter s adopts a pseudo frequency f s The corresponding relation between the scale parameter and the pseudo frequency is expressed by the following formula:
Figure FDA0003947768780000021
wherein Δ t =1/f samp Is the sampling time interval of the time domain signal, f samp To sample frequency, f c ＝ω c The/2 pi is the central frequency of the wavelet, for the time domain impact signal, the frequency component greater than zero is more concerned, and the mother wavelet function g adopted in the wavelet transformation is selected n (t) after, the mother wavelet time center t c Angular frequency center omega c Calculated from the following formula:
Figure FDA0003947768780000022
and (7): performing convolution operation on the rockfall displacement time domain signal and Gaussian wavelets;
and (8): and (3) carrying out signal feedback and self-consistency check on the displacement, the speed and the acceleration of the rockfall.
2. The automatic identification and feedback method for non-contact rockfall protection dynamic response signals according to claim 1, wherein: in the step (1), 2 high-speed cameras are erected according to the installation positions of the flexible protective nets, and the camera positions and the recording targets meet the requirements of different distances and different angles; setting a camera frame rate, and recording the whole impact process; the minimum frame rate of the camera should meet the requirement of the sampling theorem, and the maximum frame rate should ensure that the recording time can be longer than or equal to the duration of the whole impact process.
3. The method of claim 1The non-contact rockfall protection dynamic response signal automatic identification and feedback method is characterized in that: in the step (3), marking a rock falling outline in a video picture, and selecting feature points captured by the motion of the falling rocks by adjusting the contrast and the brightness of the picture; ensuring that the rockfall characteristic points are not completely shielded in the whole impact process; the automatic capture of the rockfall characteristic points and the automatic identification of rockfall motion tracks are realized by combining a motion analysis technology; converting the capture pixel points of the rockfall motion process into rockfall displacement time-domain signals f through the pixel-distance scale factors calculated in the step (2) M [t n ]Wherein N =1,2, \8230, N.
4. The automatic identification and feedback method for non-contact rockfall protection dynamic response signals according to claim 3, wherein: in the step (4), the falling rock displacement time domain signal f obtained in the step (3) is subjected to M [t n ]Performing discrete Fourier transform to obtain Fourier transform result F (omega) of falling rock displacement time domain signal k )：
Figure FDA0003947768780000031
Wherein omega k Is the frequency value of the signal; according to the discrete Fourier transform result of the rockfall displacement time-domain signal, further obtaining the frequency domain energy E of the signal freq ：
Figure FDA0003947768780000032
Recording the signal frequency band with the energy ratio of 0-90% and the energy ratio of 0-99%, and recording the frequency of the energy band as omega 90 And omega 99 。
5. The automatic identification and feedback method for non-contact rockfall protection dynamic response signals according to claim 1, wherein: in the step (7), performing convolution operation on the rockfall displacement time domain signal obtained in the step (2) and the Gaussian wavelet selected in the step (5) to obtain a Gaussian wavelet transform result GWT (u, s) of the rockfall displacement time domain signal:
Figure FDA0003947768780000033
wherein
Figure FDA0003947768780000034
Figure FDA0003947768780000035
Is the complex conjugate of g (t); based on the differential nature of the convolution operation, the gaussian wavelet transform is ultimately written in the form:
Figure FDA0003947768780000036
the above expression shows that the expression for selecting the Gaussian wavelet to perform the wavelet transformation explicitly includes the differential operation d n /du n And convolution operation f (u) g s (u); because the integral of the Gaussian function in the real number domain is not zero, the convolution operation is explained as the function f in the kernel function g s (u) in the weighted average smoothing process, gaussian wavelets with different orders are correspondingly calculated according to differential of different orders of the original rockfall displacement time-domain signal; through wavelet transformation, smoothing of the noisy vibration signal f and its n-order differentiation on the scale s are simultaneously achieved.
6. The non-contact rockfall protection dynamic response signal automatic identification and feedback method according to claim 1, wherein: in the step (8), for the actual time domain vibration signal, the accuracy of the amplitude of the differential result is required to be ensured besides the accuracy of the curve shape and the positions of wave crests and wave troughs of the wavelet transform differential result; from step (7), it is found that the amplitude of the wavelet differential result is s of the wavelet scale parameter n Correlation, and meanwhile, the amplitude result of differential operation is also influenced by the normalization processing of the wavelet function family; is composed ofEliminating the influence of the Gaussian wavelet transform process on the amplitude of the differential result, and introducing an amplitude parameter A mp Defining the wavelet differential operation of the rockfall displacement time-domain signal as:
Figure FDA0003947768780000041
the influence of noise interference in the amplified signal is amplified in the numerical differentiation process, and the noise interference in the original signal is inhibited in the numerical integration process in an inverse view; therefore, the accuracy of the wavelet differential result is measured by performing inverse integration on the wavelet differential result and comparing the coincidence degree of the wavelet differential result and the original signal, and the higher the coincidence degree of the integration result and the original signal is, the more accurate and reliable the result after differentiation is considered; because the two sections of time sequences before and after the wavelet transformation have the same length, the Euclidean distance is adopted to evaluate the coincidence degree between the rockfall displacement time domain signal and the integral result of wavelet approximate differentiation, and the Euclidean distance is calculated as follows:
Figure FDA0003947768780000042
wherein j represents equidistant sampling of wavelet differential results of the original time domain signals, m is the number of discrete points of the sampled signals, and n is the wavelet differential order;
Figure FDA0003947768780000043
for wavelet transformation of the result of an n-th order differentiation
Figure FDA0003947768780000044
Integral of the value of (a), t k To find the product node, A k For the multiplication factor, also called the accompanying node t k The right of (1); when n =0, the number of the bits is set to n =0,
Figure FDA0003947768780000045
shifting time domain signals for rockfall
Figure FDA0003947768780000046
And wavelet function g 0 (t) convolution smoothing results;
Figure FDA0003947768780000047
amplitude parameter A mp Is obtained by iterative solution, when the Euclidean distance between the first integral of the n-order wavelet differential and the n-1 order wavelet differential reaches the minimum value ED min When the coincidence degree of the curves is the highest, the amplitude parameter obtains the optimal value; finally, feeding back a rock fall impact displacement signal through the convolution operation of the rock fall displacement time domain signal and a wavelet function; feeding back a falling rock impact speed signal through a falling rock displacement time domain signal and first-order Gaussian wavelet convolution operation; and feeding back a falling rock impact acceleration signal through the convolution operation of the falling rock displacement time domain signal and a second-order Gaussian wavelet.
CN202210653638.XA 2022-06-09 2022-06-09 Non-contact rockfall protection dynamic response signal automatic identification and feedback method Active CN114898278B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210653638.XA CN114898278B (en) 2022-06-09 2022-06-09 Non-contact rockfall protection dynamic response signal automatic identification and feedback method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210653638.XA CN114898278B (en) 2022-06-09 2022-06-09 Non-contact rockfall protection dynamic response signal automatic identification and feedback method

Publications (2)

Publication Number Publication Date
CN114898278A CN114898278A (en) 2022-08-12
CN114898278B true CN114898278B (en) 2023-01-03

Family

ID=82728015

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210653638.XA Active CN114898278B (en) 2022-06-09 2022-06-09 Non-contact rockfall protection dynamic response signal automatic identification and feedback method

Country Status (1)

Country Link
CN (1) CN114898278B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001352537A (en) * 2000-06-08 2001-12-21 Ntt Data Corp Supervisory system by moving picture
CN110597149A (en) * 2019-10-10 2019-12-20 中国地质环境监测院 Interactive coupling multi-dimensional intelligent collapse and rock fall monitoring system and method
CN113804166A (en) * 2021-11-19 2021-12-17 西南交通大学 Rockfall motion parameter digital reduction method based on unmanned aerial vehicle vision
CN113963512A (en) * 2021-12-22 2022-01-21 四川省交通勘察设计研究院有限公司 Rockfall monitoring system and method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001352537A (en) * 2000-06-08 2001-12-21 Ntt Data Corp Supervisory system by moving picture
CN110597149A (en) * 2019-10-10 2019-12-20 中国地质环境监测院 Interactive coupling multi-dimensional intelligent collapse and rock fall monitoring system and method
CN113804166A (en) * 2021-11-19 2021-12-17 西南交通大学 Rockfall motion parameter digital reduction method based on unmanned aerial vehicle vision
CN113963512A (en) * 2021-12-22 2022-01-21 四川省交通勘察设计研究院有限公司 Rockfall monitoring system and method

Also Published As

Publication number Publication date
CN114898278A (en) 2022-08-12

Similar Documents

Publication Publication Date Title
CN103076602B (en) For the radar self-adaption constant false alarm rate fusion detection method of multiple goal background
Cramer et al. On the analysis of UWB communication channels
CN104407328B (en) Closing space sound localization method based on space impulse response coupling and system
CN105223482B (en) The wavelet decomposition two-value denoising method of partial-discharge ultrahigh-frequency signal waveform
CN102866260B (en) Non-contact river surface flow field imaging measuring method
CN111398910A (en) Radar signal detection method and device, electronic equipment and storage medium
CN106469313A (en) Weak and small target detection method for pipe diameter self-adaptive time-space domain filtering
CN109283576B (en) Method for automatically picking up seismic phase of P wave by taking amplitude as characteristic function
CN105116442A (en) Lithologic oil-gas reservoir weak-reflection seismic signal reconstruction method
CN106569196A (en) Ground-based radar multi-target detection method based on compressed sensing
CN104865570A (en) Rapid dynamic programming track-before-detect method
CN114898278B (en) Non-contact rockfall protection dynamic response signal automatic identification and feedback method
CN109613483B (en) Multi-target track initiation method based on Hough transformation
CN105652319A (en) Estimation method of near-surface stratum Q value of complex mediums
CN105204065A (en) Method and device for picking up preliminary wave
CN112558159B (en) Acoustic logging first arrival pickup method
CN104570074A (en) Waste trace identification method based on SVD (singular value decomposition) technology
US5511008A (en) Process and apparatus for extracting a useful signal having a finite spatial extension at all times and which is variable with time
CN116299712A (en) Unsteady state seismic rule noise suppression method and system
CN108008374B (en) Sea surface large target detection method based on energy median
CN110221289A (en) Object detection method for three coordinate Connectors for Active Phased Array Radar
Liu et al. Practical ultrasonic damage monitoring on pipelines using component analysis methods
CN107589413A (en) A kind of ground high resolution radar target multicycle detection method based on compressed sensing
CN113702937A (en) Double-channel incoherent detection method based on self-adaptive MTI filter
CN116679278B (en) Target radar detection method under strong ground clutter interference

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant