CN104484572A - Landslide displacement sudden change identification method based on wavelet analysis - Google Patents

Abstract

The invention discloses a landslide displacement sudden change identification method based on wavelet analysis. The method comprises the steps that (1) a landslide real-time accumulation displacement signal is obtained through onsite monitoring; (2) the accumulation displacement signal is subject to piecewise linear interpolation processing; (3) the processed displacement signal is subject to wavelet multi-scale decomposition and reconstruction, and a high-frequency wavelet coefficient under multi-scale is obtained; (4) the change features of the high-frequency wavelet coefficient are analyzed, and the model maximum value of the change features is calculated; (5) the occurrence time of the sudden change of the landslide body is determined. The landslide displacement sudden change identification method can reasonably judge the deformation stage of the landslide, can identify the sudden change point of the landslide entering the accelerated deformation stage especially, and has very fundamental significance in timely capturing landslide premonition, carrying out landslide warning as early as possible and performing engineering treatment.

Description

A kind of sudden change of the landslide displacement based on wavelet analysis recognition methods

Technical field

The invention belongs to a kind of landslide disaster forecasting procedure, be specifically related to a kind of landslide displacement based on wavelet analysis sudden change recognition methods.

Background technology

At present enter to landslide the Study of recognition achievement of accelerating deformation stage catastrophe point and few, traditional recognition methods mainly contains two kinds: a kind of is the stage of the deformation monitoring data on landslide and macroscopical geological analysis and skew prired phenomenon combined to carry out aggregate qualitative differentiation; Another kind is the landslide accumulation displacement monitoring sequential data after foundation filtering process, and the variation characteristic according to rate of deformation carries out rational judgment.First method varies with each individual, and subjectivity is too strong; Second method is only applicable to Landslide Deformation not by the desirable Landslide Deformation situation of any factor interference.But, in practice, due to the interference of rainfall, earthquake, Human dried bloodstains and other enchancement factor, cause the accumulation displacement duration curve complexity that comes down various, wanting to utilize existing method to identify exactly, that landslide enters the catastrophe point accelerating deformation stage is still very difficult.

Wavelet analysis is the powerful carrying out time-varying signal processing closely grown up during the last ten years.It can carry out multiscale analysis to signal, has very strong noise removal function and feature extraction functions, particularly evident to the process advantage of jump signal.Because landslide has multiple dimensioned characteristic in deformation evolutionary process, contain multi-level sudden change, be described as the wavelet analysis of " school microscop ", be particularly suitable for carrying out multiscale analysis, partial analysis and Singularity Analysis to signal.This is undoubtedly for the research of Landslide Deformation sudden change provides new method.At present, wavelet analysis method has achieved plentiful and substantial achievement in research in signal transacting, seismic prospecting, atmospheric science and many nonlinear science field.In hazard forecasting forecast field, landslide, wavelet analysis has just started the concern causing domestic and international associated specialist scholar, and its research method is single, and achievement in research is little, there is obvious limitation.

Summary of the invention

(1) technical matters that will solve

For solving the problem, the present invention proposes a kind of landslide displacement based on wavelet analysis sudden change recognition methods, reasonably can judge the deformation evolution stage of coming down, particularly enter the identification accelerating deformation stage catastrophe point, landslide omen can be caught in time and carry out landslide precaution alarm as early as possible, to carry out project treatment in time.

(2) technical scheme

Based on a landslide displacement sudden change recognition methods for wavelet analysis, in conjunction with the principle of wavelet multi-scale analysis and catastrophe point identification, the basic step concluding the recognition methods of landslide displacement small echo catastrophe point is as follows:

(1) the real-time progressive displacement monitoring signal on landslide is obtained by on-the-spot displacement monitoring;

(2) piecewise linear interpolation process is carried out to landslide accumulation displacement signal;

(3) select the wavelet basis function (as bior1.1 small echo, db5 small echo) meeting regularity conditions, Multiscale Wavelet Decomposition and reconstruct are carried out to the landslide displacement signal after interpolation processing, obtains the high-frequency wavelet coefficient on multiple yardstick;

(4) analyze the variation characteristic of high-frequency wavelet coefficient on different scale, calculate the modulus maximum of wavelet coefficient;

(5) catastrophe point and catastrophe point time of origin is determined.Variation characteristic or the wavelet coefficient modulus maximum of the high-frequency wavelet coefficient on several continuous yardstick should have consistance, just can be defined as possible catastrophe point.

Further, the small echo described in step 3 be within the scope of finite time change and its mean value is the mathematical function of zero, there is frequency and the amplitude of limited duration and sudden change; Satisfy condition small echo be called wavelet or morther wavelet.

Further, the wavelet function described in step 3 be by mother wavelet function ψ (t) after flexible and translation, then have:

${\mathrm{\ψ}}_{a,b}\left(t\right)=\frac{1}{\sqrt{a}}\mathrm{\ψ}\left(\frac{t-b}{a}\right),(a,b\∈R,a\≠0)---\left(1\right)$

In formula (1), ψ _a,bt () is called wavelet basis function, a is scale factor or contraction-expansion factor, and b is shift factor; Wavelet transformation is defined as one group of wavelet basis function ψ _a,bthe inner product of (t) and signal f (t) to be analyzed; By wavelet transformation, can on different scale the local feature of analytic signal; For continuous wavelet transform.

Further, step 3 and the wavelet coefficient described in step 4 are by the definition of multiscale analysis, by f (t) ∈ L ²(R) launch by following Spatial Coupling:

$L^2\left(R\right)=\underset{j=-\∞}{\overset{J}{\mathrm{\Σ}}}{W}_j\⊕V_j---\left(2\right)$

Wherein J is the yardstick of setting arbitrarily, V _jfor metric space, W _jfor yardstick is the wavelet space of j, then

$f\left(t\right)=\underset{j=-\∞}{\overset{J}{\mathrm{\Σ}}}\underset{k=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{d}_{j,k}{\mathrm{\ψ}}_{j,k}\left(t\right)+\underset{k=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{c}_{j,k}{\mathrm{\φ}}_{j,k}\left(t\right)---\left(3\right)$

In formula, c _j,k=< f (t), φ _j,kt () >, is called yardstick expansion coefficient; d _j,k=< f (t), ψ _j,kt () > is called Wavelet Expansions coefficient (corresponding low-frequency component), the 2nd summation is the wavelet details component (corresponding radio-frequency component) on different scale.

Further, the wavelet coefficient modulus maximum described in step 4 calculates and is mainly used in jump signal detection.Jump signal is the signal produced when sign mutation, and sign mutation has the implication of two aspects: one is that the sharply change of displacement and peak value are unusual, and two is sharply changes of frequency.For the signal of the free locality of frequency spectrum, the frequency analysis of given time can be carried out by wavelet transformation, simultaneously again can the situation of change of As time goes on tracking signal frequency.

From the angle of mathematics, the null point of first order derivative of a function corresponds to this Function Extreme Value point, and second order null point reciprocal corresponds to the flex point of this function, i.e. turning point.Therefore, if be the first order derivative or the second derivative that come from some lowpass functions for the wavelet function of wavelet transformation, so the result of wavelet transformation will embody extreme point or the turning point of signal.

Further, come down when entering acceleration deformation stage, the amplitude of the detail signal after its accumulation displacement monitoring signal is decomposed and reconstituted on multiple continuous yardstick all also exists obvious Characteristics of Mutation, and the Characteristics of Mutation of multiple yardstick all has good consistance.The detection of characteristic sum wavelet coefficient module maximum point accordingly, more adequately can identify that landslide enters the catastrophe point accelerating deformation stage.

(3) beneficial effect

The present invention compared with prior art, it has following beneficial effect: a kind of sudden change of the landslide displacement based on wavelet analysis recognition methods of the present invention, the deformation evolution stage on landslide can be differentiated exactly, particularly enter the identification accelerating deformation stage catastrophe point, for catch landslide omen in time and carry out landslide precaution alarm as early as possible and carry out project treatment all tool be of great significance.

Accompanying drawing explanation

Fig. 1 is overall flow schematic diagram of the present invention.

Detail signal schematic diagram after Tu2Shi Baishi township landslide accumulation displacement monitoring signal and continuous 4 Scale Decompositions thereof reconstruct.

Fig. 3 is the detail signal schematic diagram after late Archaean accumulation displacement monitoring signal and continuous 4 Scale Decompositions thereof reconstruct.

Fig. 4 is the detail signal schematic diagram after Danba landslide accumulation displacement monitoring signal and continuous 4 Scale Decompositions thereof reconstruct.

Embodiment

As shown in Figure 1, a kind of landslide displacement based on wavelet analysis sudden change recognition methods, in conjunction with the principle of wavelet multiresolution analysis and catastrophe point identification, the basic step concluding the recognition methods of small echo catastrophe point is as follows:

(1) the real-time progressive displacement monitoring signal on landslide is obtained by on-the-spot displacement monitoring;

(2) piecewise linear interpolation process is carried out to landslide accumulation displacement signal;

(3) select the wavelet basis function meeting regularity conditions, as bior1.1 small echo or db5 small echo, Multiscale Wavelet Decomposition and reconstruct are carried out to the landslide displacement signal after interpolation processing, obtains the high-frequency wavelet coefficient on multiple yardstick;

(4) analyze the variation characteristic of high-frequency wavelet coefficient on different scale, calculate the maximum value of wavelet coefficient;

(5) determine catastrophe point and catastrophe point time of origin, variation characteristic or the wavelet coefficient modulus maximum of the high-frequency wavelet coefficient on several continuous yardstick should have consistance, just can be defined as possible catastrophe point.

Wherein, the small echo described in step 3 be within the scope of finite time change and its mean value is the mathematical function of zero, there is frequency and the amplitude of limited duration and sudden change; Satisfy condition small echo be called wavelet or morther wavelet.

Wherein, the wavelet function described in step 2 be by mother wavelet function ψ (t) after flexible and translation, then have:

${\mathrm{\&psi;}}_{a,b}\left(t\right)=\frac{1}{\sqrt{a}}\mathrm{\&psi;}\left(\frac{t-b}{a}\right),(a,b\&Element;R,a\&NotEqual;0)---\left(1\right)$

In formula (1), ψ _a,bt () is called wavelet basis function, a is scale factor or contraction-expansion factor, and b is shift factor; Wavelet transformation is defined as one group of wavelet basis function ψ _a,bthe inner product of (t) and signal f (t) to be analyzed; By wavelet transformation, can on different scale the local feature of analytic signal.

Wherein, step 3 or the wavelet coefficient described in step 4 are by the definition of multiscale analysis, by f (t) ∈ L ²(R) launch by following Spatial Coupling:

$L^2\left(R\right)=\underset{j=-\&infin;}{\overset{J}{\mathrm{\&Sigma;}}}{W}_j\&CirclePlus;V_j---\left(2\right)$

Wherein J is the yardstick of setting arbitrarily, V _jfor metric space, W _jfor yardstick is the wavelet space of j, then

$f\left(t\right)=\underset{j=-\&infin;}{\overset{J}{\mathrm{\&Sigma;}}}\underset{k=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}{d}_{j,k}{\mathrm{\&psi;}}_{j,k}\left(t\right)+\underset{k=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}{c}_{j,k}{\mathrm{\&phi;}}_{j,k}\left(t\right)---\left(3\right)$

In formula, c _j,k=< f (t), φ _j,kt () >, is called yardstick expansion coefficient; d _j,k=< f (t), ψ _j,kt () > is called Wavelet Expansions coefficient, the 1st summation of formula (3) equal sign right-hand member is that f (t) is at W _jon approximation component, its correspondence be low-frequency component, the 2nd summation is the wavelet details component on different scale, its correspondence be radio-frequency component; For arbitrary function, it can be decomposed into detail section according to formula (3) and large scale approaches part, then large scale is approached the nearly step of part to decompose, so repeat just to obtain approaching partly and detail section on any yardstick (or resolution).

Wherein, catastrophe point described in step 5 is for landslide is when entering acceleration deformation stage, the amplitude of the detail signal after its accumulation displacement monitoring signal is decomposed and reconstituted on 4 continuous yardsticks all also exists obvious Characteristics of Mutation, and the Characteristics of Mutation of 4 yardsticks all has good consistance; Feature accordingly, more adequately can identify that landslide enters the catastrophe point accelerating deformation stage.

As shown in Figure 2, be the small echo catastrophe point identification on smooth type landslide, landslide, Yi Baishi township is example.

This landslide is positioned at Lao Jiehou mountain, Bai Shi township, Beichuan County, Sichuan Province, landslide volume about 3,000,000 m ³.Landslide area belongs to etching structure senior middle school mountain region looks, and the shallow rotten phyllite of main exposure Silurian upper system Mao County group (Smx) and slate, general about the 1000 ~ 1800m of height above sea level, relative relief is about 800m.This Landslide Deformation starts from 1986, and in January, 2007, relevant department implemented monitoring to this landslide, and on April 1st, 2007, landslide entered acceleration deformation stage, and landslide main body glided and plugged the plain boiled water river of landslide leading edge on July 28th, 2007.

According to the Theories and methods (wavelet decomposition selects bior1.1 small echo, and wavelet reconstruction selects db5 small echo) of above-mentioned wavelet recognition catastrophe point, the accumulation displacement monitoring signal of 6# monitoring point, landslide, dialogue assorted township is analyzed.

As can be seen from Figure 2, before the position of time sequence number 84 (the corresponding time is on April 4th, 2007), the amplitude of 4 yardstick detail signals is consistent substantially, and the amplitude of 4 yardstick detail signals is in constantly increasing trend afterwards, imply that landslide enters acceleration deformation stage.This and actual landslide enter the April 1 2007 time of accelerating deformation stage and only differ 3 days.

As shown in Figure 3, be the small echo catastrophe point identification on step change type landslide, for late Archaean.

This landslide is positioned at the domestic north, Xin Tan town of Hubei Province's Zigui County, lower to Three Gorges Dam 26.5km, is the ancient slumped mass of a Polyphase activity.On June 12nd, 1985 Landslide revivification, (10 ~ 30m/s) native stone of gliding at a high speed, destroys Liao Xintangu town, defines famous late Archaean.Due to numerous geologists, early oneself have employed the measures such as monitoring, has successfully carried out prediction to landslide, and landslide relates to no one's injures and deaths in district, thus becomes the model of slide prediction research.

The resurrection of late Archaean experienced by a progressivity process of deformation and failure by quantitative change to qualitative change, and is at the uniform velocity out of shape respectively at entering in August, 1979 and July nineteen eighty-two and accelerates deformation stage.

Utilize principle and the method for above-mentioned wavelet recognition sign mutation point, the monitoring data of displacement of this key landslide monitoring point A3 is analyzed.

As can be seen from Figure 3, time sequence number be 40 (the corresponding time is on August 15th, 1979) position on, there is obvious sudden change in the detail signal of 4 yardsticks in landslide, from the position that time sequence number is 110 (the corresponding time is July 15 nineteen eighty-two), the change frequency of 4 the yardstick detail signals in landslide, amplitude all present the trend of increase, the variation characteristic generation significant change of detail signal.Accordingly, can differentiate that the time entering at the uniform velocity deformation stage of coming down is on August 15th, 1979, entering the time of accelerating deformation stage is July 15 nineteen eighty-two.The time that this is actual in landslide enters at the uniform velocity deformation stage (in August, 1979) and accelerate deformation stage (July nineteen eighty-two) is consistent.

As shown in Figure 4, be the small echo catastrophe point identification on turnover type landslide, for Danba landslide.

This landslide is positioned at county town, Danba, Tibetan Autonomous Prefecture of Garze, Sichuan Province behind, and it is longitudinally about 290 meters, wide 200 ~ 250 meters, average 30 meters of thickness of sliding body, the front and rear edge discrepancy in elevation close to 200 meters, cumulative volume about 200 × 10 ⁴m3.From in August, 2004, slope body starts to occur significantly being out of shape sign.Within 2005,1 month by month correlation department starts monitoring to this landslide.Monitoring result shows, in February, 2005 initial set is accelerated, its average displacement reaches 2-3cm/d, maximum displacement is close to 5cm/d, and create at sliding mass trailing edge arc that width is about 1m and to grow up drawing crack seam, shearing crack and the anterior bulging tension gash of both sides also develop rapidly with through gradually, present the overall situation glided.The overall forward of sliding mass not only makes county town, Danba occur even collapsing destruction in large crack near the house of thousands of square metres of landslide leading edge, and also make more than half county town, Danba become explosive area, about 4600 people were once forced to withdraw explosive area.

In order to ensure Danba County the people's lives and property safety and social stability, relevant departments have organized rapidly professional contingent to carry out the whole day monitoring and warning of 24 hours to landslide on the one hand, take Emergency management engineering measure to carry out consolidation process to sliding mass on the other hand from 20 days February in 2005.By the sandbag of more than 7000 m3 that bank up in landslide leading edge with after landslide middle front part construction 6 row totally 244 prestress anchorage cables (completing construction by the end of April), the deformation velocity of sliding mass is obviously controlled, and slope stability is improved.In order to ensure the long-time stability of sliding mass and living and working in peace and contentment of county town, the Danba people, implement Technology for Comprehensive Landslide Treatment engineering.

As can be seen from Figure 4, Danba landslide is the position (the corresponding time is on February 2nd, 2005) of 12 from time sequence number, the amplitude of each yardstick detail signal increases gradually, until time sequence number is the position (the corresponding time is on February 24th, 2005) of 34, the amplitude of each detail signal reaches maximum, afterwards, amplitude reduces again gradually.

Correspondence on February 2nd, 2005 landslide enters the time of accelerating deformation stage, and landslide emergency repair on February 20th, 2005 comes into effect carrying engineering, and the deformation development trend on correspondence on February 24th, 2005 landslide starts to reverse, and Landslide Stability improves gradually.

Embodiment recited above is only be described the preferred embodiment of the present invention, not limits the spirit and scope of the present invention.Especially, for oscillation mode landslide, after can first adopting Methods for Wavelet Denoising Used to carry out de-noising to landslide displacement duration curve, recycling this patent method carries out catastrophe point identification.Under the prerequisite not departing from design concept of the present invention; the various modification that this area ordinary person makes technical scheme of the present invention and improvement; all should drop into protection scope of the present invention, the technology contents of request protection of the present invention, all records in detail in the claims.

Claims (5)

1., based on a landslide displacement sudden change recognition methods for wavelet analysis, in conjunction with the principle of wavelet multi-scale analysis and catastrophe point identification, the basic step concluding the Wavelets of landslide displacement sudden change is as follows:

(1) the real-time progressive displacement monitoring signal on landslide is obtained by on-the-spot displacement monitoring;

(2) piecewise linear interpolation process is carried out to landslide accumulation displacement signal;

(3) select the wavelet basis function meeting regularity conditions, bior1.1 small echo or db5 small echo, carry out Multiscale Wavelet Decomposition and reconstruct to the landslide displacement signal after interpolation processing, obtain the high-frequency wavelet coefficient on multiple yardstick;

(4) analyze the variation characteristic of high-frequency wavelet coefficient on different scale, calculate the modulus maximum of wavelet coefficient;

(5) determine catastrophe point and catastrophe point time of origin, variation characteristic or the wavelet coefficient modulus maximum of the high-frequency wavelet coefficient on several continuous yardstick should have consistance, just can be defined as possible catastrophe point.

2. a kind of sudden change of the landslide displacement based on wavelet analysis recognition methods according to claim 1, it is characterized in that: the small echo described in step 3 be within the scope of finite time change and its mean value is the mathematical function of zero, there is frequency and the amplitude of limited duration and sudden change; Satisfy condition small echo be called wavelet or morther wavelet.

3. a kind of landslide displacement based on wavelet analysis sudden change recognition methods according to claim 1, is characterized in that: the wavelet function described in step 2 be by mother wavelet function ψ (t) after flexible and translation, then have:

${\mathrm{\&psi;}}_{a,b}\left(t\right)=\frac{1}{\sqrt{a}}\mathrm{\&psi;}\left(\frac{t-b}{a}\right),(a,b\&Element;R,a\&NotEqual;0)$

In formula (1), ψ _a,bt () is called wavelet basis function, a is scale factor or contraction-expansion factor, and b is shift factor; Wavelet transformation is defined as one group of wavelet basis function ψ _a,bthe inner product of (t) and signal f (t) to be analyzed; By wavelet transformation, can on different scale the local feature of analytic signal.

4. a kind of sudden change of the landslide displacement based on wavelet analysis recognition methods according to claim 1, is characterized in that: step 3 and the wavelet coefficient described in step 4 are by the definition of multiscale analysis, by f (t) ∈ L ²(R) launch by following Spatial Coupling:

$L^2\left(R\right)=\underset{j=-\&infin;}{\overset{J}{\mathrm{\&Sigma;}}}{W}_j\&CirclePlus;V_j---\left(2\right)$

Wherein J is the yardstick of setting arbitrarily, V _jfor metric space, W _jfor yardstick is the wavelet space of j, then

$f\left(t\right)=\underset{j=-\&infin;}{\overset{J}{\mathrm{\&Sigma;}}}\underset{k=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}{d}_{j,k}{\mathrm{\&psi;}}_{j,k}\left(t\right)+\underset{k=-\&infin;}{\overset{\&infin;}{\mathrm{\&Sigma;}}}{c}_{j,k}{\mathrm{\&phi;}}_{j,k}\left(t\right)---\left(3\right)$

In formula, c _j,k=< f (t), φ _j,kt () >, is called yardstick expansion coefficient; d _j,k=< f (t), ψ _j,kt () > is called Wavelet Expansions coefficient, the 1st summation of formula (3) equal sign right-hand member is that f (t) is at W _jon approximation component, its correspondence be low-frequency component, the 2nd summation is the wavelet details component on different scale, its correspondence be radio-frequency component; For arbitrary function, it can be decomposed into detail section according to formula (3) and large scale approaches part, then large scale be approached the nearly step of part and decompose, so repeat just to obtain approaching partly and detail section in any yardstick or resolution.

5. a kind of sudden change of the landslide displacement based on wavelet analysis recognition methods according to claim 1, it is characterized in that: the catastrophe point described in step 5 is for landslide is when entering acceleration deformation stage, the amplitude of the detail signal after its accumulation displacement monitoring signal is decomposed and reconstituted on 4 continuous yardsticks all also exists obvious Characteristics of Mutation, and the Characteristics of Mutation of 4 yardsticks all has good consistance; Feature accordingly, more adequately can identify that landslide enters the catastrophe point accelerating deformation stage.