1. A monitoring signal filtering method based on wavelet analysis and threshold processing is characterized by comprising the following steps:

step S1: acquiring signal data s, monitoring and acquiring information monitoring data of the sensors in real time by arranging the sensors, and taking the acquired information monitoring data of the sensors as the signal data s;

step S2: selecting wavelet function to carry out n-layer wavelet decomposition on signal data s to respectively obtain approximate coefficient and n-layer detail coefficient w_jk；

Step S3: the threshold method combining soft and hard threshold compromise and modular processing is utilized to carry out the second stepTo n layers of detail coefficients w_jkCarrying out threshold processing to obtain detail coefficients w 'of all layers'_jkAnd an approximation coefficient;

step S4: utilizing detail coefficient w 'of each layer obtained after treatment'_jkAnd an approximation coefficient, performing wavelet reconstruction on the acquired signal data s to obtain signal data s' with noise suppressed.

2. The monitoring signal filtering method based on wavelet analysis and threshold processing as claimed in claim 1, wherein said information monitoring data in step S1 includes monitoring data of deformation, angle, stress, etc. of bridge structure, high speed railway roadbed, tower, side landslide.

3. The method for filtering monitor signals according to claim 1, wherein the threshold method combining soft and hard threshold compromise and modulo processing in step S3 is implemented by first using a formulaTo obtain the firstA threshold value lambda j of each layer in the n layers is obtained, wherein Nj is the number of detail coefficients of the j layer, and sigma j is the variance of the detail coefficients of the j layer;

then, the threshold value lambda j of each layer is processed by a soft and hard threshold value compromise method to obtain an approximate wavelet coefficient w'_jkThe method for compromising the soft and hard thresholds is,

wherein a is a reduction coefficient, and a is more than or equal to 0 and less than or equal to 1; a threshold value for layer j;

finally, obtaining approximate wavelet coefficient w'_jkPerforming a mold square process by, for example,

then there isObtaining a processed detail coefficient w'_jkAnd approximation coefficients.

4. The method for filtering a monitor signal based on wavelet analysis and thresholding as claimed in claim 1, wherein said step S3 isThe layer detail coefficients all take 0.

Background

With the rapid improvement of economic strength in China, the construction number of bridges, high-speed railway roadbeds, towers and side landslides is increased year by year, and the construction proportion of large and medium bridges is increased year by year. However, accidents such as bridge, high-speed railway roadbed, pole tower side landslide damage and collapse caused by a plurality of factors such as structural degradation, natural factor damage, external force and environmental mutation are increased year by year. Therefore, health conditions of bridges, high-speed railway roadbeds, towers and side landslides need to be monitored, the stress and deformation of the bridges, the high-speed railway roadbeds, the towers and the side landslides are monitored by arranging sensors at the existing bridges, the high-speed railway roadbeds, the towers and the side landslides, and health state information of structures is hidden in health monitoring signals of the bridges, the high-speed railway roadbeds and the towers and the side landslides; and the data monitored by the sensor can not objectively reflect the monitoring and health conditions of bridges, high-speed railway roadbeds and side landslides of towers, and the monitoring effect on preventing debris flow is poor and the monitoring error is large. Therefore, we improve this and propose a monitoring signal filtering method based on wavelet analysis and threshold processing.

Disclosure of Invention

In order to solve the technical problems, the invention provides the following technical scheme:

the invention relates to a monitoring signal filtering method based on wavelet analysis and threshold processing, which comprises the following steps of:

step S1: acquiring signal data s, monitoring and acquiring information monitoring data of the sensors in real time by arranging the sensors, and taking the acquired information monitoring data of the sensors as the signal data s;

step S2: selecting wavelet function to carry out n-layer wavelet decomposition on signal data s to respectively obtain approximate coefficient and n-layer detail coefficient w_jk；

Step S3: the threshold method combining soft and hard threshold compromise and modular processing is utilized to carry out the second stepTo n layers of detail coefficients w_jkCarrying out threshold processing to obtain detail coefficients w 'of all layers'_jkAnd an approximation coefficient;

step S4: after treatment withResulting detail coefficients W 'of the layers'_jkAnd an approximation coefficient, performing wavelet reconstruction on the acquired signal data s to obtain signal data s' with noise suppressed.

As a preferred technical solution of the present invention, the information monitoring data in step S1 includes monitoring data of deformation, angle, stress, and the like of a bridge structure, a high speed railway roadbed, a tower, and a side landslide.

As a preferred technical solution of the present invention, the threshold method combining the soft and hard threshold compromise and the modulus processing in step S3 is implemented by using a formulaTo obtain the firstA threshold value lambda j of each layer in the n layers is obtained, wherein Nj is the number of detail coefficients of the j layer, and sigma j is the variance of the detail coefficients of the j layer;

then, the threshold value lambda j of each layer is processed by a soft and hard threshold value compromise method to obtain an approximate wavelet coefficient w'_jkThe method for compromising the soft and hard thresholds is,

wherein a is a reduction coefficient, and a is more than or equal to 0 and less than or equal to 1; a threshold value for layer j;

finally, obtaining approximate wavelet coefficient w'_jkPerforming a mold square process by, for example,

then there isObtaining a processed detail coefficient w'_jkAnd approximation coefficients.

As one of the present inventionIn a preferred embodiment, the step S3 isThe layer detail coefficients all take 0.

The invention has the beneficial effects that: the monitoring signal filtering method based on wavelet analysis and threshold processing comprises the steps of arranging sensors on a bridge, monitoring and obtaining information monitoring data such as deformation, angles and stress of a bridge structure, a high-speed railway roadbed, a tower and a side landslide in real time, using the obtained information monitoring data such as the deformation, the angles and the stress of the bridge as signal data s, then carrying out primary filtering on the signal data s by using a wavelet decomposition method, leading to monitoring data jumping due to the influence of dynamic load of a vehicle, vibration of surrounding environment and the like during bridge monitoring, decomposing a data curve on different frequency bands by using the wavelet decomposition, leading the obtained coefficient value to be in positive correlation with the similarity of a wavelet function on the frequency scale, carrying out threshold processing on a coefficient smaller than a threshold (indicating that the similarity of the coefficient with the data curve in a corresponding time period is small), and inhibiting, (the amplitude of a part of wavelet coefficient Wj, k corresponding to noise is reduced along with the increase of scale (the larger the scale is, the smaller the frequency is, the smaller the fluctuation is), especially, the higher frequency is suppressed more, the lower frequency is suppressed less, the interference of noise (vibration caused by dynamic load, etc.) can be suppressed more accurately, and the processing of the soft threshold and the hard threshold is combined to make | w'_jkTaking | w between |)_jkL- λ and | w_jkThe numerical value between the | | can make the processed wavelet coefficient closer to a real signal, so that a better denoising effect can be obtained by adjusting the factor alpha (alpha is more than 0 and less than 1); then, the mold square processing is carried out, and the main purpose is to further restrain the part with smaller coefficient, and the part with larger coefficient is contracted or stored because of the wavelet coefficient w 'obtained after the soft and hard threshold processing'_jkThere is still a case of one-off cutting, i.e. wavelet coefficient w'_jkIs not continuous at alpha lambda, the processed signal has oscillation phenomenon, and the larger the scale, the smaller the frequency and the smaller the fluctuation are not considered at the same time) the noise component is reducedThis requires reprocessing of the wavelet coefficients decomposed at each band, so for each layer (different frequency scale) of wavelet coefficients: the smaller value (the noise component is relatively large) is restrained to a larger degree, the larger value (the noise component is relatively small) is restrained to a smaller degree, and meanwhile, the value alpha lambda is processed into a value 0, so that the processed signal data is smoother and more consistent with the original signal. And the monitored signals are further set to reflect the health condition of the bridge.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:

FIG. 1 is a graph of a bridge monitoring curve after signal data s is processed by a monitoring signal filtering method based on wavelet analysis and threshold processing according to the present invention;

FIG. 2 is a graph of signal data monitored by a sensor.

Detailed Description

The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.

Example (b): as shown in fig. 1-2, the monitoring signal filtering method based on wavelet analysis and threshold processing according to the present invention includes steps S1: monitoring and obtaining information monitoring data of the sensors in real time by arranging the sensors, and taking the obtained information monitoring data of the sensors as signal data s;

because in view of the bridge condition, more random high-frequency vibration is often noise, or dynamic load causes, regard the monitoring data who obtains as by signal, noise, outlier three constitute jointly, the noise characteristics: the high frequency coefficient accounts for a large proportion, the low frequency coefficient accounts for a small proportion, and certain specific positions w in the same frequency_jk(the k-th detail coefficient of the j-th layer after wavelet decomposition) has a large value, and these points correspond to the distortion position or important information position of the original signal, and the other part positions w_jkThe k value is small.

Wavelet coefficient w corresponding to noise_jkThe magnitude of the part decreases with increasing scale, typically by finding coefficients below λ that are set to 0 (mainly due to n (k)) and for coefficients above λ (mainly due to the signal) that are retained or shrunk, with a suitable threshold of λ, to obtain an estimated wavelet coefficient w'_jkThen w'_jkAnd (6) carrying out reconstruction.

The information monitoring data in the step S1 includes monitoring data such as deformation, angle, stress, etc. of the bridge structure, the high-speed railway roadbed, the tower, and the side landslide.

Step S2: selecting wavelet function (such as db9) to perform n-layer (such as 8-layer) wavelet decomposition on signal data s to obtain approximate coefficient and n-layer detail coefficient w_jk(ii) a (approximation and detail coefficients represent letters, different data represent symbols different, not specified in unity, where one can distinguish between the words w is the wavelet initial, j represents the jth layer, and k represents the kth coefficient of the jth layer.)

Step S3: the threshold method combining soft and hard threshold compromise and modular processing is utilized to carry out the second stepTo n layers of detail coefficients w_jkCarrying out threshold processing to obtain detail coefficients w 'of all layers'_jkAnd an approximation coefficient;

step S4: utilizing detail coefficient w 'of each layer obtained after treatment'_jkAnd an approximation coefficient, performing wavelet reconstruction on the acquired signal data s to obtain the signal data s after noise suppression.

In step S1, the information monitoring data includes monitoring data such as deformation, angle, stress, etc. of the bridge.

In step S3, the threshold method combining soft and hard threshold compromise and modulo processing is implemented by using a formulaTo obtain the firstA threshold value lambda j of each layer in the n layers is obtained, wherein Nj is the number of detail coefficients of the j layer, and sigma j is the variance of the detail coefficients of the j layer;

then, the threshold value lambda j of each layer is processed by a soft and hard threshold value compromise method to obtain an approximate wavelet coefficient w'_jkThe method for compromising the soft and hard thresholds is,

wherein a is a reduction coefficient, and a is more than or equal to 0 and less than or equal to 1; a threshold value for layer j;

finally obtaining approximate wavelet coefficient w'_jkPerforming a mold square process by, for example,

then there isThe processed detail coefficient is obtainedAnd approximation coefficients.

Wherein, in the step S3Layer detail coefficients are all taken.

Comparing fig. 1 with fig. 2, monitoring data such as deformation, angle and stress of a bridge structure, a high-speed railway roadbed, a tower and a side landslide can be obtained by arranging sensors, and the obtained monitoring data such as the deformation, the angle and the stress of the bridge structure, the high-speed railway roadbed, the tower and the side landslide can be used as signal data s, then, the signal data s is subjected to preliminary filtering by using a wavelet decomposition method, when a bridge is monitored, the monitoring data can jump due to the influence of dynamic load of a vehicle, vibration of surrounding environment and the like, and the monitoring data can be divided by waveletsThe solution can be that the data curve is decomposed on different frequency bands, the obtained coefficient value is in positive correlation with the similarity of the wavelet function on the frequency scale, the coefficient smaller than the threshold (indicating that the similarity with the data curve of the corresponding time period is small) is suppressed by threshold processing, (the amplitude of part of the wavelet coefficient Wj, k corresponding to the noise is reduced along with the increase of the scale (the scale is larger, namely the frequency is smaller, the fluctuation is smaller), especially the higher frequency is suppressed more, the lower frequency is suppressed less, the interference of the noise (vibration caused by dynamic load and the like) can be suppressed more accurately, the over-soft and hard threshold processing is combined to process | w, and | w is processed in a mode of combining with the soft and hard threshold processing_jkTaking | w between |)_jkL- λ and | w_jkThe numerical value between the | | can make the processed wavelet coefficient closer to a real signal, so that a better denoising effect can be obtained by adjusting the factor alpha (alpha is more than 0 and less than 1); then, the mold square processing is carried out, and the main purpose is to further restrain the part with smaller coefficient, and the part with larger coefficient is contracted or stored because of the wavelet coefficient w 'obtained after the soft and hard threshold processing'_jkThere is still a case of one-off cutting, i.e. wavelet coefficient w'_jkThe processed signal is not continuous at the position of alpha lambda, the processed signal also has oscillation phenomenon, and meanwhile, the factor that the noise component is reduced when the scale is larger, namely the frequency is smaller and the fluctuation is smaller, is not considered, so that the wavelet coefficient obtained by decomposition at each frequency band needs to be processed again, and therefore, the wavelet coefficient of each layer (different frequency scales) is processed: the smaller value (the noise component is relatively large) is restrained to a larger degree, the larger value (the noise component is relatively small) is restrained to a smaller degree, and meanwhile, the value alpha lambda is processed into a value 0, so that the processed signal data is smoother and more consistent with the original signal. And the monitored signals are further set to reflect the health condition of the bridge.

Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.