A deformation analysis method and system based on time series

Abstract

The present invention discloses a deformation analysis method and system based on time series, belonging to the field of radio measurement technology, the method comprises the following steps: obtaining time series data; obtaining coherent points according to the time series data; judging whether the coherent points meet the first condition; if so, filtering the signals of the coherent points through a spatial domain filter to obtain a spatial low-frequency signal and a residual signal; filtering the residual signal through a time domain filter to obtain a temporal low-frequency signal; obtaining a nonlinear deformation variable according to a nonlinear phase; obtaining the deformation of the coherent points according to the nonlinear deformation variable and the linear deformation variable. The coherent points are classified according to their features, and for the coherent points meeting the first condition, the nonlinear phase and the linear phase are separated in turn through a spatial domain filter and a time domain filter, and the final deformation of the coherent points is calculated, thereby improving the accuracy of the deformation analysis of the coherent points in complex scenes.

Description

Technical Field

[0001] The invention relates to the field of radio measurement technology, and in particular to a deformation analysis method and system based on time series. Background Art

[0002] During the operation and maintenance of the subway, the deformation of the area along the rail transit line is very complex due to the combined effects of factors such as track load, foundation conditions, construction and maintenance. The deformation characteristics of different coherent point targets in rail transit are different. They may have undergone linear deformation, nonlinear deformation, or even a complex combination of these two types of deformation. This makes it difficult for traditional time series deformation analysis methods to take into account coherent points with multiple deformation modes.

[0003] For the coherent points containing nonlinear deformation, this part of the phase is not only correlated in the time domain, but also correlated in the space domain. This characteristic makes it easy for nonlinear deformation to be misjudged as atmospheric delay phase; and it is not possible to simply use spatial domain filtering to separate the deformation phase.

[0004] Therefore, the coherent point features are classified, the coherent points containing nonlinear deformations are processed, and the nonlinear deformations are calculated. Summary of the invention

[0005] In view of the above technical problems existing in the prior art, the present invention provides a deformation analysis method based on time series, which separates nonlinear deformation and atmospheric delay through multiple filters to improve the accuracy of deformation analysis.

[0006] The invention discloses a deformation analysis method based on time series, comprising the following steps: acquiring time series data; obtaining coherent points according to the time series data; judging whether the coherent points meet the first condition: containing a linear deformation phase and a nonlinear deformation phase; if satisfied, filtering the signal of the coherent points through a spatial domain filter to obtain a spatial low-frequency signal and a residual signal, wherein the spatial low-frequency signal includes a linear deformation phase and a linear deformation amount; filtering the residual signal through a time domain filter to obtain a temporal low-frequency signal, wherein the temporal low-frequency signal includes a nonlinear phase; obtaining a nonlinear deformation amount according to the nonlinear phase; obtaining the deformation of the coherent points according to the nonlinear deformation amount and the linear deformation amount.

[0007] Preferably, if the first condition is not met, the signal of the coherent point is filtered by a spatial domain filter to obtain a spatial low-frequency signal, that is, a linear deformation phase.

[0008] Preferably, the nonlinear deformation variable is obtained based on a linear regression method;

[0009] The nonlinear deformation is expressed as:

[0010] Disp 2 = K+V ×( t-t 0 )

[0011] in, Disp 2 Expressed as a nonlinear deformation variable, K is a zero-order constant term, V is the displacement speed, t Expressed as time, t 0 Indicates the start time.

[0012] Preferably, the deformation of the coherent point is the sum of a nonlinear deformation variable and a linear deformation variable.

[0013] Preferably, the coherence point is located in an area with a larger deformation gradient.

[0014] Preferably, the method for obtaining a spatial low-frequency signal includes: after presetting an input window for spatial domain filtering, applying a filter based on spatial distribution, i.e., a spatial domain filter, to separate the spatial low-frequency signal and the residual signal; the method for obtaining a temporal low-frequency signal includes: after presetting an input window for temporal domain filtering, applying a filter based on time distribution, i.e., a temporal domain filter, to filter out the temporal low-frequency signal from the residual signal to obtain the temporal low-frequency signal.

[0015] The present invention also provides a system for implementing the above deformation analysis method, comprising: an acquisition module, a coherent point classification module, a first filtering module, a nonlinear deformation variable calculation module and a deformation analysis module.

[0016] The acquisition module is used to obtain time series data;

[0017] The coherent point classification module is used to obtain coherent points according to the time series data, and determine whether the coherent points meet the first condition:

[0018] If satisfied, the first filtering module is called, and the first filtering module is used to filter the signal of the coherent point through a spatial domain filter to obtain a spatial low-frequency signal and a residual signal, wherein the spatial low-frequency signal includes a linear deformation phase and a linear deformation amount; and filter the residual signal through a time domain filter to obtain a temporal low-frequency signal, wherein the temporal low-frequency signal includes a nonlinear phase;

[0019] The nonlinear deformation variable calculation module is used to obtain the nonlinear deformation variable according to the nonlinear phase;

[0020] The deformation analysis module is used to obtain the deformation of the coherent point according to the nonlinear deformation variable and the linear deformation variable.

[0021] Preferably, the system further comprises a second filtering module,

[0022] If the first condition is not met, the second filtering module is called, and the second filtering module is used to filter the signal of the coherent point through a spatial domain filter to obtain a spatial low-frequency signal, that is, a linear deformation phase.

[0023] Preferably, the first filtering module includes a spatial domain filtering submodule and a temporal domain filtering submodule.

[0024] The spatial domain filtering submodule is used to apply a filter based on spatial distribution after the input window of the preset spatial domain filtering to separate the spatial low-frequency signal and the residual signal;

[0025] The time domain filtering submodule is used to apply a filter based on time distribution after the input window of the preset time domain filtering to filter out the time low frequency signal from the residual signal to obtain the time low frequency signal.

[0026] Compared with the prior art, the beneficial effects of the present invention are: classification is performed based on the characteristics of coherent points, and for coherent points that meet the first condition, the nonlinear phase and the linear phase are separated in turn through spatial domain filters and time domain filters, and the final deformation of the coherent points is calculated, thereby improving the accuracy of deformation analysis of coherent points in complex scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flow chart of the deformation analysis method based on time series of the present invention;

[0028] Figure 2 It is a system logic block diagram of the present invention;

[0029] Figure 3 is the average sedimentation curve of the first coherence point;

[0030] Figure 4 is the average sedimentation curve of the second coherence point. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0033] A deformation analysis method based on time series, such as Figure 1 , including the following steps:

[0034] Step 101: Obtain time series data. The time series data can be obtained through radar images or related databases.

[0035] Step 102: Obtaining coherent points according to the time series data.

[0036] Based on the PSInSAR technology, coherence points (PS points) and their features can be obtained, where the features include linear deformation phase, DEM error and residual phase.

[0037] Step 103: Determine whether the coherent point satisfies the following first condition: including a linear deformation phase and a nonlinear deformation phase.

[0038] The features of the relevant points can be matched with the first condition, and the basic information of rail transit, leveling point monitoring data and other deformation results can be used to extract auxiliary features for the classification and matching of relevant points.

[0039] If satisfied, execute step 104: filter the signal of the coherent point through a spatial domain filter to obtain a spatial low-frequency signal and a residual signal, wherein the spatial low-frequency signal includes a linear deformation phase and a linear deformation amount. Disp 1 The residual signal includes atmospheric delay, random noise and nonlinear deformation phase.

[0040] Step 105: Filter the residual signal through a time domain filter to obtain a temporal low-frequency signal, which includes a nonlinear deformation and a nonlinear deformation variable. Disp 2 , and filter out the atmospheric delay. The atmospheric delay has the characteristics of low spatial frequency and high temporal frequency.

[0041] Step 106: Obtain the nonlinear deformation according to the nonlinear phase Disp 2 .

[0042] Step 107: According to the nonlinear deformation Disp 2 and linear deformation Disp 1 , obtain the deformation of the coherent point, and immediately obtain the final deformation on the time sequence. Specifically, the nonlinear deformation variable Disp 2 and linear deformation Disp 1 By adding them together, the deformation of the coherent point can be obtained.

[0043] If the first condition is not met, step 108 is performed: the signal of the coherent point is filtered by a spatial domain filter to obtain a spatial low-frequency signal, the spatial low-frequency signal includes a linear deformation phase and a linear deformation phase amount, that is, the deformation amount of the coherent point is obtained. Here, the coherent point does not have a nonlinear phase, so there is no need to consider the step of separating the nonlinear phase.

[0044] The coherent points are classified by their features. For the coherent points that meet the first condition, the nonlinear phase and the linear phase are separated by the spatial domain filter and the time domain filter in turn, and the final deformation of the coherent points is calculated, which improves the accuracy of the deformation analysis of the coherent points in complex scenes. In particular, when the coherent points are located in an area with a large deformation gradient, their deformation features are complicated and have obvious nonlinear deformation.

[0045] In a specific embodiment, the nonlinear deformation variable of the spatial low-frequency and temporal low-frequency signals can be obtained based on a linear regression method. The nonlinear deformation variable is expressed as:

[0046] Disp 2 = K+V ×( t-t 0 )

[0047] in, Disp 2 Expressed as a nonlinear deformation variable, K is a zero-order constant term, V is the displacement speed, t Expressed as time, t 0 The linear regression method is used to separate the nonlinear deformation phase from the residual phase after filtering, and the final average nonlinear displacement can be calculated in the time series.

[0048] The method for obtaining the spatial low-frequency signal includes: presetting an input window of the spatial domain filter, for example, 1200 meters, applying a filter based on spatial distribution, and separating the spatial low-frequency signal and the residual signal. The filter is a prior art, and the parameters of the filter can be set according to empirical values. The frequency of the linear deformation phase is lower than the frequency of the atmospheric delay and the nonlinear deformation, and therefore, can be separated by the spatial domain filter.

[0049] The method for obtaining the temporal low-frequency signal includes: after presetting the input window of the time domain filter, for example, 365 days, applying a filter based on time distribution to filter out the temporal low-frequency signal from the residual signal to obtain the temporal low-frequency signal, so as to filter out the atmospheric delay and part of the random noise.

[0050] It should be pointed out that the features in the coherent points also include random noise, which has no correlation in space and time, so it is not possible to simply use spatial domain filtering to separate the deformation phase.

[0051] In rail transit, settlement is usually used to express it, and the component of deformation in the vertical direction is settlement. Figure 3 The average sedimentation curve of the first coherence point is shown; Figure 4 The average settlement curve of the second coherent point is shown. The horizontal axis is the date / time, expressed in the form of year / month / day, and the vertical axis is the average settlement; both average settlement curves contain linear deformation and nonlinear deformation, which is a comprehensive performance of the addition of the two.

[0052] The present invention also provides a system for implementing the above deformation analysis method, such as Figure 2 As shown, it includes a collection module 1, a coherent point classification module 2, a first filtering module 3, a nonlinear deformation variable calculation module 4 and a deformation analysis module 5.

[0053] The acquisition module 1 is used to obtain time series data;

[0054] The coherent point classification module 2 is used to obtain coherent points according to the time series data, and determine whether the coherent points meet the first condition:

[0055] If the first condition is met, the first filtering module 3 is called, and the first filtering module 3 is used to filter the signal of the coherent point through a spatial domain filter to obtain a spatial low-frequency signal and a residual signal, wherein the spatial low-frequency signal includes a linear deformation phase and a linear deformation amount; and filter the residual signal through a time domain filter to obtain a temporal low-frequency signal, wherein the temporal low-frequency signal includes a nonlinear phase;

[0056] The nonlinear deformation variable calculation module 4 is used to obtain the nonlinear deformation variable of the spatial low-frequency and temporal low-frequency signals based on the linear regression method;

[0057] The deformation analysis module 5 is used to obtain the deformation of the coherent point according to the nonlinear deformation variable and the linear deformation variable.

[0058] If the first condition is not met, the second filtering module 6 is called, and the second filtering module 6 is used to filter the signal of the coherent point through a spatial domain filter to obtain a spatial low-frequency signal, that is, a linear deformation phase.

[0059] The first filtering module 3 includes a spatial domain filtering submodule 31 and a temporal domain filtering submodule 32.

[0060] The spatial domain filtering submodule 31 is used to apply a filter based on spatial distribution after the input window of the preset spatial domain filtering to separate the spatial low-frequency signal and the residual signal;

[0061] The time domain filtering submodule 32 is used to apply a filter based on time distribution after the input window of the preset time domain filtering to filter out the time low frequency signal from the residual signal to obtain the time low frequency signal.

[0062] The present invention classifies the coherent points according to their characteristics, and improves the PS-InSAR time series deformation analysis method according to the categories. A new atmospheric delay and nonlinear deformation separation method is adopted. After the characteristics of the coherent points in rail transit are classified, different spatiotemporal multi-model algorithms are used to process the time series data of the coherent points with different deformation characteristics, so that the time evolution of the surface deformation of the coherent points with different deformation characteristics can be obtained more accurately. The present invention can not only obtain linear deformation and nonlinear deformation, but also obtain the atmospheric delay phase; it is suitable for areas with large deformation gradients.

[0063] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A deformation analysis method based on time series, characterized in that: The following steps are involved: Obtain time series data through radar images; Obtaining coherence points according to the time series data; Determining whether the coherent point satisfies a first condition: including a linear deformation phase and a nonlinear deformation phase; If satisfied, filtering the signal of the coherent point through a spatial domain filter to obtain a spatial low-frequency signal and a residual signal, wherein the spatial low-frequency signal includes a linear deformation phase and a linear deformation amount; The residual signal is filtered by a time domain filter to obtain a time low-frequency signal, wherein the time low-frequency signal includes a nonlinear phase; According to the nonlinear phase, a nonlinear deformation amount is obtained; The sum of the nonlinear deformation variable and the linear deformation variable is the deformation of the coherent point; Based on the linear regression method, the nonlinear deformation variable is obtained; The nonlinear deformation is expressed as: Disp 2 ＝ K+V ×( t-t 0 ) in, Disp 2 Expressed as a nonlinear deformation variable, K is a zero-order constant term, V is the displacement speed, t Expressed as time, t 0 It is indicated as the start time; If the first condition is not met and there is no nonlinear phase, the signal of the coherent point is filtered by a spatial domain filter to obtain a spatial low-frequency signal, wherein the spatial low-frequency signal includes a linear deformation phase and a linear deformation amount.

2. The deformation analysis method according to claim 1, characterized in that: The method for obtaining the spatial low-frequency signal comprises: after presetting the input window of the spatial domain filter, applying a filter based on spatial distribution to separate the spatial low-frequency signal and the residual signal.

3. The deformation analysis method according to claim 2, characterized in that: The method for obtaining a temporal low-frequency signal comprises: after presetting an input window of a time domain filter, applying a filter based on time distribution to filter out the temporal low-frequency signal from the residual signal to obtain the temporal low-frequency signal.

4. A deformation analysis system based on time series, characterized in that: For implementing the deformation analysis method according to any one of claims 1 to 3, the system comprises: an acquisition module, a coherent point classification module, a first filtering module, a nonlinear deformation variable calculation module and a deformation analysis module, The acquisition module is used to obtain time series data; The coherent point classification module is used to obtain coherent points according to the time series data, and determine whether the coherent points meet the first condition: If satisfied, the first filtering module is called, and the first filtering module is used to filter the signal of the coherent point through a spatial domain filter to obtain a spatial low-frequency signal and a residual signal, wherein the spatial low-frequency signal includes a linear deformation phase and a linear deformation amount; and filter the residual signal through a time domain filter to obtain a temporal low-frequency signal, wherein the temporal low-frequency signal includes a nonlinear phase; The nonlinear deformation variable calculation module is used to obtain the nonlinear deformation variable according to the nonlinear phase; The deformation analysis module is used to obtain the deformation of the coherent point according to the nonlinear deformation variable and the linear deformation variable.

5. The deformation analysis system according to claim 4, characterized in that: Also includes a second filtering module, If the first condition is not met, the second filtering module is called, and the second filtering module is used to filter the signal of the coherent point through a spatial domain filter to obtain a spatial low-frequency signal, and the spatial low-frequency signal includes a linear deformation phase and a linear deformation amount.

6. The deformation analysis system according to claim 4, characterized in that: The first filtering module includes a spatial domain filtering submodule and a temporal domain filtering submodule. The spatial domain filtering submodule is used to apply a filter based on spatial distribution after the input window of the preset spatial domain filtering to separate the spatial low-frequency signal and the residual signal; The time domain filtering submodule is used to apply a filter based on time distribution after the input window of the preset time domain filtering to filter out the time low frequency signal from the residual signal to obtain the time low frequency signal.

Images