CN-121980250-A - Geological settlement monitoring data feature extraction method, system, equipment and medium

CN121980250ACN 121980250 ACN121980250 ACN 121980250ACN-121980250-A

Abstract

The invention discloses a geological settlement monitoring data characteristic extraction method, a system, equipment and a medium, wherein the method comprises the following steps of obtaining an instantaneous monitoring sequence in a monitoring time window; the instantaneous monitoring sequence is decomposed into a plurality of intrinsic mode functions by a first-level decomposition algorithm, the decomposed plurality of intrinsic mode functions are classified to identify a plurality of mode feature types, corresponding preset type features are obtained by calculating the plurality of mode feature types, and the obtained preset type features are used as independent features to be input into a downstream processing model. According to the invention, a first-level decomposition algorithm is adopted to deconstruct the mixed original instantaneous monitoring sequence into a plurality of independent eigen mode functions respectively representing long-term trend, periodic fluctuation and sudden disturbance through introducing a technical route of decomposition before calculation, and then, the characteristics are respectively calculated on the pure modes with definite physical meaning and mutual independence.

Inventors

LIU ZHUOYA
ZHANG QILI
WU JIANRONG
ZENG RONG
CHEN CHEN
DING JIANGQIAO
HUANG JUNKAI
WEN YI
FAN QIANG
YANG TAO
YE HUAYANG
ZHANG YANG
CHEN JIASHENG
LUO XIN
Yuan Xianmei

Assignees

贵州电网有限责任公司

Dates

Publication Date: 20260505
Application Date: 20251218

Claims (10)

1. The geological settlement monitoring data characteristic extraction method is characterized by comprising the following steps of: Acquiring an instantaneous monitoring sequence in a monitoring time window; decomposing the instantaneous monitoring sequence into a plurality of eigenmode functions by adopting a first-level decomposition algorithm; classifying the decomposed multiple intrinsic mode functions to identify multiple mode feature types; calculating a plurality of modal feature types to obtain corresponding preset type features; And inputting the obtained preset type of features as independent features into a downstream processing model.
2. A method of extracting features from geologic settlement monitoring data as defined in claim 1, wherein decomposing the transient monitoring sequence into a plurality of eigenmode functions comprises: And decomposing the instantaneous monitoring sequence into a preset number of eigenmode functions by solving a constraint variation problem through a first-level decomposition algorithm until the sum of bandwidths of the eigenmode functions is minimum and the sum of the eigenmode functions can reconstruct the instantaneous monitoring sequence.
3. The method for extracting features of geological subsidence monitoring data of claim 2, further comprising, prior to employing the first decomposition algorithm: optimizing the modal decomposition number and the secondary penalty factor of the primary decomposition algorithm by adopting a genetic algorithm; The fitness function of the genetic algorithm is the sum of the envelope spectrum entropy of all eigenvalue functions after the minimization decomposition.
4. A method of extracting features from geologic settlement monitoring data as defined in claim 3, wherein optimizing the modal decomposition number and the quadratic penalty factor of the first-order decomposition algorithm comprises: Coding the modal decomposition number and the secondary penalty factor as chromosomes of a genetic algorithm, and initializing a population; evaluating each chromosome in the population according to the fitness function; performing selection, crossing and mutation operations on the population to generate a new generation population; judging whether the termination condition is met, returning to the step of evaluating according to the fitness function when the termination condition is not met, and outputting the current optimal modal decomposition number and the secondary penalty factor when the termination condition is met.
5. A method of extracting features from geologic settlement monitoring data as defined in claim 4, wherein the step of classifying the decomposed plurality of eigenmode functions comprises: Calculating the center frequency of each eigenmode function, and calculating the spearman class correlation coefficient of each eigenmode function and the time vector; identifying an eigenmode function with the lowest center frequency and the absolute value of the spearman grade correlation coefficient larger than a preset trend threshold as a trend mode; Calculating a normalized Fourier spectrum for the rest eigenmode functions, calculating the spectral kurtosis, and identifying the eigenmode functions with the spectral kurtosis larger than a preset period threshold as period modes; The remaining unclassified eigenmode functions are identified as anomalous modes.
6. A method of extracting a feature of geological subsidence monitoring data as defined in claim 5, wherein the step of calculating a plurality of modal feature types comprises: calculating a first preset type characteristic through the trend mode, wherein the first preset type characteristic is used for representing the long-term change trend of the instantaneous monitoring sequence; and calculating a second preset type characteristic through the abnormal mode, wherein the second preset type characteristic is used for representing the sudden disturbance of the instantaneous monitoring sequence.
7. The method of claim 6, wherein the modal feature types include trend modalities, periodic modalities, and anomaly modalities.
8. A geological settlement monitoring data feature extraction system employing the method of any one of claims 1-7, comprising: The sequence acquisition module is used for acquiring an instantaneous monitoring sequence in a monitoring time window; The signal decomposition module is used for decomposing the instantaneous monitoring sequence into a plurality of intrinsic mode functions by adopting a first-level decomposition algorithm; The modal classification module is used for classifying the decomposed multiple eigen-modal functions and identifying multiple modal feature types; the feature calculation module is used for calculating a plurality of modal feature types to obtain corresponding preset type features; And the feature output module is used for inputting the obtained preset type features as independent features into a downstream processing model.
9. An electronic device, comprising: A memory and a processor; The memory is configured to store computer executable instructions that, when executed by the processor, implement the steps of the geological settlement monitoring data feature extraction method of any one of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which when executed by a processor perform the steps of the geological settlement monitoring data feature extraction method of any one of claims 1 to 7.

Description

Geological settlement monitoring data feature extraction method, system, equipment and medium Technical Field The invention relates to the technical field of geological settlement observation, in particular to a geological settlement monitoring data characteristic extraction method, a geological settlement monitoring data characteristic extraction system, geological settlement monitoring data characteristic extraction equipment and a geological settlement monitoring data characteristic extraction medium. Background In fields such as structural health monitoring and geological disaster warning, it is often necessary to develop and analyze long-span monitoring data in order to extract key feature information reflecting system state changes. These features are typically used to input subsequent machine learning or deep learning models for state assessment or risk prediction. The linear fitting 'sedimentation trend' of the data in the sliding time window, the corresponding 'abnormal fluctuation' and other various characteristic extraction and analysis means are adopted, so that the long-term development trend and the short-term severe fluctuation of the data can be reflected. Wherein the sedimentation trend can be represented by linear fitting slope of data in a window, and the like, and the abnormal fluctuation can be represented by variance or standard deviation of data in the window, and the like. However, this method of computing features directly on the original mixed signal has inherent technical contradictions. For example, in some schemes, an "abnormal surge" is used to calculate a confidence weight and this weight is used to adjust the "sedimentation trend characteristic value". The logic is that when the fluctuation is large, the credibility of the trend feature should be reduced. Such logic may fail in certain scenarios. For example, a real and severe geologic settlement event may appear on the monitored data as both a large trend and a strong fluctuation. At this time, the increased "abnormal fluctuation feature value" may erroneously decrease the reliability weight, but rather impair the recognition of this true settlement event, which may lead to missing report, i.e., insufficient robustness. On the other hand, for an early sedimentation with very weak and gentle change, the abnormal fluctuation characteristic value is very small, the credibility weight is very high, but the sedimentation trend characteristic value is very small, and even after weighting, the early warning can not be triggered sufficiently, so that insensitivity to details is reflected. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a geological settlement monitoring data feature extraction method, a geological settlement monitoring data feature extraction system, geological settlement monitoring data feature extraction equipment and a geological settlement monitoring data feature extraction medium, which solve the problems that the existing detection and monitoring modes are limited in time, the machine set is required to be cooled and prepared in advance, the time and material resource cost is high, meanwhile, the detection result is static data, the wall thickness change and the service life duration cannot be predicted in real time, and corresponding visual early warning is carried out. In order to solve the technical problems, the invention provides the following technical scheme: The invention provides a geological settlement monitoring data feature extraction method, which comprises the following steps of obtaining an instantaneous monitoring sequence in a monitoring time window, decomposing the instantaneous monitoring sequence into a plurality of intrinsic mode functions by adopting a first-level decomposition algorithm, classifying the decomposed plurality of intrinsic mode functions to identify a plurality of mode feature types, calculating the plurality of mode feature types to obtain corresponding preset type features, and inputting the obtained preset type features as independent features into a downstream processing model. The method for extracting the characteristics of the geological settlement monitoring data is a preferable scheme, wherein the step of decomposing the instantaneous monitoring sequence into a plurality of eigenmode functions comprises the steps of decomposing the instantaneous monitoring sequence into a preset number of eigenmode functions through a first-stage decomposition algorithm by solving a constraint variation problem until the sum of bandwidths of the eigenmode functions is minimum and the sum of the eigenmode functions can reconstruct the instantaneous monitoring sequence. The optimization method has the beneficial effects that the constraint variation problem is solved for decomposition, so that the minimum sum of bandwidths of all the eigenmode functions can be ensured, al