CN114138855A - Multivariable rainfall type landslide hazard monitoring method and system - Google Patents

Multivariable rainfall type landslide hazard monitoring method and system Download PDF

Info

Publication number
CN114138855A
CN114138855A CN202111086192.9A CN202111086192A CN114138855A CN 114138855 A CN114138855 A CN 114138855A CN 202111086192 A CN202111086192 A CN 202111086192A CN 114138855 A CN114138855 A CN 114138855A
Authority
CN
China
Prior art keywords
data
rainfall
historical monitoring
monitoring
historical
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.)
Granted
Application number
CN202111086192.9A
Other languages
Chinese (zh)
Other versions
CN114138855B (en
Inventor
方迎潮
余东亮
蒋毅
王彬彬
王爱玲
吴东容
轩恒
王垒超
杨川
刘宇婷
周广
时建辰
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.)
China Oil and Gas Pipeline Network Corp
National Pipeline Network Southwest Pipeline Co Ltd
Original Assignee
China Oil and Gas Pipeline Network Corp
National Pipeline Network Southwest Pipeline Co Ltd
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 China Oil and Gas Pipeline Network Corp, National Pipeline Network Southwest Pipeline Co Ltd filed Critical China Oil and Gas Pipeline Network Corp
Priority to CN202111086192.9A priority Critical patent/CN114138855B/en
Publication of CN114138855A publication Critical patent/CN114138855A/en
Application granted granted Critical
Publication of CN114138855B publication Critical patent/CN114138855B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2462—Approximate or statistical queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2474—Sequence data queries, e.g. querying versioned data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services

Abstract

The invention provides a multivariable rainfall type landslide hazard monitoring method and a monitoring system, wherein the monitoring method comprises the following steps: collecting historical monitoring data of each physical quantity and historical monitoring data of rainfall of landslide monitoring of a target area; preprocessing historical monitoring data of each physical quantity by taking the historical monitoring data of rainfall as a reference; calculating convolution rainfall data corresponding to historical monitoring data of rainfall, and calculating variable quantity data corresponding to the historical monitoring data of each physical quantity respectively; calculating the correlation between the convolution rainfall data and each variable quantity data; performing significance test, and reserving the variable quantity data meeting significance test conditions in each variable quantity data as a factor for predicting landslide; and performing data dimensionality reduction on the reserved variable data to form a new multi-dimensional comprehensive variable for landslide monitoring. The invention has the advantages of containing various physical quantities closely related to rainfall, solving the problem of hysteresis of reflecting landslide by rainfall data and the like.

Description

Multivariable rainfall type landslide hazard monitoring method and system
Technical Field
The invention relates to the technical field of geological disaster prevention and control, in particular to a multivariable rainfall type landslide disaster monitoring method and system.
Background
Landslide disasters have become the second natural disaster after earthquake disasters and are one of the main geological disaster types in China. Investigation data shows that 90% of landslides are induced by rainfall, and rainfall type landslides have become one of the important disasters affecting human life. For a long time, researchers try to find out the relation between rainfall and landslide instability, and at present, the researches mainly comprise two ideas, namely, obtaining the correlation rule of the rainfall induced landslide based on the statistical analysis of the rainfall induced landslide event, and then obtaining an empirical rainfall threshold model; and the other method is to construct a slope rainfall infiltration stability model based on rainfall infiltration mechanism analysis, and discuss the internal failure mechanism of landslide by means of model tests or numerical simulation and the like.
Due to the complexity of rainfall-induced landslide research, statistical analysis remains the primary means of studying rainfall-induced landslide hazards. Some scholars obtain remarkable results by performing relevant analysis on rainfall induced landslide data, but some problems exist, such as incomplete statistics on factors influencing landslide in the traditional rainfall type landslide monitoring method, usually only rainfall monitoring data is used as a basis for judging landslide, but in actual situations, factors causing landslide events are very complex and diversified; and the rainfall and the landslide have certain hysteresis generally, and the landslide monitoring result is possibly unreliable by directly using the rainfall monitoring data as the basis for judging the landslide.
The traditional rainfall type landslide monitoring method mainly monitors landslide events by establishing a relation model between rainfall and landslide, and when the rainfall reaches a certain threshold value, landslide is considered to occur.
According to the method, in 2019, the Shenyi Fan and the like take the mulberry county in Zhang Jiajie city in Hunan province as a research area, on the basis of comprehensively analyzing rainfall and landslide data of nearly 30 years, statistical analysis research is carried out on the relation between the quantity of landslides and rainfall factors, the effective rainfall intensity threshold of the area inducing the landslide under different probabilities is determined, logistic regression analysis is carried out by utilizing part of sample data, a landslide occurrence probability prediction equation of the research area is obtained, and a rainfall intensity threshold quantitative expression is given.
The factors inducing landslide are many, however, the traditional rainfall type landslide monitoring only depends on rainfall monitoring as the basis for judging landslide, and the relationship between rainfall and landslide has certain hysteresis, so that the model result is unreliable.
Disclosure of Invention
The present invention aims to address at least one of the above-mentioned deficiencies of the prior art. For example, an object of the present invention is to provide a multivariate rainfall-type landslide hazard monitoring method that takes into account not only rainfall intensity but also other physical quantity information related to rainfall. For another example, another object of the present invention is to provide a system for monitoring and monitoring a multi-variable rainfall type landslide disaster, which is efficient and fast, and takes full advantage of information on other physical quantities related to rainfall as well as rainfall intensity.
In order to achieve the above object, an aspect of the present invention provides a multivariate rainfall type landslide hazard monitoring method, including the steps of:
collecting historical monitoring data of each physical quantity and historical monitoring data of rainfall of landslide monitoring of a target area;
preprocessing historical monitoring data of each physical quantity by taking the historical monitoring data of rainfall as a reference;
convolution rainfall data corresponding to the historical monitoring data of the rainfall is calculated through the formula 1,
formula 1 is:
Figure BDA0003265638180000021
wherein, PVolume iThe data is the ith data, mm, in the convolution rainfall data; piThe data is the ith data, mm, in the historical monitoring data of rainfall; pi-jThe data are the ith-j data, mm, in the historical monitoring data of rainfall;
calculating each variable quantity data corresponding to the historical monitoring data of each physical quantity respectively;
the correlation between the convolved rainfall data and each variation data is calculated by equation 2,
the formula 2 is:
Figure BDA0003265638180000022
wherein r is a correlation coefficient and has no dimension; n is the number of rainfall data samples; pVolume iFor the ith data in the convolution rainfall data，mm；
Figure BDA0003265638180000023
Is the average of the convolution rainfall data, mm;
Figure BDA0003265638180000024
standard deviation, mm, of the convolved rainfall data; y isiFor the ith data in the variation data,
Figure BDA0003265638180000031
as an average of the variation data, SyIs the standard deviation of the variation data;
performing significance test, and reserving the variable quantity data meeting significance test conditions in each variable quantity data as a factor for predicting landslide;
and performing data dimensionality reduction on the reserved variable data to form a new multi-dimensional comprehensive variable for landslide monitoring.
In an exemplary embodiment of the present invention, the historical monitoring data of the respective physical quantities may include at least one of the historical monitoring data of soil pressure, slide pile deformation, deep displacement, stress strain.
In an exemplary embodiment of the invention, the historical monitoring data of soil pressure, slide pile deformation, deep displacement, stress strain may include at least one set of data.
In an exemplary embodiment of the present invention, the preprocessing the historical monitoring data of the respective physical quantities may include:
filling up missing data, eliminating data which do not meet the conditions, and replacing the missing data with interpolation of adjacent values to enable the historical monitoring data of each physical quantity and the historical monitoring data of rainfall to be in one-to-one correspondence in the time sequence;
wherein the data not satisfying the condition is data not satisfying formula 3,
formula 3 is:
Figure BDA0003265638180000032
wherein x isiFor the ith data in the historical monitoring data of the physical quantity,
Figure BDA0003265638180000033
is an average value of the historical monitoring data of the physical quantity, and σ is a standard deviation of the historical monitoring data of the physical quantity.
In an exemplary embodiment of the present invention, the variation data corresponding to the historical monitoring data for calculating each physical quantity may be calculated by equation 4,
formula 4 is:
yi＝|xi-xi-1|
wherein, yiFor the ith data in the variance data, xiFor the ith data, x, in the historical monitoring data of the physical quantityi-1The i-1 th data in the historical monitoring data of the physical quantity.
In one exemplary embodiment of the present invention, the significance test may be to exclude variables that do not satisfy equation 5,
formula 5 is:
p_Value＜0.05
wherein, p _ Value is a significance test result.
In an exemplary embodiment of the present invention, the data dimension reduction may include placing the retained delta data into a matrix Y of n rows and p columns:
Figure BDA0003265638180000041
wherein n is more than p, and p represents the number of the variation data; n represents the number of samples per variation data; y isnpRepresenting the nth data in the pth variable;
the matrix Z is obtained by normalizing the matrix Y by equation 6:
formula 6 is:
Figure BDA0003265638180000042
wherein the content of the first and second substances,
Figure BDA0003265638180000043
is the arithmetic mean of the p-th variation data, ynpN-th data, s, of the p-th variation datapIs the standard deviation of the p-th variation data;
Figure BDA0003265638180000044
wherein n is more than p, and p represents the number of the variation data; n represents the number of samples per variation data; znpDenotes ynpNormalized values;
based on the normalized matrix Z, the correlation coefficient matrix R is obtained,
Figure BDA0003265638180000045
where ρ isppThe correlation coefficient of the p column variable and the p column variable in the matrix Z;
eigen equation | R- λ I from correlation coefficient matrix RpObtaining p characteristic values, | 0, sorting corresponding characteristic values according to size, lambda1≥λ2≥…≥λmThe number m of the required new comprehensive variables is determined according to the accumulated contribution degree of more than 85 percent;
wherein the cumulative contribution degree is calculated by equation 7,
formula 7 is:
Figure BDA0003265638180000046
in an exemplary embodiment of the present invention, the forming of the new synthesized variable includes:
determining each eigenvalue lambdajThe unit feature vector corresponding to j 1, 2uj1, 2.. m, wherein uj＝(u1j，u2j，…umj)TAnd then it is used as a transformation matrix, and the data matrix Z is used for right-multiplying the transformation matrix to realize principal component mapping to obtain the formula 8,
formula 8 is:
Figure BDA0003265638180000051
in the formula, w1For the most informative synthetic variables of the original data contained, w2The information amount is the second, and so on.
In an exemplary embodiment of the invention, the collected historical monitoring data of the physical quantity includes 3 historical monitoring data of pipeline strain, 5 historical monitoring data of anti-slide pile deformation, 1 historical monitoring data of deep displacement and 1 historical monitoring data of rainfall, wherein each pipeline strain includes 4-direction strain monitoring of 3 points, 6 points, 9 points and 12 points, 12 historical monitoring data is provided, each deep displacement includes monitoring of depths of 2 meters, 7 meters and 11 meters, 3 historical monitoring data is provided, and 21 historical monitoring data are provided in total.
The invention also provides a multivariable rainfall type landslide hazard monitoring system, which comprises: a processor; and a memory storing a computer program which, when executed by the processor, implements the monitoring method as claimed above.
Compared with the prior art, the beneficial effects of the invention comprise the following:
the invention provides a multivariable selection method for rainfall type landslide monitoring, which is different from the traditional method for analyzing landslide events only by depending on monitoring rainfall, collects historical monitoring data of physical quantities observed by various sensors, screens out physical quantities closely related to rainfall through quantitative analysis, and then extracts main data information by using a linear data dimension reduction method to generate new comprehensive variables for landslide monitoring, so that the original physical quantity monitoring data can be efficiently described, the data processing amount is reduced, and the data processing speed is increased.
Drawings
Fig. 1 shows a schematic flow diagram of an exemplary embodiment of a multivariate rainfall-type landslide hazard monitoring method of the present invention.
Detailed Description
Hereinafter, the multivariate rainfall type landslide hazard monitoring method and monitoring system of the present invention will be described in detail with reference to the accompanying drawings and exemplary embodiments.
Fig. 1 shows a schematic flow diagram of an exemplary embodiment of a multivariate rainfall-type landslide hazard monitoring method of the present invention.
In a first exemplary embodiment of the present invention, as shown in fig. 1, a multivariate rainfall type landslide hazard monitoring method includes the steps of:
the first step is as follows: and collecting historical monitoring data of each physical quantity monitored by the landslide of the target area and historical monitoring data of rainfall. Specifically, historical monitoring data obtained by monitoring various sensors on site in a target area is collected, and the historical monitoring data can comprise physical quantities such as soil pressure, slide pile deformation, deep displacement, stress strain and the like, and historical monitoring data of rainfall. Here, each physical quantity data is different according to different monitoring indexes of different regions and different emphasis, and the data is determined according to the specific situation of the site.
The second step is that: and preprocessing the historical monitoring data of each physical quantity by taking the historical monitoring data of the rainfall as a reference. Specifically, the data preprocessing mainly includes data completion and abnormal value processing of the history data of the remaining physical quantities with reference to the history monitoring data of the rainfall so that the data quantity of each physical quantity is the same as the rainfall data quantity. And (3) data completion: for example, if the historical monitoring data of the strain is returned every 2 hours and the historical monitoring data of the rainfall is returned every 1 hour, the historical monitoring data of the strain needs to be interpolated in time series so that the time interval is also 1 hour. In the case where there is a partial missing of data, the data may be filled up by interpolation. Abnormal value processing: there may be some values (i.e., outliers or noise) that vary too much from one another in the raw observed data, so the raw data needs to be screened. The data screening method can screen data by using the deviation of each data from the average value of the data in a period of time which needs to be smaller than a set value, and in actual data processing operation, a threshold value (multiple of standard deviation) is often set to screen data. And eliminating data which do not meet the conditions, and replacing the data by interpolation of adjacent values. For other variables, the acquisition time may be 0.5 hour once, and the data amount is 2 times of the rainfall data, so that the data needs to be selected.
The third step: and converting historical monitoring data of various physical quantities and rainfall. And calculating convolution rainfall data corresponding to the historical monitoring data of the rainfall through formula 1.
Formula 1 is:
Figure BDA0003265638180000071
wherein, PVolume iThe data is the ith data, mm, in the convolution rainfall data; piFor the ith data, mm, i.e. P, in the historical monitoring data of rainfalliMay be the rainfall data observed the ith time; pi-jThe data are the ith-j data, mm, in the historical monitoring data of rainfall; and represents multiplication.
And calculating the variable quantity data corresponding to the historical monitoring data of each physical quantity. And respectively calculating variation data corresponding to the historical monitoring data of each physical quantity in the historical monitoring data of each physical quantity. Specifically, rainfall (historical monitoring data of rainfall) is one of the most important factors influencing landslide, but the influence of the rainfall on the landslide is hysteretic and attenuating, the total effect of the rainfall on the landslide can be shown after the current rainfall is convolved with the previous rainfall, the influence effect of the rainfall on the day is 100%, the rainfall is attenuated by 80% on the second day, the rainfall is attenuated by 40% on the third day, and the rainfall is attenuated by 0 on the fourth day. And for the physical quantity similar to the strain, the variation can better reflect the action effect of the physical quantity in the landslide event, and for the physical quantity, the variation of the adjacent data is used for describing the variation characteristic in the actual modeling.
The fourth step: screening the physical variation. The correlation between the convolution rainfall data and each variation data is calculated by equation 2.
The formula 2 is:
Figure BDA0003265638180000072
wherein r is a correlation coefficient and has no dimension; n is the number of rainfall data samples; pVolume iThe data is the ith data, mm, in the convolution rainfall data;
Figure BDA0003265638180000073
is the average of the convolution rainfall data, mm;
Figure BDA0003265638180000074
standard deviation, mm, of the convolved rainfall data; y isiFor the ith data in the variation data,
Figure BDA0003265638180000075
as an average of the variation data, SyIs the standard deviation of the variance data.
And (4) carrying out significance test, and keeping the variable quantity data meeting significance test conditions in each variable quantity data as a factor for predicting landslide. For example, the significance test is to exclude variables that do not satisfy equation 5.
Formula 5 is:
p_Value＜0.05
wherein, p _ Value is a significance test result. Of course, the right value of the inequality in equation 5 can be adjusted according to actual conditions.
The fifth step: and performing data dimensionality reduction on the reserved variable data to form a new multi-dimensional comprehensive variable for landslide monitoring. Specifically, the problems of high data dimensionality, information redundancy among data and the like generally exist in variable quantity data corresponding to historical monitoring data of multiple physical quantities passing significance test in the third step, a linear data dimension reduction method is adopted for the problems, main information of the data is extracted under the condition that original variable information is kept as much as possible, the dimensionality of an original data set is reduced, a new multi-dimensional comprehensive variable is formed, and later-stage landslide monitoring is facilitated.
In the present exemplary embodiment, the historical monitoring data of the respective physical quantities may include historical monitoring data of soil pressure, slide pile deformation, deep displacement, stress strain. Further, the historical monitoring data of soil pressure, slide-resistant pile deformation, deep displacement and stress strain can comprise at least one group of data. For example, the collected historical monitoring data of the physical quantity comprises 3 historical monitoring data of pipeline strain, 5 historical monitoring data of slide pile deformation, 1 historical monitoring data of deep displacement and 1 historical monitoring data of rainfall, wherein each pipeline strain comprises 4-direction strain monitoring of 3 points, 6 points, 9 points and 12 points, 12 historical monitoring data are provided, each deep displacement comprises monitoring of depths of 2 meters, 7 meters and 11 meters, 3 historical monitoring data are provided, and the total number of the historical monitoring data of 21 physical quantities is 21.
In the present exemplary embodiment, the preprocessing the historical monitoring data of each physical quantity may include: filling up missing data, eliminating data which do not meet the conditions, and replacing the missing data with interpolation of adjacent values to enable the historical monitoring data of each physical quantity and the historical monitoring data of rainfall to be in one-to-one correspondence in the time sequence;
wherein the data that does not satisfy the condition is data that does not satisfy equation 3.
Formula 3 is:
Figure BDA0003265638180000081
wherein x isiFor the ith data in the historical monitoring data of the physical quantity,
Figure BDA0003265638180000082
is an average value of the historical monitoring data of the physical quantity, and σ is a standard deviation of the historical monitoring data of the physical quantity. Of course, the present inventionWithout being limited thereto, the data culling may be performed by other means, such as an outlier detection method. In the present exemplary embodiment, the respective variation data corresponding to the historical monitoring data for calculating the respective physical quantities may be calculated by equation 4.
Formula 4 is:
yi＝|xi-xi-1|
wherein, yiFor the ith data in the variance data, xiFor the ith data, x, in the historical monitoring data of the physical quantityi-1The i-1 th data in the historical monitoring data of the physical quantity. Of course, the present invention is not limited to this, and the variation amount data corresponding to the history monitoring data of each physical quantity may be obtained by other methods. Such as slope, derivative, etc.
In the exemplary embodiment, the data dimensionality reduction may include placing the retained delta data into a matrix Y of n rows and p columns:
Figure BDA0003265638180000091
wherein n is more than p, and p represents the number of the variation data; n represents the number of samples per variation data; y isnpRepresents the nth data in the pth variable.
The matrix Z is obtained by normalizing the matrix Y by equation 6:
formula 6 is:
Figure BDA0003265638180000092
wherein the content of the first and second substances,
Figure BDA0003265638180000093
is the arithmetic mean of the p-th variation data, ynpN-th data, s, of the p-th variation datapIs the standard deviation of the p-th delta data.
Figure BDA0003265638180000094
Wherein n is more than p, and p represents the number of the variation data; n represents the number of samples per variation data; znpDenotes ynpNormalized values; a correlation coefficient matrix R is obtained based on the normalized matrix Z.
Figure BDA0003265638180000095
Where ρ isppIs the correlation coefficient of the p column and the p column variable in the matrix Z.
Eigen equation | R- λ I from correlation coefficient matrix RpObtaining p characteristic values, | 0, sorting corresponding characteristic values according to size, lambda1≥λ2≥…≥λmThe number m of the required new comprehensive variables is determined according to the accumulated contribution degree of more than 85 percent; .
Wherein the cumulative contribution degree is calculated by equation 7.
Formula 7 is:
Figure BDA0003265638180000101
in an exemplary embodiment of the present invention, the forming of the new synthesized variable includes:
determining each eigenvalue lambdajJ is 1, 2.. times.m corresponding unit feature vector uj1, 2.. m, wherein uj＝(u1j，u2j，…umj)TAnd the matrix is used as a conversion matrix, and the data matrix Z is used for right-multiplying the conversion matrix to realize principal component mapping, thereby obtaining the formula 8.
Formula 8 is:
Figure BDA0003265638180000102
in the formula, w1For the most informative synthetic variables of the original data contained, w2The information amount is the second, and so on.
The above and other objects and/or features of the present invention will become more apparent from the following detailed description taken in conjunction with a typical one of the pipeline region monitoring points.
The first step is as follows: data collection
The physical quantities collected included 3 pipe strains, 5 friction pile deformations, 1 deep displacement and 1 rainfall. Wherein each point pipeline strain comprises strain monitoring of 4 directions including 3 points, 6 points, 9 points and 12 points, and 12 monitoring variables are provided. Each deep displacement included monitoring at depths of 2, 7 and 11 meters, with 3 variables. There are thus 21 total monitoring volumes. Meanwhile, the data of the monitoring time period is from 6 months in 2018 to 11 months in 2019, the hours are taken as the monitoring period, and each variable has 24 monitoring data every day. Here, since the amount of data is too large, it is not shown here, but the subsequent calculation results are given.
The second step is that: data processing
1. And (3) data completion: for missing value portions, it can be filled in by interpolation.
2. Abnormal value processing: for each variable, the abnormal value elimination processing is performed by using the formula 3, and the value beyond the range is an abnormal value and is replaced by the interpolation of the peripheral normal value.
The third step: transformation of physical quantity
3. Physical quantity conversion: the convolution rainfall data is calculated by using equation 1 for the transition of the rainfall data. For other data than the rainfall, the amount of change in the adjacent observed values is obtained by equation 4.
The fourth step: variable screening
Calculating the convolution rainfall P using equation 2Roll of paperAnd (3) correlation with other physical quantity variation y, and carrying out significance test, wherein the test result is shown in table 1:
TABLE 1 convolution rainfall and monitoring physical quantity significance test
Figure BDA0003265638180000111
As can be seen from table 1, the p _ Value of the significance test result of the strain of the monitoring point 1050 is 0.9594, which is much larger than 0.05, so that the correlation between the variation data and the convolution rainfall is not significant, and the variation data cannot be used for the rainfall type landslide monitoring, but other variation data can be used.
The fifth step: data dimension reduction
And performing data dimension reduction on the remaining 19 variable quantity data which pass the significance test in the previous step and 20 variable quantity data which are 1 rainfall data in total. The 20 variance data are placed into a matrix Y of n rows and p columns:
Figure BDA0003265638180000112
wherein n > p, p representing the number of variations; n represents the number of samples per variation; y isnpRepresents the nth data in the pth variable.
The matrix Z is obtained by normalizing the matrix Y by equation 6:
formula 6 is:
Figure BDA0003265638180000113
wherein the content of the first and second substances,
Figure BDA0003265638180000114
is the arithmetic mean of the p-th variation data, ynpN-th data, s, of the p-th variation datapIs the standard deviation of the p-th delta data.
Figure BDA0003265638180000121
Wherein n is more than p, and p represents the number of the variation data; n represents the number of samples per variation data; znpDenotes ynpNormalized value。
A correlation coefficient matrix R is obtained based on the normalized matrix Z.
Figure BDA0003265638180000122
Where ρ isijThe correlation coefficient of the variable of the ith column and the jth column in the matrix Z is shown.
Eigen equation | R- λ I from correlation coefficient matrix Rp20Obtaining 20 characteristic values, sorting the corresponding characteristic values according to the size, and lambda1≥λ2≥…≥λ20More than or equal to 0: 30.62>22.38>12.66>9.64>7.94>6.64>6.11>5.83>5.42>3.21>0.30>0.22>0.14>0.10>0.09>0.06>0.02>0.01>0 is more than or equal to 0, and then the number m of the needed new comprehensive variables is determined to be 7 according to the accumulated contribution degree reaching more than 85 percent.
Wherein the cumulative contribution degree is calculated by equation 7.
Formula 7 is:
Figure BDA0003265638180000123
the 7 comprehensive variables can efficiently describe the original 20-dimensional data and can be used for rainfall type landslide monitoring.
Determining each eigenvalue lambdajUnit feature vector u corresponding to j 1, 2j1, 2, 7, wherein uj＝(u1j，u2j，…u7j)TAnd the matrix is used as a conversion matrix, and the data matrix Z is used for right-multiplying the conversion matrix to realize principal component mapping, thereby obtaining the formula 8.
Formula 8 is:
Figure BDA0003265638180000124
in the formula, w1For the most informative synthetic variables of the original data contained, w2The information amount is the second, and so on.
The invention provides a multivariable selection method for rainfall type landslide monitoring, which is different from the traditional method for analyzing landslide events only by depending on monitoring rainfall, collects historical monitoring data of physical quantities observed by various sensors, screens out physical quantities closely related to rainfall through quantitative analysis, and then extracts main data information by using a linear data dimension reduction method to generate new comprehensive variables for landslide monitoring, so that the original physical quantity monitoring data can be efficiently described, the data processing amount is reduced, and the data processing speed is increased.
The invention also provides a multivariable rainfall type landslide hazard monitoring system. The monitoring system may include a processor and a memory. The memory is for storing a computer program. The computer program is executed by a processor causing the processor to perform the multivariate rainfall type landslide hazard monitoring method according to the invention.
Although the present invention has been described above in connection with the exemplary embodiments and the accompanying drawings, it will be apparent to those of ordinary skill in the art that various modifications may be made to the above-described embodiments without departing from the spirit and scope of the claims.

Claims (10)

1. A multivariable rainfall type landslide hazard monitoring method is characterized by comprising the following steps:
collecting historical monitoring data of each physical quantity and historical monitoring data of rainfall of landslide monitoring of a target area;
preprocessing historical monitoring data of each physical quantity by taking the historical monitoring data of rainfall as a reference;
convolution rainfall data corresponding to the historical monitoring data of the rainfall is calculated through the formula 1,
formula 1 is:
Figure FDA0003265638170000011
wherein，PVolume iThe data is the ith data, mm, in the convolution rainfall data; piThe data is the ith data, mm, in the historical monitoring data of rainfall; pi-jThe data are the ith-j data, mm, in the historical monitoring data of rainfall;
calculating each variable quantity data corresponding to the historical monitoring data of each physical quantity respectively;
the correlation between the convolved rainfall data and each variation data is calculated by equation 2,
the formula 2 is:
Figure FDA0003265638170000012
wherein r is a correlation coefficient and has no dimension; n is the number of rainfall data samples; pVolume iThe data is the ith data, mm, in the convolution rainfall data;
Figure FDA0003265638170000013
is the average of the convolution rainfall data, mm;
Figure FDA0003265638170000014
standard deviation, mm, of the convolved rainfall data; y isiFor the ith data in the variation data,
Figure FDA0003265638170000015
as an average of the variation data, SyIs the standard deviation of the variation data;
performing significance test, and reserving the variable quantity data meeting significance test conditions in each variable quantity data as a factor for predicting landslide;
and performing data dimensionality reduction on the reserved variable data to form a new multi-dimensional comprehensive variable for landslide monitoring.
2. The multivariate rainfall type landslide hazard monitoring method of claim 1, wherein the historical monitoring data for each physical quantity comprises at least one of historical monitoring data for soil pressure, anti-slide pile deformation, deep displacement, stress strain.
3. The multivariate rainfall type landslide hazard monitoring method of claim 2, wherein the historical monitoring data of soil pressure, anti-slide pile deformation, deep displacement, stress strain comprises at least one set of data.
4. The multivariate rainfall type landslide hazard monitoring method of claim 1, wherein the preprocessing of the historical monitoring data for each physical quantity comprises:
filling up missing data, eliminating data which do not meet the conditions, and replacing the missing data with interpolation of adjacent values to enable the historical monitoring data of each physical quantity and the historical monitoring data of rainfall to be in one-to-one correspondence in the time sequence;
wherein the data not satisfying the condition is data not satisfying formula 3,
formula 3 is:
Figure FDA0003265638170000021
wherein x isiFor the ith data in the historical monitoring data of the physical quantity,
Figure FDA0003265638170000022
is an average value of the historical monitoring data of the physical quantity, and σ is a standard deviation of the historical monitoring data of the physical quantity.
5. The multivariate rainfall type landslide hazard monitoring method as claimed in claim 1, wherein each variation data corresponding to each of the historical monitoring data for calculating each physical quantity is calculated by equation 4,
formula 4 is:
yi＝|xi-xi-1|
wherein, yiFor the ith data in the variance data, xiFor the ith data, x, in the historical monitoring data of the physical quantityi-1The i-1 th data in the historical monitoring data of the physical quantity.
6. The multivariable rainfall type landslide hazard monitoring method of claim 1, wherein the significance test is to exclude variables not satisfying equation 5,
formula 5 is:
p_Value＜0.05
wherein, p _ Value is a significance test result.
7. The multivariate rainfall landslide hazard monitoring method of claim 1 wherein the data dimensionality reduction comprises placing the retained delta data into a matrix Y of n rows and p columns:
Figure FDA0003265638170000031
wherein n is more than p, and p represents the number of the variation data; n represents the number of samples per variation data; y isnpRepresenting the nth data in the pth variable;
the matrix Z is obtained by normalizing the matrix Y by equation 6:
formula 6 is:
Figure FDA0003265638170000032
wherein the content of the first and second substances,
Figure FDA0003265638170000033
is the arithmetic mean of the p-th variation data, ynpN-th data, s, of the p-th variation datapIs the standard deviation of the p-th variation data;
Figure FDA0003265638170000034
wherein n is more than p, and p represents the number of the variation data; n represents the number of samples per variation data; znpDenotes ynpNormalized values;
based on the normalized matrix Z, the correlation coefficient matrix R is obtained,
Figure FDA0003265638170000035
where ρ isppThe correlation coefficient of the p column variable and the p column variable in the matrix Z;
eigen equation | R- λ I from correlation coefficient matrix RpObtaining p characteristic values, | 0, sorting corresponding characteristic values according to size, lambda1≥λ2≥…≥λmThe number m of the required new comprehensive variables is determined according to the accumulated contribution degree of more than 85 percent;
wherein the cumulative contribution degree is calculated by equation 7,
formula 7 is:
Figure FDA0003265638170000036
8. the multivariable rainfall type landslide hazard monitoring method of claim 7, wherein said forming new composite variables comprises:
determining each eigenvalue lambdajJ is 1, 2.. times.m corresponding unit feature vector uj1, 2.. m, wherein uj＝(u1j，u2j，…umj)TAnd then it is used as a transformation matrix, and the data matrix Z is used for right-multiplying the transformation matrix to realize principal component mapping to obtain the formula 8,
formula 8 is;
Figure FDA0003265638170000041
in the formula, w1For the most informative synthetic variables of the original data contained, w2The information amount is the second, and so on.
9. The multivariate rainfall type landslide hazard monitoring method as claimed in claim 3, wherein the collected historical monitoring data of the physical quantity comprises historical monitoring data of 3 pipe strains, historical monitoring data of 5 anti-slide pile deformation, historical monitoring data of 1 deep displacement and historical monitoring data of 1 rainfall, wherein each pipe strain comprises strain monitoring of 4 directions including 3 points, 6 points, 9 points and 12 points, 12 historical monitoring data are provided, each deep displacement comprises monitoring of 2 meters, 7 meters and 11 meters in depth, 3 historical monitoring data are provided, and 21 historical monitoring data are provided in total.
10. A multivariable rainfall-type landslide hazard monitoring system, the monitoring system comprising:
a processor; and a memory storing a computer program which, when executed by the processor, implements the monitoring method of any one of claims 1-9.
CN202111086192.9A 2021-09-16 2021-09-16 Multivariable rainfall type landslide hazard monitoring method and monitoring system Active CN114138855B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111086192.9A CN114138855B (en) 2021-09-16 2021-09-16 Multivariable rainfall type landslide hazard monitoring method and monitoring system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111086192.9A CN114138855B (en) 2021-09-16 2021-09-16 Multivariable rainfall type landslide hazard monitoring method and monitoring system

Publications (2)

Publication Number Publication Date
CN114138855A true CN114138855A (en) 2022-03-04
CN114138855B CN114138855B (en) 2022-12-20

Family

ID=80393832

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111086192.9A Active CN114138855B (en) 2021-09-16 2021-09-16 Multivariable rainfall type landslide hazard monitoring method and monitoring system

Country Status (1)

Country Link
CN (1) CN114138855B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115587434A (en) * 2022-09-28 2023-01-10 重庆市二零八地质环境研究院有限公司 Evaluation method for quantitatively evaluating hysteresis effect of landslide deformation caused by rainfall
CN117131369A (en) * 2023-10-27 2023-11-28 福建福昇消防服务集团有限公司 Data processing method and system of intelligent safety management and emergency rescue integrated station

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104299367A (en) * 2014-10-23 2015-01-21 中国科学院、水利部成都山地灾害与环境研究所 Landslide hazard multi-stage comprehensive monitoring and early warning method
CN104318103A (en) * 2014-10-23 2015-01-28 中国科学院、水利部成都山地灾害与环境研究所 Landslide hazard monitoring and early warning rainfall threshold judging method
CN110703360A (en) * 2019-10-16 2020-01-17 西北大学 Three-dimensional effect model for landslide prediction based on rainfall intensity and threshold value

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104299367A (en) * 2014-10-23 2015-01-21 中国科学院、水利部成都山地灾害与环境研究所 Landslide hazard multi-stage comprehensive monitoring and early warning method
CN104318103A (en) * 2014-10-23 2015-01-28 中国科学院、水利部成都山地灾害与环境研究所 Landslide hazard monitoring and early warning rainfall threshold judging method
CN110703360A (en) * 2019-10-16 2020-01-17 西北大学 Three-dimensional effect model for landslide prediction based on rainfall intensity and threshold value

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
林巍等: "湖南慈利县滑坡灾害的临界降雨量阈值研究", 《长江科学院院报》 *
薛凯喜等: "诱发山地公路地质灾害的极端降雨事件辨识模型", 《东华理工大学学报(自然科学版)》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115587434A (en) * 2022-09-28 2023-01-10 重庆市二零八地质环境研究院有限公司 Evaluation method for quantitatively evaluating hysteresis effect of landslide deformation caused by rainfall
CN115587434B (en) * 2022-09-28 2023-06-16 重庆市二零八地质环境研究院有限公司 Evaluation method for quantitatively evaluating hysteresis effect of landslide deformation caused by rainfall
CN117131369A (en) * 2023-10-27 2023-11-28 福建福昇消防服务集团有限公司 Data processing method and system of intelligent safety management and emergency rescue integrated station
CN117131369B (en) * 2023-10-27 2023-12-22 福建福昇消防服务集团有限公司 Data processing method and system of intelligent safety management and emergency rescue integrated station

Also Published As

Publication number Publication date
CN114138855B (en) 2022-12-20

Similar Documents

Publication Publication Date Title
CN114138855B (en) Multivariable rainfall type landslide hazard monitoring method and monitoring system
CN106909756A (en) A kind of rolling bearing method for predicting residual useful life
Aue et al. Delay times of sequential procedures for multiple time series regression models
CN113506121A (en) Analysis method and device for price influence factors
Gitzel Data Quality in Time Series Data: An Experience Report.
CN116597939A (en) Medicine quality control management analysis system and method based on big data
Ji et al. Prediction of cumulative absolute velocity based on refined second-order deep neural network
Chen et al. Spatio‐temporal data mining method for joint cracks in concrete dam based on association rules
Huang et al. A cosine-based correlation information entropy approach for building automatic fault detection baseline construction
CN113095579A (en) Daily-scale rainfall forecast correction method coupled with Bernoulli-gamma-Gaussian distribution
CN112861422A (en) Deep-learning coal bed gas screw pump well health index prediction method and system
CN112668749A (en) Coal mine gas early warning method based on class mark weighting extreme learning machine
CN109002983B (en) Method and computer system for evaluating enterprise investment potential
Lai Fitting power law distributions to data
CN115759263A (en) Strategy effect evaluation method and device based on cause and effect inference
Pazhayamadom et al. Decision Interval Cumulative Sum Harvest Control Rules (DI-CUSUM-HCR) for managing fisheries with limited historical information
Kovářík et al. Implementing control charts to corporate financial management
Maity Frequency Analysis, Risk and Uncertainty in Hydroclimatic Analysis
CN114580472A (en) Large-scale equipment fault prediction method with repeated cause and effect and attention in industrial internet
Curt et al. Approach to improving the quality of data used to analyse dams–illustrations by two methods
CN108197824B (en) High dam service safety space alert domain diagnosis method
Acosta et al. Crisis Probability Curves (CPCs): A model for assessing vulnerability thresholds across space and over time
CN113326983B (en) Data processing system and method for natural gas consumption prediction
CN106294061A (en) Energy intensity Mutability analysis
Aziz PREDICTIVE ANALYSIS OF CLIMATE DISASTER DATA

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