CN117272232A - Tunnel monitoring method and device for data fusion, computer equipment and storage medium - Google Patents
Tunnel monitoring method and device for data fusion, computer equipment and storage medium Download PDFInfo
- Publication number
- CN117272232A CN117272232A CN202311391366.1A CN202311391366A CN117272232A CN 117272232 A CN117272232 A CN 117272232A CN 202311391366 A CN202311391366 A CN 202311391366A CN 117272232 A CN117272232 A CN 117272232A
- Authority
- CN
- China
- Prior art keywords
- data
- monitoring
- tunnel
- formula
- prediction
- 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.)
- Pending
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 101
- 238000000034 method Methods 0.000 title claims abstract description 56
- 230000004927 fusion Effects 0.000 title claims abstract description 23
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000007499 fusion processing Methods 0.000 claims abstract description 7
- 238000007500 overflow downdraw method Methods 0.000 claims abstract description 7
- 238000004458 analytical method Methods 0.000 claims description 15
- 239000011435 rock Substances 0.000 claims description 13
- 229910000831 Steel Inorganic materials 0.000 claims description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 12
- 210000002569 neuron Anatomy 0.000 claims description 12
- 239000010959 steel Substances 0.000 claims description 12
- 238000012806 monitoring device Methods 0.000 claims description 10
- 150000001875 compounds Chemical class 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 9
- 230000035772 mutation Effects 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 125000003275 alpha amino acid group Chemical group 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 238000003672 processing method Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 230000007774 longterm Effects 0.000 claims description 4
- 102100038736 Histone H3.3C Human genes 0.000 claims description 3
- 101001031505 Homo sapiens Histone H3.3C Proteins 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 238000006467 substitution reaction Methods 0.000 claims description 3
- 238000012876 topography Methods 0.000 claims description 3
- 238000009736 wetting Methods 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 12
- 238000004891 communication Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 230000005236 sound signal Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention provides a tunnel monitoring method, a device, computer equipment and a storage medium for data fusion, which comprise the following steps: s1, acquiring original data of tunnel deformation monitoring; s2, preprocessing the original data according to the type of the original data; s3, carrying out weighted fusion processing on the preprocessed data; s4, inputting the obtained original data into a training model to obtain prediction data at a prediction moment; s5, obtaining tunnel deformation strength at a prediction moment according to the prediction data; s6, comparing the tunnel deformation strength at the predicted moment with an early warning threshold value. By processing and fusing different data sources, the technology can comprehensively monitor and early warn tunnel deformation, improve the safety and sustainable operation of the tunnel, and ensure that factors such as data quality, a data fusion method and an early warning model are important so as to ensure the accuracy and reliability of a monitoring and early warning system.
Description
Technical Field
The present invention relates to the field of tunnel engineering prevention technologies, and in particular, to a method and apparatus for monitoring a tunnel by data fusion, a computer device, and a storage medium.
Background
With the continuous increase of subway, high-speed rail and expressway mileage in China, various mountain bodies, underground tunnels are rapidly increased, external conditions such as bad geologic bodies bring larger risks to the safe operation of the tunnels, tunnel deformation monitoring is more and more paid attention, currently, with the rapid development of underground engineering technology and construction scale, tunnel engineering built on weak broken surrounding rocks or bad geological sections is continuously increased, the load of a tunnel structure is changed due to peripheral engineering disturbance or natural condition evolution, larger deformation and even cracking are induced, and the bearing safety and the service performance of the structure are seriously affected. However, traditional tunnel deformation monitoring needs to be carried out manual monitoring on tunnel monitoring points in a tunnel operation window, and can not effectively capture micro-deformation lines inside a tunnel, so that risks brought by tunnel micro-deformation cannot be estimated, accurate early warning means for the safety of a full-section structure of the tunnel are lacked in time, and a certain risk is brought to the safe operation of the tunnel.
Disclosure of Invention
The invention aims at the technical problems and provides a tunnel monitoring method, a device, computer equipment and a storage medium for data fusion.
The technical scheme adopted for solving the technical problems is as follows:
as shown in fig. 1, a tunnel monitoring method for data fusion includes the following steps:
s1, acquiring original data of tunnel deformation monitoring;
s2, preprocessing the original data according to the type of the original data;
s3, carrying out weighted fusion processing on the preprocessed data;
s4, inputting the obtained original data into a training model to obtain prediction data at a prediction moment;
s5, obtaining tunnel deformation strength at a prediction moment according to the prediction data;
s6, comparing the tunnel deformation strength at the predicted moment with an early warning threshold value.
In step S1, the raw data includes sensor data, image data, and topography data.
In step S2, the preprocessing is a wavelet analysis processing method, and the wavelet analysis processing method includes denoising, normalizing and normalizing the original data.
And after the monitoring data subjected to wavelet analysis processing are generated, weighting the subsequences of the monitoring data by an entropy weight method.
In step S3, the weighted fusion process performs weighted and fused calculation on the data by a weighted fusion method, and the calculation formula is as follows:
y=w1×1+w2×2+w3×3+w4×4+w5×5+w6×6+w7×7+w8×8+w9×9 (one)
In the formula (I), Y is the deformation strength of the finally fused tunnel; x1, x2, x3, x4, x5, x6, x7, x8 and x9 are respectively deformation of a data source observation point, internal force of a steel support, ground subsidence, underground fissures, internal force of concrete, surrounding rock pressure, face information, face joint information and long-term data of observation point distribution; w1, w2, w3, w4, w5, w6, w7, w8 and w9 are weight settings corresponding to x1, x2, x3, x4, x5, x6, x7, x8 and x9, respectively.
In step S4, the training model includes the following steps:
h1, acquiring multiple groups of monitoring data of deformation strength of the tunnel in a preset period, and preprocessing the multiple groups of monitoring data;
h2, constructing a data set by the preprocessed monitoring data, and dividing the data set into a training set and a testing set according to a preset proportion;
h3, constructing a GCN-DE-LSTM model, and training and testing the GCN-DE-LSTM model through a training set and a testing set to obtain a prediction model;
firstly, performing feature extraction by using a graph roll neural network (GCN) to establish direct connection or indirect connection between observation points, wherein the construction of the CN-DE-LSTM model comprises the steps of;
h3.1, matrix W: the matrix comprises deformation of observation points, steel support internal force, ground subsidence, underground cracks, concrete internal force, surrounding rock pressure, face information, face joint information and data of observation point distribution;
h3.2, the degree of the node is: l (L) i,i ＝∑ j C j,i C j,i ；
The values of H3.3, adjacency matrix are: l (L) i,i ＝eiej→＝∑ k C k,i C k,j ；
And H3.4, an image operator is as follows: l (G) =c T C＝L i,i -Li,j；
H3.5, laplace processing is as follows:
in the formula (II), A is a characteristic value of L (G); i=u T U, U is the feature vector of L (G);
in the formula (III), the compound of the formula (III),is an element value;
h3.6, loop product processing: gx=u (U T g·U T x) (IV)
In the formula (IV), U is a characteristic matrix of L, U T X is the Fourier transform of the filter operator, U T g is the Fourier transform of the Laplace operator;
h3.7, graph convolution is over the fourier domain as follows:
U T L＝AU Λ
g θ (Λ)＝diay(θ)；
g θ *x＝Ug θ U T x
the eigenvalues of the substitution L are as follows:
in the formula (five), the amino acid sequence of the compound,is the maximum value of lambda in L;
h3.8, normalization treatment as follows:
the expression of the graph convolution is as follows:
in the formula (six), sigma is an activation function, H is corresponding to x, and W is corresponding to theta;
after the characteristics of the observation points are extracted, the observation points are summarized and summarized by repeating the step H as a new set, and then connection among the observation points is established;
substituting the data for establishing the connection into an LSTM model, wherein the LSTM model comprises four states which are respectively: the time of output value, the time of output value leading value, the time of memory neuron state and the leading value of memory neuron state all comprise three gates: an input gate, an output gate, and a forget gate;
the LSTM model realizes modeling of time series, and the principle steps are as follows:
c1, an input formula defining a single neuron is as follows:
x t = (G, H, W, F, N, Z (j, s, r), L) (seven)
In the formula (seven), x t For influencing factor single input set, y is deformation, G is steel support internal force, H is concrete internal force, W is surrounding rock pressure, F is ground subsidence, N is underground crack, Z is face information, j is face joint information, s is face rock type information, r is face wetting degree information, and L is observation point distribution;
c2, searching the super parameters of iteration times, learning rate and hidden layer neuron node number which are manually determined and input in the network, and searching the super parameters through a differential evolution algorithm;
the differential evolution algorithm comprises the following steps:
c2.1, initializing population
In formula (eight), x i,j (0) The ith individual j is the jth dimension, x i Representing a super-parameter set of the LSTM model;
in the formula (III), the amino acid sequence of the compound (III),and->A lower bound and an upper bound of the j-th dimension respectively;
c2.2, differential variation
V i (g+1)＝X rl (g)+F(X r2 (g)-X r3 (g) (ten)
In the formula (ten), r1 and r2 are random numbers between 1 and NP, F is a scaling factor, and g represents the g generation;
c2.3, cross
In the formula (eleven), CR is the crossover probability;
c2.4 greedy selection
And C2.5, adding an adaptive mutation operator into the algorithm for optimizing the scaling factor, wherein the adaptive mutation operator comprises the following steps:
F＝F 0 ·λ 2 (twelve)
In formula (twelve), F 0 For mutation operator, G max Is the maximum algebra of evolution.
The monitoring device includes:
and a data processing module: the method comprises the steps of carrying out noise elimination on random errors and interference factors involved in an original time sequence of each monitoring point through a wavelet analysis method, and carrying out standardization and normalization processing on data;
the entropy weight method processing module: the method comprises the steps of carrying out sequence fusion on denoising time of the same type of a plurality of different monitoring points, and finally fusing the denoising time into a subsequence of a corresponding monitoring index;
and a prediction module: the method comprises the steps of collecting monitoring data of tunnel deformation at a previous moment, inputting the monitoring data of tunnel deformation at the current moment into a prediction model, and obtaining a prediction result of the prediction model on tunnel deformation at a preset moment;
and a judging module: and the method is used for carrying out weighted summation on the predicted data of the tunnel deformation and the corresponding weight, then calculating to obtain the tunnel deformation strength at the preset moment, and judging whether the tunnel deformation strength at the preset moment exceeds the early warning threshold value.
For storing a computer program; the processor is used for implementing the steps of the tunnel monitoring method for data fusion when executing the computer program.
The computer program, when executed by a processor, performs the steps of a tunnel monitoring method for data fusion.
The beneficial effects of the invention are as follows:
1. the GCN-DE-LSTM model learns the characteristic representation of different layers from the time sequence to more comprehensively express the information in the sequence and better process the trend, periodicity and noise of the sequence, so that the prediction precision of future tunnel deformation monitoring data is improved, meanwhile, the weight duty ratio of various monitoring data is determined by adopting an entropy weight method and a weighting fusion method, the tunnel deformation strength is calculated, the advance prediction of the tunnel deformation strength is realized, thereby helping a road management department to carry out corresponding control measures, reducing the influence of tunnel deformation risks on road traffic, effectively improving the prediction precision of the tunnel deformation data and providing important early warning and decision support for tunnel traffic management.
2. And (3) monitoring a sensor: installing strain monitoring sensors at key positions such as pile foundations, steel supports and the like of the tunnel structure, and monitoring and recording deformation data in real time; satellite remote sensing monitoring: remotely monitoring ground subsidence conditions through satellite images and remote sensing technology, and acquiring deformation data by combining GPS positioning technology and other technologies; microseismic monitoring: and by embedding a microseismic monitoring instrument near the tunnel, the change of the underground water level and the crack expansion condition are observed, and real-time data of the change of the internal environment of the tunnel are provided.
Gcn feature extraction: extracting features from sensor monitoring data through a graph convolutional neural network (GCN), establishing a direct or indirect connection relation graph between observation points, capturing the relevance of tunnel deformation, and optimizing DE super parameters: optimizing the super parameters of the model through a differential evolution algorithm (DE) to further improve the prediction precision and accuracy; LSTM model application: and introducing the data related to the previous establishment into a long-short-term memory (LSTM) model, and performing sequence prediction analysis to realize the prediction of the tunnel deformation process and finish the operation.
4. The terminal equipment and the readable storage medium are used for displaying the monitoring result and storing the data for a long time, so that accidents can be effectively prevented, the safety level of tunnel operation can be improved, and an important reference basis is provided for maintenance and management.
Drawings
FIG. 1 is a schematic flow chart of a tunnel deformation monitoring and early warning method of the invention;
FIG. 2 is a schematic flow chart of the GCN-DE-LSTM model;
FIG. 3 is a basic unit diagram of the LSTM model;
FIG. 4 is a schematic structural diagram of tunnel deformation monitoring and early warning equipment;
the figure shows: 800-monitoring equipment; 801-a processor; an 802-memory; 803-multimedia component; 804-I/O interface; 805-a communication component.
Detailed Description
The invention will be further described with reference to the drawings and examples.
Example 1
As shown in fig. 1, a method for monitoring a tunnel by data fusion is characterized in that: the method comprises the following steps:
s1, acquiring original data of tunnel deformation monitoring;
s2, preprocessing the original data according to the type of the original data;
s3, carrying out weighted fusion processing on the preprocessed data;
s4, inputting the obtained original data into a training model to obtain prediction data at a prediction moment;
s5, obtaining tunnel deformation strength at a prediction moment according to the prediction data;
s6, comparing the tunnel deformation strength at the predicted moment with an early warning threshold value.
In step S1, the raw data includes sensor data, image data, and topography data.
In step S2, the preprocessing is a wavelet analysis processing method, and the wavelet analysis processing method includes denoising, normalizing and normalizing the original data.
And after the monitoring data subjected to wavelet analysis processing are generated, weighting the subsequences of the monitoring data by an entropy weight method.
In step S3, the weighted fusion process performs weighted and fused calculation on the data by a weighted fusion method, and the calculation formula is as follows:
y=w1×1+w2×2+w3×3+w4×4+w5×5+w6×6+w7×7+w8×8+w9×9 (one)
In the formula (I), Y is the deformation strength of the finally fused tunnel; x1, x2, x3, x4, x5, x6, x7, x8 and x9 are respectively deformation of a data source observation point, internal force of a steel support, ground subsidence, underground fissures, internal force of concrete, surrounding rock pressure, face information, face joint information and long-term data of observation point distribution; w1, w2, w3, w4, w5, w6, w7, w8 and w9 are weight settings corresponding to x1, x2, x3, x4, x5, x6, x7, x8 and x9, respectively.
And respectively acquiring monitoring data of the tunnel deformation strength of the tunnel at the current moment, and inputting the tunnel deformation monitoring data at the current moment into a prediction model to obtain prediction data of the tunnel deformation strength at the preset moment.
And carrying out weighted summation on the predicted data of the tunnel deformation strength and the corresponding weight, then calculating to obtain the tunnel deformation strength at the preset moment, and judging whether the tunnel deformation strength at the preset moment exceeds an early warning threshold value.
In step S4, the training model includes the following steps:
h1, acquiring multiple groups of monitoring data of deformation strength of the tunnel in a preset period, and preprocessing the multiple groups of monitoring data;
h2, constructing a data set by the preprocessed monitoring data, and dividing the data set into a training set and a testing set according to a preset proportion;
h3, constructing a GCN-DE-LSTM model, and training and testing the GCN-DE-LSTM model through a training set and a testing set to obtain a prediction model;
as shown in FIG. 2, the construction of the CN-DE-LSTM model firstly uses a graph convolutional neural network (GCN) to perform feature extraction, and establishes direct connection or indirect connection between observation points, including;
h3.1, matrix W: the matrix comprises deformation of observation points, steel support internal force, ground subsidence, underground cracks, concrete internal force, surrounding rock pressure, face information, face joint information and long-term data of observation point distribution;
h3.2, the degree of the node is: l (L) i,i ＝∑ j C j,i C j,i ；
The values of H3.3, adjacency matrix are: l (L) i,i ＝eiej→＝∑ k C k,i C k,j ；
And H3.4, an image operator is as follows: l (G) =c T C＝L i,i -Li,j；
H3.5, laplace processing is as follows:
in the formula (II), A is a characteristic value of L (G); i=u T U, U is the feature vector of L (G);
in the formula (III), the compound of the formula (III),is an element value;
h3.6, loop product processing:
g*x＝U(U T g·U T x) (IV)
In the formula (IV), U is a characteristic matrix of L, U T X is the Fourier transform of the filter operator, U T g is the Fourier transform of the Laplace operator;
h3.7, graph convolution is over the fourier domain as follows:
U T L＝AU Λ
g θ (Λ)＝diay(θ)；
g θ *x＝Ug θ U T x
the eigenvalues of the substitution L are as follows:
in the formula (five), the amino acid sequence of the compound,is the maximum value of lambda in L;
h3.8, normalization treatment as follows:
the expression of the graph convolution is as follows:
in the formula (six), sigma is an activation function, H is corresponding to x, and W is corresponding to theta;
after the characteristics of each observation point are extracted, the characteristic is used as a new set, and the step H is repeated to summarize and generalize the similar observation points, namely, the connection between the observation points is established;
substituting the data for establishing the connection into an LSTM model, wherein the LSTM model comprises four states which are respectively: the time of output value, the time of output value leading value, the time of memory neuron state and the leading value of memory neuron state all comprise three gates: input gate, output gate and forget gate as shown in fig. 3;
the LSTM model realizes dependency modeling on time sequence, and the principle steps are as follows:
c1, an input formula defining a single neuron is as follows:
x t = (G, H, W, F, N, Z (j, s, r), L) (seven)
In the formula (seven), x t For influencing factor single input set, y is deformation, G is steel support internal force, H is concrete internal force, W is surrounding rock pressure, F is ground subsidence, N is underground crack, Z is face information, j is face joint information, s is face rock type information, r is face wetting degree information, and L is observation point distribution;
c2, searching the super parameters of iteration times, learning rate and hidden layer neuron node number which are determined and input by artificial experience in the network, and searching the super parameters by a differential evolution algorithm;
the differential evolution algorithm comprises the following steps:
c2.1, initializing population
In formula (eight), x i,j (0) The ith individual j is the jth dimension, x i Representing a super-parameter set of the LSTM model;
in the formula (III), the amino acid sequence of the compound (III),and->A lower bound and an upper bound of the j-th dimension respectively;
c2.2, differential variation
V i (g+1)＝X rl (g)+F(X r2 (g)-X r3 (g) (ten)
In the formula (ten), r1 and r2 are random numbers between 1 and NP, F is a scaling factor, and g represents the g generation;
c2.3, cross
In the formula (eleven), CR is the crossover probability;
c2.4 greedy selection
And C2.5, adding an adaptive mutation operator into the algorithm for optimizing the scaling factor, wherein the adaptive mutation operator comprises the following steps:
F＝F 0 ·λ 2 (twelve)
In formula (twelve), F 0 For mutation operator, G max Is the maximum algebra of evolution.
The monitoring device includes:
and a data processing module: the method comprises the steps of carrying out noise elimination on random errors and interference factors involved in an original time sequence of each monitoring point through a wavelet analysis method, and carrying out standardization and normalization processing on data;
the entropy weight method processing module: the method comprises the steps of carrying out sequence fusion on denoising time of the same type of a plurality of different monitoring points, and finally fusing the denoising time into a subsequence of a corresponding monitoring index;
and a prediction module: the method comprises the steps of collecting monitoring data of tunnel deformation at a previous moment, inputting the monitoring data of tunnel deformation at the current moment into a prediction model, and obtaining a prediction result of the prediction model on tunnel deformation at a preset moment;
and a judging module: and the method is used for carrying out weighted summation on the predicted data of the tunnel deformation and the corresponding weight, then calculating to obtain the tunnel deformation strength at the preset moment, and judging whether the tunnel deformation strength at the preset moment exceeds the early warning threshold value.
For storing a computer program; the processor is used for implementing the steps of the tunnel monitoring method for data fusion when executing the computer program.
The computer program, when executed by a processor, performs the steps of a tunnel monitoring method for data fusion.
The GCN-DE-LSTM model learns the characteristic representation of different layers from the time sequence to more comprehensively express the information in the sequence and better process the trend, periodicity and noise of the sequence, so that the prediction precision of future tunnel deformation monitoring data is improved, meanwhile, the weight duty ratio of various monitoring data is determined by adopting an entropy weight method and a weighting fusion method, the tunnel deformation strength is calculated, the advance prediction of the tunnel deformation strength is realized, thereby helping a road management department to carry out corresponding control measures, reducing the influence of tunnel deformation risks on road traffic, effectively improving the prediction precision of the tunnel deformation data and providing important early warning and decision support for tunnel traffic management.
And (3) monitoring a sensor: installing strain monitoring sensors at key positions such as pile foundations, steel supports and the like of the tunnel structure, and monitoring and recording deformation data in real time; satellite remote sensing monitoring: remotely monitoring ground subsidence conditions through satellite images and remote sensing technology, and acquiring deformation data by combining GPS positioning technology and other technologies; microseismic monitoring: and by embedding a microseismic monitoring instrument near the tunnel, the change of the underground water level and the crack expansion condition are observed, and real-time data of the change of the internal environment of the tunnel are provided.
GCN feature extraction: extracting features from sensor monitoring data through a graph convolutional neural network (GCN), establishing a direct or indirect connection relation graph between observation points, capturing the relevance of tunnel deformation, and optimizing DE super parameters: optimizing the super parameters of the model through a differential evolution algorithm (DE) to further improve the prediction precision and accuracy; LSTM model application: and introducing the data related to the previous establishment into a long-short-term memory (LSTM) model, and performing sequence prediction analysis to realize the prediction of the tunnel deformation process and finish the operation.
The terminal equipment and the readable storage medium are used for displaying the monitoring result and storing the data for a long time, so that accidents can be effectively prevented, the safety level of tunnel operation can be improved, and an important reference basis is provided for maintenance and management.
As shown in fig. 4, a monitoring device (800) of a tunnel monitoring method for data fusion, where the monitoring device (800) includes a processor (801), a memory (802), a multimedia component (803), an I/O interface (804) and a communication component (805), and overall operation of the monitoring device (800) is controlled by the processor (801) to complete all or part of the steps in the tunnel deformation monitoring and early warning method.
The memory (802) is used to store various types of data to support operation on the monitoring device (800), which may include instructions for any application or method executing on the monitoring device (800), as well as application-related data, and the memory (802) may be implemented by any type of storage device or combination of any type of storage devices, such as internal memory/memory (RAM), external memory/disk storage: including Hard Disk Drives (HDDs) and Solid State Drives (SSDs), flash memory: including USB flash drives, flash cards, etc., optical disk storage: such as CD, DVD, blu-ray disc, etc.
The multimedia component (803) comprises a display, which may be a touch screen for example, and an audio component for outputting and/or inputting audio signals, which received audio signals may be further stored in the memory (802) or transmitted through the communication component (805).
The I/O interface (804) provides an interface between the processor (801) and other interface modules, which may be a keyboard, mouse, buttons, etc., which may be virtual buttons or physical buttons.
The communication component (805) is used for conducting wired or wireless communication between the tunnel deformation early warning device (800) and other devices; and a wireless module: such as Wi-Fi module, bluetooth module and mobile communication module (such as 4G, 5G module) for implementing wireless data communication and interconnection functions; network interface: such as an ethernet port, a USB port, and a serial port, for connecting devices to perform wired communication with a network or other external devices.
The monitoring device (800) may use electronic components such as a digital signal processing device, a digital signal processor, a programmable logic device, a field programmable gate array, an application specific integrated circuit, a controller, a microprocessor, a microcontroller, etc. to perform the tunnel deformation monitoring and early warning method described above.
A computer readable storage medium comprising program instructions which, when executed by a processor, implement the steps of the tunnel deformation monitoring and warning method described above, for example, the computer readable storage medium may be the memory (802) comprising program instructions described above, the program instructions being executable by the processor (801) to perform the tunnel deformation monitoring and warning method described above.
Readable storage media include, but are not limited to, magnetic media: including magnetic tape, magnetic disk, floppy disk, etc., optical media: including CD, DVD, blu-ray disc, etc., flash memory media: including USB flash drives, flash cards, solid State Drives (SSDs), etc., cloud storage: the data is stored on a remote server over the internet.
The tunnel deformation data monitoring method comprises the following steps: and (3) monitoring a sensor: installing strain monitoring sensors at key positions such as pile foundations and steel supports of a tunnel structure, monitoring and recording deformation data in real time, and monitoring by satellite remote sensing: through satellite images and remote sensing technology, ground subsidence condition is monitored remotely, deformation data is obtained by combining GPS positioning technology and other technologies, and microseism monitoring is carried out: and by embedding a microseismic monitoring instrument near the tunnel, the change of the underground water level and the crack expansion condition are observed, and real-time data of the change of the internal environment of the tunnel are provided.
Denoising the original monitoring data; carrying out standardization and normalization processing on the obtained data, and generating a plurality of corresponding subsequences T1, T2, … … and Tp according to the type of the original monitoring data, wherein any subsequence is composed of the monitoring data with the same type from each monitoring point; and carrying out weighted fusion on the processed data by using a weighted fusion method.
The tunnel deformation early warning implementation method comprises the following steps: GCN feature extraction: features are extracted from sensor monitoring data through a graph convolutional neural network (GCN), a direct or indirect connection relation graph between observation points is established, and the relevance of tunnel deformation is captured. Optimizing DE super parameters: optimizing the super parameters of the model by utilizing a differential evolution algorithm (DE) to further improve the accuracy and precision of prediction, and applying an LSTM model: and introducing the data related to the previous establishment into a long-short-term memory (LSTM) model, and performing sequence prediction analysis to realize the prediction of the tunnel deformation process and finish the operation.
The above description is not intended to limit the invention to the particular embodiments disclosed, but on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (9)
1. A tunnel monitoring method for data fusion is characterized in that: the method comprises the following steps:
s1, acquiring original data of tunnel deformation monitoring;
s2, preprocessing the original data according to the type of the original data;
s3, carrying out weighted fusion processing on the preprocessed data;
s4, inputting the obtained original data into a training model to obtain prediction data at a prediction moment;
s5, obtaining tunnel deformation strength at a prediction moment according to the prediction data;
s6, comparing the tunnel deformation strength at the predicted moment with an early warning threshold value.
2. The method for tunnel monitoring for data fusion according to claim 1, wherein: in step S1, the raw data includes sensor data, image data, and topography data.
3. The method for tunnel monitoring for data fusion according to claim 1, wherein: in step S2, the preprocessing is a wavelet analysis processing method, and the wavelet analysis processing method includes denoising, normalizing and normalizing the original data.
4. A method of tunnel monitoring for data fusion as defined in claim 3, wherein: and after the monitoring data subjected to wavelet analysis processing are generated, weighting the subsequences of the monitoring data by an entropy weight method.
5. The method for tunnel monitoring for data fusion according to claim 1, wherein: in step S3, the weighted fusion process performs weighted and fused calculation on the data by a weighted fusion method, and the calculation formula is as follows:
y=w1×1+w2×2+w3×3+w4×4+w5×5+w6×6+w7×7+w8×8+w9×9 (one)
In the formula (I), Y is the deformation strength of the finally fused tunnel; x1, x2, x3, x4, x5, x6, x7, x8 and x9 are respectively deformation of a data source observation point, internal force of a steel support, ground subsidence, underground fissures, internal force of concrete, surrounding rock pressure, face information, face joint information and long-term data of observation point distribution; w1, w2, w3, w4, w5, w6, w7, w8 and w9 are weight settings corresponding to x1, x2, x3, x4, x5, x6, x7, x8 and x9, respectively.
6. The method for tunnel monitoring for data fusion according to claim 1, wherein: in step S4, the training model includes the following steps:
h1, acquiring multiple groups of monitoring data of deformation strength of the tunnel in a preset period, and preprocessing the multiple groups of monitoring data;
h2, constructing a data set by the preprocessed monitoring data, and dividing the data set into a training set and a testing set according to a preset proportion;
h3, constructing a GCN-DE-LSTM model, and training and testing the GCN-DE-LSTM model through a training set and a testing set to obtain a prediction model;
firstly, performing feature extraction by using a graph roll neural network (GCN) to establish direct connection or indirect connection between observation points, wherein the construction of the CN-DE-LSTM model comprises the steps of;
h3.1, matrix W: the matrix comprises deformation of observation points, steel support internal force, ground subsidence, underground cracks, concrete internal force, surrounding rock pressure, face information, face joint information and data of observation point distribution;
h3.2, the degree of the node is: l (L) i,i ＝Σ j C j,i C j,i ；
The values of H3.3, adjacency matrix are: l (L) i,i ＝eiej→＝Σ k C k,i C k,j ；
And H3.4, an image operator is as follows: l (G) =c T C＝L i,i -Li,j；
H3.5, laplace processing is as follows:
in the formula (II), A is a characteristic value of L (G); i=u T U, U is the feature vector of L (G);
in the formula (III), the compound of the formula (III),is an element value;
h3.6, loop product processing: gx=u (U T g·U T x) (IV)
In the formula (IV), U is a characteristic matrix of L, U T X is the Fourier transform of the filter operator, U T g is the Fourier transform of the Laplace operator;
h3.7, graph convolution is over the fourier domain as follows:
U T L＝AU Λ
g θ (Λ)＝diay(θ)；
g θ *x＝Ug θ U T x
the eigenvalues of the substitution L are as follows:
in the formula (five), the amino acid sequence of the compound,is the maximum value of lambda in L;
h3.8, normalization treatment as follows:
the expression of the graph convolution is as follows:
in the formula (six), sigma is an activation function, H is corresponding to x, and W is corresponding to theta;
after the characteristics of the observation points are extracted, the observation points are summarized and summarized by repeating the step H as a new set, and then connection among the observation points is established;
substituting the data for establishing the connection into an LSTM model, wherein the LSTM model comprises four states which are respectively: the time of output value, the time of output value leading value, the time of memory neuron state and the leading value of memory neuron state all comprise three gates: an input gate, an output gate, and a forget gate;
the LSTM model realizes modeling of time series, and the principle steps are as follows:
c1, an input formula defining a single neuron is as follows:
x t = (G, H, W, F, N, Z (j, s, r), L) (seven)
In the formula (seven), x t For influencing factor single input set, y is deformation, G is steel support internal force, H is concrete internal force, W is surrounding rock pressure, F is ground subsidence, N is underground crack, Z is face information, j is face joint information, s is face rock type information, r is face wetting degree information, and L is observation point distribution;
c2, searching the super parameters of iteration times, learning rate and hidden layer neuron node number which are manually determined and input in the network, and searching the super parameters through a differential evolution algorithm;
the differential evolution algorithm comprises the following steps:
c2.1, initializing population
In formula (eight), x i,j (0) The ith individual j is the jth dimension, x i Representing a super-parameter set of the LSTM model;
in the formula (III), the amino acid sequence of the compound (III),and->A lower bound and an upper bound of the j-th dimension respectively;
c2.2, differential variation
V i (g+1)＝X rl (g)+F(X r2 (g)-X r3 (g) (ten)
In the formula (ten), r1 and r2 are random numbers between 1 and NP, F is a scaling factor, and g represents the g generation;
c2.3, cross
In the formula (eleven), CR is the crossover probability;
c2.4 greedy selection
And C2.5, adding an adaptive mutation operator into the algorithm for optimizing the scaling factor, wherein the adaptive mutation operator comprises the following steps:
F＝F 0 ·λ 2 (twelve)
In formula (twelve), F 0 For mutation operator, G max Is the maximum algebra of evolution.
7. A monitoring device for a tunnel monitoring method for data fusion according to any one of claims 1 to 6, wherein the monitoring device comprises:
and a data processing module: the method comprises the steps of carrying out noise elimination on random errors and interference factors involved in an original time sequence of each monitoring point through a wavelet analysis method, and carrying out standardization and normalization processing on data;
the entropy weight method processing module: the method comprises the steps of carrying out sequence fusion on denoising time of the same type of a plurality of different monitoring points, and finally fusing the denoising time into a subsequence of a corresponding monitoring index;
and a prediction module: the method comprises the steps of collecting monitoring data of tunnel deformation at a previous moment, inputting the monitoring data of tunnel deformation at the current moment into a prediction model, and obtaining a prediction result of the prediction model on tunnel deformation at a preset moment;
and a judging module: and the method is used for carrying out weighted summation on the predicted data of the tunnel deformation and the corresponding weight, then calculating to obtain the tunnel deformation strength at the preset moment, and judging whether the tunnel deformation strength at the preset moment exceeds the early warning threshold value.
8. A computer device comprising a memory and a processor, wherein the memory: for storing a computer program; the processor being adapted to carry out the steps of the method of any one of claims 1 to 5 when said computer program is executed.
9. A computer storage medium, characterized by: the storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method of any of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311391366.1A CN117272232A (en) | 2023-10-25 | 2023-10-25 | Tunnel monitoring method and device for data fusion, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311391366.1A CN117272232A (en) | 2023-10-25 | 2023-10-25 | Tunnel monitoring method and device for data fusion, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117272232A true CN117272232A (en) | 2023-12-22 |
Family
ID=89200861
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311391366.1A Pending CN117272232A (en) | 2023-10-25 | 2023-10-25 | Tunnel monitoring method and device for data fusion, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117272232A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118094205A (en) * | 2024-04-28 | 2024-05-28 | 中建铁路投资建设集团有限公司 | Intelligent tunnel structure health monitoring system based on deep learning |
| CN118627016A (en) * | 2024-08-13 | 2024-09-10 | 广州地铁设计研究院股份有限公司 | Track monitoring data processing method, equipment and storage medium based on three-dimensional structure |
-
2023
- 2023-10-25 CN CN202311391366.1A patent/CN117272232A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118094205A (en) * | 2024-04-28 | 2024-05-28 | 中建铁路投资建设集团有限公司 | Intelligent tunnel structure health monitoring system based on deep learning |
| CN118627016A (en) * | 2024-08-13 | 2024-09-10 | 广州地铁设计研究院股份有限公司 | Track monitoring data processing method, equipment and storage medium based on three-dimensional structure |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhou et al. | Dynamic prediction for attitude and position in shield tunneling: A deep learning method | |
| Xin et al. | A signal recovery method for bridge monitoring system using TVFEMD and encoder-decoder aided LSTM | |
| CN117272232A (en) | Tunnel monitoring method and device for data fusion, computer equipment and storage medium | |
| Wang et al. | Dynamic prediction of mechanized shield tunneling performance | |
| CN110751108B (en) | Subway distributed vibration signal similarity determination method | |
| Fathnejat et al. | A data-driven structural damage identification approach using deep convolutional-attention-recurrent neural architecture under temperature variations | |
| Leng et al. | A hybrid data mining method for tunnel engineering based on real-time monitoring data from tunnel boring machines | |
| CN104766129A (en) | Subway shield construction surface deformation warning method based on temporal and spatial information fusion | |
| Lee et al. | Cultural heritage and the intelligent internet of things | |
| Son et al. | Deep learning-based anomaly detection to classify inaccurate data and damaged condition of a cable-stayed bridge | |
| CN112036042A (en) | Power equipment abnormality detection method and system based on variational modal decomposition | |
| CN113687433A (en) | Bi-LSTM-based magnetotelluric signal denoising method and system | |
| Huang et al. | Physics-guided deep neural network for structural damage identification | |
| CN115146842A (en) | Multivariate time series trend prediction method and system based on deep learning | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium | |
| Mamat et al. | Artificial neural networks in slope of road embankment stability applications: a review and future perspectives | |
| CN113553356A (en) | Drilling parameter prediction method and system | |
| Zhao et al. | A hybrid recognition model of microseismic signals for underground mining based on CNN and LSTM networks | |
| CN116881819A (en) | Stay cable working state monitoring method based on isolated forest | |
| Cao et al. | Seismic damage identification of high arch dams based on an unsupervised deep learning approach | |
| Xiong et al. | GTRF: A general deep learning framework for tuples recognition towards supervised, semi-supervised and unsupervised paradigms | |
| Giglioni et al. | Deep autoencoders for unsupervised damage detection with application to the Z24 benchmark bridge | |
| Ye et al. | Confining pressure forecasting of shield tunnel lining during construction based on LSTM-PSO models combined with the multi-output recursive strategy | |
| Wu et al. | Multisource information fusion for real-time prediction and multiobjective optimization of large-diameter slurry shield attitude | |
| Li et al. | Displacement prediction of open-pit mine slope based on SSA-ELM |
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 |