CN-121168288-B - Dangerous chemical wharf breakwater monitoring system and method based on multi-source data integration

CN121168288BCN 121168288 BCN121168288 BCN 121168288BCN-121168288-B

Abstract

The invention discloses a jetty breakwater monitoring system and a jetty breakwater monitoring method based on multi-source data integration, wherein the method firstly obtains multi-mode monitoring data of a first surface structure of a breakwater, an environmental load, an underwater structure of the breakwater and a second surface structure of the breakwater through a multi-source sensor; calculating four safety monitoring coefficients of the first surface structure, the environment, the underwater structure and the second surface structure by using a model comprising a space-time feature extraction module, a dynamic feature analysis module and a multi-source data fusion module; the method is characterized by creatively introducing eight layers of fine grain cross fusion mechanisms, realizing deep interaction and fusion of multi-mode features by means of cross-mode attention, gate control aggregation and the like, acquiring a dynamic threshold value by establishing a digital twin model, carrying out self-adaptive adjustment based on fine grain feature distribution, and finally realizing accurate early warning from single index to multi-mode associated risk by means of multi-level early warning logic. The invention improves the state sensing and risk early warning capability of the complex system of the dangerous chemical wharf.

Inventors

LI ZHENXING
Yu Shishuang
HOU JIAQUAN
LI WEI
YANG JINGPENG
ZHENG WENJIN
CHEN JINQIAO
ZHANG JINHAO
ZHAO YUTAO

Assignees

中交华南勘察测绘科技有限公司
中海油江苏天然气有限责任公司

Dates

Publication Date: 20260508
Application Date: 20251121

Claims (9)

1. The jetty breakwater monitoring method for the hazardous chemicals based on multi-source data integration is characterized by comprising the following steps of: S1, acquiring a current structure settlement value and a structure displacement value to obtain first breakwater surface structure monitoring data, acquiring a real-time wave intensity value, a total quantity of dangerous chemical ships and a quantity of dangerous chemical ships on the pre-shore to obtain environment monitoring basic data, acquiring a dyke body scouring value and a collapse value to obtain breakwater underwater structure monitoring data, and acquiring cracks and block displacement of the surface layer of the breakwater to obtain second breakwater surface structure monitoring data; s2, carrying out space-time feature extraction on the first breakwater surface structure monitoring data through a multi-mode feature analysis model, calculating to obtain a first structural safety monitoring coefficient, carrying out dynamic feature analysis on the environment monitoring basic data through the multi-mode feature analysis model, calculating to obtain an environment risk monitoring coefficient, carrying out underwater performance structure feature analysis on the breakwater underwater structure monitoring data through the multi-mode feature analysis model, calculating to obtain a breakwater underwater safety structure monitoring coefficient, carrying out multi-source data fusion analysis on the second breakwater surface structure monitoring data through the multi-mode feature analysis model, calculating to obtain a second structural safety monitoring coefficient, and defining the first structural safety monitoring coefficient, the environment risk monitoring coefficient, the breakwater underwater safety structure monitoring coefficient and the second structural safety monitoring coefficient as breakwater health monitoring analysis data; s3, establishing a breakwater digital twin model, acquiring a breakwater underwater safety structure monitoring coefficient threshold value, an environmental risk monitoring coefficient threshold value, a first structural safety monitoring coefficient threshold value and a second structural safety monitoring coefficient threshold value through the breakwater digital twin model, and defining the breakwater underwater safety structure monitoring coefficient threshold value, the environmental risk monitoring coefficient threshold value, the first structural safety monitoring coefficient threshold value and the second structural safety monitoring coefficient threshold value as breakwater health monitoring threshold value data; and S4, carrying out safety risk early warning according to breakwater health monitoring analysis data and breakwater health monitoring threshold value data.
2. The jetty monitoring method based on multi-source data integration according to claim 1, wherein in S1, acquiring the first jetty surface structure monitoring data includes: The method comprises the steps of arranging N monitoring points on a breakwater, arranging a GNSS receiver on each monitoring point, receiving satellite signals in real time, calculating three-dimensional coordinates of each monitoring point, obtaining displacement values of the monitoring points of the breakwater according to the three-dimensional coordinates to obtain structural displacement values, determining k monitoring time nodes of the N monitoring points in a current monitoring period, separating each two monitoring time nodes by a standard monitoring time period, respectively obtaining structural sedimentation values corresponding to the k monitoring time nodes through the GNSS receivers, respectively marking the structural sedimentation values as first sedimentation rate to k sedimentation rate, carrying out weighted average calculation on the first sedimentation rate to k sedimentation rate to obtain structural sedimentation values, and defining the current structural sedimentation values and the structural displacement values as first breakwater surface structure monitoring data.
3. The jetty monitoring method based on multi-source data integration according to claim 1, wherein in S1, acquiring the environmental monitoring basic data includes: The method comprises the steps of acquiring a wave intensity value corresponding to a breakwater in real time through a wave sensor to obtain a real-time wave intensity value, acquiring the total quantity of dangerous chemical ships parked at a current dock and the quantity of dangerous chemical ships not parked at the dock through a camera image recognition algorithm, marking the total quantity of dangerous chemical ships parked at the current dock as the total quantity of dangerous chemical ships, marking the quantity of dangerous chemical ships not parked at the dock as the quantity of dangerous chemical ships to be pre-landed, and defining the real-time wave intensity value, the total quantity of dangerous chemical ships and the quantity of dangerous chemical ships to be pre-landed as environment monitoring basic data.
4. The jetty monitoring method for the jetty based on multi-source data integration according to claim 1, wherein in S1, the second jetty surface structure monitoring data is acquired, and specifically includes: based on a navigation technology, determining an initial navigation path of the unmanned aerial vehicle, and configuring a high-definition camera and a three-dimensional laser radar on the unmanned aerial vehicle; The unmanned aerial vehicle flies according to the initial navigation path, and basic information and initial image data of the breakwater surface structure are collected through the carried high-definition camera; Acquiring regional key points of the breakwater surface structure according to the basic information to obtain a plurality of regional key point coordinates; Inputting the initial image data, the basic information and the coordinates of the key points of the plurality of areas into a navigation parameter planning model, and outputting a navigation path track set of the unmanned aerial vehicle; The unmanned aerial vehicle regularly acquires images according to the navigation path track set, the images shot by the unmanned aerial vehicle are analyzed by utilizing an image recognition algorithm, the crack value and the block displacement value of the surface layer of the breakwater are recognized, and the crack value and the block displacement value are defined as second breakwater surface structure monitoring data.
5. The jetty monitoring method based on multi-source data integration according to claim 1, wherein in S2, the method comprises: Establishing a multi-mode feature analysis model, wherein the multi-mode feature analysis model comprises a space-time feature extraction module, a dynamic feature analysis module, an underwater performance structure analysis module, a multi-source data fusion module and a fine grain cross fusion layer, and the fine grain cross fusion layer is totally provided with 8 layers; S21, acquiring first breakwater surface structure monitoring data, environment monitoring basic data, breakwater underwater structure monitoring data and second breakwater surface structure monitoring data; S22, analyzing the first breakwater surface structure monitoring data through a space-time feature extraction module to obtain a first structure safety monitoring coefficient Ks; s23, analyzing the environment monitoring basic data through a dynamic characteristic analysis module to obtain an environment risk monitoring coefficient Ke; s24, analyzing the underwater structure monitoring data of the breakwater through an underwater performance structure analysis module to obtain a monitoring coefficient Km of the underwater safety structure of the breakwater; s25, analyzing the second breakwater surface structure monitoring data through a multi-source data fusion module to obtain a second structure safety monitoring coefficient Kt; S26, performing primary fusion analysis through a first fine grain cross fusion layer to realize multi-scale feature enhancement in a mode: and extracting multi-scale space-time characteristics of the first breakwater surface structure monitoring data: Fs = [ , , ..., ] = MLP(CNN(LSTM( ) A) a plurality of the components, wherein, The raw data sequence is monitored for the breakwater structure, The dimension of (2) is [ T1, ds ], T1 is the time step, ds is the structural feature dimension; the LSTM is used for capturing time dependence of the first breakwater surface structure monitoring data, the CNN is used for extracting a local mode of the first breakwater surface structure monitoring data, the MLP is used for characteristic transformation and dimension reduction, and the Fs is used for representing a multi-scale space-time characteristic vector; representing the nth element of the multi-scale space-time feature vector; extracting multi-resolution dynamic characteristics of the environment monitoring basic data: Fe = [ , , ..., ] = Wavelet(GRU( ) And) a combination of one or more of the above, The system is characterized by comprising an environment monitoring basic original data sequence, wherein the dimension of the environment monitoring basic data sequence is [ T, de ], de is the dimension of an environment characteristic, GRU is a gating cyclic unit neural network, wavelet is Wavelet transformation and is used for extracting time-frequency characteristics, and Fe represents a multi-resolution dynamic characteristic vector; An mth element for representing the multi-resolution dynamic feature vector; S27, performing second fusion analysis through a second fine grain cross fusion layer to realize cross-modal attention feature interaction, wherein the input of the second fine grain cross fusion layer is the output of the first fine grain cross fusion layer: building a structure-environment cross-modal attention matrix Ase, and calculating fine-grained structure characteristics under the influence of environmental conditions : Ase = softmax((Wq*Fs)*(Wk*Fe) T /qqq); = Fs + Ase * (Wv * Fe); Wherein Wq epsilon , Wk ∈ , Wv ∈ Wq, wk, wv respectively represent a learnable weight matrix, d represents an implicit layer dimension, qqq represents a scaling factor, T represents a matrix transpose, and R represents a real set; computing fine grain environmental features under structural state influence : Aes = Where Aes represents the transposed matrix of Ase; = Fe + Aes * (Wv * Fs); s28, fine grain structural features based on cross fusion And fine grain environmental features Reconstructing the monitoring coefficient through a third fine grain cross fusion layer and performing a third fusion analysis to realize gated multi-mode feature aggregation: = σ( Ws * pool( ) + bs ); = σ( We * pool( ) + be ); gate = sigmoid( Wg * [pool( ), pool( )] + bg); = Km + λ * gate * ( + ); align = cosine_similarity( pool( ), pool( )); = Kt + μ * align * ( + ); Wherein Ws represents structural characteristics pool used for pooling ) Mapping to first structural safety monitoring coefficient We represents the environmental characteristics pool used for pooling ) Mapping to environmental risk monitoring coefficients Wg represents the weight matrix used for splicing the structure and the environmental characteristics [ pool ] );pool( ) The method comprises the steps of mapping a weight matrix of a gating signal gate, bs, be, bg respectively representing bias vectors of a leachable model, gate representing the gating signal, pool representing a pooling function, sigma representing an activating function, lambda representing a fusion intensity coefficient, mu representing an alignment gain coefficient, align representing a characteristic alignment degree; Representing the refined first structural safety monitoring coefficient; representing the thinned environmental risk monitoring coefficient; representing the corrected monitoring coefficient of the underwater safety structure of the dyke body; The modified second structural safety monitoring coefficient is represented, wherein cosine similarity represents a cosine similarity function; s29, monitoring the coefficient of the corrected underwater safety structure of the dyke body Environmental risk monitoring coefficient after refinement First structure safety monitoring coefficient after refinement Second structural safety monitoring coefficient after correction All fine grain structural features And fine grain environmental features Collectively defined as breakwater health monitoring analysis data.
6. The jetty monitoring method based on multi-source data integration of claim 5, wherein S22 comprises: S221, performing space-time correlation analysis on the first breakwater surface structure monitoring data through a space-time feature extraction module; S222, acquiring space-time feature vectors of a current structure sedimentation value and a structure displacement value according to an analysis result; s223, acquiring a structure settlement allowing threshold Dd0 and a structure displacement allowing threshold Vf0 through a database; S224, calculating the space-time characteristic vector and the corresponding threshold value through multiple modes to obtain an initial value Ks 0 of the first structural safety monitoring coefficient: Ks 0 = β (Dd/Dd 0) +γ (Vf/Vf 0), where β, γ are weight coefficients, and β+γ = 1, dd represents structural sedimentation eigenvalues, vf represents structural displacement eigenvalues; S225, introducing a fine-grained environmental context sensor through a fourth fine-grained cross fusion layer to perform fourth fusion analysis on environmental context vectors: extracting short-term fluctuation features Eshort and long-term trend features Elong of the environmental data: Eshort = TCN( ) Wherein TCN represents a time convolution network for processing recent data; Representing a short-term environmental data sequence; Elong = Transformer( ) Wherein, the transducer represents a transducer neural network for processing long-term history data; Representing a long-term environmental data sequence; Computing environment context vector Cenv: Cenv = Attention ([ Eshort, elong ]), wherein Attention represents the Attention function; S226, generating a fine-grained first structural safety monitoring coefficient Ksfine of environmental perception according to the environmental context vector, and performing fifth fusion analysis on the fine-grained first structural safety monitoring coefficient through a fifth fine-grained cross fusion layer to realize self-adaptive weighting, so as to obtain a final first structural safety monitoring coefficient: Ksfine = LSTM([ ,Cenv]); = sigmoid([ Cenv), wherein Representing adaptive weights; Ks = * Ks 0 + (1 - ) Ksfine, ks represents a first structural safety monitoring coefficient.
7. The jetty monitoring method based on multi-source data integration according to claim 5, wherein in S25, comprising: s253, carrying out multi-mode feature extraction on the second breakwater surface structure monitoring data through a multi-source data fusion module to generate multi-mode features; S254, acquiring a crack threshold Tt0 and a block displacement threshold Pp0 through a database; S255, calculating the multi-mode characteristic and the corresponding safety threshold value through a BP deep neural network to obtain a second structural safety monitoring coefficient initial value Kt 0 : kt 0 =δ1 (Tt/tt0) +δ2 (Pp/Pp 0), where δ1, δ2 are weights, tt represents a crack characteristic value, and Pp represents a block displacement characteristic value; S256, setting a multi-mode fine-grain region of interest (ROI) through a sixth fine-grain cross fusion layer, and carrying out a sixth fusion analysis through the multi-mode fine-grain region of interest (ROI): for n monitoring points of breakwater according to the space position Crack and crack Displacement of block Calculating attention weight: = exp(MLP([ , , ]))/ Wherein Representing the attention weight corresponding to the ith monitoring point; s257, performing a seventh fusion analysis through a seventh fine grain cross fusion layer: Performing cross-modal alignment on the structural settlement features Xvib monitored by the first breakwater surface structure and the crack development features Pfluct monitored by the second breakwater surface structure, and calculating an alignment matrix Avibp: Avibp = crossattention (Xvib, pfluct), wherein crossattention represents a cross-attention function; calculating a fine-grained second breakwater surface safety coefficient Ktfine under physical constraint: = + bf1 * MLP( ) Where bf1 is the physical influence coefficient, i represents the ith monitoring point, Fine-grained second breakwater surface safety factor Ktfine representing the ith monitoring point; a second structural safety monitoring coefficient initial value Kt 0 representing an ith monitoring point; an alignment matrix Avibp representing the i-th monitoring point; S258, performing eighth fusion analysis through an eighth fine grain cross fusion layer: = wherein Kt represents a second structural safety monitoring coefficient.
8. The jetty monitoring method based on multi-source data integration according to claim 5, wherein the S3 comprises: s31, acquiring historical operation data of the breakwater, and establishing a digital twin model of the breakwater by using the historical operation data of the breakwater through a BIM modeling tool; S32, respectively acquiring a displacement allowable threshold, a crack expansion threshold, a wave intensity threshold, a dangerous chemical ship capacity threshold, a pre-shore ship quantity threshold, a settlement allowable threshold, a crack threshold and a block displacement threshold through a digital twin model; S34, calculating a structural settlement threshold value and a structural displacement threshold value through multiple modes to obtain a first structural safety monitoring coefficient basic threshold Ks0base; S35, carrying out dynamic characteristic calculation through a wave intensity threshold value, a dangerous chemical ship capacity threshold value and a pre-shore ship quantity threshold value to obtain an environmental risk monitoring coefficient basic threshold value Ke0base; S36, calculating the underwater safety performance characteristics of the dyke body scouring threshold value and the collapse threshold value to obtain a dyke body underwater safety structure monitoring coefficient basic threshold value Km0base; s37, carrying out multi-source data fusion calculation on the crack threshold value and the block displacement threshold value to obtain a second structural safety monitoring coefficient basic threshold value Kt0base; S38, respectively adjusting a first structural safety monitoring coefficient threshold value and an environmental risk monitoring coefficient threshold value based on the structural fine granularity characteristic and the environmental fine granularity characteristic, and respectively generating a first dynamic first structural safety monitoring coefficient threshold value and a first environmental risk monitoring coefficient threshold value: Calculating fine grain structural features Dynamically adjusting a first structural safety monitoring coefficient threshold: Ks0 = Ks0base * (1 + ηs * (|| - μs/σs)), where ηs is a sensitivity coefficient, μs represents a mean value, σs represents a standard deviation, ks0 represents a first dynamic first structural safety monitoring coefficient threshold; Computing fine grain environmental features Dynamically adjusting an environmental risk monitoring coefficient threshold value: Ke0 = Ke0base * (1 + ηe * (std( )/ mean( ) -c), wherein ηe represents an environmental fluctuation coefficient, ke0 represents a first environmental risk monitoring coefficient threshold value, std represents a standard deviation function, mean represents an average function; s39, generating a structure-environment coupling threshold value by using the cross-modal attention matrix Ase of S27 and the first dynamic first structure safety monitoring coefficient threshold value coupling: Kse0 = (Ase * Ke0) + ((1 - Ase) * Ks0); Using the gate signal gate, the alignment align, and the first environmental risk monitoring coefficient threshold of S28, coupling to generate a bank-environment association threshold: Kmt0 = gate * (Km0base + Kt0base) + (1 - gate) * align * max(Km0base, Kt0base); S40, the basic thresholds Ks0base, ke0base, km0base, kt0base, dynamic adjustment thresholds Ks0, ke0, and coupling thresholds Kse0, kmt0 are collectively defined as breakwater health monitoring threshold data.
9. A jetty monitoring system based on multi-source data integration, which is characterized in that the system is used for executing the method of any one of claims 1-8, and comprises a monitoring data acquisition module, a multi-mode analysis module, a threshold analysis module and an early warning analysis module; The monitoring data acquisition module is used for acquiring current structure settlement values and structure displacement values to obtain first breakwater surface structure monitoring data, acquiring real-time wave intensity values, total dangerous chemical ship quantity and number of dangerous chemical ships on the pre-shore to obtain environment monitoring basic data, acquiring a dyke body scouring value and collapse value to obtain breakwater underwater structure monitoring data, and acquiring cracks and block displacement of the breakwater surface layer to obtain second breakwater surface structure monitoring data; The multi-mode analysis module performs space-time feature extraction on the first breakwater surface structure monitoring data, calculates to obtain a first structural safety monitoring coefficient, performs dynamic feature analysis on the environment monitoring basic data through the multi-mode feature analysis model, calculates to obtain an environment risk monitoring coefficient, performs underwater performance structure feature analysis on the breakwater underwater structure monitoring data through the multi-mode feature analysis model, calculates to obtain a breakwater underwater safety structure monitoring coefficient, performs multi-source data fusion analysis on the second breakwater surface structure monitoring data through the multi-mode feature analysis model, calculates to obtain a second structural safety monitoring coefficient, and defines the first structural safety monitoring coefficient, the environment risk monitoring coefficient, the breakwater underwater safety structure monitoring coefficient and the second structural safety monitoring coefficient as breakwater health monitoring analysis data; the threshold analysis module is used for establishing a breakwater digital twin model, acquiring a breakwater underwater safety structure monitoring coefficient threshold value, an environmental risk monitoring coefficient threshold value, a first structural safety monitoring coefficient threshold value and a second structural safety monitoring coefficient threshold value through the breakwater digital twin model, and defining the breakwater underwater safety structure monitoring coefficient threshold value, the environmental risk monitoring coefficient threshold value, the first structural safety monitoring coefficient threshold value and the second structural safety monitoring coefficient threshold value as breakwater health monitoring threshold value data; And the early warning analysis module carries out safety risk early warning according to the breakwater health monitoring analysis data and the breakwater health monitoring threshold value data.

Description

Dangerous chemical wharf breakwater monitoring system and method based on multi-source data integration Technical Field The invention relates to the technical field of dangerous chemical wharf breakwater intelligent monitoring, in particular to a dangerous chemical wharf breakwater monitoring system and method based on multi-source data integration. Background The breakwater is used as a front barrier of the wharf to directly bear complex environmental loads such as waves, tide, ship impact and the like, and the structural health state of the breakwater is directly related to the safety of the whole wharf. However, the traditional breakwater monitoring method has obvious limitations that on one hand, a single sensor or an isolated system is adopted for monitoring, if only structural displacement or corrosion rate is concerned, collaborative sensing and fusion analysis on multi-source information such as a structure, environment, load and a rear tank area are lacked, the overall safety state of the system is difficult to comprehensively grasp, on the other hand, the data analysis means is relatively simple and depends on threshold value alarming or simple statistics, space-time correlation, multi-mode coupling relation and early hidden fault characteristics contained in data cannot be deeply mined, early warning capability is insufficient, and false alarm and missing report rate are high. In recent years, a digital twin technology provides a new idea for infrastructure health management, and by constructing virtual mapping of physical entities, deep fusion and state simulation prediction of data are realized. Meanwhile, artificial intelligence technologies such as multi-mode data analysis and deep learning have great potential in the field of structural health monitoring, and can process monitoring data in heterogeneous, high-dimensional and time sequence and extract deep features. However, the existing researches focus on analysis of a single structure or a single data mode, and for complex systems such as dangerous chemical wharfs containing multiple high-risk elements such as breakwater structures, complex ocean environments and the like, a set of overall solutions capable of realizing multi-source data deep integration, multi-mode feature cross fusion, dynamic threshold intelligent adjustment and systematic risk accurate early warning are not available. In particular, how to quantify the dynamic interaction effect between environmental load and structural response, and how to realize the crossing from single parameter overrun alarm to multi-factor coupling risk early warning, remain a prominent challenge faced by the current technology. Therefore, the prior art lacks accurate sensing and intelligent early warning of the full life cycle, full elements and full risks of the breakwater and related facilities of the breakwater, and lacks accurate state sensing and risk early warning capability of the complex system of the breakwater. Disclosure of Invention The invention aims to provide a jetty monitoring system and a jetty monitoring method based on multi-source data integration, which are used for solving the problems in the prior art. The application is specifically as follows: A jetty breakwater monitoring method based on multi-source data integration comprises the following steps: S1, acquiring a current structure settlement value and a structure displacement value, acquiring first breakwater surface structure monitoring data, acquiring real-time wave intensity value, total quantity of dangerous chemical ships and quantity of dangerous chemical ships on the pre-shore, acquiring environment monitoring basic data, acquiring a dyke body scouring value and a collapse value, acquiring breakwater underwater structure monitoring data, acquiring cracks and block displacement of the surface layer of the breakwater, acquiring second breakwater surface structure monitoring data, and defining the first breakwater surface structure monitoring data, the environment monitoring basic data, the breakwater underwater structure monitoring data and the second breakwater surface structure monitoring data as breakwater health monitoring basic data; s2, carrying out space-time feature extraction on the first breakwater surface structure monitoring data through a multi-mode feature analysis model, calculating to obtain a first structural safety monitoring coefficient, carrying out dynamic feature analysis on the environment monitoring basic data through the multi-mode feature analysis model, calculating to obtain an environment risk monitoring coefficient, carrying out underwater performance structure feature analysis on the breakwater underwater structure monitoring data through the multi-mode feature analysis model, calculating to obtain a breakwater underwater safety structure monitoring coefficient, carrying out multi-source data fusion analysis on the second breakwater surface structure monitoring data through the multi-m