Dynamic safety analysis method for traffic construction project based on multi-source data fusion

Abstract

The application discloses a traffic construction project site dynamic safety analysis method based on multi-source data fusion, relates to the technical field of traffic construction engineering safety management and intelligent monitoring, and comprises the following steps: establishing a real-time data acquisition channel, extracting key characteristic parameters of each construction stage, and constructing a standardized characteristic vector; based on the standardized characteristic vector and the spatial topological relation matrix of the site, a time-space correlation model between multi-source data is established, and a risk transmission path is identified; by calculating the time-space cross-correlation function between characteristics, key risk nodes and transmission paths are identified; a risk assessment model based on the construction stage of a hierarchical early warning mechanism is established; and electronic fence parameters are dynamically optimized according to the risk level and the construction stage. Through multi-source data fusion and dynamic risk assessment, the application realizes real-time monitoring and early warning of the safety of a traffic construction project site, can effectively identify a risk transmission path, dynamically adjusts electronic fence parameters, and reduces safety risks.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of traffic construction project safety management and intelligent monitoring, and particularly relates to a traffic construction project site dynamic safety analysis method based on multi-source data fusion. BACKGROUND

[0002] At present, the traffic construction project site safety management has the following technical bottlenecks:

[0003] The traditional site selection evaluation method is mostly based on one-time static analysis, which cannot adapt to the dynamically changing risk environment in the construction process. There is a lack of dynamic response mechanism for risk evolution in the construction process. The existing electronic fence design mostly adopts a fixed boundary mode, which cannot be self-adaptively adjusted according to the construction stage progress and risk level changes, resulting in a mismatch between safety protection and actual risk. Various monitoring data (meteorological, geological, personnel positioning, etc.) are scattered and independent, lacking an effective multi-source data fusion mechanism, and it is difficult to form a overall risk assessment. SUMMARY

[0004] The present application provides a traffic construction project site dynamic safety analysis method based on multi-source data fusion, which solves the problems of the prior art.

[0005] In a first aspect, the present application provides a traffic construction project site dynamic safety analysis method based on multi-source data fusion, which specifically includes the following steps:

[0006] Establish a real-time data acquisition channel, extract key feature parameters of each construction stage, and construct a standardized feature vector;

[0007] Based on the standardized feature vector and the spatial topological relationship matrix of the site, a spatio-temporal correlation model between multi-source data is established to identify the risk transmission path: by calculating the spatio-temporal cross-correlation function between features, key risk nodes and transmission paths are identified;

[0008] Establish a risk assessment model based on the construction stage of the hierarchical early warning mechanism;

[0009] Dynamically optimize the electronic fence parameters according to the risk level and the construction stage.

[0010] Further, the establishment of the real-time data acquisition channel, the extraction of the key feature parameters of each construction stage, and the construction of the standardized feature vector specifically include:

[0011] Input: original geographic spatial data, historical disaster data, meteorological data, construction plan data, real-time monitoring data; wherein the original geographic spatial data includes DEM, slope, slope direction, water system, geological data;

[0012] Uniform the coordinates of corresponding spatial data of different data types to CGCS2000, obtain a spatial topological relationship matrix, the spatial topological relationship matrix being used for describing distances or correlations between different positions;

[0013] Remove outliers and missing values;

[0014] Perform standardization processing, normalize numerical data to the interval [0, 1], align multi-source data in space, and obtain standardized feature vectors of each data: i (t) = (x i (t) - μ i ) / σ i , where x i (t) is the ith original monitoring value, μ i and σ i are historical mean and standard deviation, μ i = (1 / N)∑x i (t), σ i = √[(1 / N)∑(x i (t)-μ i )²], N represents the number of data samples used for calculating historical statistics, and is usually the monitoring data of the last 30-90 days, to ensure the timeliness and stability of the statistics.

[0015] Further, the spatial topological relationship matrix of the residence based on the standardized feature vector establishes a spatio-temporal correlation model between multi-source data, and identifies a risk transmission path: by calculating the spatio-temporal cross-correlation function between features, a key risk node and a transmission path are identified, and specifically include:

[0016] Access the standardized feature vector after preprocessing and the spatial topological relationship matrix, and fuse multiple features into a comprehensive risk index through spatio-temporal correlation analysis: the fusion risk index R fused (t) = ∑ i ∑ j w ij * ρ(f i (t), f j (t-τ))* f i (t) * f j (t-τ), where w ij = exp(-d ij / λ) is a spatial attenuation weight, d ij is the spatial distance between the positions of features i and j, λ is an attenuation coefficient, ρ(f i (t), f j (t-τ)) is the correlation coefficient of feature i at time t and feature j at time t-τ, and τ is a time lag parameter; the fusion risk index Rfused (t) for reflecting the risk status under the joint action of multiple factors.

[0017] Further, the construction phase-based risk assessment model for establishing the hierarchical early warning mechanism further comprises: collecting construction phase data and identifying, inputting: fusing risk indicators R fused (t), construction phase identification S phase , according to L risk (t) = Φ(α * R fused (t) + β * H(S phase ) + γ * T(t)), wherein Φ is a sigmoid function that maps the output to between 0 and 1, H(S phase) is a construction phase risk base, different construction phases have different basic risk levels, which are realized by pre-setting parameters, for example, the roadbed construction phase is 0.3, the bridge erection phase is 0.7, and T(t) = dR fused (t) / dt is the change trend of the fused risk indicator, and α, β, and γ are weight coefficients.

[0018] Construction phase H(Sphase) value Risk characteristics:

[0019] Site preparation 0.2 Low risk, mainly ground work;

[0020] Foundation construction 0.5 Medium risk, involving deep foundation pits;

[0021] Main structure 0.7 High risk, more high-altitude work;

[0022] Equipment installation 0.6 Medium-high risk, involving heavy equipment;

[0023] High-risk phase (S phase ≥0.6): α=0.7, β=0.2, γ=0.1;

[0024] Medium-risk phase (0.3<S phase <0.6): α=0.5, β=0.3, γ=0.2;

[0025] Low-risk phase (S phase ≤0.3): α=0.3, β=0.4, γ=0.3;

[0026] Hierarchical early warning mechanism: divide the risk level calculated by the risk assessment model:

[0027] Low risk: L risk (t) ≤ 0.3;

[0028] Medium risk: 0.3<Lrisk (t) ≤ 0.6;

[0029] High risk: 0.6 < L risk (t) ≤ 0.8;

[0030] Very high risk: L risk (t) > 0.8.

[0031] Further, the electronic fence parameters are dynamically optimized according to the risk level and the construction stage, specifically including:

[0032] Input: risk level L risk (t), personnel distribution P dist , construction area A work According to the target function: min J(Θ) = ω1 * Risk_Exposure + ω2 * Operation_Cost + ω3 *Adjustment_Cost, the constraint condition is set as the position data of all electronic fences are within the boundary B of the construction project; wherein, Risk_Exposure = ∫∫ B L risk (x,y) * ρ(x,y) dxdy, represents the product of risk and personnel distribution density in the fence area, i.e. risk exposure, Operation_Cost represents operation cost, which is proportional to fence perimeter and monitoring density, Adjustment_Cost represents adjustment cost; c1: unit length fence construction and maintenance cost, value range 100-300 yuan / m, c2: unit monitoring point construction and operation cost, value range 5000-10000 yuan / individual, perimeter(B) is the fence length, D_mon is the number of monitoring points; Θ(t) is the cost function when this adjustment;

[0033] The weights ω1, ω2, ω3 are dynamically adjusted according to the construction stage and the risk level, in high risk, the adjustment cost weight is reduced, in low risk, the operation cost weight is increased, for example, in high risk, ω1 (risk exposure weight) is increased, ω3 (adjustment cost weight) is reduced, to preferentially reduce risk. ω1 = 0.4 + 0.3 × L risk (t), risk exposure weight increases with risk level, ω2 = 0.3 - 0.2 × L risk (t), operation cost weight decreases with risk level, ω3 = 0.3 - 0.1 × L risk (t), adjustment cost weight decreases with risk level,

[0034] Further, the manual adjustment record of the electronic fence parameter optimization stage is also converted into a learning sample, specifically including: collecting the record of manual adjustment, including the system recommended parameter before adjustment and the manual actual parameter after adjustment, recording the context information at the time of adjustment, including: risk level, construction stage, personnel distribution; converting the context information into a feature vector, including risk level, construction stage code, personnel distribution characteristics, taking the system recommended parameter and the manual actual parameter as the target variable; using a supervised learning model for supervised learning, taking each manual adjustment as a training sample to optimize the adjustment strategy, wherein the loss function adopts a mean square error loss function for minimizing the difference between the system recommended parameter and the manual actual parameter, and a regularization term is added to prevent overfitting, and a gradient descent algorithm is used to train the supervised learning model to minimize the loss function;

[0035] The collected manual adjustment record dataset is divided into a training set and a validation set, the trained supervised learning model is verified for accuracy using the validation set, and then when the system needs to adjust the electronic fence parameter, the trained model is used to generate a recommended parameter according to the current context features.

[0036] Further, it further specifically includes interpolating the continuous manual adjustment parameters to generate intermediate states.

[0037] The present application realizes real-time monitoring and early warning of traffic construction project site safety through multi-source data fusion and dynamic risk assessment, can effectively identify risk transmission paths, dynamically adjust electronic fence parameters, and reduce safety risks. At the same time, by introducing artificial feedback learning, the system can continuously optimize the adjustment strategy, reduce manual intervention, and improve safety management efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0038] The drawings described herein are used to provide further understanding of the embodiments of the present application, and form a part of the present application, and do not constitute a limitation on the embodiments of the present application. In the drawings:

[0039] Figure 1 A flowchart of a traffic construction project site dynamic safety analysis method based on multi-source data fusion is provided for an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0040] The exemplary embodiments will be described in detail here, and their examples are shown in the drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application.

[0041] The traffic construction project site dynamic safety analysis method based on multi-source data fusion provided by the present application aims to solve the above technical problems of the prior art.

[0042] The technical solutions of the present application and how the technical solutions solve the above technical problems will be described in detail below with specific examples. The following specific examples can be combined with each other, and the same or similar concepts or processes can not be described again in some examples. The embodiments of the present application will be described below with reference to the accompanying drawings.

[0043] Embodiment 1:

[0044] As shown in the following table, the present embodiment provides a traffic construction project site dynamic safety analysis method based on multi-source data fusion, including the following steps: Figure 1

[0045] Step 1: Establish a real-time data acquisition channel, extract key feature parameters of each construction stage, and construct a standardized feature vector;

[0046] Step 2: Based on the standardized feature vector, the spatial topological relationship matrix of the site, establish a spatio-temporal correlation model between multi-source data, and identify the risk transmission path: by calculating the spatio-temporal cross-correlation function between features, identify the key risk nodes and transmission path;

[0047] Step 3: Establish a risk assessment model based on the construction stage of the hierarchical early warning mechanism;

[0048] Step 4: Dynamically optimize the electronic fence parameters according to the risk level and construction stage;

[0049] Based on steps 1-4, in a highway project, real-time acquisition of various types of monitoring data at the construction site is implemented as follows:

[0050] Step 1:

[0051] Input original data:

[0052] Meteorological data: G(t) = {Rainfall: 45mm / h, Wind speed: 8m / s, Temperature: 28℃};

[0053] Geological data: D(t) = {Displacement rate: 2.1mm / d, Groundwater level: 3.2m};

[0054] Personnel data: P(t) = {Personnel density: 0.8 person / ㎡, Aggregation degree: 0.6};

[0055] Construction data: C(t) = {Stage: Subgrade construction, Progress: 65%};

[0056] Standardization calculation:

[0057] ​f_rain(t) = (45 - 25) / 15 = 1.33 / / historical mean 25mm / h, standard deviation 15;

[0058] f_displacement(t) = (2.1 - 1.5) / 0.8 = 0.75 / / historical mean 1.5mm / d, standard deviation 0.8;

[0059] f_density(t) = (0.8 - 0.5) / 0.3 = 1.0 / / historical mean 0.5 person / m2, standard deviation 0.3;

[0060] Output feature vector:

[0061] F(t) = [1.33, 0.75, 1.0, 0.6, 0.65,...];

[0062] Step 2:

[0063] Input data:

[0064] F(t) = [rain: 1.33, displacement: 0.75, density: 1.0,...];

[0065] F(t-6h) = [rain: 0.8, displacement: 0.6, density: 0.9,...] / / data 6 hours ago;

[0066] Temporal and spatial weight calculation:

[0067] w_rain-displacement = exp(-150 / 500) = 0.74 / / monitoring point distance 150m, λ=500m;

[0068] ρ(rain(t), displacement(t-6)) = 0.68 / / time-lag correlation coefficient;

[0069] Fused risk calculation:

[0070] Rfused(t) = 0.74×0.68×1.33×0.6 + 0.82×0.72×0.75×0.8 +... =0.40 + 0.44 +... = 2.35;

[0071] Step 3:

[0072] Input parameters:

[0073] Rfused(t) = 2.35;

[0074] Sphase = "bridge erection" → H(S phase) = 0.7 / / high risk stage base;

[0075] T(t) = (2.35 - 1.8) / 6 = 0.092 / / risk change rate in 6 hours;

[0076] Risk level calculation:

[0077] Lrisk(t) = 1 / (1+exp(-(0.6×2.35 + 0.3×0.7 + 0.1×0.092)))

[0078] = 1 / (1+exp(-(1.41 + 0.21 + 0.009)))

[0079] = 1 / (1+exp(-1.629)) = 1 / (1+0.196) = 0.836;

[0080] Risk level determination: L risk = 0.836>0.8 → very high risk;

[0081] Step 4:

[0082] Current fence parameters:

[0083] Θ(t-1) = {boundary length: 1200m, buffer zone: 50m, monitoring points: 8};

[0084] Risk exposure calculation:

[0085] Risk_Exposure = ∫∫ L risk (x,y)·ρ(x,y) dxdy = 0.836 × 0.8 × 1200 ×50 = 40128;

[0086] Operation cost:

[0087] Operation_Cost = 200×1200 + 5000×8 = 240000 + 40000 = 280000;

[0088] Adjustment cost:

[0089] Adjustment_Cost = ‖Θ(t) - Θ(t-1)‖² = (1250-1200)² = 2500;

[0090] Objective function optimization:

[0091] min J(Θ) = 0.5x40128 + 0.3x280000 + 0.2x2500 = 20064 + 84000 + 500 = 104564;

[0092] Optimized parameters:

[0093] Θ(t) = {Boundary length: 1250m, Buffer: 60m, Monitoring points: 10}

[0094] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, i.e., they may be located in one place or distributed over multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the embodiment of the present application.

[0095] In addition, each functional module in the embodiment of the present application can be integrated into a processing module, or each module can exist physically alone, or two or more modules can be integrated into one module. The integrated module can be realized in the form of hardware or in the form of hardware plus software functional module.

[0096] Those skilled in the art will appreciate that embodiments of the application can be provided as a method or system. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects.

[0097] It should also be noted that the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles or devices including a series of elements not only include those elements, but also include other elements not explicitly listed, or inherent to such processes, methods, articles or devices. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0098] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the scope of claims of the present application.

[0099] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.

[0100] It is to be understood that the application is not limited to the precise structures hereinabove described and shown in the drawings, for purposes of illustration and education only, and that various modifications and changes can be made therein without departing from the scope thereof. The scope of the application is indicated by the claims appended hereto.

Claims

1. A method for dynamic safety analysis of transportation construction project sites based on multi-source data fusion, characterized in that, Includes the following steps: Step 1: Establish a real-time data acquisition channel, extract key feature parameters for each construction stage, and construct standardized feature vectors; Step 2: Establish a spatiotemporal correlation model among multi-source data based on standardized feature vectors and the spatial topology matrix of the site to identify risk transmission paths: By calculating the spatiotemporal cross-correlation function between features, key risk nodes and transmission paths are identified. Specifically, this includes: accessing the preprocessed standardized feature vectors and spatial topology matrix, and through spatiotemporal correlation analysis, fusing multiple features into a comprehensive risk indicator: the fused risk indicator R. fused (t) = ∑ i ∑ j w ij * ρ(f i (t), f j (t-τ)) * f i (t) * f j (t-τ), where w ij = exp(-d ij / λ) is the spatial decay weight, d ij ρ(f) is the spatial distance between the locations of features i and j, λ is the attenuation coefficient, and ρ(f) is the attenuation coefficient. i (t), f j (t-τ) is the correlation coefficient between feature i at time t and feature j at time t-τ, where τ is the time delay parameter; the fusion risk index R fused (t) is used to reflect the risk situation under the combined effect of multiple factors; Step 3: Establish a risk assessment model based on the construction phase for a tiered early warning mechanism; Step 4: Dynamically optimize the electronic fence parameters based on the risk level and construction stage, specifically including: Input: Risk level L risk (t), personnel distribution P dist Construction Area A work The objective function is calculated as follows: min J(Θ) = ω1 * Risk_Exposure + ω2 * Operation_Cost + ω3 * Adjustment_Cost, with the constraint that all geofence location data must be within the construction project boundary B; where Risk_Exposure = ∫∫ B L risk (x,y) * ρ(x,y) dxdy represents the product of risk and population density within the fenced area, i.e., risk exposure. Operation_Cost represents the operating cost, which is proportional to the fence perimeter and monitoring density. Adjustment_Cost represents the adjustment cost. The weights ω1, ω2, and ω3 are dynamically adjusted based on the construction stage and risk level. During high-risk periods, the weight of adjustment cost is reduced, while during low-risk periods, the weight of operating cost is increased. Furthermore, the manual adjustment records during the optimization of electronic fence parameters are transformed into learning samples. Specifically, this includes: collecting records of manual adjustments, including the system-recommended parameters before adjustment and the actual manual parameters after adjustment, recording the contextual information during adjustment, including risk level, construction stage, and personnel distribution; converting the contextual information into feature vectors, including risk level, construction stage encoding, and personnel distribution characteristics, using the system-recommended parameters and the actual manual parameters as target variables; and using a supervised learning model for supervised learning, where each... The manual adjustments are used as training samples to optimize the adjustment strategy. The loss function adopts the mean squared error loss function to minimize the difference between the system's recommended parameters and the actual manual parameters, and a regularization term is added to prevent overfitting. The gradient descent algorithm is used to train the supervised learning model, and the model parameters are adjusted to minimize the loss function. The collected manual adjustment record dataset is divided into a training set and a validation set. The accuracy of the trained supervised learning model is verified using the validation set. Then, when the system needs to adjust the electronic fence parameters, the trained model is used to generate recommended parameters based on the current context features. Specifically, this also includes interpolating the continuous manual adjustment parameters to generate intermediate states.

2. The method for dynamic safety analysis of transportation construction project sites based on multi-source data fusion according to claim 1, characterized in that, The establishment of a real-time data acquisition channel, extraction of key feature parameters for each construction stage, and construction of standardized feature vectors specifically include: Inputs: raw geospatial data, historical disaster data, meteorological data, construction plan data, and real-time monitoring data; among which, raw geospatial data includes DEM, slope, aspect, water system, and geological data; The coordinates of corresponding spatial data of different data types are unified to CGCS2000 to obtain a spatial topology matrix, which is used to describe the distance or association between different locations. Remove outliers and missing values; Standardization is performed to normalize numerical data to the [0,1] interval, and multi-source data are spatially aligned to obtain the standardized feature vector of each data: fᵢ(t) = (xᵢ(t) - μᵢ) / σᵢ, where xᵢ(t) is the i-th original monitoring value, μᵢ and σᵢ are the historical mean and standard deviation, μᵢ = (1 / N)∑xᵢ(t), σᵢ = √[(1 / N)∑(xᵢ(t)-μᵢ)²], and N represents the number of data samples used to calculate historical statistics.

3. The method for dynamic safety analysis of transportation construction project sites based on multi-source data fusion according to claim 2, characterized in that, The risk assessment model based on the construction phase for establishing a tiered early warning mechanism further includes: collecting and labeling construction phase data, and inputting: a fusion risk index R. fused (t), Construction Stage Marker S phase According to L risk (t) = Φ(α* R fused (t) + β * H(S phase ) + γ * T(t)), where Φ is the sigmoid function that maps the output to the range of 0-1, H(S phase) This is the risk baseline for the construction phase. Different construction phases have different basic risk levels, which are achieved by pre-setting parameters: T(t) = dR fused (t) / dt represents the changing trend of the integrated risk indicator, and α, β, and γ are weighting coefficients. Tiered early warning mechanism: This involves classifying the risk levels calculated by the risk assessment model. Low risk: L risk (t) ≤ 0.3; Medium risk: 0.3 < L risk (t) ≤ 0.6; High risk: 0.6 < L risk (t) ≤ 0.8; Extremely high risk: L risk (t) > 0.

8.

4. The method for dynamic safety analysis of transportation construction project sites based on multi-source data fusion according to claim 2, characterized in that, in, The time period for acquiring historical data of μᵢ and σᵢ is set to seven or thirty days, depending on the amount of historical data collected and the collection period.

Images