Construction site safety early warning method and system based on multi-dimensional data mining

Abstract

The invention discloses a construction site safety early warning method and system based on multi-dimensional data mining, and relates to the technical field of construction site early warning, and the method comprises the steps: collecting and preprocessing multi-dimensional data, and forming a multi-dimensional data set; performing feature extraction on the multi-dimensional data set by using a convolutional neural network, and fusing time sequence information through a space-time dynamic weighting mechanism to generate a multi-dimensional risk feature set; based on the multi-dimensional risk feature set, the risk features are converted into nodes, the incidence relation between the risk features is converted into edges, a causal correlation structure of the nodes and the edges is formed, and a dynamic risk map is constructed; analyzing the association strength among the risk factors by using the dynamic risk map, evaluating the overall risk level, and generating a risk event list; performing trend prediction on the risk events in the risk event list based on the time sequence to obtain a risk trend index; and performing dynamic early warning according to the risk trend index, and triggering a safety response. According to the invention, through multi-dimensional data mining and analysis, intelligentization of construction site safety early warning is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of construction site early warning, and in particular to a construction site safety early warning method and system based on multidimensional data mining. Background Art

[0002] In traditional construction site safety management, risk warning mainly relies on manual experience and static monitoring methods, which leads to insufficient prediction of risk events and inability to achieve dynamic and real-time risk trend analysis and warning. Traditional methods are usually based on data from only one dimension, such as environmental data or equipment status, and lack the comprehensive use of multi-dimensional data. In addition, these methods cannot effectively capture the correlation between risk factors and their dynamic characteristics over time, resulting in inaccurate risk assessment results and difficulty in timely identification of potential high-risk areas.

[0003] Especially for the prediction of risk trends in time series, traditional methods mostly use simple statistical analysis or fixed models, which cannot handle the dynamic changes of multiple risk events in complex scenarios. The characteristic values ​​of risk events are easily affected by many factors, and traditional models are difficult to capture their inherent time and space dependence characteristics, resulting in low prediction accuracy of future risk trends and lagging early warning mechanisms. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a construction site safety early warning method based on multidimensional data mining to solve the problem of lack of comprehensive analysis of multidimensional data and accurate prediction of future risk trends in traditional construction site safety management.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a construction site safety early warning method based on multidimensional data mining, which includes collecting multidimensional data and preprocessing it to form a multidimensional data set;

[0008] Convolutional neural networks are used to extract features from multidimensional data sets, and time series information is fused through a spatiotemporal dynamic weighting mechanism to generate a multidimensional risk feature set;

[0009] Based on the multi-dimensional risk feature set, the risk features are converted into nodes, and the correlation between risk features is converted into edges, forming a causal correlation structure between nodes and edges, and constructing a dynamic risk map;

[0010] Use dynamic risk maps to analyze the correlation strength between risk factors, assess the overall risk level, and generate a list of risk events;

[0011] Based on the time series, the risk events in the risk event list are trend predicted to obtain the risk trend index;

[0012] Provide dynamic warnings based on risk trend indexes and trigger security responses.

[0013] As a preferred solution of the construction site safety early warning method based on multidimensional data mining of the present invention, wherein: the collection is carried out by using a drone and a sensor to collect data in coordination;

[0014] The multidimensional data includes environmental data, worker data, equipment data, risk events, time and coordinates;

[0015] The environmental data include temperature and humidity, dust concentration, nitrogen dioxide concentration, noise intensity and wind speed;

[0016] The worker data includes the worker's heart rate, location, and behavior;

[0017] The equipment data includes vibration, pressure and displacement of the equipment;

[0018] Said risk events include worker fatigue, worker heat stroke, equipment failure and environmental degradation;

[0019] The preprocessing includes data cleaning, normalization and spatiotemporal alignment.

[0020] As a preferred solution of the construction site safety early warning method based on multidimensional data mining described in the present invention, a convolutional neural network is used to extract features from a multidimensional data set, and time series information is fused through a spatiotemporal dynamic weighting mechanism to generate a multidimensional risk feature set. The specific steps are as follows:

[0021] The multidimensional data set is input into the convolutional neural network, and the convolutional layer of the convolutional neural network is used to extract features from the multidimensional data to obtain spatial features;

[0022] Use the pooling layer of the convolutional neural network to reduce the dimension and denoise the spatial features to extract key spatial features;

[0023] According to the time series information of the collected data, a time weight decreasing according to an exponential decay function is assigned to each key spatial feature;

[0024] Using a weighted fusion method, the time series information is combined with the spatial features obtained by the convolutional neural network to generate a multidimensional risk feature set.

[0025] As a preferred solution of the construction site safety early warning method based on multidimensional data mining described in the present invention, the risk features are converted into nodes based on a multidimensional risk feature set, and the correlation between risk features is converted into edges to form a causal correlation structure between nodes and edges, and a dynamic risk map is constructed. The specific steps are as follows:

[0026] Extract risk factors from the multidimensional risk feature set and use each risk factor as a node;

[0027] Each node contains the characteristic value of the risk factor and spatiotemporal information;

[0028] Analyze the correlation between risk factors and convert them into edges between nodes;

[0029] Using the Granger causal analysis method, remove the edges with low correlation and obtain the causal inference results;

[0030] The causal inference results are combined with time series information to generate a dynamic risk map that changes over time.

[0031] As a preferred solution of the construction site safety early warning method based on multidimensional data mining described in the present invention, the following specific steps are used to analyze the correlation strength between risk factors, evaluate the overall risk level, and generate a risk event list.

[0032] Identify specific high-risk areas and potential risk events based on risk factors and correlations in the dynamic risk map;

[0033] Calculate the risk weight of each node based on the correlation between nodes and edges in the dynamic risk graph;

[0034] Calculate the overall site risk level based on the number and risk weights of risk factors in high-risk areas;

[0035] The risk factors belonging to risk events that appear in high-risk areas are arranged according to the possibility of occurrence to generate a list of risk events.

[0036] As a preferred solution of the construction site safety early warning method based on multidimensional data mining described in the present invention, wherein: based on the time series, the risk events in the risk event list are trend predicted to obtain the risk trend index. The specific steps are as follows:

[0037] Divide the risk event list and the corresponding time series data according to the time window to form a training set, a test set, and a validation set;

[0038] As a preferred solution of the construction site safety early warning method based on multidimensional data mining described in the present invention, wherein: dynamic early warning is performed according to the risk trend index and a safety response is triggered. The specific steps are as follows:

[0039] Analyze the risk trend index to determine low risk threshold and medium risk threshold;

[0040] When the risk trend index is less than or equal to the low risk threshold, a low-level warning is issued to remind workers to pay attention;

[0041] When the risk trend index is greater than the low risk threshold and less than or equal to the medium risk threshold, a medium-level warning is issued, related operations are suspended, and equipment and environmental status are checked;

[0042] When the risk trend index is greater than the medium-risk threshold, a high-level warning is issued, personnel are evacuated urgently, drones are launched to inspect high-risk areas, and risk changes are monitored in real time.

[0043] In a second aspect, the present invention provides a construction site safety early warning system based on multidimensional data mining, including a data acquisition module, a feature set module, a risk map module, a risk assessment module, a trend prediction module and an early warning module;

[0044] The data acquisition module is used to collect multidimensional data and perform preprocessing to form a multidimensional data set;

[0045] The feature set module is used to extract features from a multidimensional data set using a convolutional neural network and fuse time series information through a spatiotemporal dynamic weighting mechanism to generate a multidimensional risk feature set;

[0046] The risk map module is used to convert risk features into nodes based on a multi-dimensional risk feature set, convert the correlation between risk features into edges, form a causal correlation structure between nodes and edges, and construct a dynamic risk map;

[0047] The risk assessment module is used to analyze the correlation strength between risk factors using a dynamic risk map, assess the overall risk level, and generate a risk event list;

[0048] The trend prediction module is used to perform trend prediction on risk events in the risk event list based on the time series to obtain a risk trend index;

[0049] The early warning module is used to issue dynamic early warnings based on the risk trend index and trigger security responses.

[0050] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the construction site safety warning method based on multidimensional data mining as described in the first aspect of the present invention.

[0051] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the construction site safety warning method based on multidimensional data mining as described in the first aspect of the present invention.

[0052] The beneficial effects of the present invention are as follows: the present invention realizes the intelligent and dynamic monitoring and early warning of construction site safety through multidimensional data mining and time series analysis. Combined with multidimensional data such as environment, workers, equipment, etc. collected by drones and sensors, the safety risk factors of the construction site are fully covered. The convolutional neural network and spatiotemporal dynamic weighting mechanism are used to extract features, and the dynamic risk map is constructed in combination with Granger causal analysis, which can intuitively present the correlation between risk factors and their changes over time, providing support for accurate risk assessment. The ARIMA time series model is introduced to predict the trend of risk events and generate a risk trend index, which can capture the dynamic changes of risks over time and provide a scientific basis for risk early warning. Low, medium and high level dynamic early warnings are carried out according to the risk trend index, and corresponding safety response measures are triggered to improve the initiative and effectiveness of construction site safety management and reduce the harm of risk events to personnel and equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0054] Figure 1 This is a flow chart of the construction site safety early warning method based on multidimensional data mining in Example 1.

[0055] Figure 2 This is a module diagram of the construction site safety early warning system based on multidimensional data mining in Example 1. DETAILED DESCRIPTION

[0056] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0057] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a construction site safety early warning method based on multidimensional data mining, comprising the following steps:

[0058] S1, collect multidimensional data and preprocess them to form a multidimensional data set, including the following steps:

[0059] Use drones and sensors to collaboratively collect multi-dimensional data, including environmental data, worker data, equipment data, risk events, time and coordinates. Environmental data includes temperature and humidity, dust concentration, nitrogen dioxide concentration, noise intensity and wind speed. Worker data includes workers' heart rate, position and behavior. Equipment data includes equipment vibration, pressure and displacement. Risk events include worker fatigue, worker heat stroke, equipment failure and environmental deterioration. Preprocessing includes data cleaning, normalization and spatiotemporal alignment.

[0060] Specifically, drones are used to collect environmental data over a large area, including temperature and humidity, dust concentration, nitrogen dioxide concentration, noise intensity, and wind speed. Sensors are deployed in key areas of the construction site to monitor the status of equipment (vibration, pressure, displacement) and workers (heart rate, position, behavior) in real time. Drones and sensors work together through wireless networks to collect data regularly or on demand, covering the entire site and key areas. The frequency of data collection is adjusted as needed, for example, every 5 minutes, or the frequency of data collection is increased in key processes.

[0061] Data cleaning is to check the integrity of collected data, eliminate data with serious missing values, and handle outliers. For example, according to statistical rules, data that exceeds a reasonable range is eliminated, duplicate data is removed, and each piece of data is ensured to be unique.

[0062] Normalization processing is to convert data of different dimensions (such as temperature and vibration) into dimensionless values ​​through normalization methods (such as min-max normalization or z-score normalization) to ensure that all data are in the same range for subsequent model processing.

[0063] Time and space alignment is to align data according to timestamps to ensure that data of different dimensions can match at the same time point, spatially cluster data according to coordinates, integrate multidimensional data at the same location, and interpolate missing data in time and space (such as linear interpolation). Data is initially screened by edge computing devices and then uploaded to the data center.

[0064] It should be noted that the collected multidimensional data comprehensively covers multiple risk factors such as environment, workers, and equipment, which can reflect the overall safety situation of the construction site and establish a spatiotemporal dynamic model through time and coordinate information. The comprehensiveness and accuracy of multidimensional data provide a solid data foundation for subsequent feature extraction, risk assessment, and trend prediction.

[0065] S2. Use convolutional neural networks to extract features from multidimensional data sets and fuse time series information through a spatiotemporal dynamic weighting mechanism to generate a multidimensional risk feature set, including the following steps:

[0066] The multidimensional data set is input into the convolutional neural network, and the convolutional layer of the convolutional neural network is used to extract features from the multidimensional data to obtain spatial features. Specifically, the multidimensional data set is first sorted, and the environmental data, worker data, equipment data, etc. are arranged according to the time and space dimensions to form a structured multidimensional input tensor.

[0067] Convolution kernels are used to perform convolution operations on multidimensional data to extract local spatial information. The convolution layer moves on the input tensor through a sliding window to capture feature patterns at different locations. Each convolution kernel corresponds to a specific feature pattern, such as the temperature and humidity change pattern in environmental data or the feature pattern of worker behavior.

[0068] The pooling layer of the convolutional neural network is used to reduce the dimension and denoise the spatial features and extract key spatial features. Specifically, the feature map output by the convolutional layer is reduced in dimension through the maximum pooling or average pooling operation, and the most significant features (such as the maximum value or average value) are extracted in each pooling area to reduce redundant features, remove noise and local outliers, enhance the model's ability to express key features, generate simplified and more compact spatial features, and provide a basis for subsequent weighted fusion processing.

[0069] According to the time series information of the collected data, a time weight that decreases according to the exponential decay function is assigned to each key spatial feature. Specifically, each spatial feature is assigned an initial time weight according to its corresponding timestamp. Using the exponential decay function, features closer to the current time point are assigned higher weights, and features farther away from the current time point are assigned lower weights. The importance of spatial features is adjusted in combination with the time weight to ensure that the model focuses on current and recent key information.

[0070] Using the weighted fusion method, the time series information is combined with the spatial features obtained by the convolutional neural network to generate a multidimensional risk feature set. Specifically, each spatial feature is multiplied by the corresponding time weight to generate a weighted feature, which is stacked in time order to form a spatiotemporal feature matrix. The spatiotemporal feature matrix is ​​nonlinearly transformed using a fully connected layer to further fuse time and space information and output a multidimensional risk feature set, which contains spatiotemporal dynamic information of multidimensional features such as environment, workers, and equipment.

[0071] It should be noted that the spatial feature extraction capability of convolutional neural networks combined with a dynamic time weighting mechanism can effectively capture the spatiotemporal correlation characteristics of multidimensional data. The time weight is assigned by an exponential decay function, which enhances the model's ability to focus on recent data while reducing the interference of long-term data on risk prediction. The multidimensional risk feature set provides accurate and efficient basic data for subsequent causal analysis and dynamic risk map construction.

[0072] S3. Based on the multi-dimensional risk feature set, the risk features are converted into nodes, and the correlation between risk features is converted into edges to form a causal correlation structure between nodes and edges, and a dynamic risk map is constructed, including the following steps:

[0073] Risk factors are extracted from the multidimensional risk feature set, and each risk factor is used as a node. Specifically, each feature in the multidimensional risk feature set is classified to identify risk factors in categories such as environment, workers, and equipment. Each risk factor, such as temperature and humidity, worker heart rate, and equipment vibration, is used as an independent node, and the attributes of the node include the characteristic value, category, and associated weight of the risk factor.

[0074] Each node contains the characteristic value and spatiotemporal information of the risk factor. Specifically, the current characteristic value and change trend of the corresponding risk factor are stored in the node, including the timestamp information of the risk factor, which is used to dynamically update the graph, record the spatial coordinate information of the node, and identify the specific location of the risk factor.

[0075] Analyze the correlation between risk factors and convert them into edges between nodes. Specifically, calculate the correlation between risk factors, for example, use the Pearson correlation coefficient or mutual information method, connect risk factors with high correlation as edges, and assign edge weights. The attributes of the edges include correlation strength and direction, and the directionality reflects the causal relationship.

[0076] Using the Granger causality analysis method, we remove edges with low correlation and obtain causal inference results. Specifically, we use Granger causality analysis to detect whether the causal relationship between risk factors is significant. By selecting typical risk factors and edges to set thresholds, we remove edges with insignificant causal relationships or low correlations and output the optimized causal association structure.

[0077] The causal inference results are combined with time series information to generate a dynamic risk map that changes over time. Specifically, the nodes and edges in the causal association structure are dynamically updated, and the changes in the risk factor relationship are reflected in the time series information. The dynamic risk map contains nodes, edges and their attribute values ​​that change over time. The map can show the risk distribution and causal relationship of the construction site in real time.

[0078] It should be noted that the dynamic risk map can intuitively reflect the correlation strength and changing trends between risk factors through causal analysis and time series updates. This map provides an important basis for accurately identifying high-risk areas and potential risk events, and provides spatiotemporal dynamic support for subsequent risk assessment and early warning.

[0079] S4. Use the dynamic risk map to analyze the correlation strength between risk factors, evaluate the overall risk level, and generate a list of risk events, including the following steps:

[0080] According to the risk factors and correlations in the dynamic risk map, specific high-risk areas and potential risk events are identified. Specifically, the nodes and edges with high correlation strength in the dynamic risk map are screened, and the high-risk areas are identified according to the spatial coordinate clustering of the nodes. Potential risk events such as worker fatigue and equipment failure are extracted from the high-risk areas.

[0081] Based on the correlation between nodes and edges in the dynamic risk graph, the risk weight of each node is calculated. Specifically, the initial weight of the risk factor is calculated based on the characteristic value of the node (such as worker heart rate, equipment vibration), and the risk weight of the node is adjusted based on the correlation strength of the edge, and the final risk weight of each node is output.

[0082] The risk level of the entire construction site is calculated based on the number and risk weight of risk factors in the high-risk area. Specifically, the number of risk factors in the high-risk area is counted, the weighted average of the risk weights of all nodes in the high-risk area is calculated, and the risk level of the entire construction site is determined based on the preset risk level classification standards.

[0083] The risk factors that appear in the high-risk area and belong to risk events are arranged according to the probability of occurrence to generate a risk event list. Specifically, according to the weight and time trend of the risk factor, the probability of occurrence is predicted, and the high-probability risk factors are sorted by priority to generate a risk event list, and a list containing information such as the type, location, and probability of occurrence of the risk event is output.

[0084] It should be noted that through the analysis of nodes and edges in the dynamic risk map, high-risk areas can be accurately located and the overall risk level of the construction site can be quantified. The risk event list provides decision makers at the construction site with a clear risk priority ranking, making it easier to take targeted safety response measures.

[0085] S5. Predict the trend of risk events in the risk event list based on the time series to obtain the risk trend index, including the following steps:

[0086] The risk event list and the corresponding time series data are divided according to the time window to form a training set, a test set, and a validation set. Specifically, the time window is divided, and the historical data of each risk event is divided into a training set, a test set, and a validation set to ensure that the training set data covers a sufficient time range. The test set and validation set are used for model evaluation.

[0087] The time series data of each risk event is expressed as,

[0088] {Y(t)|t=1,2,…,T w};

[0089] Among them, Y(t) is the characteristic value of the risk event at time t, t is the index of the time point, T wFor the T w a point in time.

[0090] Perform stationarity detection and differential processing on the time series data of each risk event to construct the input features of the time series model. The differential formula is expressed as:

[0091] ΔY(t)=Y(t)-Y(t-1), t>1;

[0092] Among them, ΔY(t) is the difference between the characteristic value of the risk event at time t and the characteristic value of the risk event at time t-1, Y(t) is the characteristic value of the risk event at time t, and Y(t-1) is the characteristic value of the risk event at time t-1.

[0093] The ARIMA time series analysis model is trained using the training set, test set, and validation set to obtain the risk trend model, which is expressed as:

[0094]

[0095] Where T is the risk trend index, n is the total number of risk events, c is the fixed base value of the trend, i is the number index of the risk event, p is the order of the autoregressive term, j is the counter index of the autoregressive term, k is the counter index of the moving average term, ∈ is the random error, and Y i (t) is the characteristic value of the i-th risk event at time t, R i is the weight of the risk event, φ j is the autoregressive coefficient, θ k is the moving average coefficient.

[0096] It should be noted that the use of the ARIMA model to capture the temporal dynamic characteristics of risk events can accurately predict future risk trends, and the comprehensive trend index provides a quantitative basis for construction site safety warnings, facilitating further graded responses.

[0097] S6. Dynamically warn based on the risk trend index and trigger a security response, including the following steps:

[0098] Analyze the risk trend index to determine the low risk threshold and the medium risk threshold. Specifically, divide the risk trend index into three risk ranges: low, medium, and high.

[0099] When the risk trend index is less than or equal to the low risk threshold, a low-level warning is issued to remind workers to pay attention. Specifically, a low-level warning message is sent to construction site workers, suggesting that workers strengthen their safety awareness and pay attention to changes in the surrounding environment.

[0100] When the risk trend index is greater than the low risk threshold and less than or equal to the medium risk threshold, a medium-level warning is issued, related operations are suspended, and equipment and environmental conditions are checked. Specifically, operations in high-risk areas are stopped, safety inspectors are dispatched to check equipment and environmental conditions, and drones are launched to inspect high-risk areas.

[0101] When the risk trend index is greater than the medium risk threshold, a high-level warning will be issued, personnel will be evacuated urgently, drones will be launched to inspect high-risk areas, and risk changes will be monitored in real time. Specifically, all personnel in high-risk areas will be evacuated urgently, drones will be launched to monitor high-risk areas in real time, and emergency management departments will be notified to take further measures.

[0102] It should be noted that through the graded early warning mechanism, safety response measures can be dynamically adjusted according to the risk trend index. The three-level early warnings of low, medium and high cover the full life cycle management of different risk levels, effectively reducing the probability of safety accidents on construction sites.

[0103] This embodiment also provides a construction site safety early warning system based on multidimensional data mining, including: a data acquisition module, a feature set module, a risk map module, a risk assessment module, a trend prediction module and an early warning module;

[0104] A data acquisition module is used to collect multidimensional data and perform preprocessing to form a multidimensional data set;

[0105] The feature set module is used to extract features from multidimensional data sets using convolutional neural networks and fuse time series information through a spatiotemporal dynamic weighting mechanism to generate a multidimensional risk feature set;

[0106] The risk graph module is used to convert risk features into nodes and the correlation between risk features into edges based on a multi-dimensional risk feature set, forming a causal correlation structure between nodes and edges and constructing a dynamic risk graph;

[0107] The risk assessment module is used to analyze the correlation strength between risk factors using dynamic risk maps, evaluate the overall risk level, and generate a list of risk events;

[0108] The trend prediction module is used to predict the trend of risk events in the risk event list based on the time series and obtain the risk trend index;

[0109] The early warning module is used to issue dynamic early warnings based on risk trend indexes and trigger security responses.

[0110] This embodiment also provides a computer device, which is suitable for the construction site safety early warning method based on multidimensional data mining, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the construction site safety early warning method based on multidimensional data mining proposed in the above embodiment.

[0111] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0112] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the construction site safety early warning method based on multidimensional data mining proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0113] In summary, the present invention realizes the intelligent and dynamic monitoring and early warning of construction site safety through multidimensional data mining and time series analysis. Combined with the multidimensional data of the environment, workers, equipment, etc. collected by drones and sensors, the safety risk factors of the construction site are fully covered. The convolutional neural network and the spatiotemporal dynamic weighting mechanism are used to extract features, and the dynamic risk map is constructed in combination with Granger causal analysis, which can intuitively present the correlation between risk factors and their changes over time, providing support for accurate risk assessment. The ARIMA time series model is introduced to predict the trend of risk events and generate a risk trend index, which can capture the dynamic changes of risks over time and provide a scientific basis for risk early warning. According to the risk trend index, low, medium and high level dynamic early warnings are carried out, and corresponding safety response measures are triggered to improve the initiative and effectiveness of construction site safety management and reduce the harm of risk events to personnel and equipment. .

[0114] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A construction site safety early warning method based on multidimensional data mining, characterized in that: include, Collect multidimensional data and preprocess them to form a multidimensional data set; Convolutional neural networks are used to extract features from multidimensional data sets, and time series information is fused through a spatiotemporal dynamic weighting mechanism to generate a multidimensional risk feature set; Based on the multi-dimensional risk feature set, the risk features are converted into nodes, and the correlation between risk features is converted into edges, forming a causal correlation structure between nodes and edges, and constructing a dynamic risk map; Use dynamic risk maps to analyze the correlation strength between risk factors, assess the overall risk level, and generate a list of risk events; Based on the time series, the risk events in the risk event list are trend predicted to obtain the risk trend index; Provide dynamic warnings based on risk trend indexes and trigger security responses.

2. The construction site safety early warning method based on multidimensional data mining according to claim 1, characterized in that: The collection is carried out by using drones and sensors in collaboration; The multidimensional data includes environmental data, worker data, equipment data, risk events, time and coordinates; The environmental data include temperature and humidity, dust concentration, nitrogen dioxide concentration, noise intensity and wind speed; The worker data includes the worker's heart rate, location, and behavior; The equipment data includes vibration, pressure and displacement of the equipment; Said risk events include worker fatigue, worker heat stroke, equipment failure and environmental degradation; The preprocessing includes data cleaning, normalization and spatiotemporal alignment.

3. The construction site safety early warning method based on multidimensional data mining as claimed in claim 2, characterized in that: Convolutional neural networks are used to extract features from multidimensional data sets, and time series information is fused through a spatiotemporal dynamic weighting mechanism to generate a multidimensional risk feature set. The specific steps are as follows: The multidimensional data set is input into the convolutional neural network, and the convolutional layer of the convolutional neural network is used to extract features from the multidimensional data to obtain spatial features; Use the pooling layer of the convolutional neural network to reduce the dimension and denoise the spatial features to extract key spatial features; According to the time series information of the collected data, a time weight decreasing according to an exponential decay function is assigned to each key spatial feature; Using a weighted fusion method, the time series information is combined with the spatial features obtained by the convolutional neural network to generate a multidimensional risk feature set.

4. The construction site safety early warning method based on multidimensional data mining as claimed in claim 3, characterized in that: Based on the multi-dimensional risk feature set, the risk features are converted into nodes, and the correlation between risk features is converted into edges to form a causal relationship structure between nodes and edges, and a dynamic risk map is constructed. The specific steps are as follows: Extract risk factors from the multidimensional risk feature set and use each risk factor as a node; Each node contains the characteristic value of the risk factor and spatiotemporal information; Analyze the correlation between risk factors and convert them into edges between nodes; Granger causal analysis method is used to remove edges with low correlation and obtain causal inference results; The causal inference results are combined with time series information to generate a dynamic risk map that changes over time.

5. The construction site safety early warning method based on multidimensional data mining according to claim 4, characterized in that: Use the dynamic risk map to analyze the correlation strength between risk factors, evaluate the overall risk level, and generate a list of risk events. The specific steps are as follows: Identify specific high-risk areas and potential risk events based on risk factors and correlations in the dynamic risk map; Calculate the risk weight of each node based on the correlation between nodes and edges in the dynamic risk graph; Calculate the overall site risk level based on the number and risk weights of risk factors in high-risk areas; The risk factors belonging to risk events that appear in high-risk areas are arranged according to the possibility of occurrence to generate a list of risk events.

6. The construction site safety early warning method based on multidimensional data mining according to claim 5, characterized in that: Based on the time series, the risk events in the risk event list are trend predicted to obtain the risk trend index. The specific steps are as follows: Divide the risk event list and the corresponding time series data according to the time window to form a training set, a test set, and a validation set; Perform stationarity detection and differential processing on the time series data of each risk event to construct the input features of the time series model; The ARIMA time series analysis model is trained using the training set, test set, and validation set to obtain the risk trend model.

7. The construction site safety early warning method based on multidimensional data mining according to claim 6, characterized in that: Dynamic warning is carried out according to the risk trend index, and security response is triggered. The specific steps are as follows: Analyze the risk trend index to determine low risk threshold and medium risk threshold; When the risk trend index is less than or equal to the low risk threshold, a low-level warning is issued to remind workers to pay attention; When the risk trend index is greater than the low risk threshold and less than or equal to the medium risk threshold, a medium-level warning is issued, related operations are suspended, and equipment and environmental status are checked; When the risk trend index is greater than the medium-risk threshold, a high-level warning is issued, personnel are evacuated urgently, drones are launched to inspect high-risk areas, and risk changes are monitored in real time.

8. A construction site safety early warning system based on multidimensional data mining, based on the construction site safety early warning method based on multidimensional data mining according to any one of claims 1 to 7, characterized in that: Including data collection module, feature set module, risk map module, risk assessment module, trend prediction module and early warning module; The data acquisition module is used to collect multidimensional data and perform preprocessing to form a multidimensional data set; The feature set module is used to extract features from a multidimensional data set using a convolutional neural network and fuse time series information through a spatiotemporal dynamic weighting mechanism to generate a multidimensional risk feature set; The risk map module is used to convert risk features into nodes based on a multi-dimensional risk feature set, convert the correlation between risk features into edges, form a causal correlation structure between nodes and edges, and construct a dynamic risk map; The risk assessment module is used to analyze the correlation strength between risk factors using a dynamic risk map, assess the overall risk level, and generate a risk event list; The trend prediction module is used to perform trend prediction on risk events in the risk event list based on the time series to obtain a risk trend index; The early warning module is used to issue dynamic early warnings based on the risk trend index and trigger security responses.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the construction site safety early warning method based on multidimensional data mining described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the construction site safety early warning method based on multidimensional data mining described in any one of claims 1 to 7 are implemented.

Images