Hoisting safety risk identification method based on big data

Abstract

The invention relates to a hoisting safety risk identification method based on big data, and the method comprises the steps: obtaining multi-source data of equipment state, environment change and personnel behavior from a hoisting site, carrying out the preprocessing of the multi-source data, and obtaining a unified spatio-temporal data sequence; according to the unified spatio-temporal data sequence, using a convolutional neural network to extract spatio-temporal characteristics of the device state and the environment change, determining whether the spatio-temporal characteristics exceed a preset threshold, and obtaining a risk level; according to the risk level, analyzing time sequence association between personnel behaviors and equipment states through a long-short-term memory network, and obtaining an evolution trend of potential hidden dangers; and obtaining risk probability distribution according to the evolution trend of the potential hazard, and analyzing the risk probability distribution to obtain a high risk type. The safety and efficiency of hoisting operation can be improved, and the accident risk is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of safety management, and particularly to a method for identifying hoisting safety risks based on big data. Background Art

[0002] As a core link in engineering construction, hoisting operations carry the installation and transportation tasks of major equipment and components, and their safety is directly related to project progress and the safety of personnel's lives and property. In modern engineering projects, hoisting operations often involve complex environments, equipment, and personnel collaboration, and any oversight in any link may trigger serious accidents. Therefore, improving the safety management level of hoisting operations has become an urgent need to ensure project quality and efficiency.

[0003] However, current safety management means often seem inadequate when dealing with complex operation scenarios. Many methods rely too much on post-event analysis or single-link monitoring, making it difficult to comprehensively capture various potential risks in the entire operation process, and lacking the ability to make dynamic judgments after integrating information from different sources. This limitation makes it often impossible to detect safety hazards in a timely manner before they occur, especially in high-risk scenarios where multiple factors are intertwined, and the lag of management measures is particularly prominent.

[0004] In this field, the technical challenges mainly focus on how to effectively integrate and analyze multi-source information. Hoisting operations involve aspects such as equipment operating status, changes in the surrounding environment, and personnel operation behaviors. These information sources are extensive and in various formats, lacking a unified correlation with each other. Since these scattered information cannot be formed into an overall judgment basis, it is difficult for managers to accurately identify which links may have problems in actual operations. For example, abnormal equipment operation parameters may be superimposed with adverse weather conditions, further amplifying risks, but existing technologies are difficult to combine the two for analysis, and thus cannot predict potential chain reactions in advance.

[0005] Therefore, how to integrate multi-dimensional information such as equipment, environment, and personnel in real time throughout the hoisting operation process and accurately identify potential safety risks from it has become a key issue in improving the safety management level of operations. Solving this problem not only requires technological breakthroughs but also the ability to dynamically perceive and respond to complex risks in actual business scenarios. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for identifying hoisting safety risks based on big data, which can improve the safety and efficiency of hoisting operations and effectively reduce accident risks.

[0007] To achieve the above object, the present invention provides the following solution:

[0008] A method for identifying hoisting safety risks based on big data, comprising:

[0009] Multi-source data on equipment status, environmental changes, and personnel behavior are acquired from the hoisting site. The multi-source data is then preprocessed to obtain a unified spatiotemporal data sequence.

[0010] Based on the unified spatiotemporal data sequence, a convolutional neural network is used to extract the spatiotemporal features of device status and environmental changes, determine whether they exceed a preset threshold, and obtain the risk level.

[0011] Based on the risk level, the evolution trend of potential hazards is obtained by analyzing the temporal correlation between personnel behavior and equipment status through long short-term memory network analysis.

[0012] Based on the evolution trend of the potential hazards, obtain the risk probability distribution, analyze the risk probability distribution, and identify the types of high risks.

[0013] Optionally, preprocessing the multi-source data includes:

[0014] The multi-source data is cleaned to obtain a preliminary dataset;

[0015] The Kalman filter algorithm is used to remove noise and optimize inconsistencies in the preliminary dataset to obtain an intermediate data sequence.

[0016] If there is a timestamp mismatch in the intermediate data sequence, time correction is performed to obtain a corrected data sequence with consistent time.

[0017] Based on the correction data sequence, a spatial mapping method is used to perform position calibration, adjust the spatial distribution of the hoisting site, and obtain the unified spatiotemporal data sequence.

[0018] Optionally, a convolutional neural network can be used to extract the spatiotemporal features of device status and environmental changes, determine whether they exceed a preset threshold, and obtain the risk level, including:

[0019] A convolutional neural network is used to process the input unified spatiotemporal data sequence, extract spatiotemporal features related to device status and environmental changes, and obtain a feature set;

[0020] Based on the feature set, analyze the key points related to equipment operation. If the key points deviate from the preset threshold, they are recorded as abnormal state points, and an abnormal state set is determined.

[0021] By analyzing the feature set, the fluctuations related to the environment are analyzed. If the fluctuation amplitude exceeds the preset range, it is marked as an environmental anomaly point, and an environmental anomaly set is obtained.

[0022] The correlation degree between the set of environmental anomalies and the set of anomalies is calculated. If the correlation degree is higher than the preset standard, it is identified as a potential risk point, and a set of risk points is obtained.

[0023] Based on the set of risk points, each risk point is classified and processed to obtain the risk level.

[0024] Optionally, analyzing the temporal correlation between human behavior and equipment status using long short-term memory networks includes:

[0025] If the risk level is higher than the preset level, extract the time-series records of personnel behavior and equipment status to obtain a time-series relationship dataset;

[0026] Long Short-Term Memory (LSTM) networks are used to perform in-depth analysis on the temporal relationship dataset to obtain the evolution trend of the potential risks.

[0027] Optionally, obtaining the risk probability distribution based on the evolution trend of the potential hazards includes:

[0028] Based on the evolution trend of the potential hazards, key nodes of trend changes are extracted, and the feature set of the evolution trend is determined;

[0029] The feature set of the evolution trend is input into a Bayesian network model to obtain the risk probability distribution.

[0030] Optionally, based on the risk probability distribution, obtaining the specific location and type of high risk includes:

[0031] If the risk probability distribution is higher than the warning level, then obtain similar historical scenarios from the historical operation database;

[0032] Based on similar historical scenarios, the risk probability distribution level in historical operations is compared with the distribution level in the current scenario to determine the difference in the distribution of potential risks.

[0033] Based on the distribution differences, the high-risk type is identified.

[0034] Optionally, retrieving similar historical scenarios from the historical task database includes:

[0035] Reference data for historical scenarios is obtained from a pre-established historical task database. Based on the reference data, the current scenario is compared with historical scenarios to determine the similarity value. Based on the similarity value, the historical similar scenario is obtained.

[0036] Optionally, based on the distribution differences, the types of high-risk individuals can be identified as follows:

[0037] If the distribution difference exceeds a preset range, then obtain the difference data corresponding to the current scene;

[0038] The differential data is classified using the support vector machine algorithm to obtain the high-risk type.

[0039] The beneficial effects of this invention are as follows: This invention acquires real-time data on equipment status, environmental changes, and personnel behavior through a sensor network. It employs a data fusion algorithm to eliminate noise and inconsistencies, forming a unified spatiotemporal data sequence, thus solving the operational problem of complex and difficult-to-analyze multi-source information at hoisting sites. This invention utilizes convolutional neural networks to extract spatiotemporal features, combines this with long short-term memory networks to analyze temporal correlations, accurately judges the evolution trend of potential hazards, extracts similar scenario data from historical databases to identify high-risk process types, and finally generates alarm signals and transmits them to on-site equipment, ensuring the accuracy of safety response commands. Through intelligent analysis and real-time early warning, this invention significantly improves the safety and efficiency of hoisting operations and effectively reduces the risk of accidents. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a flowchart of a hoisting safety risk identification method based on big data, according to an embodiment of the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] like Figure 1 As shown in the figure, this embodiment proposes a method for identifying hoisting safety risks based on big data, including:

[0045] Multi-source data on equipment status, environmental changes, and personnel behavior are acquired from the hoisting site. The multi-source data is then preprocessed to obtain a unified spatiotemporal data sequence.

[0046] Based on the unified spatiotemporal data sequence, a convolutional neural network is used to extract the spatiotemporal features of device status and environmental changes, determine whether they exceed a preset threshold, and obtain the risk level.

[0047] Based on the risk level, the evolution trend of potential hazards is obtained by analyzing the temporal correlation between personnel behavior and equipment status through long short-term memory network analysis.

[0048] Based on the evolution trend of the potential hazards, obtain the risk probability distribution, analyze the risk probability distribution, and identify the types of high risks.

[0049] Furthermore, the preprocessing of the multi-source data includes:

[0050] The multi-source data is cleaned to obtain a preliminary dataset;

[0051] The Kalman filter algorithm is used to remove noise and optimize inconsistencies in the preliminary dataset to obtain an intermediate data sequence.

[0052] If there is a timestamp mismatch in the intermediate data sequence, time correction is performed to obtain a corrected data sequence with consistent time.

[0053] Based on the correction data sequence, a spatial mapping method is used to perform position calibration, adjust the spatial distribution of the hoisting site, and obtain the unified spatiotemporal data sequence.

[0054] Specifically, in equipment status monitoring at hoisting sites, sensor networks can collect multi-source information in real time, including equipment operating parameters, ambient temperature and humidity, and personnel operation behavior. Assume a hoisting site is equipped with multiple sensors, including equipment vibration sensors, wind speed sensors, and personnel positioning sensors, collecting data once per minute. The raw data may contain noise or missing values. A pre-established data processing module performs initial cleaning, such as removing obvious outliers from vibration data. For example, if a vibration frequency collected far exceeds the normal range of 100 Hz, that data point is directly discarded, resulting in a relatively clean preliminary data set. For the fusion of multi-source information, the Kalman filter algorithm can be used to optimize noise removal and address data inconsistency issues. Suppose there is a discrepancy between equipment vibration data and wind speed data at a certain point in time. The Kalman filter, through prediction and update mechanisms, reduces the noise impact of the vibration data and adjusts the equipment status assessment results by combining wind speed data, forming a more accurate intermediate data sequence. This effectively improves data reliability and provides a stable foundation for subsequent analysis.

[0055] In one possible implementation, if a timestamp mismatch is found in the intermediate data sequence—for example, vibration data recorded at 10:00:05 while wind speed data is recorded at 10:00:10—the time alignment module will estimate the data at the missing time point using interpolation to ensure the time consistency of the corrected data sequence. This avoids analysis errors caused by time deviations and improves the accuracy of data processing. For spatial distribution adjustments, the spatial mapping method can map the data collected by sensors to the actual location at the hoisting site. Assuming a sensor is installed at the front end of the hoisting arm with coordinates x=10 meters and y=5 meters, spatial mapping aligns the data with the site layout, forming a unified mapped data sequence. This helps to accurately locate the anomaly and improves the targeting of monitoring.

[0056] Furthermore, a convolutional neural network is used to extract the spatiotemporal features of device status and environmental changes, determine whether they exceed a preset threshold, and obtain the risk level, including:

[0057] A convolutional neural network is used to process the input unified spatiotemporal data sequence, extract spatiotemporal features related to device status and environmental changes, and obtain a feature set;

[0058] Based on the feature set, analyze the key points related to equipment operation. If the key points deviate from the preset threshold, they are recorded as abnormal state points, and an abnormal state set is determined.

[0059] By analyzing the feature set, the fluctuations related to the environment are analyzed. If the fluctuation amplitude exceeds the preset range, it is marked as an environmental anomaly point, and an environmental anomaly set is obtained.

[0060] The correlation degree between the set of environmental anomalies and the set of anomalies is calculated. If the correlation degree is higher than the preset standard, it is identified as a potential risk point, and a set of risk points is obtained.

[0061] Based on the set of risk points, each risk point is classified and processed to obtain the risk level.

[0062] Specifically, convolutional neural networks (CNNs) can perform deep analysis of input spatiotemporal data, capturing the spatiotemporal characteristics of equipment operation and environmental changes. For example, at a hoisting site, sensors collect data every minute, including equipment load and ambient wind speed. The CNN extracts load change trends and wind speed fluctuation patterns from this data, forming a feature set. This approach helps extract useful pattern information from complex data. For equipment status monitoring, key points of equipment operation are analyzed based on the feature set. Assuming the normal range of equipment load is set to 5 to 20 tons, if the load reaches 25 tons at a certain time point, significantly exceeding the preset threshold, it is recorded as an abnormal state point. Further collection of similar anomalies at multiple time points forms an abnormal state set. This method can quickly identify potential problems in equipment operation. In environmental change analysis, attention is paid to environmental fluctuations within the feature set. Assuming the normal wind speed range is 2 to 10 meters per second, if the wind speed suddenly increases to 18 meters per second at a certain time point, exceeding the preset range, it is marked as an environmental anomaly point, and these are aggregated to form an environmental anomaly set. This analysis effectively captures the potential impact of the environment on equipment operation.

[0063] By utilizing sets of abnormal states and environmental anomalies, the correlation between the two is analyzed. If, within a certain time period, abnormal equipment load and abnormal wind speed occur simultaneously, and their correlation exceeds a preset standard of 80%, they are identified as potential risk points and aggregated into a risk point set. This correlation analysis helps uncover deeper connections behind the anomalies. Each risk point is then categorized based on the risk point set. If a risk point has both high abnormal values ​​for equipment load and wind speed, and the categorization meets the high-risk standard, a "high-risk" label is generated, ultimately forming a risk level distribution. This categorization method clearly distinguishes the severity of different risks.

[0064] Furthermore, analyzing the temporal correlation between human behavior and equipment status through long short-term memory networks includes:

[0065] If the risk level is higher than the preset level, extract the time-series records of personnel behavior and equipment status to obtain a time-series relationship dataset;

[0066] Long Short-Term Memory (LSTM) networks are used to perform in-depth analysis on the temporal relationship dataset to obtain the evolution trend of the potential risks.

[0067] Specifically, for extracting time-series records of personnel behavior and equipment status, operator action records and equipment load change data can be obtained within 10 minutes before and after the anomaly occurs, forming a time-series relationship dataset. For example, if operators frequently adjust the hoisting angle within a certain time period, and the equipment load value fluctuates synchronously, this information can be integrated as the basis for preliminary analysis. This method helps to clarify the causes and consequences of the event. When using Long Short-Term Memory (LSTM) networks for deep analysis, the time-series relationship dataset can be input into the model to analyze the evolution trend of potential hazards. For example, if the model output shows that equipment load fluctuations have gradually intensified over the past hour and show signs of continuous deterioration, it can be identified as a potential hazard. This analytical approach can uncover patterns of change from historical data, providing a basis for subsequent judgments.

[0068] Furthermore, based on the evolution trend of the potential hazards, the risk probability distribution is obtained as follows:

[0069] Based on the evolution trend of the potential hazards, key nodes of trend changes are extracted, and the feature set of the evolution trend is determined;

[0070] The feature set of the evolution trend is input into a Bayesian network model to obtain the risk probability distribution.

[0071] Specifically, in this embodiment, based on the risk probability distribution and combined with a multidimensional analysis perspective, the probability distribution is stratified. If the probability value of a certain dimension exceeds a preset threshold, it is prioritized to determine the distribution range of high-risk dimensions. For the distribution range of high-risk dimensions, a real-time monitoring data update mechanism is introduced. Through continuous data input, the weights of the probability distribution are adjusted to determine the dynamic results of the risk assessment. The stratified processing of the initial risk probability distribution can be analyzed from three dimensions: equipment, environment, and personnel. If the probability value of the equipment dimension exceeds a preset threshold of 0.8, it is prioritized, indicating that its high-risk distribution range is concentrated in the hydraulic system. This stratified marking helps to accurately pinpoint the focus. In adjusting the distribution range of high-risk dimensions, a real-time monitoring data update mechanism is introduced, continuously inputting data such as hydraulic pressure values ​​to adjust the probability distribution weights. For example, if the hydraulic pressure value remains consistently high within one hour, the dynamic assessment result will increase the risk level. This dynamic adjustment more closely reflects the actual situation.

[0072] Furthermore, based on the aforementioned risk probability distribution, obtaining the specific locations and types of high-risk risks includes:

[0073] If the risk probability distribution is higher than the warning level, then obtain similar historical scenarios from the historical operation database;

[0074] Based on similar historical scenarios, the risk probability distribution level in historical operations is compared with the distribution level in the current scenario to determine the difference in the distribution of potential risks.

[0075] Based on the distribution differences, the high-risk type is identified.

[0076] Furthermore, retrieving similar historical scenarios from the historical task database includes:

[0077] Reference data for historical scenarios is obtained from a pre-established historical task database. Based on the reference data, the current scenario is compared with historical scenarios to determine the similarity value. Based on the similarity value, the historical similar scenario is obtained.

[0078] Furthermore, based on the aforementioned distribution differences, the types of high-risk individuals include:

[0079] If the distribution difference exceeds a preset range, then obtain the difference data corresponding to the current scene;

[0080] The differential data is classified using the support vector machine algorithm to obtain the high-risk type.

[0081] Specifically, the process of extracting reference data from a historical operation database can be envisioned as a database containing hoisting operation records from the past five years, covering information such as equipment status, environmental parameters, and operation logs. Suppose that the load data of a hoisting device in the current scenario exceeds the warning standard, the system will automatically filter out past operation records with similar load anomalies, forming a preliminary historical dataset. This method can quickly identify relevant reference information, providing a foundation for subsequent analysis.

[0082] For the feature comparison process between historical datasets and the current scenario, computational tools can be used to analyze the similarity of key indicators such as equipment load and ambient wind speed. Assuming the peak equipment load in the current scenario is 80 tons, and the peak load in a historical operation was 78 tons, with wind speed fluctuations within a similar range, the similarity value might reach 0.9, exceeding the preset threshold of 0.85. This comparison method helps find the closest reference case to the current situation, ensuring the analysis's relevance. When conducting deep correlation analysis, the focus can be on the distribution characteristics of key risk probabilities. Assuming that abnormal loads in historical data are often accompanied by hydraulic system pressure fluctuations, and the current scenario shows a similar trend, a preliminary risk assessment framework can be constructed, focusing on potential problems in the hydraulic system. This correlation analysis provides a clear direction for subsequent assessments. Comparative analysis of risk probability distribution levels reveals that the probability of abnormal hydraulic pressure in the current scenario is 0.75, while the probability of a similar scenario in historical data is 0.6, a difference exceeding the tolerance range of 0.1. This difference assessment helps identify potential special risk points in the current scenario, providing a basis for further optimization.

[0083] When using the Support Vector Machine (SVM) algorithm to classify discrepancies in data, indicators such as hydraulic pressure and equipment load can be used as input features to classify high-risk and low-risk categories, thereby optimizing the parameters of the risk assessment model. This classification process can more accurately characterize the risk distribution features and improve the model's adaptability.

[0084] When generating alarm signal sources based on risk level and priority, a signal generation-based mechanism can be envisioned. High-risk alarm signals are set as high-frequency vibration signals and sent to field equipment via the main transmission network. Assuming a signal strength of 80 units, this ensures signal coverage of critical areas at the hoisting site. If signal transmission fails due to interference in the main network, it switches to a backup network for retransmission until the equipment returns a confirmation message. When receiving alarm signals and determining the interactive status at the equipment end, if a hoisting device fails to receive the signal in time due to network latency, the system automatically detects the abnormal status and retransmits the signal via the backup network. After confirmation that reception was successful, the system records the device status as normal, ensuring no alarm signals are missed. If the alarm confirmation status is not achieved, the system triggers a secondary alert signal through the device interactive mechanism. If a device fails to confirm the alarm within 5 minutes, the system sends a stronger audible alert signal until the operator manually confirms. This mechanism effectively avoids information transmission interruptions. When generating safety response commands and transmitting them through the command output terminal, load reduction commands can be generated for abnormal load issues and issued through the main command output terminal. If feedback indicates that the instruction was not executed, the system will resend it from the backup output to ensure that the instruction is implemented. When recording the results of high-risk process handling and updating data, assuming that the load anomaly issue has been resolved by reducing the load to 50 tons, the system will enter the handling result into the database and update the risk type of that location to low risk, forming a closed-loop monitoring system. This approach helps to continuously optimize the risk management process.

[0085] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for identifying hoisting safety risks based on big data, characterized in that, include: Multi-source data on equipment status, environmental changes, and personnel behavior are acquired from the hoisting site. The multi-source data is then preprocessed to obtain a unified spatiotemporal data sequence. Based on the unified spatiotemporal data sequence, a convolutional neural network is used to extract the spatiotemporal features of device status and environmental changes, determine whether they exceed a preset threshold, and obtain the risk level. Based on the risk level, the evolution trend of potential hazards is obtained by analyzing the temporal correlation between personnel behavior and equipment status through long short-term memory network analysis. Based on the evolution trend of the potential hazards, obtain the risk probability distribution, analyze the risk probability distribution, and identify the types of high risks.

2. The method for identifying hoisting safety risks based on big data according to claim 1, characterized in that, Preprocessing the multi-source data includes: The multi-source data is cleaned to obtain a preliminary dataset; The Kalman filter algorithm is used to remove noise and optimize inconsistencies in the preliminary dataset to obtain an intermediate data sequence. If there is a timestamp mismatch in the intermediate data sequence, time correction is performed to obtain a corrected data sequence with consistent time. Based on the correction data sequence, a spatial mapping method is used to perform position calibration, adjust the spatial distribution of the hoisting site, and obtain the unified spatiotemporal data sequence.

3. The method for identifying hoisting safety risks based on big data according to claim 1, characterized in that, A convolutional neural network is used to extract the spatiotemporal features of equipment status and environmental changes, determine whether they exceed a preset threshold, and obtain the risk level, including: A convolutional neural network is used to process the input unified spatiotemporal data sequence, extract spatiotemporal features related to device status and environmental changes, and obtain a feature set; Based on the feature set, analyze the key points related to equipment operation. If the key points deviate from the preset threshold, they are recorded as abnormal state points, and an abnormal state set is determined. By analyzing the feature set, the fluctuations related to the environment are analyzed. If the fluctuation amplitude exceeds the preset range, it is marked as an environmental anomaly point, and an environmental anomaly set is obtained. The correlation degree between the set of environmental anomalies and the set of anomalies is calculated. If the correlation degree is higher than the preset standard, it is identified as a potential risk point, and a set of risk points is obtained. Based on the set of risk points, each risk point is classified and processed to obtain the risk level.

4. The method for identifying hoisting safety risks based on big data according to claim 1, characterized in that, Analyzing the temporal correlation between human behavior and equipment status using long short-term memory networks includes: If the risk level is higher than the preset level, extract the time-series records of personnel behavior and equipment status to obtain a time-series relationship dataset; Long Short-Term Memory (LSTM) networks are used to perform in-depth analysis on the temporal relationship dataset to obtain the evolution trend of the potential risks.

5. The method for identifying hoisting safety risks based on big data according to claim 1, characterized in that, Based on the evolution trend of the potential hazards, the risk probability distribution is obtained as follows: Based on the evolution trend of the potential hazards, key nodes of trend changes are extracted, and the feature set of the evolution trend is determined; The feature set of the evolution trend is input into a Bayesian network model to obtain the risk probability distribution.

6. The method for identifying hoisting safety risks based on big data according to claim 1, characterized in that, Based on the aforementioned risk probability distribution, the specific locations and types of high-risk risks are obtained, including: If the risk probability distribution is higher than the warning level, then obtain similar historical scenarios from the historical operation database; Based on similar historical scenarios, the risk probability distribution level in historical operations is compared with the distribution level in the current scenario to determine the difference in the distribution of potential risks. Based on the distribution differences, the high-risk type is identified.

7. The method for identifying hoisting safety risks based on big data according to claim 6, characterized in that, Retrieving similar historical scenarios from the historical assignment database includes: Reference data for historical scenarios is obtained from a pre-established historical task database. Based on the reference data, the current scenario is compared with historical scenarios to determine the similarity value. Based on the similarity value, the historical similar scenario is obtained.

8. The method for identifying hoisting safety risks based on big data according to claim 6, characterized in that, Based on the aforementioned distribution differences, the types of high-risk individuals include: If the distribution difference exceeds a preset range, then obtain the difference data corresponding to the current scene; The high-risk type is obtained by classifying the differential data using the support vector machine algorithm.

Images