Mine ecological restoration engineering safety risk identification method

Abstract

The invention discloses a safety risk identification method for a mine ecological restoration project, relates to the technical field of project safety risk identification, and can monitor the environment condition of a construction site and the behaviors of constructors in real time, find problems in time and take corresponding measures, thereby improving the response speed and the processing efficiency. Automation and unmanned operation are achieved, automatic monitoring is achieved, dependence on manual supervision is reduced, the possibility of human errors is reduced, and meanwhile the workload of supervisors is relieved; the safety is improved, and non-standard behaviors and potential dangers can be found and corrected in time through comprehensive monitoring of the construction site, so that the risk of accidents is reduced; and the supervision cost is reduced, and the demand on human resources is reduced through an automatic and unmanned supervision mode, so that the total cost of safety supervision is reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of engineering safety risk identification, and specifically is a method for identifying safety risks in mine ecological restoration projects. Background Art

[0002] Mine ecological restoration projects are field constructions, with uneven quality of workers, high personnel mobility, and random and irregular construction, which are important reasons for safety risks. At the same time, under the influence of the external environment, the working environment is complex and harsh, and it is difficult for workers to concentrate during operation. In addition, if the weather changes suddenly, strong winds or heavy rains will cause rolling stones on the slopes, and the scaffolding or towering mechanical equipment will shake, which will bring danger to the workers.

[0003] At present, the safety control of mine ecological restoration construction mostly relies on human management, and there is a lack of auxiliary equipment to manage the entire process of construction safety. The subjective factors of people have a great impact on construction safety. Some key indicators still rely on manual supervision, and the supervision capacity of safety inspectors is limited. It is difficult to implement on-site supervision of multiple construction operations carried out at the same time, forming safety monitoring loopholes and easily leading to construction accidents. At the same time, the existing monitoring methods in other fields are not suitable for the field of ecological restoration construction. The installation process is cumbersome, the sensor deployment method lacks flexibility, the layout is slow, and the data information cannot be transmitted in a timely and effective manner. The monitoring object can often only be a single target such as not wearing a helmet or not wearing a reflective vest, and it is impossible to provide comprehensive mine ecological restoration safety risk control data. Based on this, a solution is provided. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; To this end, the present invention proposes a method for identifying safety risks of mine ecological restoration projects, which specifically includes the following steps: Conduct data fusion and risk assessment before work. Before workers enter the work area, face recognition, dangerous area identification, meteorological disasters, and landslide and rockfall data collection are first performed. The collected data is then fused and a preliminary risk assessment is performed based on the fused data. The fusion and modeling of multi-source information on construction safety risks is carried out by collating and collecting data and pictures of mine ecological restoration projects, analyzing safety risks and influencing factors, integrating multi-source information on construction safety risks, and uniformly modeling the data flows, business flows, and control flows that affect construction safety according to the types of information system data; Rapid acquisition and automatic identification of safety risks, identification of safety hazards in real-time construction process, identification of existing risks, evaluation using analytic hierarchy process, and obtaining the evaluation index score of construction risk factors according to the classification standard of construction risk evaluation index; Simultaneously detect and identify unsafe behaviors of construction workers, thereby integrating safety risk data and assessing potential risks.

[0005] Compared with the prior art, the present invention has the following beneficial effects: This application can monitor the environmental conditions of the construction site and the behavior of construction workers in real time, discover problems in a timely manner and take corresponding measures, thereby improving response speed and processing efficiency; Automation and unmanned: Automated monitoring reduces dependence on manual supervision, reduces the possibility of human error, and also reduces the workload of supervisors; Improve safety. Through comprehensive monitoring of the construction site, irregular behaviors and potential dangers can be discovered and corrected in a timely manner, thereby reducing the risk of accidents. Reduce supervision costs. Automated and unmanned supervision models reduce the demand for human resources, thereby reducing the overall cost of safety supervision. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION

[0007] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0008] See also Figure 1 The present application provides a method for identifying safety risks of mine ecological restoration projects, which specifically includes the following steps: 1. Data fusion and risk assessment before operation; providing the necessary data basis for subsequent analysis; Pre-operation data fusion and risk assessment refers to the collection of data such as face recognition, dangerous area identification, meteorological disasters, landslides and rolling stones before workers enter the work area to work; Step 1, data collection; Facial recognition data collection: Use high-definition facial recognition equipment to collect workers’ facial features before they enter the work area, and record workers’ identity information and other data; Dangerous area identification data collection: Through the sensors and image recognition systems installed in the work area, the terrain, topography, equipment layout, etc. of the work area are monitored in real time to identify the location, scope and other data of the dangerous area; Meteorological disaster data collection: Use meteorological monitoring instruments to collect meteorological data such as wind speed, rainfall, temperature, humidity, etc. in the work area; Landslide and rockfall data collection: With the help of geological monitoring sensors, the geological conditions of the work area are monitored and data related to landslide and rockfall are collected; Step 2: Data integration and fusion; The various types of collected data are screened and cleaned through the threshold method to remove invalid and erroneous data; Standardize the filtered and cleaned data to make them have a unified data format and measurement standards; According to the following data fusion formula, the processed data are fused to obtain the fused data; the data fusion formula is as follows: Assign a weight to each type of collected video image data, including: face recognition data indicator Dr, dangerous area recognition data indicator Dd, meteorological disaster data indicator Dw, landslide and rolling stone data indicator Ds; the corresponding weights are w r 、w d 、w w and w s, And satisfy: w r +w d +w w +w s =1; various video image data are fused by means of the following method, the specific fusion method is: Df=w r Dr+w d Dd+w w Dw+w s Ds; The fused data index is Df ， A data set that comprehensively reflects various risk factors to assess the overall security risk level; The risk assessment index R' is calculated based on the corresponding weights of historical risk operation data index Dh, operation guidance data index Dz and fusion data index Df, α, β and γ respectively; historical risk operation data includes historical accident records, number of dangerous situations, severity of accidents and other historical data; operation guidance data includes operation procedures, operating guidelines, safety standards, etc., which are used to guide operations and evaluate whether operations meet safety standards The risk assessment index R' is: R' = αDh + βDz + γDf; Step 3: Risk Assessment Determine the risk level based on the calculated risk assessment index R'.

[0009] Step 4: Risk warning and pre-control Based on the risk assessment results, when the risk assessment indicator exceeds the preset threshold, the system will issue a risk warning signal.

[0010] Based on the risk warning results, formulate corresponding risk prevention and control measures, such as adjusting the operation plan, adding safety protection facilities, strengthening personnel training, etc., to reduce operation risks.

[0011] Second, the fusion and modeling of multi-source information on construction safety risks provides a deeper understanding of data and contextual information for risk assessment by building a knowledge graph; By collating and collecting data, pictures and other information from projects such as mine ecological restoration, analyzing safety risks and influencing factors, and integrating multi-source information on construction safety risks, we mainly target the types of information system data and uniformly model the data flows, business flows, and control flows that affect construction safety.

[0012] Step 1: Data flow modeling. The generation of data is mainly related to cameras and sensors. Therefore, the management and control of data flow mainly focuses on on-site videos and sensors. The device ID is used as the association field between the data of the same device object to integrate the data of the unified device.

[0013] Step 2: Business flow modeling is mainly related to specific construction business. The modeling work will be carried out around the construction stage and construction specifications, with the business stage number as the associated field for the handover of different businesses.

[0014] Step 3: Control flow modeling is mainly related to personnel decision-making. It will model personnel and specific decisions, and associate the user account that makes the decision with the decision data.

[0015] Step 4: After locking the device ID, business number information, and user account information, the video data, sensor data, business information, and personnel management information are linked to each other to form a complete business closed loop.

[0016] Step 5: After establishing a complete unified naming and coding rule, standardize the naming of the data information in the data flow, business flow, and control flow, and improve the missing information and data structure.

[0017] Step 6: Use the PFCA model to model construction safety smart management and control; with the goal of achieving the specific functions of the project organization and business processes as the core, use advanced information technology and modern management measures to maximize the organization's technical and management integration, establish a process-oriented organizational structure and information communication channels, create a favorable organizational atmosphere, and make the project organization respond quickly and produce agilely.

[0018] 3. Rapidly acquire and automatically identify safety risks, and use the results of the first two steps to conduct risk assessment and decision support to achieve accurate management and control of construction risks; 1) Conduct a special risk assessment first Organize the mine ecological restoration construction process and common safety influencing factors, and fully understand the safety hazards in the construction process, such as personnel, equipment, materials, systems and environment; the main contents include risk source identification and risk assessment.

[0019] Step 1: Risk source identification. Risk assessors obtain relevant information and combine it with on-site project investigations, based on expert investigation and system analysis. Expert investigation refers to the use of experience and understanding of the industry by managers to provide insights and predictions about potential risks; system analysis is a systematic approach to comprehensively analyze all components of the project, including physical equipment, operating procedures, personnel behavior, environmental factors, etc. Identify, analyze and archive possible risk sources or risk events.

[0020] Step 2: Summarize the construction safety risk assessment index system, select personnel, electromechanical equipment, materials, management system, environment and other factors as the core factors of the evaluation index, and obtain the evaluation index value of each risk factor based on the risk factor assessment standard, expert advice and previous research experience. Quantify the probability of occurrence and impact of risk events to form a risk matrix, so as to sort and control risks.

[0021] Step 3: Risk assessment, using the analytic hierarchy process to evaluate, according to the classification standards of construction risk evaluation indicators and the opinions of consulting experts, to obtain the evaluation index scores of construction risk factors. First, determine the evaluation indicators, which are the basis for evaluating the degree of risk. Then, through expert scoring or data analysis, compare each indicator pairwise to construct a judgment matrix. Next, calculate the eigenvalues ​​and corresponding eigenvectors of the judgment matrix to obtain the weight of each indicator. Consult expert opinions: Combine the knowledge and experience of experts to evaluate the risk factors and obtain the preliminary score values ​​of each risk factor. Substitute the quantified values ​​of the directional concepts into the forward cloud generator, and repeatedly calculate to obtain the membership values ​​at each risk level. Based on a large amount of sample data, the membership function is determined by statistical methods. For example, if the value of an indicator is within a certain interval, its membership can be determined by counting the proportion of samples in the interval. The membership matrix is ​​A, and the entropy weight method is used to determine the weight W of each evaluation indicator. Finally, A and W are substituted into the designed comprehensive risk assessment expression to calculate the risk value. The steps for determining the weight W of each evaluation indicator using the entropy weight method are as follows: 1) Data standardization: Standardize the original data to eliminate the dimensional impact between different indicators.

[0022] 2) Calculate information entropy: Calculate the information entropy of each indicator based on the standardized data to reflect the degree of discreteness of each indicator.

[0023] 3) Calculate weight: Calculate the weight of each indicator based on information entropy. The larger the weight, the higher the importance of the indicator in the evaluation system. Standardize the original data to eliminate the dimensional impact between different indicators.

[0024] The comprehensive risk assessment expression is usually a weighted summation formula in the following form:

[0025] in, is the comprehensive risk value; n is the number of total indicators in the integrated indicator; is the weight of the ith indicator, which can be determined by methods such as cross-validation; is the i-th indicator for the input feature set The prediction results; is the number of features considered in feature engineering; is the importance weight of the jth feature, which can be determined by the feature selection algorithm; is the transformation result of the jth feature on the input data set y, which can be a nonlinear transformation; it is the base of the natural logarithm. In the feature engineering stage, not only the original data should be considered, but also the prediction ability of the model should be enhanced through data transformation and feature selection:

[0026] in, It is a feature set after feature engineering; It is a primitive feature; It is a new feature obtained through feature selection and transformation, which can be a polynomial feature, interactive feature, etc. In the indicator integration stage, a variety of machine learning algorithms are used, and weights are assigned to each indicator:

[0027] in, is the performance indicator accuracy of the i-th indicator on the validation set; is the average value of all performance indicators.

[0028] Through the above formula, we can calculate the comprehensive risk value and thus evaluate the risk level of the project; 2) Detect and identify unsafe behaviors of construction workers: Step 1: Sort out the unsafe behaviors of construction workers. Classify typical unsafe behaviors in construction into three categories: first, unsafe behaviors caused by the on-site environment or social environment; second, unsafe factors in construction organization and management, which are mainly reflected in the construction process management, the safety status management of construction materials, safety training and safety investment; third, unsafe factors at the construction site, which are mainly reflected in the operation safety, clothing safety and safety awareness of the construction workers at the construction site.

[0029] Step 2: Use vision-based detection equipment for construction worker violations. Segment the image into meaningful target area blocks, apply specific neural network models, such as the YOLO series; identify construction worker violations, and identify some representative and meaningful objects in these sub-block scenes to determine which type of behavior the scene belongs to, and mark the most likely illegal operation behavior in the area.

[0030] Step 3: Establish unsafe behaviors and their detection framework. By analyzing the main functional requirements of the video monitoring system during ecological restoration construction, counting unsafe behaviors, and studying image preprocessing algorithms suitable for ecological restoration construction. The Retinex algorithm is mainly introduced to comprehensively process construction images, and the system database for describing, analyzing and designing functional points is studied. Finally, the system is developed through the OpenCV image recognition library to achieve the purpose of identifying unsafe behaviors.

[0031] 3) Conduct security risk data integration and risk assessment Step 1: The decision-making basis for business management and control includes various business management systems and information platforms, construction safety intelligent management and control models, and various safety-related normative documents based on the current national and corporate standards. These are the data sources and interaction objects for intelligent management and control.

[0032] Step 2: First, use a data flow diagram (DFD) to represent the input, processing, and output of data, clarify the data input and output and processing name of each processing step, and avoid over-refinement. At the same time, use id-mapping technology to map data from different sources to a unified device identifier to associate the data flow of the same device object. Through the ETL process, we integrate data from cameras and sensors into a central database for subsequent analysis and processing.

[0033] Then, natural language processing (NLP) technology is applied to extract structured data, namely entities, relationships, and attributes, from unstructured text and store them in a graph database such as Neo4j to manage and query complex relational data. Indexes are created in the graph database to improve query efficiency.

[0034] On this basis, entities and relations are extracted from the cleaned data, and deep learning models are used for named entity recognition (NER) and relation extraction to build a knowledge graph framework. Knowledge fusion technology is used to integrate knowledge from different sources to solve knowledge conflicts and redundancy problems. At the same time, the extracted entities are linked with entities in the knowledge base to enrich and expand the knowledge graph.

[0035] Furthermore, semantic annotation of data instances is performed, and keyword semantic annotation is performed in combination with manual semi-automatic annotation to enhance the semantic understanding of data. By defining the spatiotemporal data association state matrix, the semantic association state, temporal association state, and spatial association state of content objects between instances are calculated to achieve the association organization of multimodal spatiotemporal data.

[0036] Finally, deep learning technology is used to extract relationships and reason about knowledge to identify potential risk patterns and associations. A risk assessment model is built to combine historical data and real-time data to assess risk levels and develop corresponding preventive measures. Risk prevention plans are automatically generated, including adjusting work plans, adding safety protection facilities, and strengthening personnel training to reduce operational risks.

[0037] Step 3: Realize intelligent mining of multi-source information such as data flow, business flow, and control flow, so as to achieve accurate assessment of potential risks and provide decision-making assistance for risk prevention during construction.

[0038] Data collection: Collect data streams related to the construction project, including but not limited to environmental data, equipment status, personnel behavior, historical accident records, etc.; environmental data includes environmental parameters such as temperature, humidity, wind speed, etc., which are acquired in real time through sensors and normalized to eliminate dimensional effects; equipment status includes equipment operating parameters, fault records, and maintenance logs; personnel behavior is acquired by tracking the behavior patterns of construction personnel through video and sensor acquisition; historical accident records are extracted from the database, including accident type, time, location, and results; Data integration: Integrate data from different sources to form a unified data warehouse.

[0039] Feature engineering: Extract features from raw data, such as the frequency of equipment failures, the number of personnel violations, etc.; these features may be relevant to risk assessment.

[0040] Risk model construction: Use machine learning or statistical methods to build risk assessment models.

[0041] Risk assessment: Apply risk models to assess potential risks during construction. Risk assessment models may use a variety of algorithms, such as logistic regression, decision trees, random forests, etc. Its basic form is: Risk score = log(1 / (1−p)) = β0+β1x1+β2x2+…+βnxn, where p is the probability of risk occurring, β is the model parameter, and x is the feature.

[0042] Decision support: Provide decision support based on risk assessment results, such as adjusting construction plans, strengthening safety measures, etc.

[0043] The above embodiments are only used to illustrate the technical method 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 method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for identifying safety risks in mine ecological restoration projects, characterized in that: include: Conduct data fusion and risk assessment before work. Before workers enter the work area, face recognition, dangerous area identification, meteorological disasters, and landslide and rockfall data collection are first performed. The collected data is then fused and a preliminary risk assessment is performed based on the fused data. The fusion and modeling of multi-source information on construction safety risks is carried out by collating and collecting data and pictures of mine ecological restoration projects, analyzing safety risks and influencing factors, integrating multi-source information on construction safety risks, and uniformly modeling the data flows, business flows, and control flows that affect construction safety according to the types of information system data; Rapid acquisition and automatic identification of safety risks, identification of safety hazards in real-time construction process, identification of existing risks, evaluation using analytic hierarchy process, and obtaining the evaluation index score of construction risk factors according to the classification standard of construction risk evaluation index; Simultaneously detect and identify unsafe behaviors of construction workers, thereby integrating safety risk data and assessing potential risks.

2. A method for identifying safety risks of mine ecological restoration projects according to claim 1, characterized in that: The specific methods of data fusion and risk assessment before operation are as follows: Conduct data collection to obtain face recognition data, dangerous area recognition data, meteorological disaster data, and landslide and rockfall data; Screen and clean all types of collected data to remove invalid and erroneous data; Standardize the filtered and cleaned data to make them have a unified data format and measurement standards; According to the following data fusion formula, the processed data are fused to obtain the fused data. The fusion formula is: Df=w r Dr+w d Dd+w w Dw+w s Ds; In the formula, the face recognition data index is Dr, the dangerous area recognition data index is Dd, the meteorological disaster data index is Dw, the landslide and rolling stone data index is Ds, and the weights are w and r +w d +w w +w s =1, the fused data index is Df; Perform risk assessment based on fused data.

3. A method for identifying safety risks of mine ecological restoration projects according to claim 2, characterized in that: The specific methods of risk assessment are as follows: The fused data, historical risk operation data, and operation instruction data are input into the safety risk management and control system of intelligent behavior recognition; the risk assessment formula is used for evaluation, and the specific formula is: R' = αDh + βDz + γDf; In the formula, the risk assessment index is R', the weights of the historical risk operation data index Dh and the operation guidance data index Dz are α and β respectively, and γ is the weight of Df.

4. A method for identifying safety risks of mine ecological restoration projects according to claim 3, characterized in that: Based on the risk assessment results, when the risk assessment indicator exceeds the preset threshold, the system will issue a risk warning signal.

5. A method for identifying safety risks in mine ecological restoration projects according to claim 1, characterized in that: The specific methods of multi-source information fusion and modeling of security risks are as follows: Carry out targeted data flow modeling, business flow modeling, and control flow modeling. By locking the device ID, business number information, and user account information, the video data, sensor data, business information, and personnel control information are linked to each other to form a complete business closed loop. After establishing a complete unified naming and coding rule, standardize the naming of data information in data flow, business flow, and control flow, and improve missing information and data structure; The PFCA model is used to carry out construction safety intelligent management and control modeling.

6. A method for identifying safety risks in mine ecological restoration projects according to claim 1, characterized in that: The specific methods for rapid acquisition and automatic identification of security risks are: First, carry out special risk assessment. The administrator inputs the preliminary evaluation index score of the construction risk factor, substitutes the quantified value of the directional concept into the forward cloud generator, and repeatedly calculates the membership value under each risk level. The membership matrix is ​​A, and then uses the entropy weight method to determine the weight W of each evaluation index; finally, A and W are substituted into the designed comprehensive risk assessment expression to calculate the risk value; Then, the YOLO series of neural network models are used to detect and identify unsafe behaviors of construction workers, and an unsafe behavior and its detection framework are established. By analyzing the main functional requirements of the video surveillance system of the ecological restoration construction process, unsafe behaviors are counted, and the Retinex algorithm is introduced to comprehensively process the construction images. The system database for describing, analyzing and designing functional points is studied. Finally, the system is developed through the OpenCV image recognition library to achieve the purpose of identifying unsafe behaviors.

7. A method for identifying safety risks in mine ecological restoration projects according to claim 6, characterized in that: The steps of determining the weight W of each evaluation index by entropy weight method are as follows: Standardize the original data to eliminate the dimensional impact between different indicators; According to the standardized data, the information entropy of each indicator is calculated to reflect the discrete degree of each indicator; The weight of each indicator is calculated according to the information entropy. The larger the weight, the higher the importance of the indicator in the evaluation system. The original data is standardized to eliminate the dimensional influence between different indicators.

8. A method for identifying safety risks in mine ecological restoration projects according to claim 7, characterized in that: The comprehensive risk assessment is determined by: in, is the comprehensive risk value; n is the number of total indicators in the integrated indicator; is the weight of the ith indicator, determined by the cross-validation method; is the i-th indicator for the input feature set The prediction results; is the number of features considered in feature engineering; is the importance weight of the jth feature, determined by the feature selection algorithm; It is the transformation result of the j-th feature pair input data set y, which is the base of the natural logarithm; In the feature engineering stage, data transformation and feature selection are also used to enhance the model’s predictive power: in, It is a feature set after feature engineering; It is a primitive feature; It is a new feature obtained through feature selection and transformation. The new features include polynomial features and interactive features. In the indicator integration stage, a variety of machine learning algorithms are used and weights are assigned to each indicator: in, is the performance indicator accuracy of the ith indicator on the validation set, is the average value of all performance indicators.

9. A method for identifying safety risks in mine ecological restoration projects according to claim 6, characterized in that: After detecting and identifying unsafe behaviors, safety risk data fusion and risk assessment are also required.

10. A method for identifying safety risks in mine ecological restoration projects according to claim 7, characterized in that: The specific methods of security risk data fusion and risk potential assessment are as follows: Integrate data flow, business flow and control flow, rely on text information to interact, establish a database based on text information, and thus build a knowledge graph framework related to security risks. Finally, through the semantic association of text and the intrinsic relationship of time and space coupling, provide data analysis results and risk prevention plans.

Images