Construction engineering safety risk grading management and control method and system based on AI

Abstract

The invention relates to the technical field of engineering safety management, in particular to an AI-based construction engineering safety risk hierarchical management and control method and system. Comprising the following steps: S1, acquiring various construction data; s2, constructing a three-dimensional quantitative model, calculating a risk value, setting a plurality of security risk levels, and obtaining a corresponding security risk level according to the risk value; s3, a rule base is constructed, a knowledge graph is constructed, and the rule base comprises a forcing rule, an enterprise definition rule and a suggestion rule; s4, corresponding rules are matched according to the safety risk levels, the priority score of each rule is calculated, and the corresponding rules are executed in sequence according to the priority scores from high to low; s5, according to the sequence obtained in the step S4, after the rule is executed, the relevant parameters for calculating the risk value in the step S2 and the relevant parameters for calculating the priority score in the step S4 are dynamically corrected; and the accuracy of construction engineering safety risk grading management and control is improved.

Description

Technical Field

[0001] The present invention relates to the field of engineering safety management technology, and in particular to an AI-based construction engineering safety risk classification management and control method and system. Background Art

[0002] In construction scenarios in the fields of water conservancy, hydropower, wind power, photovoltaics, and construction, there are often high-risk links such as high-altitude operations, confined space operations, heavy machinery group collaboration, and blasting operations. In operations in high-risk links, precise control of safety risks is required to ensure the safety of workers.

[0003] Most of the existing safety risk management methods rely on manual experience and cannot cover the hidden risks in dynamic construction scenarios, resulting in risk omissions, delayed responses, etc.; there is also a disconnect between risk classification and management measures, resulting in "high level and low control" or "low level and high consumption" situations; cross-stage and cross-departmental data silos cannot achieve closed-loop control of risks throughout the entire cycle, resulting in low accuracy in the classification and management of construction project safety risks.

[0004] Therefore, there is an urgent need to provide an AI-based construction project safety risk classification management method and system to improve the accuracy of construction project safety risk classification management compared with existing technologies. Summary of the Invention

[0005] The present invention solves the technical problems existing in the prior art and provides an AI-based construction project safety risk classification management method and system.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows: An AI-based construction project safety risk classification management method includes the following steps: S1. Obtain various construction data; S2. Build a three-dimensional quantitative model, calculate the risk value based on the construction data obtained in step S1, set multiple safety risk levels, and obtain the corresponding safety risk level based on the risk value; S3. Build a rule base and a knowledge graph. The rule base includes mandatory rules, enterprise-defined rules, and recommended rules. S4. Based on the security risk level obtained in step S2, the corresponding rules are matched in the knowledge graph constructed in step S3, and the priority score of each rule is calculated. The corresponding rules are executed in descending order of priority scores; S5. After executing the rules in the order obtained in step S4, dynamically modify the relevant parameters for calculating the risk value in step S2 and the relevant parameters for calculating the priority score in step S4.

[0007] Furthermore, the three-dimensional quantitative model is expressed as follows: ; In the above formula, represents the risk value, represents the risk probability, Indicates the degree of exposure, Indicates the severity of the consequences.

[0008] Furthermore, the risk probability P is calculated as follows: ; In the above formula, represents the initial probability, Indicates the dynamic correction factor.

[0009] Furthermore, the dynamic correction factor is calculated from the data collected in real time: ; ; In the above formula, represents the weight coefficient of the i-th real-time collected data, obtained through the hierarchical analysis method; represents the normalized deviation value of the i-th real-time collected data, Express The amplification value, i ranges from 1 to I, I represents the number of all real-time collected data, represents the risk sensitivity coefficient of the i-th real-time collected data; The data collected in real time include rainfall, wind speed, temperature, humidity, and deformation and displacement of foundation pit slopes.

[0010] Furthermore, the exposure level is calculated as follows: ; In the above formula, Indicates the number of people in the danger zone, represents the average exposure time in the hazardous area, express The corresponding weight value, express The corresponding weight value.

[0011] Furthermore, the severity of the consequences is given by: ; ; In the above formula, Indicates the area affected by an accident in the construction area. represents the chain reaction evaluation score, express The corresponding weight value, Respectively represent the weight values ​​corresponding to C, represents the theoretical impact area under ideal conditions, Calculated by corresponding accident model; represents the terrain correction factor, Greater than 1, indicating an uphill slope. Less than 1 indicates downhill, Equal to 1 means flat land; represents the meteorological correction factor; Indicates the obstacle correction factor.

[0012] Furthermore, the priority score is calculated as follows: ; In the above formula, represents the priority score, 、 、 Represent different weight values, Take 0.6, Take 0.25, Take 0.15; Indicates the score corresponding to the rule. The score corresponding to the mandatory rule is 100, the score corresponding to the enterprise-defined rule is 50, and the score corresponding to the recommended rule is 10. represents the normalized risk value of the rule, Indicates the normalized consequence value of the rule.

[0013] Furthermore, in step S5, 、 、 、 、 、 、 Perform dynamic correction to the S4 step 、 、 Make dynamic corrections.

[0014] Furthermore, in step S5, the method for dynamic correction is: (1) Analyze the risk events that occurred and determine the key parameters and rules related to the risk events; (2) Calculate the effectiveness based on the determined parameters and rules, which can be calculated using the following formula: ; In the above formula, Indicates validity, Represents the risk value before the measures are implemented, It represents the risk value of one observation period after the implementation of the measures; (3) For the parameters to be dynamically corrected, update them according to the following formula: ; In the above formula, represents the updated parameters, Indicates the parameters before updating, represents the learning rate, Indicates expected effectiveness; Selected from 、 、 、 、 、 、 、 、 、 ; (4) When When it is greater than 0, right Make positive adjustments when When it is less than 0, right Make negative adjustments.

[0015] An AI-based construction project safety risk classification and control system includes a multi-source data input module, a risk classification module, an intelligent decision-making module, a management and control strategy matching module and a closed-loop optimization module. The multi-source data input module is used to execute the content in step S1, the risk classification module is used to execute the content in step S2, the intelligent decision-making module is used to execute the content in step S3, the management and control strategy matching module is used to execute the content in step S4, and the closed-loop optimization module is used to execute the content in step S5.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention collects construction data in real time, calculates risk values ​​based on the real-time collected construction data, obtains the risk level of the area, and then builds a rule base to match different rules in the rule base according to different risk levels. The execution order of different rules is determined by the priority score of each rule. At the same time, after each rule is executed, the effectiveness of the measures is evaluated, and in each evaluation, the relevant parameters for risk value calculation and priority score calculation are dynamically corrected, which can better make the risk assessment and rule priority score fit the reality, correct deviations, and thus improve the accuracy of construction project safety risk classification management and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0018] The technical solution of the present invention will be clearly described below in conjunction with the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, the present invention provides an AI-based construction project safety risk classification management method, comprising the following steps: S1. Acquire a variety of construction data. Specifically, use a multi-source sensor network and adopt the MQTT protocol to collect construction environment data in seconds, and align the collected construction environment data with the construction schedule in the BIM model in time and space. The multi-source sensor network includes geological radar, anemometer, UWB personnel positioning tags, equipment IoT terminals, etc. The construction environment data includes geological displacement, wind speed, temperature, humidity, equipment status and other data.

[0020] In step S1, the various construction data obtained are cleaned and standardized, abnormal data such as mutation values ​​are eliminated, and missing values ​​are filled with algorithms. At the same time, the data are normalized to ensure that data from different sources have consistent scales and are time series aligned to avoid decision-making bias due to data asynchrony.

[0021] S2. Construct a three-dimensional quantitative model, calculate the risk value through the three-dimensional quantitative model, set the first risk threshold, the second risk threshold, and the third risk threshold. When the risk value is greater than or equal to the first risk threshold, the safety risk level of the construction project is evaluated as level one. When the risk value is less than the first risk threshold and greater than or equal to the second risk threshold, the safety risk level of the construction project is evaluated as level two. When the risk value is less than the second risk threshold and greater than or equal to the third risk threshold, the safety risk level of the construction project is evaluated as level three. When the risk value is less than the third risk threshold, the safety risk level of the construction project is evaluated as level four.

[0022] The first, second and third risk thresholds are associated with the overall risk profile of the project. For example, during critical project stages (such as dam closure and wind turbine installation), or when near-misses occur frequently in the near future, the system can automatically lower all thresholds by 10-20%, entering a "strict control" mode, allowing more scenarios to trigger higher levels of control.

[0023] The three-dimensional quantitative model is expressed as follows: ; In the above formula, represents the risk value, represents the risk probability, Indicates the degree of exposure, Indicates the severity of the consequences.

[0024] (1) The risk probability is calculated by the following formula: ; In the above formula, represents the initial probability, Indicates the dynamic correction factor.

[0025] The method for obtaining the initial probability is: based on historical accident data, multiple risk scenarios are divided according to the construction stage, accident records of similar projects in the past 5-10 years are integrated, the frequency of accidents under different risk scenarios is counted, and the frequency of accidents in each risk scenario is recorded as the initial probability of the risk scenario.

[0026] The dynamic correction coefficient is calculated from the data collected in real time: ; In the above formula, represents the weight coefficient of the i-th real-time collected data, represents the normalized deviation value of the i-th real-time collected data, Express The amplification value of i ranges from 1 to I, and I represents the number of all real-time collected data.

[0027] The data collected in real time include rainfall, wind speed, temperature, humidity, and deformation and displacement of foundation pit slopes.

[0028] Calculated by the following formula: ; In the above formula, represents the real-time detection value of the i-th real-time collected data, represents the preset safety threshold of the i-th real-time collected data, Dynamic adjustments can be made through industry standards or construction plans.

[0029] Indicates increased risk; Indicates that the real-time collected data is greater than the corresponding preset safety threshold, usually according to treatment, or as a risk reduction factor in specific scenarios.

[0030] It is expressed by the following formula: ; In the above formula, represents the risk sensitivity coefficient of the i-th real-time collected data, greater than 0, Determines the rate at which risk increases with parameter deviation, and high-risk parameters correspond to The bigger.

[0031] It is obtained through the hierarchical analysis method, specifically: the target layer is the dynamic correction coefficient, the criterion layer is I real-time collected data, and the scheme layer is the weight coefficient of I real-time collected data; the data in the criterion layer are compared pairwise to determine their relative importance, thereby constructing a judgment matrix. The 1-9 scaling method is usually used to determine the relative importance, and then the weight coefficient of each real-time collected data is obtained.

[0032] (2) The degree of exposure is calculated using the following formula: ; In the above formula, Indicates the number of people in the danger zone, represents the average exposure time in the hazardous area, express The corresponding weight value, express The corresponding weight value.

[0033] (3) The severity of the consequences is obtained by the following formula: ; In the above formula, Indicates the area affected by an accident in the construction area. represents the chain reaction evaluation score, express The corresponding weight value, They represent the weight values ​​corresponding to C respectively.

[0034] A is calculated by the following formula: ; In the above formula, represents the theoretical impact area under ideal conditions, Calculated by corresponding accident model; represents the terrain correction factor, Greater than 1, indicating an uphill slope. Less than 1 indicates downhill, Equal to 1 means flat land; represents the meteorological correction factor; Indicates the obstacle correction factor. When there are embankments and retaining walls, the impact range will be effectively reduced. Less than 1.

[0035] for The corresponding accident models include collapse accident model, equipment collapse model and explosion accident model.

[0036] (1) The collapse accident model is expressed by the following formula: ; ; In the above formula, Indicates the impact distance, represents the lateral diffusion angle of the slip body, It can be obtained through geological exploration reports. represents the slope height, represents the soil internal friction angle, Indicates the slope angle.

[0037] (2) The equipment collapse model is constructed using the tower as an example, and is specifically expressed by the following formula: ; In the above formula, Indicates the tower height, represents the length of the blade, Indicates the width of the tower base.

[0038] (3) The explosion accident model is expressed by the following formula: ; ; In the above formula, Indicates the damage radius of the explosion. Indication and overpressure The relevant empirical coefficient, Indicates the shock wave overpressure value corresponding to different hazard levels. Indicates the TNT equivalent of the explosive that caused the explosion.

[0039] S3. Build a rule base and a knowledge graph. In the knowledge graph, the safety risk levels of the four construction projects built in step S2 correspond to multiple rules respectively, which facilitates the automatic association of subsequent rules.

[0040] Using semantic parsing algorithms, we extract key parameters and logical relationships from industry specifications and construct multiple rules, including "when the slope displacement rate is greater than 5mm / h, it is judged as a level 2 risk" and "when the gust of wind is greater than 10m / s, the lifting operation is suspended."

[0041] The rules in the rule base include mandatory rules, enterprise-defined rules and recommended rules. Mandatory rules are directly triggered control measures, such as "when the wind speed and rain are 12m / s, work must be stopped"; enterprise-defined rules are rules customized by each enterprise based on work experience, such as "displacement exceeding 5mm requires manual review"; recommended rules are recommended solutions generated by the model, such as "when the displacement rate is greater than 2mm / h, it is recommended to increase support."

[0042] S4. Based on the safety risk level of the construction project evaluated in step S2, the corresponding rules are matched in the knowledge graph constructed in step S3. Specifically, the Drools rule engine is used to load IF-THEN rules to achieve matching between the construction project, the safety risk level of the construction project, and the rules in the rule library.

[0043] For matching rules, the priority score of each rule is first calculated, and then the corresponding rules are executed in descending order of priority score. If the priority scores of two rules are within the priority score difference threshold, the principle of "life safety first" is followed, and the safer rule is executed first. The priority score difference threshold indicates that the two priority scores are relatively close.

[0044] The priority score is calculated according to the following formula: ; In the above formula, represents the priority score, 、 、 Represent different weight values, Take 0.6, Take 0.25, Take 0.15; Indicates the score corresponding to the rule. The score corresponding to the mandatory rule is 100, the score corresponding to the enterprise-defined rule is 50, and the score corresponding to the recommended rule is 10. represents the normalized risk value of the rule, Indicates the normalized consequence value of the rule.

[0045] S5. According to the rules selected in step S4, the rules are executed in the construction project, and the relevant parameters in the three-dimensional quantitative model in step S2 and the relevant parameters in the priority score calculation formula in step S4 are dynamically corrected; at the same time, the optimized rules are updated to the knowledge graph constructed in step S3.

[0046] In this step, specifically in step S2 、 、 、 、 、 、 Perform dynamic correction, specifically in step S4 、 、 Make dynamic corrections.

[0047] The specific dynamic correction method is: (1) First analyze the risk event that occurred and determine the key parameters and rules related to the risk event.

[0048] (2) Calculate the effectiveness based on the determined parameters and rules, which can be calculated using the following formula: ; In the above formula, Indicates validity, Represents the risk value before the measures are implemented, It represents the risk value of one observation period after the implementation of the measure.

[0049] When it is greater than 0, it means the measure is effective. When it is less than 0, it means the risk is worsening.

[0050] (3) For the parameters to be dynamically corrected, update them according to the following formula: ; In the above formula, represents the updated parameters, Indicates the parameters before updating, represents the learning rate, Take 0.01-0.1, Indicates expected effectiveness, Take 0 or 0.5.

[0051] for 、 、 、 、 、 、 、 、 、 .

[0052] (4) When When it is greater than 0, right Make positive adjustments when When it is less than 0, right Make negative adjustments.

[0053] The present invention also provides an AI-based construction project safety risk grading management and control system, including a multi-source data input module, a risk grading module, an intelligent decision-making module, a management and control strategy matching module and a closed-loop optimization module. The multi-source data input module is used to execute the content in step S1, the risk grading module is used to execute the content in step S2, the intelligent decision-making module is used to execute the content in step S3, the management and control strategy matching module is used to execute the content in step S4, and the closed-loop optimization module is used to execute the content in step S5.

[0054] The present invention collects construction data in real time, calculates risk values ​​based on the real-time collected construction data, obtains the risk level of the area, and then builds a rule base to match different rules in the rule base according to different risk levels. The execution order of different rules is determined by the priority score of each rule. At the same time, after each rule is executed, the effectiveness of the measures is evaluated, and in each evaluation, the relevant parameters for risk value calculation and priority score calculation are dynamically corrected, which can better make the risk assessment and rule priority score fit the reality, correct deviations, and thus improve the accuracy of construction project safety risk classification management and control.

[0055] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. An AI-based construction project safety risk classification management method, characterized by: The following steps are involved: S1. Obtain various construction data; S2. Build a three-dimensional quantitative model, calculate the risk value based on the construction data obtained in step S1, set multiple safety risk levels, and obtain the corresponding safety risk level based on the risk value; S3. Build a rule base and a knowledge graph. The rule base includes mandatory rules, enterprise-defined rules, and recommended rules. S4. Based on the security risk level obtained in step S2, the corresponding rules are matched in the knowledge graph constructed in step S3, and the priority score of each rule is calculated. The corresponding rules are executed in descending order of priority scores; S5. After executing the rules in the order obtained in step S4, dynamically modify the relevant parameters for calculating the risk value in step S2 and the relevant parameters for calculating the priority score in step S4.

2. The AI-based construction project safety risk classification management method according to claim 1 is characterized in that: The three-dimensional quantitative model is expressed as follows: ； In the above formula, represents the risk value, represents the risk probability, Indicates the degree of exposure, Indicates the severity of the consequences.

3. The AI-based construction project safety risk classification management method according to claim 2 is characterized in that: The risk probability P is calculated by the following formula: ； In the above formula, represents the initial probability, Indicates the dynamic correction factor.

4. The AI-based construction project safety risk classification management method according to claim 3 is characterized in that: The dynamic correction coefficient is calculated from the data collected in real time: ； ； In the above formula, represents the weight coefficient of the i-th real-time collected data, obtained through the hierarchical analysis method; represents the normalized deviation value of the i-th real-time collected data, Express The amplification value, i ranges from 1 to I, I represents the number of all real-time collected data, represents the risk sensitivity coefficient of the i-th real-time collected data; The data collected in real time include rainfall, wind speed, temperature, humidity, and deformation and displacement of foundation pit slopes.

5. The AI-based construction project safety risk classification management method according to claim 4 is characterized in that: Exposure is calculated using the following formula: ； In the above formula, Indicates the number of people in the danger zone, represents the average exposure time in the hazardous area, express The corresponding weight value, express The corresponding weight value.

6. The AI-based construction project safety risk classification management method according to claim 5 is characterized in that: The severity of the consequences is given by: ； ； In the above formula, Indicates the area affected by an accident in the construction area. represents the chain reaction evaluation score, express The corresponding weight value, Respectively represent the weight values ​​corresponding to C, represents the theoretical impact area under ideal conditions, Calculated by corresponding accident model; represents the terrain correction factor, Greater than 1, indicating an uphill slope. Less than 1 indicates downhill, Equal to 1 means flat land; represents the meteorological correction factor; Indicates the obstacle correction factor.

7. The AI-based construction project safety risk classification management method according to claim 6 is characterized in that: The priority score is calculated according to the following formula: ； In the above formula, represents the priority score, 、 、 Represent different weight values, Take 0.6, Take 0.25, Take 0.15; Indicates the score corresponding to the rule. The score corresponding to the mandatory rule is 100, the score corresponding to the enterprise-defined rule is 50, and the score corresponding to the recommended rule is 10. represents the normalized risk value of the rule, Indicates the normalized consequence value of the rule.

8. The AI-based construction project safety risk classification management method according to claim 7 is characterized in that: In step S5, the 、 、 、 、 、 、 Perform dynamic correction to the S4 step 、 、 Make dynamic corrections.

9. The AI-based construction project safety risk classification management method according to claim 8 is characterized in that: In step S5, the method for dynamic correction is: (1) Analyze the risk events that occurred and determine the key parameters and rules related to the risk events; (2) Calculate the effectiveness based on the determined parameters and rules, which can be calculated using the following formula: ； In the above formula, Indicates validity, Represents the risk value before the measures are implemented, It represents the risk value of one observation period after the implementation of the measures; (3) For the parameters to be dynamically corrected, update them according to the following formula: ； In the above formula, represents the updated parameters, Indicates the parameters before updating, represents the learning rate, Indicates expected effectiveness; Selected from 、 、 、 、 、 、 、 、 、 ; (4) When When it is greater than 0, right Make positive adjustments when When it is less than 0, right Make negative adjustments.

10. An AI-based construction project safety risk classification management and control system, characterized by: An AI-based construction project safety risk grading control method according to any one of claims 1 to 9 is used, which includes a multi-source data input module, a risk grading module, an intelligent decision-making module, a control strategy matching module and a closed-loop optimization module. The multi-source data input module is used to execute the content in step S1, the risk grading module is used to execute the content in step S2, the intelligent decision-making module is used to execute the content in step S3, the control strategy matching module is used to execute the content in step S4, and the closed-loop optimization module is used to execute the content in step S5.

Images