Safety construction management method based on digitization

Abstract

The invention discloses a safety construction management method based on digitization, and belongs to the field of construction management. The safety construction management method comprises the steps that disk saw operation data of a construction site are collected; preprocessing the collected disk saw operation data; according to the preprocessed disk saw operation data, the safe operation state of the disk saw is judged through a support vector machine (SVM) algorithm; according to the safe operation state, performing risk assessment by adopting a risk assessment method based on combination of an evidence theory and a fuzzy set theory to obtain a safety risk level; and according to the safe operation state and the safety risk level, carrying out safety management on the disk saw through a multi-level early warning control strategy based on reinforcement learning. In view of low risk assessment precision in disk saw construction in the prior art, the risk assessment precision is improved through risk assessment combining an evidence theory and a fuzzy set theory and by using a reinforcement learning early warning control strategy and the like.

Description

Technical Field

[0001] The present application relates to the field of construction management, and in particular to a digital-based safe construction management method. Background Art

[0002] With the rapid development of the construction industry and the continuous expansion of the scale of projects, construction safety issues have become increasingly prominent and have become a bottleneck restricting the sustainable and healthy development of the industry. According to statistics, construction safety accidents account for more than 20% of the total number of production safety accidents in the country, among which a large number of accidents occurred in the operation of construction machinery such as circular saws. Circular saws are widely used in construction due to their high efficiency and convenience, but they also have many safety hazards and difficulties in risk control. Once an accident occurs, it often causes serious casualties and economic losses.

[0003] Traditional circular saw construction safety management mainly relies on manual experience and qualitative analysis, which has defects such as strong subjectivity, large randomness, and obvious lag. On the one hand, the construction site environment is complex and there are many interference factors. It is difficult for operators to accurately identify and evaluate the safety risks of circular saws. They often take countermeasures only after accidents occur, and safety management is passive and lagging. On the other hand, the safety status of circular saws is affected by many factors, such as saw blade speed, saw blade height, protective cover position, etc. It is difficult to monitor and judge in real time and comprehensively by manual labor alone. Risk assessment and early warning often become formalities, lacking scientificity and effectiveness.

[0004] In order to improve the level of circular saw construction safety management, some scholars and enterprises have carried out fruitful research and practice. The "Metal Cutting Machine Tool Safety Protection Technical Manual" published by the Machinery Industry Press systematically summarizes the safety protection design methods and operating procedures of circular saws and other machine tools; the "Construction Safety Risk Grading Control and Hidden Danger Investigation and Management System" of the China Academy of Building Research has established safety inspection standards and management processes for construction machinery including circular saws; a construction company in Shenzhen has developed an "Intelligent Safety Monitoring System for Circular Saws", which realizes real-time monitoring and visual management of equipment status through sensor collection and big data analysis. These research results have improved the current safety management status of circular saws to a certain extent, but there are still problems such as low risk assessment accuracy and unintelligent early warning control, which makes it difficult to meet the growing safety management needs. Summary of the invention

[0005] In response to the problem of low risk assessment accuracy in circular saw construction in the prior art, the present application provides a digital-based safe construction management method, which improves the risk assessment accuracy through risk assessment combining evidence theory and fuzzy set theory, and utilizes reinforcement learning early warning control strategies.

[0006] The purpose of this application is achieved through the following technical solutions.

[0007] The present application provides a digital-based safe construction management method, comprising: collecting circular saw operation data at a construction site; the operation data comprising saw blade rotation speed, saw blade height and protective cover position; preprocessing the collected circular saw operation data; judging the safe operation state of the circular saw by a support vector machine (SVM) algorithm based on the preprocessed circular saw operation data; the input of the support vector machine (SVM) algorithm is the circular saw operation data, and the output is the safe operation state of the circular saw; according to the safe operation state, a risk assessment method based on a combination of evidence theory and fuzzy set theory is used to conduct risk assessment to obtain a safety risk level; according to the safe operation state and the safety risk level, the circular saw is safely managed by a multi-level early warning control strategy based on reinforcement learning.

[0008] Among them, the circular saw is mainly composed of motors, saw blades, protective covers and other components. Its working principle is that the circular saw blade rotates at a high speed to cut the workpiece. Circular saws are widely used in wood processing, construction and other fields, but there are also certain safety risks. The saw blade speed refers to the speed at which the circular saw blade rotates, usually expressed in revolutions per minute (rpm). The speed is one of the key parameters affecting cutting efficiency and safety. Too high a speed may cause problems such as overheating, deformation or breakage of the saw blade, while too low a speed will reduce cutting efficiency. The saw blade height refers to the height of the saw blade exposed from the work surface. The appropriate saw blade height can ensure the cutting depth and improve cutting efficiency and quality. However, if the saw blade height is too large, it will increase the exposed area of ​​the saw blade and increase the risk of accidental contact. The protective cover is a key safety protection device of the circular saw, which can cover the saw blade when the saw blade is not in use to prevent accidental contact. The protective cover position refers to the coverage of the saw blade by the protective cover. The appropriate protective cover position can isolate the saw blade to the greatest extent and reduce accidental risks. In this application, the risk assessment method based on the combination of evidence theory and fuzzy set theory uses evidence theory to express and combine the importance of different risk indicators with uncertainty, and uses fuzzy set theory to describe the membership of the risk level of each indicator. The evidence combination results and the fuzzy assessment results are integrated to obtain a comprehensive risk level. This method can handle both random uncertainty and fuzzy uncertainty at the same time, thereby improving the accuracy and reliability of the assessment.

[0009] Furthermore, based on the preprocessed circular saw operation data, the safe operation state of the circular saw is judged by a support vector machine (SVM) algorithm, including: selecting a radial basis kernel function RBF as a kernel function of the SVM algorithm to construct an SVM classification model; using a grid search method to optimize the hyperparameters of the SVM classification model, wherein the hyperparameters include a penalty coefficient C and a parameter γ of the radial basis kernel function RBF; based on the preprocessed circular saw operation data, a feature vector x is constructed; and through a decision function of the SVM classification model, a distance between the feature vector x and the classification hyperplane is calculated to obtain the safe operation state of the circular saw.

[0010] Furthermore, the decision function expression of the SVM classification model is: , where x represents the feature vector of the circular saw operation data. x = [ x 1 , x 2 , x 3 ] ,in, is the saw blade speed, is the saw blade height, is the position of the protective cover; w represents the normal vector of the SVM classification hyperplane, which represents the direction and slope of the hyperplane and is obtained through SVM model training; Represents the nonlinear transformation of the RBF kernel function on the feature vector x, mapping x to a high-dimensional feature space. , where γ is the parameter of the RBF kernel function, which controls the width of the Gaussian function. is the training sample; b represents the intercept of the SVM classification hyperplane, which represents the distance between the hyperplane and the origin and is obtained through SVM model training; λ represents the safety state correction factor, which is used to adjust the original SVM classification result to make it more consistent with the actual safety state of the circular saw. The value of λ can be set based on experience or obtained through data analysis and optimization; g(x) represents the safety state correction function, which adjusts the original SVM classification result according to the characteristics of the circular saw operation data.

[0011] Furthermore, ,in, are the weight coefficients of the three characteristic parameters, indicating their influence on the safety status, which can be determined based on expert experience or data analysis; The security state correction subfunction for the three characteristic parameters: , , ,in, They are the safety thresholds of saw blade speed, saw blade height, and protective cover position. They are the upper and lower limits of the safety range of the three characteristic parameters respectively.

[0012] By introducing the safety state correction factor λ and the correction function , according to the actual characteristics of the circular saw operation data, the original SVM classification results can be appropriately adjusted to make the decision function more in line with the actual safety status of the circular saw. When a certain feature parameter exceeds the safety range, its corresponding correction sub-function will produce a larger value, thereby reducing the value of the entire decision function, indicating that the feature parameter has a negative impact on the safety status; when all feature parameters are within the safety range, the value of the correction function is close to 0, and the value of the decision function is mainly determined by the original SVM classification results. In this way, the accuracy and reliability of the circular saw safety status judgment can be further improved on the basis of SVM classification.

[0013] Furthermore, the safety risk level is obtained, including: determining the initial risk level according to the safe operation status of the circular saw; establishing a risk assessment index system including the three indicators of accident possibility, accident consequence severity and saw blade exposure; and setting the assessment level and fuzzy membership function of each indicator; setting the importance weights of the three indicators of accident possibility, accident consequence severity and saw blade exposure according to historical data, and using the probability assignment function BPA in the evidence theory to quantify the importance weights, and obtaining the evidence system of the importance of each indicator. According to the body of evidence , using Dempster's combination rule to calculate the comprehensive evidence ; According to the circular saw operation data, the fuzzy evaluation vector of each indicator is obtained by matching the membership function; among them, the evaluation level of the i-th indicator is , the membership value obtained by matching is , then the fuzzy evaluation vector of the corresponding index is ; Using the fuzzy weighted average operator, the comprehensive evidence and the fuzzy evaluation vector Fusion is performed to obtain the fuzzy membership distribution of the comprehensive risk assessment level. : U ( x ) = ∑ i = 1 3 [ m ( A i ) × R i ( x ) ] , where x is the risk level variable, is the comprehensive evidence of the importance of indicator i, is the fuzzy evaluation vector of index i; the fuzzy membership distribution U(x) is converted into a quantitative safety risk level value by using the centroid defuzzification method in fuzzy mathematics ; Based on the initial risk level and security risk level value ,The Yager fusion rule in DS evidence theory is used to obtain the final safety risk level that comprehensively considers the safety status and risk assessment of the circular saw.

[0014] Among them, in evidence theory, BPA is a basic probability assignment function used to describe the uncertainty of evidence. It divides the uncertainty of evidence into several non-overlapping subsets, and assigns a basic probability value between [0, 1] to each subset to indicate the degree to which the evidence supports the subset. The sum of the probability assignments of all subsets is equal to 1. The possibility of an accident refers to the probability of a safety accident occurring under given conditions. It is a key indicator of risk assessment and reflects the random uncertainty of the occurrence of accidents. In this scheme, the possibility of an accident is used as an important dimension for assessing the safety risk of circular saws.

[0015] The severity of accident consequences refers to the severity of the negative consequences such as casualties, equipment damage, and environmental impact caused by a safety accident. It is another important indicator for measuring the size of the risk and reflects the potential harm of the accident. The saw blade exposure reflects the area of ​​the saw blade exposed during the operation of the circular saw, which directly affects the risk level of accidental contact with the saw blade.

[0016] Dempster combination rule is an evidence fusion method in evidence theory, which is used to integrate information from multiple different evidence sources. It obtains a comprehensive evidence by calculating the degree of consistency between different evidences, reflecting the propositions supported by each piece of evidence. Fuzzy weighted average operator is a commonly used fuzzy information integration method. It obtains a comprehensive evaluation result by assigning weights to each fuzzy evaluation result and performing weighted average on them. The weight is usually determined according to the importance of the evaluation index, reflecting the degree of influence of different indicators on the overall evaluation. The centroid method is a commonly used fuzzy mathematical defuzzification method, which is used to convert a fuzzy set into a deterministic numerical value. It calculates the centroid abscissa of the fuzzy set membership function as the representative value of the fuzzy set. The centroid value comprehensively considers the degree of membership of each element and can reasonably reflect the overall position of the fuzzy set.

[0017] The Yager fusion rule is an important extension of the DS evidence theory, which is used to solve the problem of contra-intuitive results of the Dempster combination rule when there is a serious conflict of evidence. It avoids the appearance of overly negative results by introducing an unknown set and assigning the probability of conflicting evidence to this set.

[0018] Furthermore, a risk assessment index system including three indicators, namely, accident possibility, accident consequence severity and saw blade exposure, is established; and the assessment level and fuzzy membership function of each indicator are set, including: the assessment level of accident possibility is {very small, small, medium, large, very large}, and the membership function adopts a triangular membership function; the assessment level of accident consequence severity is {very low, low, medium, high, very high}, and the membership function adopts a trapezoidal membership function; the assessment level of saw blade exposure is {very weak, weak, medium, strong, very strong}, and the membership function adopts a Gaussian membership function.

[0019] Furthermore, the Dempster combination rule is used to calculate the comprehensive evidence , including: calculating the conflict degree K between evidence bodies: K = ∑ B 1 ∩ B 2 ∩ B 3 =∅ [ m 1 ( B 1 ) × m 2 ( B 2 ) × m 3 ( B 3 ) ] , where B1, B2, and B3 are subsets of the three indicator importance evaluation frameworks, and ; That is, K is the sum of the incompatible confidences between the evidence bodies; calculate the comprehensive evidence body of the non-zero subset A : m ( A ) = 1 1 − K × ∑ B 1 ∩ B 2 ∩ B 3 = A [ m 1 ( B 1 ) × m 2 ( B 2 ) × m 3 ( B 3 ) ] , where A is the non-zero subset of the indicator importance evaluation framework, are subsets of the three indicator importance assessment frameworks, and ;Right now It is the sum of the normalized products of the confidence scores of each evidence body on subset A.

[0020] Among them, the evaluation framework of indicator importance is set as Θ={A1, A2, ..., An}, where A1, A2, ..., An represent n levels of indicator importance, such as {very low, low, medium, high, very high}. Then the subset of this framework includes all possible combinations of indicator importance levels, for example: {A1}, {A2}, ..., {An}: single importance level; {A1, A2}, {A1, A3}, ...: two-by-two combination; {A1, A2, A3}, {A1, A2, A4}, ...: three-by-three combination; ...; {A1, A2, ..., An-1}, {A2, A3, ..., An}: combination of n-1 levels; Θ: combination of all levels; the evidence bodies m1, m2, m3 of indicator importance are actually probability assignment functions defined on these subsets. They represent the degree of support of each piece of evidence (such as expert opinion) for various combinations of importance levels.

[0021] According to the theory of evidence, if the support of a subset of evidence is 0, that is, the probability is 0, then the subset is called the zero subset of evidence. On the contrary, if the probability of a subset is greater than 0, it is called the non-zero subset of evidence. For the evaluation framework of indicator importance Θ, its non-zero subset refers to the combination of indicator importance levels that are supported by at least one piece of evidence. For example, if the non-zero subsets corresponding to the three evidence bodies are: m1: {A1}, {A2, A3}, {A4, A5}; m2: {A1, A2}, {A3}, {A4}; m3: {A2}, {A3, A4}, {A5}; then the non-zero subsets of the entire evaluation framework are the union of these subsets: {A1}, {A2}, {A3}, {A4}, {A5}, {A1, A2}, {A2, A3}, {A3, A4}, {A4, A5}; when calculating the comprehensive evidence body m(A), A traverses all these non-zero subsets. The calculation of K is also performed on these non-zero subsets. The purpose of this treatment is to reduce the amount of calculation, because the product of m1, m2, and m3 corresponding to the zero subset must be 0, which does not affect the result.

[0022] Furthermore, the centroid defuzzification method in fuzzy mathematics is used to transform the fuzzy membership distribution U(x) into a quantitative safety risk level value x*, including: calculating the centroid abscissa x* of the fuzzy membership distribution U(x): , where x is the risk level variable, is the membership distribution function of comprehensive risk assessment, [a, b] is the domain interval of x, a and b are the lower and upper bounds of the domain respectively; x* represents the centroid position of the membership distribution, reflecting the overall position of the comprehensive risk level in the domain.

[0023] The domain refers to the value range of the elements in the fuzzy set. It defines the boundary and range of the fuzzy set. In this scheme, the domain interval [a, b] refers to the value range of the security risk level.

[0024] Furthermore, according to the safe operation status and safety risk level, the circular saw is safely managed through a multi-level early warning control strategy based on reinforcement learning, including: constructing a Markov decision process model MDP for circular saw safety early warning control; using the Q-learning algorithm to solve the Markov decision process model MDP to obtain the optimal early warning control strategy ; According to the optimal early warning control strategy , generate corresponding early warning control instructions.

[0025] Furthermore, a Markov decision process model MDP for circular saw safety early warning control is constructed, including:

[0026] According to the safe operation status and safety risk level of the circular saw, the safety status of the circular saw is divided into three states: safe, warning and dangerous, forming a state space ;in, Indicates a safe state, the corresponding SVM classification result is safe, and the risk level is low risk; Indicates the warning state, the corresponding SVM classification result is safe, and the risk level is medium risk; Indicates a dangerous state, the corresponding SVM classification result is dangerous, and the risk level is high risk or very high risk;

[0027] According to the safety status of the circular saw, set the corresponding early warning control action to form an action space ;in, Indicates maintaining the current status and not taking early warning control measures; Indicates the issuance of warning signals, including sound and light alarms, human-computer interaction prompts, etc.; Indicates forced speed reduction, which limits the maximum speed of the circular saw through the controller; Indicates emergency braking, the controller cuts off the power of the circular saw and applies braking force; according to historical operation data, the frequency of the circular saw transferring to other states after taking different early warning control actions in different safety states is counted to obtain the state transfer frequency matrix ; According to the state transition frequency matrix Calculate state transition probability : , set the reward function for early warning control , represents the immediate reward obtained by the circular saw after taking action a in state s; ,in, represents a security risk reward, Represents a production efficiency reward, Indicates energy loss reward; is the weight coefficient of the three reward components, satisfying ; ,in, It represents the risk cost under state s, and different cost values ​​can be set according to the risk level; It represents the probability of an accident occurring after taking action a in state s, which can be estimated based on historical data and expert experience; The calculation formula is: ,in, It represents the time cost saved after taking action a in state s. Different cost values ​​can be set according to the impact of different actions on production efficiency. It represents the probability of resuming normal production after taking action a in state s, which can be estimated based on the state transition probability; The calculation formula is: ,in, It represents the energy cost consumed after taking action a in state s. Different cost values ​​can be set according to the energy consumption characteristics of different actions.

[0028] Among them, the source loss reward refers to the energy cost such as additional electricity consumed due to forced speed reduction, emergency braking, etc. of the equipment after taking the early warning control action a in state s. It is a negative reward, which means that the more energy loss, the lower the reward value. In the reward function R(s, a), the energy loss reward is used as a component to comprehensively evaluate the effect of the early warning control action together with the safety risk reward and the production efficiency reward. The purpose of introducing the energy loss reward is to balance safety and economy and avoid excessively sacrificing energy efficiency in pursuit of absolute safety.

[0029] The probability of resuming normal production refers to the possibility that the circular saw will return to normal production after taking early warning control action a in state s. It is a conditional probability that indicates the chance of the system returning to normal from an abnormal state given the current state and control action. The probability of resuming normal production reflects the effectiveness of early warning control and is a key factor in the production efficiency reward Refficiency(s, a). If a control action can restore the system to normal with a high probability, it means that the production efficiency reward for the action is high. Specifically, it is to calculate the frequency of successful recovery to a safe state after taking action a in state s, and then divide it by the total frequency of taking action a in state s.

[0030] Energy cost refers to the economic cost of the energy such as electricity actually consumed by the circular saw after taking early warning control action a in state s. It is related to factors such as the energy consumption characteristics of the equipment, energy prices, and the duration of the control action. When calculating the energy loss reward Renergy(s, a), the energy cost Cenergy(s, a) is a basic variable. It can usually be estimated by multiplying the rated power of the equipment, the duration of the control action, and the unit energy price. The introduction of energy cost enables the reward function to more comprehensively evaluate the effect of early warning control and balance the impact of safety, production efficiency, and energy loss. This helps the Markov decision process model learn an optimal early warning control strategy.

[0031] Compared with the prior art, the advantages of this application are:

[0032] This application uses the support vector machine (SVM) algorithm to identify the safe operating state of the circular saw. By selecting the radial basis kernel function (RBF) as the kernel function, using the grid search method to optimize the hyperparameters, and introducing the safety state correction factor and correction function, the accuracy of the safety state judgment is improved from the data and algorithm level. The SVM algorithm can effectively process high-dimensional small sample data, the RBF kernel function can realize nonlinear classification, the grid search and cross-validation can optimize the model parameters, and the safety state correction mechanism can adaptively adjust the judgment results according to the actual working conditions of the circular saw.

[0033] This application innovatively proposes a risk assessment method based on the combination of evidence theory and fuzzy set theory, constructs a risk assessment index system that includes three indicators: accident possibility, accident consequence severity, and saw blade exposure, uses fuzzy membership function to quantitatively evaluate each indicator, uses the probability assignment function in evidence theory to model the importance of indicators, integrates evidence information and assessment results through Dempster combination rule and fuzzy weighted average operator, and finally uses the center of gravity defuzzification method to obtain a quantitative risk level. This method comprehensively considers the uncertainty factors in the risk assessment process and improves the assessment accuracy;

[0034] This application constructs a Markov decision process MDP model based on the safe operation status and risk assessment level of the circular saw. By defining the state space, action space, state transition probability and reward function, the dynamic random characteristics of the early warning control process are characterized. On this basis, the Q-learning reinforcement learning algorithm is used to solve the MDP model. Through continuous exploration and learning, the optimal early warning control strategy is obtained and the corresponding control instructions are generated. This method can autonomously select the optimal early warning control action according to the real-time safety status and risk level of the circular saw. Through multi-level early warning and linkage control, timely intervention measures can be taken before the accident occurs, thereby effectively avoiding or mitigating the consequences of the accident and improving the timeliness and effectiveness of safety early warning control. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The present application will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0036] Figure 1 is an exemplary flow chart of a digital-based safe construction management method according to some embodiments of the present application;

[0037] Figure 2 is an exemplary flow chart of generating a safe operating state of a circular saw according to some embodiments of the present application;

[0038] Figure 3 is an exemplary flow chart of generating a final security risk level according to some embodiments of the present application;

[0039] Figure 4 This is an exemplary flowchart for setting early warning control instructions according to some embodiments of the present application. DETAILED DESCRIPTION

[0040] The method and system provided in the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0041] like Figure 1As shown, the circular saw operation data at the construction site is collected; the operation data includes saw blade speed, saw blade height and protective cover position; the collected circular saw operation data is preprocessed; based on the preprocessed circular saw operation data, the safe operation state of the circular saw is judged by a support vector machine SVM algorithm; the input of the support vector machine SVM algorithm is the circular saw operation data, and the output is the safe operation state of the circular saw; according to the safe operation state, a risk assessment method based on the combination of evidence theory and fuzzy set theory is used to conduct risk assessment to obtain a safety risk level; according to the safe operation state and the safety risk level, a multi-level early warning control strategy based on reinforcement learning is used to manage the circular saw safely;

[0042] Collect the operation data of the circular saw at the construction site; the operation data includes the saw blade speed, saw blade height and protective cover position; specifically, according to the data type and accuracy requirements to be collected, select appropriate sensors, such as Hall sensors for measuring saw blade speed, laser distance sensors for measuring saw blade height, proximity switches for detecting the position of protective covers, etc. Install the sensor at the corresponding position of the circular saw to ensure that the sensor's measurement range, installation angle, fixing method, etc. meet the requirements of data collection. Design the sensor data acquisition circuit and acquisition module, including sensor signal conditioning circuit, analog-to-digital conversion circuit, data storage circuit, etc. According to the output characteristics of the sensor and the time interval of data collection, reasonably set parameters such as sampling frequency, quantization accuracy, and data cache size. According to the site environment and data transmission distance, select the appropriate communication protocol and communication module, such as Zigbee, Bluetooth and other wireless communication protocols for short distances, and LoRa, NB-IoT and other low-power wide area network protocols for long distances. The communication module is responsible for transmitting the collected sensor data to the edge gateway or cloud server.

[0043] Preprocess the collected circular saw operation data; specifically, parse the received raw data packets according to the communication protocol and data format, and extract valid data such as saw blade speed, saw blade height, and protective cover position. Convert the data into a unified physical quantity unit and numerical range to facilitate subsequent data analysis and modeling. Since the sampling frequency and timestamp of different sensors may be inconsistent, it is necessary to synchronize and align multi-source heterogeneous data. Based on the timestamp or sampling sequence number, interpolate, fit or resample the data of different sensors to make their time axes consistent, which is convenient for data fusion and association analysis. Since the physical quantities measured by different sensors have different dimensions and numerical ranges, in order to eliminate the dimension effect and facilitate data comparison and model training, the data needs to be normalized or standardized. Commonly used methods include maximum and minimum value normalization, zero mean unit variance standardization, etc., so that data with different characteristics are distributed on the same scale.

[0044] like Figure 2As shown in the figure, according to the preprocessed circular saw operation data, the safe operation state of the circular saw is judged by the support vector machine SVM algorithm; the input of the support vector machine SVM algorithm is the circular saw operation data, and the output is the safe operation state of the circular saw; the radial basis kernel function RBF is selected as the kernel function of the SVM algorithm to construct the SVM classification model; the form of the RBF kernel function is: , where x and are two feature vectors, and γ is a parameter that controls the width of the Gaussian function. The RBF kernel function maps the original feature space to an infinite-dimensional high-dimensional feature space, making the originally linearly inseparable data linearly separable in the high-dimensional space, thereby improving the classification performance. Based on the preprocessed circular saw operation data, a training sample set is constructed. ,in is the feature vector of the i-th sample, is the corresponding safety state label, -1 indicates a dangerous state, +1 indicates a safe state, and N is the total number of training samples. The SVM classification model is trained using the training sample set D to obtain the normal vector w and intercept b of the classification hyperplane, as well as the parameter γ of the RBF kernel function.

[0045] Use the grid search method to optimize the hyperparameters of the SVM classification model, where the hyperparameters include the penalty coefficient C and the parameter γ of the radial basis kernel function RBF; divide the value ranges of C and γ into several discrete intervals to form a hyperparameter grid, such as C∈{0.1, 1, 10, 100}, γ∈{0.01, 0.1, 1, 10}. For each combination (C, γ) in the hyperparameter grid, train the corresponding SVM classification model, and evaluate its classification performance on the validation set, such as using the cross-validation method to calculate evaluation indicators such as average classification accuracy and F1 value. Select the hyperparameter combination with the best classification performance , as the optimal hyperparameter of the SVM classification model, is used for subsequent safety status discrimination.

[0046] According to the preprocessed circular saw operation data, the feature vector x is constructed; the three key feature parameters of saw blade speed, saw blade height and protective cover position are extracted from the preprocessed operation data and recorded as . Combine the three feature parameters into a feature vector x = [ x 1 , x 2 , x 3 ] , as the input of the SVM classification model. In order to eliminate the dimensional influence of different features, the feature vector can be normalized, such as using maximum and minimum value normalization to scale the value range of each feature to the interval [0, 1].

[0047] Through the decision function of the SVM classification model, the distance between the feature vector x and the classification hyperplane is calculated to obtain the safe operation state of the circular saw. The decision function expression of the SVM classification model is: , where x represents the feature vector of the circular saw operation data, x = [ x 1 , x 2 , x 3 ] ,in is the saw blade speed, is the saw blade height, is the position of the protective cover; w represents the normal vector of the SVM classification hyperplane, which represents the direction and slope of the hyperplane and is obtained through SVM model training; Represents the nonlinear transformation of the RBF kernel function on the feature vector x, mapping x to a high-dimensional feature space. , where γ is the parameter of the RBF kernel function, which controls the width of the Gaussian function. is the training sample; b represents the intercept of the SVM classification hyperplane, which represents the distance between the hyperplane and the origin and is obtained through SVM model training; λ represents the safety status correction factor, which is used to adjust the original SVM classification result to make it more consistent with the actual safety status of the circular saw. The value of λ can be set based on experience or obtained through data analysis optimization; It represents the safety state correction function, which adjusts the original SVM classification result according to the characteristics of the circular saw operation data; ,in, are the weight coefficients of the three characteristic parameters, indicating their influence on the safety status, which can be determined based on expert experience or data analysis; The security state correction subfunction for the three characteristic parameters: , , ,in, They are the safety thresholds of saw blade speed, saw blade height, and protective cover position. They are the upper and lower limits of the safety range of the three characteristic parameters respectively.

[0048] like Figure 3 As shown in the figure, according to the safe operation status, a risk assessment method based on the combination of evidence theory and fuzzy set theory is used to conduct risk assessment and obtain the safety risk level; according to the safe operation status of the circular saw, the initial risk level is determined; according to the output results of the SVM classification model, the safe operation status of the circular saw is divided into three levels: safe, warning, and dangerous, corresponding to the three initial risk levels of low risk, medium risk, and high risk. The initial risk level reflects the current safety status of the circular saw and is the basis for subsequent risk assessment. By setting the initial risk level, the real-time safety status of the equipment can be considered in the risk assessment process, which improves the pertinence and timeliness of the assessment results.

[0049] A risk assessment index system including three indicators, namely, accident possibility, severity of accident consequences and saw blade exposure, is established; and the evaluation level and fuzzy membership function are set for each indicator; specifically, accident possibility index: evaluation level: {very small, small, medium, large, very large}; membership function: triangular membership function is adopted; according to historical accident data and expert experience, a quantitative scoring interval of the accident possibility index is set, such as [0, 10]. The scoring interval is divided into five sub-intervals, corresponding to five evaluation levels: very small [0, 2], small (2, 4], medium (4, 6], large (6, 8], and very large (8, 10]. A triangular membership function is defined to describe the membership of the score value to each evaluation level. The parameters of the triangular function include the peak point and the left and right boundary points, which correspond to the score values ​​with membership of 1 and 0, respectively. For example, the triangular function of the "very small" level can be defined as: peak point: 0, membership is 1; right boundary point: 2, membership is 0; function expression: μ(x) = 1-x / 2, x∈[0, 2]; similarly, triangular membership functions of other levels can be defined. The triangular function has a simple shape and high computational efficiency, and is suitable for describing fuzzy sets with clear boundaries. By adjusting the peak point and boundary point, the shape of the membership function can be flexibly controlled to meet different evaluation requirements.

[0050] Accident consequence severity index: Assessment level: {very low, low, medium, high, very high}, Membership function: Using the trapezoidal membership function, according to the qualitative description and quantitative analysis of the accident consequences, set the quantitative scoring interval of the severity index, such as [0, 10]. Divide the scoring interval into five sub-intervals, corresponding to five assessment levels: very low [0, 2], low [2, 4], medium [4, 6], high [6, 8], very high [8, 10]. Define the trapezoidal membership function to describe the membership of the score value to each assessment level. The parameters of the trapezoidal function include the upper and lower bottom boundary points and the left and right boundary points, which correspond to the score values ​​with membership of 1 and 0, respectively. For example, the "medium" level trapezoidal function can be defined as: lower base left boundary point: 4, membership is 0; upper base left boundary point: 5, membership is 1; upper base right boundary point: 6, membership is 1; lower base right boundary point: 7, membership is 0; function expression: μ(x)=0, x<4 or x>7; μ(x)=(x-4), x∈[4,5]; μ(x)=1, x∈[5,6]; μ(x)=(7-x), x∈[6,7]; Similarly, other levels of trapezoidal membership functions can be defined. The trapezoidal function can represent the core area and transition area of ​​the fuzzy set, which is suitable for describing the fuzziness of continuous variables. By adjusting the upper and lower bases and boundary points, the size of the core area and the width of the transition area of ​​the membership function can be controlled to improve the flexibility of the description.

[0051] Saw blade exposure index: Evaluation level: {very weak, weak, medium, strong, very strong}, membership function: Gaussian membership function is used. According to the questionnaire and behavioral observation of saw blade exposure, the quantitative scoring interval of risk awareness index is set, such as [0, 10]. The scoring interval is divided into five sub-intervals, corresponding to five evaluation levels: very weak [0, 2], weak (2, 4], medium (4, 6], strong (6, 8], very strong (8, 10]. Define the Gaussian membership function to describe the membership of the score value to each evaluation level. The parameters of the Gaussian function include mean and variance, which control the center position and width of the function respectively. For example, the Gaussian function of the "strong" level can be defined as: mean: 7, corresponding to the maximum membership value of 1; variance: 0.8, controlling the width of the function; function expression: ; Similarly, other levels of Gaussian membership functions can be defined. The Gaussian function is a bell-shaped curve, with the highest membership at the center and gradually decreasing on both sides, which is suitable for describing the ambiguity of human subjective judgment. By adjusting the mean and variance, the concentration and coverage of the membership function can be controlled, and the ability to describe risk awareness can be improved.

[0052] Using the historical accident data of circular saws, the influence of three indicators on safety risks is analyzed: the historical accident data of circular saws are collected, including the time, location, cause, consequence, personnel involved and other information of the accident. The accident data is statistically analyzed to calculate the frequency and contribution of each indicator in the accident. Accident probability index: the total number of accidents is counted, and the proportion of accidents with different probability levels (such as very small, small, medium, large, and very large) is calculated. Accident consequence severity index: the proportion of accidents with different severity levels (such as very low, low, medium, high, and very high) is counted, and the influence of accident consequences on casualties, equipment damage, production delays, etc. is analyzed. Saw blade exposure index: the proportion of human factors in accidents is counted, and the frequency and consequences of illegal operations, ignoring warnings, and improper use of protective devices caused by weak saw blade exposure are analyzed. According to the frequency and contribution analysis results, the influence of the three indicators on safety risks is ranked, and their importance weights are obtained. The weight can be expressed as a percentage or a fraction, such as {accident probability: 0.5, accident consequence severity: 0.3, saw blade exposure: 0.2. The larger the weight, the greater the impact of the indicator on security risks, and it should be given a higher priority in risk assessment and early warning control.

[0053] The probability assignment function BPA in evidence theory is used to quantify the indicator weight into evidence body: Evidence theory, also known as Dempster-Shafer theory, is a mathematical theory for processing uncertain information. It defines the evidence body (basic probability as sign ment, BPA) and assigns confidence to the power set of events to express the confidence level of different event combinations. The BPA function satisfies the following properties: , the confidence of the empty set is 0. , A is all subsets in the power set, and the sum of the confidences is 1. , the confidence of any subset A is between 0 and 1. According to the indicator weights, a BPA function is constructed to assign weights to single-element subsets of indicators. Assuming the weights of the three indicators are 0.5, 0.3, and 0.2 respectively, the corresponding BPA function can be expressed as: m1({accident probability})=0.5, m2({accident consequence severity})=0.3, m3({saw blade exposure})=0.2, and the single-element subset represents the independent confidence in the indicator, that is, the degree of influence of the indicator on the safety risk is considered to be the corresponding weight value. For other subsets of indicators, if there is no additional evidence to support it, the confidence is 0. For example: m({accident probability, accident consequence severity})=0, m({accident probability, saw blade exposure})=0, m({accident consequence severity, saw blade exposure})=0, m({accident probability, accident consequence severity, saw blade exposure})=0. This application uses the BPA function in evidence theory to convert indicator weights into evidence bodies to express the degree of confidence in different indicator combinations. Evidence theory can flexibly model uncertainty information and quantify the impact of indicator weights on security risks through the allocation and combination of confidence. By assigning confidence to the power set of indicators through the BPA function, it not only expresses the confidence in a single indicator, but also expresses the confidence in a combination of indicators. This method can more comprehensively characterize the uncertainty of indicator weights and reflects the advantages of evidence theory in dealing with incomplete and imprecise information.

[0054] Calculate the conflict degree K between the evidence bodies, where K represents the sum of the incompatible confidences between the evidence bodies. The greater the conflict degree, the greater the contradiction between the evidence bodies, and the lower the confidence of the comprehensive result. Calculate the comprehensive evidence body m(A) of the non-zero subset A, where m(A) is the sum of the products of the normalized confidences of each evidence body on the subset A. m(A) combines the support level of each evidence body and represents the comprehensive confidence in the subset A. Assuming m1({accident probability})=0.5, m2({accident consequence severity})=0.3, m3({saw blade exposure})=0.2, the comprehensive evidence m(A) can be calculated as: K=m1({accident probability})m2({accident consequence severity})m3({saw blade exposure})=0.03; m({accident probability, accident consequence severity})=[m1({accident probability})m2({accident consequence severity})+m1({accident probability})m3({saw blade exposure})+m 2({accident consequence severity})m3({saw blade exposure})] / (1-K)=0.35; m({accident probability, saw blade exposure})=[m1({accident probability})m3({saw blade exposure})] / (1-K)=0.10; m({accident consequence severity, saw blade exposure})=[m2({accident consequence severity})m3({saw blade exposure})] / (1-K)=0.06; The Dempster combination rule can integrate information from multiple sources of evidence to obtain a unified understanding. By calculating the conflict degree and combined confidence, the consistency of the evidence body can be evaluated, and the importance of different indicator combinations can be quantified, making risk assessment more comprehensive and reliable.

[0055] According to the operation data of the circular saw, the fuzzy evaluation vector of each indicator is obtained by matching the membership function: according to the real-time operation parameters of the circular saw (such as current, vibration, etc.), the fuzzy membership function is used to calculate the membership of each parameter value at different evaluation levels. For the i-th indicator, let its evaluation level be , the corresponding membership value is , then the fuzzy evaluation vector of the indicator is For example, assuming that the evaluation level of the accident probability index is {very small, small, medium, large, very large}, and the corresponding membership value is {0.2, 0.5, 0.3, 0.0, 0.0}, then its fuzzy evaluation vector is R1=(0.2, 0.5, 0.3, 0.0, 0.0). The membership function matching converts the operating parameters into fuzzy evaluation vectors, quantifies the possibility of parameter values ​​at different risk levels, and provides fuzzy information representation for comprehensive risk assessment.

[0056] Using the fuzzy weighted average operator, the comprehensive evidence body m(A) and the fuzzy evaluation vector Fusion is performed to obtain the fuzzy membership distribution U(x) of the comprehensive risk assessment level: The fuzzy weighted average operator linearly combines the evidence body and the fuzzy evaluation vector to obtain the membership distribution of the risk level. Suppose the importance of the comprehensive evidence body of the i-th indicator is , the fuzzy evaluation vector is , then the fuzzy membership distribution of the comprehensive risk assessment level is U ( x ) = ∑ i = 1 3 [ m ( A i ) × R i ( x ) ] Membership distribution It combines the importance of indicators and fuzzy evaluation information, reflects the possibility of different risk levels, and provides a basis for the final risk level decision. Fuzzy weighted average realizes the integration of evidence theory and fuzzy set theory, uses evidence to allocate indicator weights, and uses membership function for fuzzy evaluation, so that risk assessment takes into account both subjective and objective factors, and the evaluation results are more reasonable and credible.

[0057] Using the centroid defuzzification method in fuzzy mathematics, the fuzzy membership distribution Converted into a quantitative safety risk level value :The centroid method converts the fuzzy quantity into a definite quantity by calculating the centroid abscissa of the membership distribution, and obtains the quantitative risk level value. The calculation formula of the centroid abscissa x is: , where a and b are the upper and lower bounds of the risk level domain. x reflects the overall position of the comprehensive risk level in the domain. The larger the x, the higher the risk level. The smaller the value, the lower the risk level. The centroid defuzzification method is simple and intuitive. It directly reflects the overall characteristics of the membership distribution through the centroid position, quantifies the fuzzy risk into a certain level value, and facilitates the comparison and ranking of risks.

[0058] The initial risk level and the quantitative risk level value Considered as two sources of evidence: The initial risk level comes from the safety status information of the equipment, such as sensor data, fault records, etc., which reflects the current operating status and potential risks of the equipment. Quantified risk level value It comes from comprehensive risk assessment models, such as the evidence theory-fuzzy assessment model, which comprehensively considers factors such as the possibility of accidents, the severity of consequences, and personnel risk awareness, and gives a quantitative score for the risk level. The two sources of evidence provide a basis for judging the safety risk level of circular saws from the perspectives of equipment status and comprehensive assessment.

[0059] Using Yager rule for evidence fusion: Yager rule is a conflict evidence fusion method in DS evidence theory, which is suitable for situations where there is a high conflict between evidence. Suppose the evidence body corresponding to the initial risk level is , quantify the risk level value The corresponding body of evidence is , the fusion process of Yager rule is as follows: Calculate the conflict coefficient k between the evidence bodies: ,in , for the non-empty subset C, calculate the confidence after fusion : , where A∩B=C, for the entire domain , calculate the confidence after fusion : , the fused confidence m(C) represents the final support level for subset C, Represents the uncertainty of the entire domain. Through the evidence fusion of Yager rule, the initial risk level and the quantitative risk level value are combined. The judgment results are integrated to obtain the final risk level that takes into account both the equipment status and the comprehensive assessment.

[0060] Interpret the fusion results and get the final risk level: Determine the final risk level based on the confidence distribution after fusion. You can select the risk level corresponding to the subset with the highest confidence, or divide the risk level according to the preset threshold based on the size of the confidence. For example, if the subset with the highest confidence after fusion is "high risk", the current safety risk level of the circular saw is judged to be high risk, and emergency measures need to be taken to deal with it. The final risk level comprehensively considers the equipment status information and the comprehensive risk assessment results, and can more comprehensively and objectively reflect the safety risk level of the circular saw, providing a reliable basis for subsequent early warning decisions.

[0061] This application integrates the equipment status information and the comprehensive risk assessment results by fusing the initial risk level and the quantitative risk level value, and obtains the final risk judgment that takes both factors into account, thereby improving the comprehensiveness and accuracy of the risk assessment. The Yager rule distributes the confidence of the conflicting evidence to the entire domain by redistributing the confidence, effectively reducing the impact of the conflict between evidence on the fusion result, making the risk level judgment more robust and reliable. The fused confidence distribution not only gives the final risk level judgment result, but also reflects the degree to which the evidence supports the result, as well as the uncertainty of the entire domain, providing decision makers with richer information.

[0062] like Figure 4 As shown in the figure, according to the safe operation status and safety risk level, the circular saw is safely managed through a multi-level early warning control strategy based on reinforcement learning; the Markov decision process model MDP of circular saw safety early warning control is constructed; according to the safe operation status and safety risk level of the circular saw, the safety status of the circular saw is divided into three states: safe, warning and dangerous, forming a state space ; Use the SVM classification model to classify the real-time operation data of the circular saw and judge the current safe operation status, which is divided into safe and dangerous categories. Use the evidence theory-fuzzy evaluation model to evaluate the safety risk of the circular saw and obtain the risk level, which is divided into four levels: low risk, medium risk, high risk and extremely high risk. Combine the SVM classification results and the risk assessment level to define three safety states: Indicates a safe state. The corresponding SVM classification result is safe and the risk level is low risk. Indicates the warning state. The corresponding SVM classification result is safe and the risk level is medium risk. Indicates a dangerous state, the corresponding SVM classification result is dangerous, and the risk level is high risk or extremely high risk. The state classification comprehensively considers the real-time operating status and potential risk level of the equipment, which can more comprehensively and accurately reflect the safety status of the equipment and provide a basis for early warning control decisions.

[0063] According to the safety status of the circular saw, set the corresponding early warning control action to form an action space ; According to different safety states, four early warning control actions are defined: Indicates maintaining the current state without taking early warning control measures. Applicable to safe state , the equipment is operating normally, the risk level is low, and no intervention is required. Indicates the issuance of warning signals, including sound and light alarms, human-machine interaction prompts, etc. Applicable to warning status , the equipment is operating normally, but the risk level is high and the operator needs to be reminded. Indicates forced speed reduction, limiting the maximum speed of the circular saw through the controller. Applicable to dangerous conditions , the equipment is operating abnormally and the risk level is high. It is necessary to limit equipment performance and reduce the risk of accidents. Indicates emergency braking, which cuts off the power of the circular saw through the controller and applies braking force. Applicable to dangerous conditions , the equipment is running extremely abnormally and the risk level is extremely high. It is necessary to stop the equipment immediately to avoid accidents. The early warning control action takes intervention measures of different intensities according to the severity of the safety status, realizing multi-level early warning control.

[0064] Collect the historical operation data of the circular saw, including information such as device status, control actions, and state transitions. For each state-action pair (s, a), count the frequency of transitions to different states s' after taking action a, and obtain the frequency matrix The frequency matrix reflects the empirical distribution of transitions to different states after taking a specific action in a specific state, and embodies the statistical law of state transition.

[0065] According to the state transition frequency matrix Calculate state transition probability :For each state-action pair (s, a), the frequency matrix Normalize each row of to get the state transition probability . State transition probability It represents the probability of transitioning to state s' after taking action a in state s. It converts the frequency distribution into probability distribution, reflecting the uncertainty of state transition. The state transition probability is one of the core parameters of the MDP model, which determines the dynamic change of the state when the agent interacts with the environment.

[0066] Setting the reward function for early warning control , represents the immediate reward obtained by the circular saw after taking action a in state s; ,

[0067] in, represents a security risk reward, Represents a production efficiency reward, Indicates energy loss reward; is the weight coefficient of the three reward components, satisfying ; ,in, It represents the risk cost under state s, and different cost values ​​can be set according to the risk level; It represents the probability of an accident occurring after taking action a in state s, which can be estimated based on historical data and expert experience;

[0068] The calculation formula is: ,in, It represents the time cost saved after taking action a in state s. Different cost values ​​can be set according to the impact of different actions on production efficiency. It represents the probability of resuming normal production after taking action a in state s, which can be estimated based on the state transition probability; The calculation formula is: ,in, It represents the energy cost consumed after taking action a in state s. Different cost values ​​can be set according to the energy consumption characteristics of different actions.

[0069] The MDP model formalizes the circular saw safety warning control problem into a sequential decision-making problem by defining elements such as state space, action space, state transition probability and reward function. The state space comprehensively considers the equipment operation status and risk level, the action space covers multi-level warning control measures, and the reward function takes into account multiple goals such as safety, production, and energy consumption. It can fully characterize the internal mechanism and decision-making process of circular saw safety warning control.

[0070] The Q-learning algorithm is used to solve the Markov decision process model MDP to obtain the optimal early warning control strategy ; Q-learning is a model-free, heterogeneous temporal difference learning algorithm that gradually approaches the optimal strategy by iteratively updating the state-action value function Q(s, a). Initialize Q(s, a) to an arbitrary value (such as all zeros), set the learning rate α and the discount factor γ. Repeat the following steps until Q(s, a) converges or the maximum number of iterations is reached: Select action a according to the ε-greedy strategy, that is, select the action with the largest Q value with a probability of 1-ε, and randomly explore with a probability of ε. Execute action a and observe the next state s' and the immediate reward r. Update Q(s, a) according to the Q-learning update formula: Q ( s , a ) ← Q ( s , a ) + α [ r + γ max α ' Q ( s ', α ') − Q ( s , α ) ] , update the current state s to , repeat the above steps. After the learning process is completed, for each state s, select the action a with the largest Q value as the optimal strategy π(s): The Q-learning algorithm continuously interacts with the environment, accumulates experience samples of state transfer and reward feedback, dynamically adjusts the state-action value function, and finally learns the optimal early warning control strategy. The algorithm can learn autonomously in an unknown environment, has strong adaptability and robustness, and can cope with random fluctuations in the safety state of the circular saw and make the best decision.

[0071] According to the optimal early warning control strategy , generate corresponding early warning control instructions; according to the optimal strategy Select the corresponding warning control action :The safety state s and risk level of the circular saw are calculated in real time through the safety state discrimination model (such as the SVM classification model) and the risk assessment model (such as the evidence theory-fuzzy assessment model). The real-time state s is input into the optimal early warning control strategy π to obtain the corresponding optimal control action a, that is, The optimal strategy π can be implemented by table lookup or function mapping, and the optimal action a can be directly output according to the state s, avoiding the need to recalculate the Q value and action selection every time, thus improving the real-time nature of decision making. Transformed into specific early warning control instructions, such as warning signals, speed reduction instructions, braking instructions, etc.: Generate corresponding early warning control instructions according to the type of optimal action a. For example, if a is a warning signal, generate an audible and visual alarm instruction; if To force a speed reduction, a speed limit instruction and a target speed value are generated. The early warning control instruction needs to comply with the communication protocol and instruction format of the circular saw control system, such as Modbus, CANopen, etc. The instruction content includes parameters such as instruction type, target value, and execution time. The early warning control instruction is encapsulated into a standardized data packet, such as JSON and XML format, to facilitate transmission and parsing between different systems.

Claims

1. A digital-based safe construction management method, characterized in that: include: Collecting the circular saw operation data at the construction site; the operation data includes saw blade speed, saw blade height and protective cover position; Preprocessing the collected circular saw operation data; According to the pre-processed circular saw operation data, the safe operation state of the circular saw is judged by a support vector machine (SVM) algorithm; the input of the support vector machine (SVM) algorithm is the circular saw operation data, and the output is the safe operation state of the circular saw; According to the safe operation status, a risk assessment method based on the combination of evidence theory and fuzzy set theory is used to conduct risk assessment and obtain the safety risk level; According to the safe operation status and safety risk level, the circular saw is safely managed through a multi-level early warning control strategy based on reinforcement learning.

2. The digital-based safe construction management method according to claim 1 is characterized by: The safe operation status of the circular saw is judged by the support vector machine (SVM) algorithm, including: Select the radial basis kernel function RBF as the kernel function of the SVM algorithm and build the SVM classification model; A grid search method is used to optimize the hyperparameters of the SVM classification model, wherein the hyperparameters include a penalty coefficient C and a parameter γ of a radial basis kernel function RBF; Construct a feature vector x according to the preprocessed circular saw operation data; Through the decision function of the SVM classification model, the distance between the feature vector x and the classification hyperplane is calculated to obtain the safe operation state of the circular saw.

3. The digital-based safe construction management method according to claim 2 is characterized in that: The decision function expression of the SVM classification model is: ; Where x represents the feature vector formed by the circular saw operation data; w represents the normal vector of the SVM classification hyperplane; represents the nonlinear transformation of the RBF kernel function on the feature vector x; b represents the intercept of the SVM classification hyperplane; λ represents the safety state correction factor; Represents the security state correction function.

4. The digital-based safe construction management method according to claim 3 is characterized by: Security status correction function The expression is: ; in, is the weight coefficient of the three characteristic parameters; The security state correction subfunction for the three characteristic parameters: ; ; ; in, They are the safety thresholds of saw blade speed, saw blade height, and protective cover position. They are the upper and lower limits of the safety range of the three characteristic parameters respectively.

5. The digital-based safe construction management method according to any one of claims 2 to 4, characterized in that: The risk assessment method based on the combination of evidence theory and fuzzy set theory is used to conduct risk assessment, including: Determine the initial risk level based on the safe operating status of the circular saw; Establish a risk assessment index system including three indicators: accident possibility, accident consequence severity and saw blade exposure; and set the assessment level and fuzzy membership function for each indicator; According to historical data, the importance weights of the three indicators of accident possibility, accident consequence severity and saw blade exposure are set, and the probability assignment function BPA in evidence theory is used to quantify the importance weights to obtain the evidence of the importance of each indicator. ; According to the body of evidence , using Dempster's combination rule to calculate the comprehensive evidence ; According to the circular saw operation data, the fuzzy evaluation vector of each indicator is obtained by matching the membership function; among them, the evaluation level of the i-th indicator is , the membership value obtained by matching is , then the fuzzy evaluation vector of the corresponding index is ; Using the fuzzy weighted average operator, the comprehensive evidence and the fuzzy evaluation vector Fusion is performed to obtain the fuzzy membership distribution of the comprehensive risk assessment level. : ; in, is the comprehensive evidence of the importance of indicator i, is the fuzzy evaluation vector of index i; Using the centroid defuzzification method in fuzzy mathematics, the fuzzy membership distribution Converted into a quantitative safety risk level value ; According to the initial risk level and security risk level value ,The Yager fusion rule in DS evidence theory is used to obtain the final safety risk level that comprehensively considers the safety status and risk assessment of the circular saw.

6. The digital-based safe construction management method according to claim 5 is characterized by: Set the evaluation level and fuzzy membership function of each indicator, including: The evaluation levels of accident possibility include very small, small, medium, large and very large, and the membership function adopts the triangular membership function; The evaluation levels of the severity of accident consequences include very low, low, medium, high and very high, and the membership function adopts a trapezoidal membership function; The evaluation levels of saw blade exposure include very weak, weak, medium, strong and very strong, and the membership function adopts Gaussian membership function.

7. The digital-based safe construction management method according to claim 5 is characterized by: Using Dempster's combination rule to calculate the comprehensive evidence body ,include: Calculate the conflict degree K between evidence bodies: ; in, They are subsets of the three indicator importance assessment frameworks; Compute the comprehensive body of evidence for the non-zero subset A : ; Among them, A is the non-zero subset of the indicator importance evaluation framework.

8. The digital-based safe construction management method according to claim 5 is characterized by: Fuzzy membership distribution Converted into a quantitative safety risk level value ,include: Calculate fuzzy membership distribution The horizontal coordinate of the center of gravity x*: ; in, is the membership distribution function of comprehensive risk assessment, [a, b] is the domain interval of x, and a and b are the lower and upper bounds of the domain, respectively.

9. The digital-based safe construction management method according to claim 8 is characterized in that: The circular saw is safely managed through a multi-level early warning control strategy based on reinforcement learning, including: Construct the Markov decision process model MDP of circular saw safety warning control; The Q-learning algorithm is used to solve the Markov decision process model MDP to obtain the optimal early warning control strategy ; According to the optimal early warning control strategy , generate corresponding early warning control instructions.

10. The digital-based safe construction management method according to claim 9 is characterized in that: The Markov decision process model MDP of circular saw safety early warning control is constructed, including: According to the safe operation status and safety risk level of the circular saw, the safety status of the circular saw is divided into three states: safe, warning and dangerous, forming a state space ;in, Indicates a safe state; Indicates a warning state; Indicates a dangerous condition; According to the safety status of the circular saw, set the corresponding early warning control action to form an action space ;in, Indicates maintaining the current status and not taking early warning control measures; Indicates a warning signal; Indicates forced speed reduction; Indicates emergency braking; According to the historical operation data, the frequency of the circular saw transferring to other states after taking different early warning control actions in different safety states is counted to obtain the state transfer frequency matrix ; According to the state transition frequency matrix Calculate state transition probability : ; Setting the reward function for early warning control , represents the immediate reward obtained by the circular saw after taking action a in state s; ; in, represents a security risk reward, Represents a production efficiency reward, Indicates energy loss reward; is the weight coefficient of the three reward components; ; in, represents the risk cost under state s; represents the probability of an accident occurring after taking action a in state s; The calculation formula is: ; in, Indicates the time cost saved after taking action a in state s; represents the probability of resuming normal production after taking action a in state s; The calculation formula is: ; in, It represents the energy cost consumed after taking action a in state s.

Images