Construction hoist riding safety supervision system and method based on big data

Abstract

The invention discloses a building hoist riding safety supervision system and method based on big data, and relates to the technical field of hoist safety. According to the invention, various sensors are arranged at the key position points and the periphery of the construction hoist, so that the operation state, personnel information and environmental parameters of the construction hoist are comprehensively monitored; the method comprises the functions of constructor identity authentication, basic anomaly recognition, GRU dynamic safety threshold model construction and fault probability prediction. A face recognition and RFID work card dual verification mechanism is adopted; abnormal behaviors are detected in real time through the basic abnormity recognition model, and operation is stopped and an alarm is given when abnormity is detected; adjusting a safe load and a speed threshold according to the real-time data through a GRU dynamic safe threshold model; predicting the fault probability of the construction hoist through the fault probability function; the safety management level of the construction hoist is improved, the use safety is enhanced, and safety accidents are effectively prevented.

Description

Technical Field

[0001] The present invention relates to the field of elevator safety technology, and more specifically, to a construction elevator riding safety supervision system and method based on big data. Background Art

[0002] With the acceleration of urbanization, the construction industry has an increasing demand for construction hoists. As an indispensable vertical transportation equipment in high-rise building construction, the safe operation of construction hoists is crucial to protecting the lives of construction workers and improving construction efficiency.

[0003] Chinese patent CN110817634B discloses an information-based monitoring and management system for elevators; the present invention realizes intelligent management of construction elevators through a monitoring platform, an electronic tag module, a management device in the car, and an operation control device in the operation room; the operation control device integrates radio frequency identification, operation buttons, main control, monitoring and communication modules, and wirelessly communicates with the electronic tag module set in the operator's safety helmet to obtain personnel information in real time; the system interacts with the monitoring platform data through the network to realize monitoring and early warning of the number and weight of people in the car and the wearing of safety helmets.

[0004] In the existing technology, although the safety of elevator use and the management efficiency have been improved to a certain extent, it has only achieved the monitoring of the number of people in the car, their weight and the wearing of safety helmets; the operating data, personnel information and environmental parameters in the existing technology are scattered, and it is impossible to comprehensively judge the risks; it can only alarm after a fault occurs, and cannot achieve advance warning; the safety threshold is fixed and cannot adapt to changes in complex construction scenarios. Summary of the invention

[0005] 1. Technical issues to be resolved In response to the problems in the related technology, the present invention provides a construction elevator riding safety supervision method based on big data. The present invention combines deep learning with sensor technology to achieve comprehensive risk judgment, provide early warning before an accident occurs, dynamically adjust the safe load threshold and safe speed threshold, and adapt to changes in complex construction scenarios.

[0006] (II) Technical solution In order to solve the above technical problems, the present invention is achieved through the following technical solutions: S1. Deploy sensors at key locations and around the construction elevator; S2. Construction personnel identity authentication; S3. Build a basic anomaly recognition model, collect real-time basic data through sensors, combine the basic anomaly recognition model to detect whether there is basic abnormal behavior, and issue instructions based on the detection results; S4. Construct a GRU dynamic safety threshold model and set parameters; collect historical safety feature data, combine with the optimization algorithm, adjust the parameters of the GRU dynamic safety threshold model, and obtain the final GRU dynamic safety threshold model; collect real-time safety feature data, and combine with the final GRU dynamic safety threshold model to obtain the real-time load threshold and real-time speed threshold; S5. Construct a fault probability function; collect historical construction elevator hardware status data and historical fault data, adjust the parameters of the fault probability function, and obtain the final fault probability function; collect real-time construction elevator hardware facility status data to obtain the predicted fault probability; make decisions based on the predicted fault probability.

[0007] Preferably, the S1 comprises the following steps: S11. Deploy a load sensor at the bottom of the construction elevator car; deploy a speed sensor on the drive motor of the construction elevator; deploy a binocular camera inside and outside the construction elevator car; and deploy infrared array sensors on both sides of the construction elevator door frame; S12, deploy wind speed sensors on the top bracket, deploy temperature and humidity sensors in the control cabinet, deploy laser radar and RFID card readers; The above steps achieve comprehensive monitoring of the operating status of the construction elevator by deploying multiple sensors at key locations of the construction elevator. These sensors can collect data on car load, operating speed, ambient wind speed, temperature and humidity in real time, as well as detect conditions inside and outside the car through binocular cameras and infrared array sensors to ensure personnel safety. LiDAR and RFID readers are used for precise positioning and identification, which overall improves the safety performance, operating efficiency and intelligence level of the construction elevator.

[0008] Preferably, S2 comprises the following steps: S21. Construction workers register their faces and bind their qualification information to their RFID badges. S22. When construction workers enter the construction elevator car, they use the camera sensor to perform face recognition, and the RFID reader to read the RFID work badge to double-verify and confirm their identity and authority. After the authentication is passed, they can enter; The above steps ensure that only certified construction personnel can use the lift, effectively preventing unauthorized personnel from entering, improving the safety management level of the construction site, and facilitating the management and tracking of construction personnel; Preferably, S3 comprises the following steps: S31. Build a basic anomaly recognition model, and deploy YOLO behavior recognition function and door status detection function in the basic anomaly recognition model; S32, using binocular camera sensors and infrared array sensors to collect the number of construction workers, whether they are wearing helmets, whether they are carrying dangerous goods, and whether the door is completely closed, to obtain real-time basic data; S33, input the real-time basic data into the basic abnormality recognition model to obtain the basic detection result. If there is a basic abnormal behavior in the basic detection result, the construction elevator stops running and issues an alarm; otherwise, the current state is maintained; The above steps build a basic anomaly recognition model and integrate the YOLO behavior recognition function and door status detection function, using sensors to collect real-time basic data in real time. Once abnormal behavior is detected, such as not wearing a safety helmet or the door is not completely closed, the construction elevator will immediately stop running and sound an alarm. This effectively improves the safety performance of the construction elevator, can prevent potential safety accidents, ensure the safety of construction workers, and improve the level of intelligent management of the construction site.

[0009] Preferably, S4 comprises the following steps: S41, construct a GRU dynamic safety threshold model, set the initial parameter vector of the initial GRU dynamic safety threshold model to k1, and the learning rate to j; set the training data ratio to b1 and the test data ratio to b2; S42, set the security feature set a={a1,a2,…,a i ,…,a n}, where a i represents the i-th feature that affects the safety of the construction elevator, and n represents the total number of features that affect the safety of the construction elevator; the features that affect the safety of the construction elevator include temperature, humidity, wind speed, and operating parameters of the construction elevator; S43. According to the features in the security feature set, historical security feature data is collected to obtain a historical security feature data matrix A, as follows: ; Among them, A in represents the nth feature data of the i-th record in the historical security feature data, and m represents the total number of historical security feature data collected; Collect the safety load threshold and safety speed threshold corresponding to each historical safety feature data to obtain the safety threshold matrix B, as follows: ; Among them, B i1 and B i2 They respectively represent the safe load threshold and safe speed threshold of the historical safety feature data corresponding to the i-th historical safety feature data; S44, according to b1 and b2, dividing the historical security feature data matrix into a historical security feature data training matrix and a historical security feature data test matrix, and dividing the security threshold matrix into a security threshold training matrix and a security threshold test matrix; S45, training the GRU dynamic security threshold model using the historical security feature data training matrix and the security threshold training matrix, and obtaining a post-security threshold dynamic model after the training is completed; S46, using the historical safety feature data test matrix and the safety threshold test matrix to test the post-safety threshold dynamic model, after the test is completed, the final safety threshold dynamic model is obtained; S47, collecting real-time safety feature data, inputting the real-time safety feature data into the final safety threshold dynamic model, and obtaining the real-time safety load threshold c1 and the real-time safety speed threshold e1; Limit the real-time load and real-time speed to the real-time safe load threshold c1 and the real-time safe speed threshold e1; The above steps construct a GRU dynamic safety threshold model and use historical safety feature data (including temperature, humidity, wind speed and construction elevator operating parameters) for training and testing to dynamically determine the real-time safety load threshold and speed threshold. During the model training process, the initial parameters, learning rate and data division ratio are set to ensure the accuracy and adaptability of the model. The real-time collected safety feature data is input into the model to generate a real-time safety threshold to limit the real-time load and speed of the construction elevator to ensure its operation within a safe range. The safety performance of the construction elevator is significantly improved, and the safety threshold can be dynamically adjusted according to environmental changes and operating conditions, effectively preventing safety hazards such as overloading and speeding, and improving the intelligent and personalized safety management level of the construction elevator.

[0010] Preferably, the S45 comprises the following steps: S451, setting a training error threshold u1 and a maximum number of training iterations; S452, repeatedly training the GRU dynamic security threshold model using the historical security feature data training matrix and the security threshold training matrix, and after each round of training, calculating the training error u2 of the GRU dynamic security threshold model, and adjusting the initial parameter vector k1 according to the training error; S453, when u2≤u1 or the maximum number of iterations is reached, a post-GRU dynamic security threshold model with a parameter vector of k2 is obtained; The above steps set the training error threshold and the maximum number of iterations, and use the historical safety feature data and safety threshold data for repeated training; after each round of training, the model training error is calculated and the parameters are adjusted accordingly until the error reaches the preset threshold or the maximum number of iterations is completed, thereby obtaining the post-GRU dynamic safety threshold model; the accuracy and efficiency of the model training are ensured, so that the model can more accurately and dynamically predict the safety threshold, further improving the safety and reliability of the construction elevator operation.

[0011] Preferably, the S46 comprises the following steps: S461, setting a test accuracy threshold v1, using a historical security feature data test matrix and a security threshold test matrix to test the post-security threshold dynamic model, and obtaining a test accuracy v2 of the post-GRU dynamic security threshold model; S462, if v2 ≥ v1, the post-GRU dynamic security threshold model is used as the final GRU dynamic security threshold model, otherwise, the parameter vector k2 of the post-GRU dynamic security threshold model is further optimized using the optimization algorithm to obtain the final GRU dynamic security threshold model; The above steps evaluate the model by setting a test accuracy threshold and using the historical security feature data test matrix and the security threshold test matrix; if the test accuracy of the model reaches or exceeds the preset threshold, it is determined as the final model; otherwise, the optimization algorithm is used to further optimize the model parameters until the requirements are met, ensuring the high accuracy and reliability of the final GRU dynamic security threshold model.

[0012] Preferably, in S462, the optimization algorithm is used to further optimize the parameter vector k2 of the post-GRU dynamic safety threshold model to obtain the final GRU dynamic safety threshold model, including the following steps: S4621, initializing the first-order moment vector d and the second-order moment vector f, and setting the maximum number of test iterations; S4622. Calculate the gradient of the current parameter vector through back propagation. The calculation formula is as follows: ; Calculate the gradient of the exponential moving average and update the first-order moment vector. The formula is as follows: ; Calculate the exponential moving average of the squared gradient and update the second-order moment vector. The formula is as follows: ; Update the parameter vector, the formula is as follows, ; where r t represents the gradient of the current parameter vector, t represents the time step, t-1 represents the previous time step, and δ represents the loss function; dt represents the current first-order moment vector, f t represents the current second-order moment vector, β1 and β1 represent the first-order decay rate and the second-order decay rate respectively, k t represents the current parameter vector, j represents the learning rate, and ε represents the numerical stability constant; S4623, repeat S4622, when the maximum number of iterations is reached, obtain the optimal parameter vector, use the optimal parameter vector as the parameter vector of the post-GRU dynamic safety threshold model, and obtain the final GRU dynamic safety threshold model; The above steps obtain the optimal parameter vector by initializing the first-order moment vector and the second-order moment vector, calculating the gradient, updating the moment vector and the parameter vector in each iteration until the maximum number of iterations is reached; the exponential moving average of the gradient is used to smooth the gradient fluctuation, thereby improving the stability and convergence speed of the optimization process; the accuracy and robustness of the model prediction are significantly improved, ensuring that the construction elevator can operate more safely and efficiently in a complex and changeable construction environment.

[0013] Preferably, the S5 comprises the following steps: S51. According to the failure probability function for predicting the failure probability of the construction elevator, the formula is as follows: ; Wherein, L represents the fault probability function, g1 and g2 represent the brake wear data and motor current stability data of the construction elevator, respectively, and h1 and h2 represent the weight coefficients of the brake wear data and motor current stability data, respectively; S52, set a prediction period, collect the motor current data and brake wear data of the historical construction elevator, and obtain the historical construction elevator hardware data; collect whether the historical construction elevator has faults within the prediction period, obtain the historical fault data, use the historical fault data to adjust the weight coefficient of the fault probability function, and obtain the final fault probability function. The formula is as follows: ; Wherein, M represents the final fault probability function, p1 and p2 represent the weight coefficients of the brake wear data and the motor current stability data in the final fault probability function respectively; S53, setting a fault probability threshold, collecting real-time construction elevator hardware data, and combining the final fault probability function to obtain a predicted fault probability value of the construction elevator within a prediction period; If the predicted fault probability value is ≥ the fault probability threshold, the construction elevator is inspected and repaired in advance; otherwise, the current state is maintained; The above steps establish a fault probability function and use historical construction elevator hardware data (including brake wear data and motor current stability data) and fault data to train and adjust the weight coefficient of the fault probability function, so as to predict the failure probability of the construction elevator in the future cycle; set a fault probability threshold, if the predicted fault probability exceeds the threshold, the construction elevator is inspected and repaired in advance, otherwise the current state is maintained; effectively predict and prevent potential failures of the construction elevator, reduce unexpected downtime, improve the operating efficiency and reliability of the construction elevator, and ensure construction safety.

[0014] A construction elevator riding safety supervision system based on big data, used to implement the above-mentioned construction elevator riding safety supervision method based on big data, including a data acquisition module, an identity authentication module, a basic abnormal behavior detection module, a safety threshold dynamic adjustment module and a fault prediction module; The data acquisition module is used to deploy sensors and collect real-time data through the sensors; The identity authentication module is used to identify the faces of construction workers through binocular cameras inside and outside the car; read the construction worker's badge information through the RFID card reader, and confirm the identity and authority of the construction worker by combining the face recognition and RFID recognition results; The basic abnormal behavior detection module is used to deploy the YOLO behavior recognition function and the door state detection function to obtain a basic abnormal recognition model; the collected real-time basic data is input into the basic abnormal recognition model, and the presence of abnormal behavior is determined according to the detection results, such as not wearing a safety helmet, carrying dangerous goods, and the door is not completely closed. If abnormal behavior is detected, the construction elevator is stopped and an alarm is issued; The safety threshold dynamic adjustment module is used to construct a GRU dynamic safety threshold model, use historical safety feature data and corresponding safety thresholds, and adjust the model parameters in combination with an optimization algorithm to obtain a final GRU dynamic safety threshold model; input real-time safety feature data into the model, predict real-time safety load thresholds and speed thresholds, and limit the real-time load and speed of the elevator within the predicted safety threshold range; The fault prediction module is used to construct a fault probability function based on brake wear data and motor current stability; use historical fault data to adjust the weight coefficient of the fault probability function; combine real-time construction elevator hardware data and the final fault probability function to predict the fault probability value, and make a decision based on the predicted fault probability value.

[0015] (III) Beneficial effects The present invention has the following beneficial effects: The present invention realizes comprehensive monitoring of the operating status, personnel information and environmental parameters of the construction elevator by deploying multiple sensors at key positions and around the construction elevator, effectively solving the problems of data dispersion and inability to comprehensively judge risks in the prior art, and significantly improving the safety management level of the construction elevator.

[0016] By introducing a dual verification mechanism of face recognition and RFID work badge, the present invention realizes accurate authentication of the identity of construction personnel, ensuring that only personnel with corresponding qualifications can enter the construction elevator, thereby effectively preventing the occurrence of illegal operations and safety accidents, and further enhancing the safety of the use of the construction elevator.

[0017] The present invention constructs a basic abnormality recognition model, which can detect basic abnormal behaviors in the construction elevator in real time, and immediately stop the operation of the construction elevator and issue an alarm when abnormal behaviors are detected, thereby achieving advance warning, avoiding accidents, and greatly improving the operation safety of the construction elevator.

[0018] By constructing a GRU dynamic safety threshold model, the present invention can dynamically adjust the safety load threshold and speed threshold according to the safety feature data collected in real time, adapt to the changes in complex construction scenes, ensure that the construction elevator operates within a safe range, and effectively prevent safety accidents caused by overloading or overspeeding.

[0019] The present invention constructs a failure probability function, collects historical and real-time construction elevator hardware status data, predicts the failure probability of the construction elevator, and performs detection and maintenance in advance, thereby achieving preventive maintenance of the construction elevator and reducing the possibility of failure.

[0020] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0022] Figure 1 It is a flow chart of the construction elevator riding safety supervision method based on big data of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0024] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0025] Embodiment 1 See also Figure 1 The present invention discloses a construction elevator riding safety supervision method based on big data, comprising the following steps: S1. Deploy sensors at key locations and around the construction elevator; The S1 comprises the following steps: S11. Deploy a load sensor at the bottom of the construction elevator car; deploy a speed sensor on the drive motor of the construction elevator; deploy a binocular camera inside and outside the construction elevator car; and deploy infrared array sensors on both sides of the construction elevator door frame; S12, deploy wind speed sensors on the top bracket, deploy temperature and humidity sensors in the control cabinet, deploy laser radar and RFID card readers; S2. Construction personnel identity authentication; The S2 comprises the following steps: S21. Construction workers register their faces and bind their qualification information to their RFID badges. S22. When construction workers enter the construction elevator car, they use the camera sensor to perform face recognition, and the RFID reader to read the RFID work badge to double-verify and confirm their identity and authority. After the authentication is passed, they can enter; S3. Build a basic anomaly recognition model, collect real-time basic data through sensors, combine the basic anomaly recognition model to detect whether there is basic abnormal behavior, and issue instructions based on the detection results; The S3 comprises the following steps: S31. Build a basic anomaly recognition model, and deploy YOLO behavior recognition function and door status detection function in the basic anomaly recognition model; S32, using binocular camera sensors and infrared array sensors to collect the number of construction workers, whether they are wearing helmets, whether they are carrying dangerous goods, and whether the door is completely closed, to obtain real-time basic data; S33, input the real-time basic data into the basic abnormality recognition model to obtain the basic detection result. If there is a basic abnormal behavior in the basic detection result, the construction elevator stops running and issues an alarm; otherwise, the current state is maintained; S4. Construct a GRU dynamic safety threshold model and set parameters; collect historical safety feature data, combine with the optimization algorithm, adjust the parameters of the GRU dynamic safety threshold model, and obtain the final GRU dynamic safety threshold model; collect real-time safety feature data, and combine with the final GRU dynamic safety threshold model to obtain the real-time load threshold and real-time speed threshold; The S4 comprises the following steps: S41, construct a GRU dynamic safety threshold model, set the initial parameter vector of the initial GRU dynamic safety threshold model to k1, and the learning rate to j; set the training data ratio to b1 and the test data ratio to b2; S42, set the security feature set a={a1,a2,…,a i ,…,a n}, where a i represents the i-th feature that affects the safety of the construction elevator, and n represents the total number of features that affect the safety of the construction elevator; the features that affect the safety of the construction elevator include temperature, humidity, wind speed, and operating parameters of the construction elevator; S43. According to the features in the security feature set, historical security feature data is collected to obtain a historical security feature data matrix A, as follows: ; Among them, A in represents the nth feature data of the i-th record in the historical security feature data, and m represents the total number of historical security feature data collected; Collect the safety load threshold and safety speed threshold corresponding to each historical safety feature data to obtain the safety threshold matrix B, as follows: ; Among them, B i1 and B i2 They respectively represent the safe load threshold and safe speed threshold of the historical safety feature data corresponding to the i-th historical safety feature data; S44, according to b1 and b2, dividing the historical security feature data matrix into a historical security feature data training matrix and a historical security feature data test matrix, and dividing the security threshold matrix into a security threshold training matrix and a security threshold test matrix; S45, training the GRU dynamic security threshold model using the historical security feature data training matrix and the security threshold training matrix, and obtaining a post-security threshold dynamic model after the training is completed; The S45 comprises the following steps: S451, setting a training error threshold u1 and a maximum number of training iterations; S452, repeatedly training the GRU dynamic security threshold model using the historical security feature data training matrix and the security threshold training matrix, and after each round of training, calculating the training error u2 of the GRU dynamic security threshold model, and adjusting the initial parameter vector k1 according to the training error; S453, when u2≤u1 or the maximum number of iterations is reached, a post-GRU dynamic security threshold model with a parameter vector of k2 is obtained; S46, using the historical safety feature data test matrix and the safety threshold test matrix to test the post-safety threshold dynamic model, after the test is completed, the final safety threshold dynamic model is obtained; The S46 comprises the following steps: S461, setting a test accuracy threshold v1, using a historical security feature data test matrix and a security threshold test matrix to test the post-security threshold dynamic model, and obtaining a test accuracy v2 of the post-GRU dynamic security threshold model; S462, if v2 ≥ v1, the post-GRU dynamic security threshold model is used as the final GRU dynamic security threshold model, otherwise, the parameter vector k2 of the post-GRU dynamic security threshold model is further optimized using the optimization algorithm to obtain the final GRU dynamic security threshold model; In S462, the parameter vector k2 of the post-GRU dynamic safety threshold model is further optimized using an optimization algorithm to obtain the final GRU dynamic safety threshold model, including the following steps: S4621, initializing the first-order moment vector d and the second-order moment vector f, and setting the maximum number of test iterations; S4622. Calculate the gradient of the current parameter vector through back propagation. The calculation formula is as follows: ; Calculate the gradient of the exponential moving average and update the first-order moment vector. The formula is as follows: ; Calculate the exponential moving average of the squared gradient and update the second-order moment vector. The formula is as follows: ; Update the parameter vector, the formula is as follows, ; where r t represents the gradient of the current parameter vector, t represents the time step, t-1 represents the previous time step, and δ represents the loss function; d t represents the current first-order moment vector, f trepresents the current second-order moment vector, β1 and β1 represent the first-order decay rate and the second-order decay rate respectively, k t represents the current parameter vector, j represents the learning rate, and ε represents the numerical stability constant; S4623, repeat S4622, when the maximum number of iterations is reached, obtain the optimal parameter vector, use the optimal parameter vector as the parameter vector of the post-GRU dynamic safety threshold model, and obtain the final GRU dynamic safety threshold model; S47, collecting real-time safety feature data, inputting the real-time safety feature data into the final safety threshold dynamic model, and obtaining the real-time safety load threshold c1 and the real-time safety speed threshold e1; Limit the real-time load and real-time speed to the real-time safe load threshold c1 and the real-time safe speed threshold e1; S5. Construct a fault probability function; collect historical construction elevator hardware status data and historical fault data, adjust the parameters of the fault probability function, and obtain the final fault probability function; collect real-time construction elevator hardware facility status data to obtain the predicted fault probability; make a decision based on the predicted fault probability; The S5 comprises the following steps: S51. According to the failure probability function for predicting the failure probability of the construction elevator, the formula is as follows: ; Wherein, L represents the fault probability function, g1 and g2 represent the brake wear data and motor current stability data of the construction elevator, respectively, and h1 and h2 represent the weight coefficients of the brake wear data and motor current stability data, respectively; S52, set a prediction period, collect the motor current data and brake wear data of the historical construction elevator, and obtain the historical construction elevator hardware data; collect whether the historical construction elevator has faults within the prediction period, obtain the historical fault data, use the historical fault data to adjust the weight coefficient of the fault probability function, and obtain the final fault probability function. The formula is as follows: ; Wherein, M represents the final fault probability function, p1 and p2 represent the weight coefficients of the brake wear data and the motor current stability data in the final fault probability function respectively; S53, setting a fault probability threshold, collecting real-time construction elevator hardware data, and combining the final fault probability function to obtain a predicted fault probability value of the construction elevator within a prediction period; If the predicted failure probability value is ≥ the failure probability threshold, the construction elevator is inspected and repaired in advance; otherwise, the current state is maintained.

[0026] Embodiment 2 A construction elevator riding safety supervision system based on big data, used to implement the above-mentioned construction elevator riding safety supervision method based on big data, including a data acquisition module, an identity authentication module, a basic abnormal behavior detection module, a safety threshold dynamic adjustment module and a fault prediction module; The data acquisition module is used to deploy sensors and collect real-time data through the sensors; The identity authentication module is used to identify the faces of construction workers through binocular cameras inside and outside the car; read the construction worker's badge information through the RFID card reader, and confirm the identity and authority of the construction worker by combining the face recognition and RFID recognition results; The basic abnormal behavior detection module is used to deploy the YOLO behavior recognition function and the door state detection function to obtain a basic abnormal recognition model; the collected real-time basic data is input into the basic abnormal recognition model, and the presence of abnormal behavior is determined according to the detection results, such as not wearing a safety helmet, carrying dangerous goods, and the door is not completely closed. If abnormal behavior is detected, the construction elevator is stopped and an alarm is issued; The safety threshold dynamic adjustment module is used to construct a GRU dynamic safety threshold model, use historical safety feature data and corresponding safety thresholds, and adjust the model parameters in combination with an optimization algorithm to obtain a final GRU dynamic safety threshold model; input real-time safety feature data into the model, predict real-time safety load thresholds and speed thresholds, and limit the real-time load and speed of the elevator within the predicted safety threshold range; The fault prediction module is used to construct a fault probability function based on brake wear data and motor current stability; use historical fault data to adjust the weight coefficient of the fault probability function; combine real-time construction elevator hardware data and the final fault probability function to predict the fault probability value, and make a decision based on the predicted fault probability value.

[0027] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0028] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A construction elevator riding safety supervision method based on big data, characterized in that: The following steps are involved: S1. Deploy sensors at key locations and around the construction elevator; S2. Construction personnel identity authentication; S3. Build a basic anomaly recognition model, collect real-time basic data through sensors, combine the basic anomaly recognition model to detect whether there is basic abnormal behavior, and issue instructions based on the detection results; S4. Build a GRU dynamic security threshold model and set parameters; Collect historical security feature data, combine with optimization algorithm, adjust the parameters of GRU dynamic security threshold model, and obtain the final GRU dynamic security threshold model; Collect real-time safety feature data and combine it with the final GRU dynamic safety threshold model to obtain the real-time load threshold and real-time speed threshold; S5. Construct a fault probability function; collect historical construction elevator hardware status data and historical fault data, adjust the parameters of the fault probability function, and obtain the final fault probability function; collect real-time construction elevator hardware facility status data to obtain the predicted fault probability; make decisions based on the predicted fault probability.

2. The method for monitoring the safety of construction elevator riding based on big data according to claim 1 is characterized in that: The S1 comprises the following steps: S11. Deploy a load sensor at the bottom of the construction elevator car; deploy a speed sensor on the drive motor of the construction elevator; deploy a binocular camera inside and outside the construction elevator car; and deploy infrared array sensors on both sides of the construction elevator door frame; S12. Deploy a wind speed sensor on the top bracket, a temperature and humidity sensor in the control cabinet, and a lidar and RFID card reader.

3. The method for monitoring the safety of construction elevator riding based on big data according to claim 1 is characterized in that: The S2 comprises the following steps: S21. Construction workers register their faces and bind their qualification information to their RFID badges. S22. When construction workers enter the construction elevator car, facial recognition is performed through the camera sensor, and the RFID identifier reads the RFID work badge to double-verify and confirm their identity and authority. They can enter after passing the authentication.

4. The method for monitoring the riding safety of a construction elevator based on big data according to claim 1 is characterized in that: The S3 comprises the following steps: S31. Build a basic anomaly recognition model, and deploy YOLO behavior recognition function and door status detection function in the basic anomaly recognition model; S32, using binocular camera sensors and infrared array sensors to collect the number of construction workers, whether they are wearing helmets, whether they are carrying dangerous goods, and whether the door is completely closed, to obtain real-time basic data; S33. Input the real-time basic data into the basic abnormality recognition model to obtain the basic detection result. If there is basic abnormal behavior in the basic detection result, the construction elevator stops running and issues an alarm; otherwise, the current state is maintained.

5. The method for monitoring the safety of construction elevator riding based on big data according to claim 1 is characterized in that: The S4 comprises the following steps: S41, construct a GRU dynamic safety threshold model, set the initial parameter vector of the initial GRU dynamic safety threshold model to k1, and the learning rate to j; set the training data ratio to b1 and the test data ratio to b2; S42, set the security feature set a={a1,a2,…,a i ,…,a n }, where a i represents the i-th feature that affects the safety of the construction elevator, and n represents the total number of features that affect the safety of the construction elevator; S43. According to the features in the security feature set, historical security feature data is collected to obtain a historical security feature data matrix A, as follows: ； Among them, A in represents the nth feature data of the i-th record in the historical security feature data, and m represents the total number of historical security feature data collected; Collect the safety load threshold and safety speed threshold corresponding to each historical safety feature data to obtain the safety threshold matrix B, as follows: ； Among them, B i1 and B i2 They respectively represent the safe load threshold and safe speed threshold of the historical safety feature data corresponding to the i-th historical safety feature data; S44, according to b1 and b2, dividing the historical security feature data matrix into a historical security feature data training matrix and a historical security feature data test matrix, and dividing the security threshold matrix into a security threshold training matrix and a security threshold test matrix; S45, training the GRU dynamic security threshold model using the historical security feature data training matrix and the security threshold training matrix, and obtaining a post-security threshold dynamic model after the training is completed; S46, using the historical safety feature data test matrix and the safety threshold test matrix to test the post-safety threshold dynamic model, after the test is completed, the final safety threshold dynamic model is obtained; S47, collecting real-time safety feature data, inputting the real-time safety feature data into the final safety threshold dynamic model, and obtaining the real-time safety load threshold c1 and the real-time safety speed threshold e1; The real-time load and real-time speed are limited to operate within the real-time safe load threshold c1 and the real-time safe speed threshold e1.

6. The method for monitoring the safety of construction elevator riding based on big data according to claim 5 is characterized in that: The S45 comprises the following steps: S451, setting a training error threshold u1 and a maximum number of training iterations; S452, repeatedly training the GRU dynamic security threshold model using the historical security feature data training matrix and the security threshold training matrix, and after each round of training, calculating the training error u2 of the GRU dynamic security threshold model, and adjusting the initial parameter vector k1 according to the training error; S453. When u2≤u1 or the maximum number of iterations is reached, a post-GRU dynamic security threshold model with a parameter vector of k2 is obtained.

7. The method for monitoring the safety of construction elevator riding based on big data according to claim 5 is characterized in that: The S46 comprises the following steps: S461, setting a test accuracy threshold v1, using a historical security feature data test matrix and a security threshold test matrix to test the post-security threshold dynamic model, and obtaining a test accuracy v2 of the post-GRU dynamic security threshold model; S462. If v2≥v1, the post-GRU dynamic safety threshold model is used as the final GRU dynamic safety threshold model. Otherwise, the parameter vector k2 of the post-GRU dynamic safety threshold model is further optimized using the optimization algorithm to obtain the final GRU dynamic safety threshold model.

8. The method for monitoring the safety of construction elevator riding based on big data according to claim 7 is characterized in that: In S462, the parameter vector k2 of the post-GRU dynamic safety threshold model is further optimized using an optimization algorithm to obtain the final GRU dynamic safety threshold model, including the following steps: S4621, initializing the first-order moment vector d and the second-order moment vector f, and setting the maximum number of test iterations; S4622. Calculate the gradient of the current parameter vector through back propagation. The calculation formula is as follows: ； Calculate the gradient of the exponential moving average and update the first-order moment vector. The formula is as follows: ； Calculate the exponential moving average of the squared gradient and update the second-order moment vector. The formula is as follows: ； Update the parameter vector, the formula is as follows, ； where r t represents the gradient of the current parameter vector, t represents the time step, t-1 represents the previous time step, and δ represents the loss function; d t represents the current first-order moment vector, f t represents the current second-order moment vector, β1 and β1 represent the first-order decay rate and the second-order decay rate respectively, k t represents the current parameter vector, j represents the learning rate, and ε represents the numerical stability constant; S4623. Repeat S4622. When the maximum number of iterations is reached, the optimal parameter vector is obtained. The optimal parameter vector is used as the parameter vector of the post-GRU dynamic safety threshold model to obtain the final GRU dynamic safety threshold model.

9. The method for monitoring the riding safety of a construction elevator based on big data according to claim 8, characterized in that: The S5 comprises the following steps: S51. According to the failure probability function for predicting the failure probability of the construction elevator, the formula is as follows: ； Wherein, L represents the fault probability function, g1 and g2 represent the brake wear data and motor current stability data of the construction elevator, respectively, and h1 and h2 represent the weight coefficients of the brake wear data and motor current stability data, respectively; S52, set a prediction period, collect the motor current data and brake wear data of the historical construction elevator, and obtain the historical construction elevator hardware data; collect whether the historical construction elevator has faults within the prediction period, obtain the historical fault data, use the historical fault data to adjust the weight coefficient of the fault probability function, and obtain the final fault probability function. The formula is as follows: ； Wherein, M represents the final fault probability function, p1 and p2 represent the weight coefficients of the brake wear data and the motor current stability data in the final fault probability function respectively; S53, setting a fault probability threshold, collecting real-time construction elevator hardware data, and combining the final fault probability function to obtain a predicted fault probability value of the construction elevator within a prediction period; If the predicted failure probability value is ≥ the failure probability threshold, the construction elevator is inspected and repaired in advance; otherwise, the current state is maintained.

10. A construction elevator riding safety supervision system based on big data, used to implement a construction elevator riding safety supervision method based on big data as described in any one of claims 1 to 9, characterized in that: The system includes a data acquisition module, an identity authentication module, a basic abnormal behavior detection module, a security threshold dynamic adjustment module and a fault prediction module.

Images