A smart helmet management method and related equipment based on the Internet of Things

Abstract

The present invention provides a smart helmet management method and related equipment based on the Internet of Things, the method comprising the following steps: obtaining personnel entry information through an entrance radio frequency identification device at the entrance of a construction site, and generating a list of construction personnel; obtaining the first personnel positioning information through the radio frequency identification device, and updating the list of construction personnel; obtaining the second personnel positioning information of the target construction personnel through a positioning communication base station, and continuously updating the list of construction personnel using the second personnel positioning information; real-time monitoring of the latest position update time of all target construction personnel in the list of construction personnel; if any target construction personnel has not updated the position for a long time, sending a position update voice message to the smart helmet worn by the unknown construction personnel, and generating an emergency management strategy for the unknown construction personnel based on the list of construction personnel. The present invention has the effect of realizing efficient management of large-scale construction sites.

Description

Technical Field

[0001] The present invention belongs to the technical field of construction site safety management, and in particular relates to an intelligent safety helmet management method and related equipment based on the Internet of Things. Background Art

[0002] In recent years, with the acceleration of urbanization and the large-scale development of infrastructure construction, the construction industry has shown a booming development trend. Large-scale construction projects continue to emerge, the scale of construction sites is getting larger and larger, and more and more people are involved. At the same time, society's requirements for building safety are constantly increasing, and various regulations and standards are becoming increasingly stringent. In this case, how to improve construction efficiency while ensuring safety has become a major challenge facing construction companies.

[0003] In traditional construction site management, due to the lack of effective technical means, it is difficult for managers to understand the specific location and status of each worker in a timely and accurate manner. This not only increases safety hazards, but also brings many inconveniences to daily management. The traditional construction site management model has been unable to meet the needs of the development of the modern construction industry. With the continuous expansion of the scale of construction projects and the increasing complexity, the traditional management method relying on manual inspections and paper registration has become inadequate. On the one hand, large construction sites are vast and have a large number of personnel, and it is difficult to achieve comprehensive coverage by manual labor alone; on the other hand, the construction environment is complex and changeable, and potential safety risks are everywhere, requiring more accurate and timely supervision. In addition, the labor-intensive characteristics of the construction industry also determine the complexity and importance of personnel management. However, the rapid development of information technology has made it possible to solve this problem. The emergence of emerging technologies such as the Internet of Things, big data, and artificial intelligence has brought new opportunities for the intelligent transformation of the construction industry. In this context, the development of an intelligent construction site management system based on advanced technology is not only a need for technological innovation, but also an inevitable trend in the development of the industry. Summary of the invention

[0004] The present invention provides an intelligent safety helmet management method and related equipment based on the Internet of Things to solve the problem that traditional manual safety management has been difficult to adapt to large-scale construction sites.

[0005] In a first aspect, the present invention provides a smart helmet management method based on the Internet of Things, which is applied to a smart construction site management system configured at a target construction site, wherein the smart construction site management system comprises three positioning communication base stations, multiple radio frequency identification devices, and smart helmets for construction workers in the target construction site to wear; the three positioning communication base stations are arranged around the regional midpoint of the core construction area in the target construction site, and the multiple radio frequency identification devices are respectively arranged at the edge of the construction area of ​​the target construction site and at the entrance of the construction site, and the device setting positions of all the radio frequency identification devices are located outside the communication range of the positioning communication base stations; the smart helmet has a built-in radio frequency identification chip, a positioning communication module, and a voice broadcast module:

[0006] The method comprises the following steps:

[0007] Acquire the personnel entry information of the target construction personnel entering the target construction site through the entrance radio frequency identification device at the construction site entrance, and generate a construction personnel list according to the personnel entry information, wherein the construction personnel list includes a real-time location update list and a location update time list of all the target construction personnel;

[0008] When the first personnel location information of the target construction personnel is obtained through any one of the radio frequency identification devices, the construction personnel list is updated using the first personnel location information;

[0009] When an activation signal sent by the positioning communication module is obtained through any one of the positioning communication base stations, the second personnel positioning information of the positioning communication module corresponding to the target construction personnel is obtained by using all the positioning communication modules and by using a signal arrival time difference calculation method;

[0010] Continuously updating the construction personnel list using the second personnel location information;

[0011] Real-time monitoring of the latest location update time of all the target construction workers in the construction worker list;

[0012] If the time difference between the latest location update time of any target construction worker and the current system time exceeds the preset first time threshold, the corresponding target construction worker will be marked as an unknown construction worker, and a location update voice message will be sent to the smart safety helmet worn by the unknown construction worker, and an emergency management strategy for the unknown construction worker will be generated based on the construction worker list.

[0013] Optionally, when the first personnel location information of the target construction personnel is acquired through any one of the radio frequency identification devices, updating the construction personnel list by using the first personnel location information comprises the following steps:

[0014] When the first personnel location information of the target construction personnel is obtained by the entrance radio frequency identification device, it is determined whether the first personnel location information obtained by the entrance radio frequency identification device has duplicate information with the construction personnel list;

[0015] If the first personnel location information acquired by the entrance radio frequency identification device has duplicate information with the construction personnel list, deleting the duplicate construction personnel information from the construction personnel list;

[0016] If there is no duplication between the first personnel location information acquired by the entrance radio frequency identification device and the construction personnel list, adding the first personnel location information acquired by the entrance radio frequency identification device to the construction personnel list;

[0017] When the first personnel positioning information of the target construction personnel is acquired by all other edge RFID devices except the entrance RFID device, the construction personnel list is updated according to the first personnel positioning information acquired by the edge RFID device.

[0018] Optionally, the step of using all the positioning communication modules and obtaining the second personnel positioning information of the positioning communication module corresponding to the target construction personnel through a signal arrival time difference calculation method comprises the following steps:

[0019] Acquire the positioning signal of the positioning communication module through each of the positioning communication base stations respectively;

[0020] Based on the signal arrival time difference of the positioning signal arriving at each of the positioning communication base stations, the initial distance difference between the positioning communication module and each of the positioning communication base stations is calculated;

[0021] Using a non-line-of-sight error identification algorithm to identify whether the initial distance difference is in a non-line-of-sight state;

[0022] If the initial distance difference is not in a non-line-of-sight state, taking the initial distance difference as the final distance difference;

[0023] If the initial distance difference is in the non-line-of-sight state, a Kalman filter method is used to correct the non-line-of-sight error of the initial distance difference to obtain the final distance difference;

[0024] The final distance difference of all the positioning communication base stations is combined and calculated through the CHAN positioning algorithm to obtain the second personnel positioning information of the positioning communication module corresponding to the target construction personnel.

[0025] Optionally, the target construction personnel include safety management personnel, and the generating of the emergency management strategy for the unknown construction personnel based on the construction personnel list includes the following steps:

[0026] Extracting the real-time location update list and the location update time list of the unknown construction workers from the construction worker list;

[0027] Combining the real-time location update list and the location update time list of the unknown construction worker, and predicting a first predicted moving path of the unknown construction worker by constructing an autoregressive prediction model;

[0028] Obtaining a first predicted moving destination of the unknown construction worker based on the first predicted moving path and according to the time difference;

[0029] Based on the first predicted moving destination, traverse the real-time location update list of all the safety managers in the construction personnel list except the unknown construction personnel, and select the safety manager whose latest real-time location in the real-time location update list is closest to the first predicted moving destination as the target safety manager;

[0030] Generate a first emergency management strategy for the unknown construction worker, wherein the first emergency management strategy is configured as follows:

[0031] Combined with the basic construction worker information corresponding to the unknown construction worker in the construction worker list, as well as the time difference corresponding to the unknown construction worker and the first predicted moving end point, a first safety alarm voice message is generated, and the first safety alarm voice message is pushed to the smart safety helmet of the target safety manager.

[0032] Optionally, combining the real-time location update list and the location update time list of the unknown construction worker and predicting the first predicted moving path of the unknown construction worker by constructing an autoregressive prediction model comprises the following steps:

[0033] Preprocessing the real-time location update list and the location update time list of the unknown construction worker to obtain a location coordinate time series of the unknown construction worker;

[0034] The augmented Dickey-Fuller test method is used to verify the sequence stationarity of the position coordinate time series;

[0035] If the position coordinate time series is not stationary, performing a differential process on the position coordinate time series, and counting the differential order of the differential process;

[0036] If the position coordinate time series is stationary, the difference order corresponding to the position coordinate time series is set to 0;

[0037] Determining the autoregressive model parameters of the position coordinate time series using the autocorrelation function;

[0038] Combining the autoregressive model parameters and the difference order fitting to construct the autoregressive integrated moving average model of the position coordinate time series;

[0039] Predicting multiple future movement coordinates of the unknown construction worker based on a preset prediction step and through the autoregressive integrated moving average model;

[0040] All of the future movement coordinates are integrated to generate a first predicted movement path of the unknown construction worker.

[0041] Optionally, the method further comprises the following steps:

[0042] If the time difference between the latest location update time of any one of the target construction workers and the current system time exceeds a preset second time threshold, the corresponding target construction worker will be marked as an emergency construction worker, the location update voice information will be sent to the smart safety helmet worn by the emergency construction worker, and an emergency management strategy for the emergency construction worker will be generated based on the construction worker list, and the second time threshold is greater than the first time threshold.

[0043] Optionally, the generating of the emergency management strategy for the emergency construction workers based on the construction workers list comprises the following steps:

[0044] Extracting the real-time location update list and the location update time list of the emergency construction workers from the construction worker list;

[0045] Preprocessing the real-time location update list and the location update time list of the emergency construction personnel to obtain the historical movement path of the emergency construction personnel;

[0046] Acquire a construction site map of the target construction site, and extract map topology features from the construction site map;

[0047] Extracting a movement path data set from a preset information database based on the historical movement path, wherein the information database pre-stores the historical comprehensive movement paths of all construction workers in the target construction site, and the path similarity between all the historical comprehensive movement paths in the movement path data set and the historical movement paths exceeds a preset similarity threshold;

[0048] Building a path precision prediction model based on a convolutional neural network model, and training the path precision prediction model in combination with the map topology features and the mobile path dataset;

[0049] Inputting the historical moving path into the trained path precision prediction model to obtain a second predicted moving path output by the path precision prediction model;

[0050] Based on the second predicted moving path and according to the time difference, a second predicted moving destination of the emergency construction worker is calculated;

[0051] Based on the second predicted moving end point, traverse the real-time location update list of all the safety managers except the emergency construction personnel in the construction personnel list, select multiple safety managers as emergency safety managers, and the distance between the latest real-time location in the real-time location update list of the emergency safety managers and the second predicted moving end point is less than a preset distance threshold;

[0052] Generate a second emergency management strategy for the emergency construction personnel, wherein the second emergency management strategy is configured as follows:

[0053] Combined with the basic information of the emergency construction worker corresponding to the construction worker in the construction worker list, as well as the time difference corresponding to the emergency construction worker and the second predicted moving end point, a second safety alarm voice message is generated, and the second safety alarm voice message is pushed to the smart safety helmet of the emergency safety manager.

[0054] Optionally, the path precision prediction model includes an environmental feature extraction layer for extracting environmental semantic information from the map topological features, a convolutional network layer for extracting spatial dimension features of the moving path, and a long short-term memory network layer for extracting temporal dimension features of the moving path.

[0055] In a second aspect, the present invention also provides an intelligent safety helmet management system based on the Internet of Things, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the intelligent safety helmet management method based on the Internet of Things as described in the first aspect is implemented.

[0056] In a third aspect, the present invention further provides a computer-readable storage medium, in which a computer program is stored. When the computer program is loaded and executed by a processor, the smart helmet management method based on the Internet of Things described in the first aspect is adopted.

[0057] The beneficial effects of the present invention are:

[0058] The present invention realizes the all-round, real-time and accurate tracking of the location of construction personnel, greatly improving the efficiency and accuracy of construction site safety management. By combining radio frequency identification technology and positioning communication base stations, the system can continuously obtain the location information of construction personnel in different areas and scenarios, avoiding the blind spots and lags of traditional manual inspection methods. Secondly, the method establishes a dynamically updated list of construction personnel, including real-time location and update time information, providing managers with a comprehensive and timely dynamic view of personnel, which is conducive to rapid decision-making and resource allocation. Furthermore, the system can automatically identify and mark those personnel who have not updated their location information for a long time, and timely remind them through the voice broadcast function of the smart helmet, greatly reducing safety hazards. This active early warning mechanism is more effective than traditional passive management and can intervene before potential dangers occur. In addition, the method can also automatically generate emergency management strategies based on the location information of construction personnel, improving the ability to respond to emergencies and the efficiency of handling them. From a technical point of view, the method makes full use of advanced technologies such as the Internet of Things, radio frequency identification, and positioning communication, organically combines them, and constructs an intelligent and networked management system, which greatly improves the scientific and technological content and intelligent level of construction site management. This not only improves management efficiency, but also reduces labor costs and reduces human errors. From the perspective of safety management, this method uses technical means to achieve full-process monitoring and timely warning of construction personnel, greatly improving the safety factor of the construction site and effectively reducing the occurrence rate of safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a system topology diagram of a smart construction site management system in one of the implementation modes of the present application.

[0060] Figure 2 This is a flow chart of a smart helmet management method based on the Internet of Things in one of the implementation modes of the present application. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0062] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0063] Reference Figure 1 The present invention discloses an intelligent safety helmet management method based on the Internet of Things, which is applied to a smart construction site management system configured in a target construction site. The smart construction site management system includes three positioning communication base stations, multiple radio frequency identification devices, and intelligent safety helmets for construction workers in the target construction site; the three positioning communication base stations are arranged around the midpoint of the core construction area in the target construction site, and the multiple radio frequency identification devices are respectively arranged at the edge of the construction area of ​​the target construction site and at the entrance of the construction site, and the device setting positions of all radio frequency identification devices are located outside the communication range of the positioning communication base station; the intelligent safety helmet has a built-in radio frequency identification chip, a positioning communication module, and a voice broadcast module.

[0064] Figure 2 FIG. 1 is a flow chart of an intelligent helmet management method based on the Internet of Things in one embodiment. It should be understood that although Figure 2 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 2 At least part of the steps in the above method may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps. Figure 2 As shown, the intelligent helmet management method based on the Internet of Things disclosed in the present invention specifically includes the following steps:

[0065] S101. Acquire the personnel entry information of target construction personnel entering the target construction site through the entrance radio frequency identification device at the construction site entrance, and generate a construction personnel list according to the personnel entry information.

[0066] Among them, when the construction workers wearing smart helmets pass through the entrance, the built-in RFID chip of the helmet will be scanned by the entrance RFID device and its unique identifier will be read. The entrance RFID device transmits the read identifier information together with the current timestamp as the personnel entry information to the central server of the smart construction site management system. After receiving the personnel entry information, the central server will match the unique identifier of the RFID chip with the specific construction personnel information according to the pre-established personnel information database, so as to confirm the identity of the target construction personnel entering the construction site. Subsequently, the system will generate a dynamically updated list of construction personnel based on these personnel entry information. The list contains two key sublists: real-time location update list and location update time list. The real-time location update list records the latest known location coordinates of each target construction worker, while the location update time list records the timestamp of the last update of the location information of each target construction worker. This method can grasp the entry and exit of all construction workers in the construction site and their approximate location distribution in real time, providing basic data support for subsequent safety management and personnel scheduling.

[0067] S102. When the first personnel positioning information of the target construction personnel is obtained through any radio frequency identification device, the construction personnel list is updated using the first personnel positioning information.

[0068] Specifically, multiple RFID devices arranged at the edge and key areas of the construction site will continuously scan the surrounding area. When the RFID chip on the smart helmet worn by the construction worker is detected, the device will read the chip information and record the current time and device location to form the first personnel positioning information. This information will be transmitted to the central server in real time. After receiving the information, the server first determines the corresponding construction worker based on the unique identifier of the RFID chip, and then updates the location information and location update time of the person in the construction worker list. For example, if the construction worker with ID A001 is detected by the RFID device RD-NE01 at the northeast corner of the construction site at 14:30:15, the system will update the location of A001 to the installation location coordinates of RD-NE01 (such as longitude 119.3456, latitude 32.0123), and record the location update time as 14:30:15. This method can quickly update the location information of construction workers when they enter the edge area or key area of ​​the construction site, which helps to monitor the flow and distribution of personnel in real time and improve the management efficiency and safety of the construction site.

[0069] S103. When an activation signal sent by a positioning communication module is obtained through any positioning communication base station, the second personnel positioning information of the target construction personnel corresponding to the positioning communication module is obtained by using all positioning communication modules and a signal arrival time difference calculation method.

[0070] Among them, the specific implementation process is as follows: First, the built-in positioning communication module of the smart helmet will send an activation signal periodically (such as every 30 seconds). When at least three positioning communication base stations receive this activation signal at the same time, each base station will record the exact time of arrival of the signal. Then, the base station transmits the received signal strength and arrival time information to the central processing unit. The central processing unit uses the signal arrival time difference (TDOA) algorithm to calculate the precise position of the signal source (ie, the positioning communication module). The basic principle of the TDOA algorithm is to use the time difference between the arrival of the signal at different base stations to determine the position of the signal source. Assuming there are three base stations A (x1, y1, z1), B (x2, y2, z2) and C (x3, y3, z3), the signal source position is (x, y, z), and the time when the signal arrives at the three base stations is t1, t2, and t3 respectively, then the following set of equations can be established:

[0071] ,

[0072] ,

[0073] Where c is the signal propagation speed. By solving this nonlinear equation set, the three-dimensional coordinates (x, y, z) of the signal source can be obtained.

[0074] S104. Continuously update the construction personnel list using the second personnel location information.

[0075] Specifically, the system will find the corresponding construction worker based on the unique identifier of the positioning communication module, and then replace the old coordinates of the person in the real-time location update list with the newly calculated location coordinates, and record the current system time as the latest location update time in the location update time list. For example, if the system calculates the location of the construction worker with ID B002 as (119.3789, 32.0456, 10.5) through the TDOA algorithm at 15:45:30, the system will immediately update B002's location information to the new coordinates and record the location update time as 15:45:30. This update process is continuous, and as long as the positioning communication module continues to send activation signals, the system will continue to calculate new locations and update the list. In this way, the system can achieve real-time and accurate tracking of the location of construction workers in the core construction area.

[0076] S105. Monitor the latest location update time of all target construction workers in the construction worker list in real time.

[0077] Among them, each entry in the list of construction personnel is traversed regularly (such as every second), the latest location update time of each construction personnel is read, and then this time is compared with the current system time. The comparison method is to calculate the difference between two time points, that is, the current system time minus the latest location update time. This difference reflects how much time has passed since the last location update. For example, assuming that the current system time is 16:00:00, and the latest location update time of a construction personnel ID C003 is 15:58:30, then the calculated time difference is 90 seconds. Through this continuous monitoring, the system can grasp the "freshness" of each construction personnel's location information at any time, providing an important basis for subsequent safety management decisions. This real-time monitoring mechanism can help managers to promptly discover possible abnormal situations. For example, if a construction personnel has not updated the location information for a long time, it may mean that the personnel is in danger or needs help. By promptly discovering these potential problems, the system can greatly improve the safety management level of the construction site and reduce the possibility of accidents.

[0078] S106. If the time difference between the latest location update time of any target construction worker and the current system time exceeds the preset first time threshold, the corresponding target construction worker will be marked as an unknown construction worker, and a location update voice message will be sent to the smart safety helmet worn by the unknown construction worker, and an emergency management strategy for the unknown construction worker will be generated based on the construction worker list.

[0079] Among them, first, the system will mark the construction worker as an "unknown construction worker", indicating that his current status and location are unknown. This mark will be specially displayed in the construction worker list to attract the attention of the management personnel. Secondly, the system will send a location update voice message to the smart helmet worn by the unknown construction worker. This voice message is played through the built-in voice broadcast module of the helmet, and the content can be "Please note that your location information has not been updated for more than 5 minutes. Please move to the signal coverage area or check whether your device is working properly." This voice reminder can help construction workers realize possible communication problems and take corresponding actions. Finally, the system will generate an emergency management strategy for the unknown construction worker based on the construction worker list. This strategy may include the following aspects: 1) Immediately notify the on-site safety manager and provide the last known location and personal information of the unknown construction worker; 2) Start the nearest search program and mobilize other nearby construction workers to assist in the search; 3) If the person is still not found within the scheduled time, the alarm level will be upgraded, and a larger-scale search and rescue operation may need to be launched. Through this multi-level emergency response mechanism, the system can respond quickly when construction workers may be in danger and maximize the safety of construction workers.

[0080] In one embodiment, when the first personnel location information of the target construction personnel is obtained through any radio frequency identification device, updating the construction personnel list using the first personnel location information includes the following steps:

[0081] When the first personnel location information of the target construction personnel is obtained through the entrance radio frequency identification device, it is determined whether the first personnel location information obtained by the entrance radio frequency identification device is repeated with the construction personnel list;

[0082] If the first personnel location information acquired by the entrance radio frequency identification device has duplicate information with the construction personnel list, the construction personnel information with duplicate information is deleted from the construction personnel list;

[0083] If there is no duplication of the first person location information obtained by the entrance radio frequency identification device and the construction personnel list, the first person location information obtained by the entrance radio frequency identification device is added to the construction personnel list;

[0084] When the first personnel positioning information of the target construction personnel is acquired through all other edge RFID devices except the entrance RFID device, the construction personnel list is updated according to the first personnel positioning information acquired by the edge RFID device.

[0085] In this embodiment, when the first personnel location information of the target construction worker is obtained through the entrance radio frequency identification device, it is necessary to immediately perform information duplication judgment. First, the entrance radio frequency identification device captures the unique identifier of the construction worker wearing the smart helmet, and records the current timestamp and device location coordinates, which together constitute the first personnel location information. Subsequently, the system compares this newly acquired information with all entries currently stored in the construction worker list. The key to the comparison is the unique identifier, because each smart helmet has its own specific radio frequency identification chip number. The system traverses each record in the construction worker list to check whether there is a unique identifier that is the same as the newly acquired information. This process can be achieved through efficient database query operations. If the query result is empty, it means that there is no information duplication; if the query returns a record, it means that there is information duplication. This judgment mechanism can effectively prevent the information of the same construction worker from being recorded repeatedly, ensuring the accuracy of the data and the efficiency of the system operation. At the same time, this step also lays the foundation for subsequent information processing, enabling the system to accurately track the entry and exit and position changes of each construction worker.

[0086] If the first person location information obtained by the entrance RFID device is duplicated with the construction personnel list, it means that the construction personnel with duplicate information are out of the site, so the duplicate construction personnel information needs to be deleted from the construction personnel list immediately. After the deletion operation is completed, the system will immediately perform a data integrity check to ensure that the deletion operation does not cause accidental loss or damage to other data. At the same time, the system will record a detailed log of the deletion operation, including the time of deletion, the specific information content deleted, etc., in order to facilitate possible subsequent data recovery needs.

[0087] If there is no duplication between the first person location information obtained by the entrance radio frequency identification device and the list of construction personnel, the system will immediately perform an add operation to add the new information to the list of construction personnel. This process involves the insertion operation of the database, and the integrity and correctness of the newly added data need to be ensured. The specific implementation method is as follows: First, the system will format the first person location information obtained to ensure that it meets the requirements of the database table structure. This usually includes fields such as the unique identifier of the construction personnel (such as RFID number), name, position, entry time, last location update time, current location coordinates, etc. Then, the system will use the database insert statement to perform the add operation. After the insert operation is completed, the system will immediately perform data verification to ensure that the new record has been correctly added to the database. At the same time, the system will update the relevant cache or memory data structure to ensure the efficiency of real-time query and processing. The importance of this addition step is to timely record the information of the construction personnel who have just entered the construction site, and provide basic data for subsequent real-time location tracking and safety management.

[0088] When the first person location information of the target construction worker is obtained through all other edge RFID devices except the entrance RFID device, the system will immediately update the list of construction workers according to the newly obtained information. This process involves the core functions of real-time data update and location tracking. The specific implementation method is as follows: First, the edge RFID device captures the unique identifier of the construction worker's smart helmet and records the current timestamp and device location coordinates. This information will be immediately transmitted to the central processing unit. After receiving the information, the central processing unit will first locate the corresponding record in the list of construction workers according to the unique identifier. Then, the system will update the last location update time and current location coordinate fields in the record. This update operation can be implemented through a database update statement. After the update operation is completed, the system will immediately verify whether the update is successful and refresh the relevant cache or memory data structure. In addition, the system will perform a series of safety checks based on the new location information, such as determining whether the construction worker has entered a prohibited area or a dangerous area. If an abnormal situation is found, the system will immediately trigger the corresponding alarm mechanism. The importance of this update step is to track the movement trajectory of construction workers in the construction site in real time to ensure that managers can keep track of everyone's location at any time. By timely updating location information, the system can provide a more accurate personnel distribution map, help optimize work arrangements and improve construction efficiency. It can also quickly locate and rescue in emergency situations, greatly improving the safety management level of the construction site.

[0089] In one embodiment, using all positioning communication modules and using a signal arrival time difference calculation method to obtain the second personnel positioning information of the positioning communication module corresponding to the target construction personnel includes the following steps:

[0090] Acquire the positioning signal of the positioning communication module through each positioning communication base station respectively;

[0091] Based on the signal arrival time difference of the positioning signal arriving at each positioning communication base station, the initial distance difference between the positioning communication module and each positioning communication base station is calculated;

[0092] A non-line-of-sight error recognition algorithm is used to identify whether the initial distance difference is in a non-line-of-sight state;

[0093] If the initial distance difference is not in the non-line-of-sight state, the initial distance difference is taken as the final distance difference;

[0094] If the initial distance difference is in a non-line-of-sight state, the Kalman filter method is used to correct the non-line-of-sight error of the initial distance difference to obtain the final distance difference;

[0095] The final distance difference of all positioning communication base stations is combined and calculated through the CHAN positioning algorithm to obtain the second personnel positioning information of the target construction personnel corresponding to the positioning communication module.

[0096] In this embodiment, in the smart construction site management system, multiple positioning communication base stations are arranged at different locations of the construction site to form a network covering the entire construction area. Each base station is equipped with a high-precision time synchronization system and a signal receiver. When the positioning communication module worn by the construction worker transmits a signal, these base stations will receive the signal at the same time. The transmission and reception of the signal follow a specific protocol, usually using ultra-wideband (UWB) technology, which has the advantages of high precision, low power consumption, and strong anti-interference ability. The signal contains the unique identifier and timestamp information of the positioning communication module. After receiving the signal, the base station will immediately record the exact time of arrival of the signal with an accuracy of nanoseconds. In this process, each base station will independently receive and process the signal, and then transmit the received information to the central processing unit. For example, assuming that there are three base stations A, B, and C on the construction site, when a construction worker's positioning communication module transmits a signal, the four base stations receive the signal at t1, t2, and t3 respectively. These time information are sent to the central processing unit together with the location coordinates of the base station to provide basic data for subsequent distance calculations.

[0097] Based on the signal arrival time difference of the positioning signal arriving at each positioning communication base station, the initial distance difference between the positioning communication module and each positioning communication base station is calculated. This calculation process utilizes the time difference of arrival (TDOA) technology. First, the central processing unit selects a base station as a reference base station, usually the first base station to receive the signal. Then, the time difference between other base stations and the reference base station is calculated. Assuming that base station A is the reference base station, (t2-t1) and (t3-t1) will be calculated. These time differences are multiplied by the signal propagation speed (close to the speed of light, about 3x10^8m / s) to obtain the initial distance difference. For example, if (t2-t1)=10ns, then the corresponding distance difference is This means that the distance from the positioning communication module to base station B is 3 meters more than the distance to base station A. For each pair of base stations, such a distance difference will be obtained. These distance differences constitute a set of hyperbolic equations, and the actual position of the positioning communication module is at the intersection of these hyperbolas. The calculation formula can be expressed as: , where di,j is the distance difference between base stations i and j, c is the signal propagation speed, tj and ti are the time when the signal arrives at base stations j and i respectively. The advantage of this method is that it does not require knowing the exact time of signal transmission, but only the time difference of each base station receiving the signal, so the requirements for time synchronization are relatively low, which is more suitable for practical applications.

[0098] Using a non-line-of-sight error identification algorithm to identify whether the initial distance difference is in a non-line-of-sight state is a key step in the positioning process, aiming to improve positioning accuracy. The non-line-of-sight (NLOS) state refers to a situation where the signal propagation path is blocked by obstacles and cannot reach the receiver directly. In complex construction environments, NLOS situations often occur, which may lead to serious positioning errors. Non-line-of-sight error identification algorithms are usually based on statistical methods and machine learning techniques. A commonly used method is based on the analysis of signal strength and arrival time. In the line-of-sight (LOS) case, the relationship between signal strength and distance usually conforms to a specific attenuation model, and the arrival time is proportional to the distance. In contrast, signals in NLOS situations usually exhibit greater attenuation and delay. The algorithm calculates the deviation between the actual measurement value and the theoretical model. If the deviation exceeds the preset threshold, it is determined to be an NLOS state. For example, the following simplified formula can be used: , where RSSI is the received signal strength indicator, ToA is the signal arrival time, and α is the weight factor. If NLOS_score exceeds a certain threshold, it is determined to be in NLOS state.

[0099] If the initial distance difference is not in the non-line-of-sight state, the initial distance difference is directly used as the final distance difference. In the line-of-sight (LOS) state, the signal can reach the receiver directly from the transmitter without encountering significant obstacle interference. In this case, the distance difference initially calculated usually has a high accuracy and does not require complex error correction. In specific implementation, the system will first confirm that the distance differences between all relevant base station pairs are judged to be in the LOS state. This usually requires independent LOS / NLOS judgments on the signal propagation paths between each pair of base stations. Only when all paths are confirmed to be LOS can the initial distance difference be used directly. In actual operation, a fault tolerance threshold may be set to allow a small number of paths to have slight NLOS characteristics, and as long as it does not exceed the preset threshold, it is still considered to be in the overall LOS state. The advantage of this method is that it has fast calculation speed and good real-time performance.

[0100] If the initial distance difference is in a non-line-of-sight state, the Kalman filter method is used to correct the non-line-of-sight error of the initial distance difference to obtain the final distance difference. Kalman filtering is a recursive state estimation algorithm that is particularly suitable for processing dynamic systems containing random noise. In an NLOS environment, the signal propagation path is blocked by obstacles, resulting in the measurement value containing additional positive deviations. Kalman filtering can effectively estimate and eliminate this deviation by fusing historical data and current measurements. In specific implementation, it is first necessary to establish a state space model of the system. Assuming that x(k) represents the true distance difference at time k, z(k) represents the measurement value, w(k) and v(k) represent process noise and measurement noise respectively, the system model can be expressed as:

[0101] ,

[0102] Where A is the state transfer matrix, H is the observation matrix, and b(k) is the deviation caused by NLOS. The core of Kalman filtering is the two steps of prediction and update. In the prediction step, the algorithm estimates the current state based on the state of the previous moment:

[0103] ,

[0104] where x̂ represents the estimate, P is the error covariance matrix, and Q is the process noise covariance. In the update step, the algorithm modifies the prediction by incorporating the new measurements:

[0105] ,

[0106] Where K is the Kalman gain and R is the measurement noise covariance. This process is iterative, continuously optimizing the estimated value of the distance difference. A key advantage of the Kalman filter is that it can adaptively adjust the weights of historical data and new measurements, and maintain good performance when the system changes dynamically. In practical applications, it is also necessary to adjust the filter parameters, such as the noise covariance matrix Q and R, according to the specific NLOS environment characteristics. In addition, in order to deal with possible non-Gaussian noise, variants such as extended Kalman filtering or unscented Kalman filtering are sometimes used. The effect of correcting NLOS errors using the Kalman filter method is significant. It can not only effectively reduce positioning errors, but also provide smooth trajectory estimates. In this way, the system can maintain a high positioning accuracy even in complex signal propagation environments, providing reliable data support for site safety management and efficiency optimization.

[0107] Combining the final distance difference of all positioning communication base stations and calculating the second personnel positioning information of the target construction personnel corresponding to the positioning communication module through the Chan positioning algorithm is the last step of the positioning process, and it is also the key link to combine the results of all previous steps to obtain the final positioning result. The Chan algorithm is an efficient closed-form solution algorithm, which is particularly suitable for processing positioning problems based on time difference of arrival (TDOA). It obtains the position of the target through two-step least squares estimation, which can provide high-precision positioning results and ensure computational efficiency. The specific implementation process is as follows: First, assume that there are N base stations, and select one of them as the reference base station (usually the first base station). For each pair of base stations, there is a final distance difference di,1 (i=2,3,...,N). The first step of the Chan algorithm is to establish a linear equation system: Ax=b. Where A is a (N-1)x3 matrix, x is a 3x1 vector containing the target coordinates and an auxiliary variable, and b is a (N-1)x1 vector. The specific form of this equation system is:

[0108] ,

[0109] Where (xi,yi) is the coordinate of the i-th base station, (x,y) is the coordinate of the target, and r is the distance from the target to the reference base station. By solving this set of equations, a preliminary position estimate can be obtained. The second step of the Chan algorithm is to correct this preliminary estimate, taking into account the covariance of the measurement error. This step involves weighted least squares estimation, which can further improve the positioning accuracy. Finally, the coordinates (x,y) of the target output by the algorithm are the coordinate information of the construction personnel corresponding to the second personnel positioning information. The advantage of the Chan algorithm is that it can overcome the nonlinear problem in TDOA measurement to a certain extent, provide performance close to maximum likelihood estimation, and have low computational complexity. In the application of smart construction sites, the Chan algorithm can quickly process distance difference data from multiple base stations and output high-precision positioning results in real time. This is particularly important for scenarios where the positions of a large number of construction personnel need to be updated frequently.

[0110] In one implementation manner, the target construction personnel include safety management personnel, and generating an emergency management strategy for unknown construction personnel based on the construction personnel list includes the following steps:

[0111] Extracting a real-time location update list and a location update time list of unknown construction workers from the construction worker list;

[0112] Combining the real-time location update list and the location update time list of the unknown construction workers, and predicting the first predicted movement path of the unknown construction workers by constructing an autoregressive prediction model;

[0113] Based on the first predicted moving path and according to the time difference, a first predicted moving destination of the unknown construction worker is calculated;

[0114] Based on the first predicted moving end point, traverse the real-time location update list of all other safety managers in the construction personnel list except the unknown construction personnel, and select the safety manager in the real-time location update list whose latest real-time location is closest to the first predicted moving end point as the target safety manager;

[0115] Generate a first emergency management strategy for unknown construction workers. The first emergency management strategy is configured as follows:

[0116] Combined with the basic information of the unknown construction worker corresponding to the construction worker in the construction worker list, as well as the time difference and the first predicted moving end point corresponding to the unknown construction worker, a first safety alarm voice message is generated, and the first safety alarm voice message is pushed to the smart safety helmet of the target safety manager.

[0117] In this embodiment, first, the real-time location update list and location update time list of unknown construction workers are extracted from the construction worker list. This process involves database query and data processing. Assuming that the construction worker list is stored in a relational database, SQL query statements can be used to filter out relevant information of specific construction workers, including x-coordinates, y-coordinates, and update times. After obtaining the data, it is organized into two lists: a location update list [(x1, y1), (x2, y2), ..., (xn, yn)] and a time update list [t1, t2, ..., tn].

[0118] Next, an autoregressive prediction model is constructed to predict the first predicted moving path by combining the real-time location update list and location update time list of the unknown construction workers. The autoregressive model is a time series prediction method that assumes that future values ​​have a linear relationship with past values. Here, a vector autoregression (VAR) model can be used because both x and y coordinates need to be predicted simultaneously. The mathematical expression of the VAR model is: Yt=c+A1Yt-1+A2Yt-2+...+ApYt-p+et, where Yt is the position vector [xt,yt] at time t, c is a constant term, Ai is a coefficient matrix, p is the lag order, and et is the error term. The model training process includes the following steps: 1) Data preprocessing: Convert the location and time data into equally spaced time series, and interpolation may be required. 2) Model order selection: Use information criteria (such as AIC or BIC) to select the optimal lag order p. 3) Parameter estimation: Estimate the model parameters using the least squares method. 4) Model diagnosis: Check the white noise of the residuals and the stability of the model. After training is complete, the model can be used for multi-step prediction to obtain the predicted path for a period of time in the future. For example, assuming the current time is t, and the path for the next 10 minutes needs to be predicted with an interval of 1 minute, 10 predictions are required: , , ..., Here f represents the prediction function of the VAR model. The prediction result will form a series of points [(x't+1, y't+1), (x't+2, y't+2), ..., (x't+10, y't+10)], which is the first predicted moving path.

[0119] Based on the first predicted moving path, the first predicted moving destination of the unknown construction worker is calculated according to the time difference. This step involves time interpolation and path extrapolation. Assuming the time difference is Δt and the current time is t_current, the target time point is t_target=t_current+Δt. Next, you need to find or calculate the position at time t_target on the predicted path. If t_target happens to fall at a certain time point on the predicted path, take that point directly as the moving destination. Otherwise, interpolation or extrapolation calculation is required. The linear interpolation method can be used: Assuming that t_target falls between ti and ti+1 of the predicted path, the coordinates of the end point can be calculated according to the time ratio: , and the same is true for y_target. If t_target exceeds the time range of the predicted path, extrapolation is required. You can use the last two points of the predicted path for linear extrapolation: , and y_target in the same way. Here xn and tn are the coordinates and time of the last point of the predicted path respectively.

[0120] Based on the first predicted moving end point, the real-time location update list of all other safety managers except the unknown construction personnel in the construction personnel list is traversed, and the safety manager whose latest real-time location in the real-time location update list is closest to the first predicted moving end point is selected as the target safety manager. This step involves spatial distance calculation and nearest neighbor search. First, the latest location information of all safety managers needs to be extracted from the database. Next, the distance between the position of each safety manager and the first predicted moving end point is calculated. On a two-dimensional plane, the Euclidean distance formula can be used: distance=sqrt((x2-x1)^2+(y2-y1)^2), where (x1,y1) is the position of the safety manager and (x2,y2) is the predicted moving end point. For large-scale scenarios, considering the influence of the earth's curvature, the Haversine formula may be used to calculate the spherical distance. After calculating all distances, the safety manager with the smallest distance is selected as the target safety manager. If there are multiple safety managers with the same and closest distance, other factors can be considered for selection, such as workload or professional skills. In order to improve efficiency, especially when there are a large number of personnel, spatial indexing technology (such as R-tree) can be used to accelerate the nearest neighbor search. In addition, a maximum distance threshold can be set. If all security managers exceed this distance, a higher level alarm may need to be triggered. The effect of implementing this step is to find the security manager who is most suitable for handling potential security issues and prepare personnel for subsequent emergency response.

[0121] Finally, the first emergency management strategy for unknown construction workers is generated. The strategy is configured as follows: Combine the basic information of the unknown construction workers in the construction worker list, the time difference corresponding to the unknown construction workers, and the first predicted moving end point, generate the first safety alarm voice information, and push the first safety alarm voice information to the smart helmet of the target safety manager. First, it is necessary to extract the basic information of the unknown construction workers from the database, such as name, work number, department, etc. Then, this information is integrated with the previously calculated time difference and predicted moving end point. Next, the alarm text is generated using natural language generation (NLG) technology. You can use predefined templates, such as "Warning: {worker_name} of work number {worker_id} may arrive at location ({x},{y}) in {time_diff} minutes, please pay attention and take necessary safety measures." Here, the {} is a variable, which will be replaced by the actual value. After the text is generated, it is converted into a voice file using text-to-speech (TTS) technology. You can use services such as Google Cloud Text-to-Speech API, or locally deployed open source TTS engines such as Mozilla TTS. When generating voice, it is necessary to consider the speaking speed, volume and clarity to ensure that it can be heard clearly even in a noisy construction environment. The generated voice file needs to be pushed to the smart helmet of the target safety manager via a wireless network. Network latency and reliability need to be considered during the push process, and message confirmation and retransmission mechanisms may need to be implemented. In addition, text messages can also be sent at the same time as a backup. On the smart helmet side, corresponding software is needed to receive and play voice alarms, and considering different environmental noise levels, the playback volume may need to be adjusted dynamically. The effect of implementing this step is to convey potential safety risk information to the most appropriate safety manager in a timely and clear manner, enabling them to respond quickly and take necessary safety measures, thereby effectively preventing possible safety accidents.

[0122] In one embodiment, combining the real-time location update list and the location update time list of the unknown construction workers and predicting the first predicted moving path of the unknown construction workers by constructing an autoregressive prediction model includes the following steps:

[0123] Preprocess the real-time location update list and location update time list of unknown construction workers to obtain the location coordinate time series of unknown construction workers;

[0124] The augmented Dickey-Fuller test method is used to verify the stationarity of the position coordinate time series;

[0125] If the position coordinate time series is not stationary, the position coordinate time series is subjected to difference processing, and the difference order of the difference processing is counted;

[0126] If the position coordinate time series is stationary, the difference order corresponding to the position coordinate time series is set to 0;

[0127] Determine the autoregressive model parameters of the position coordinate time series using the autocorrelation function;

[0128] Combining autoregressive model parameters and difference order fitting, the autoregressive integrated moving average model of the position coordinate time series is constructed;

[0129] Based on the preset prediction step length and through the autoregressive integrated moving average model, multiple future movement coordinates of unknown construction workers are predicted;

[0130] All future movement coordinates are integrated to generate the first predicted movement path of the unknown construction worker.

[0131] In this embodiment, the real-time location update list and location update time list of unknown construction workers are preprocessed to obtain the location coordinate time series of unknown construction workers. The core of this step is to convert the original data into a format suitable for time series analysis. First, the integrity and consistency of the original data need to be checked. The real-time location update list usually contains X and Y coordinates, and the location update time list contains the corresponding timestamps. Any missing or obviously erroneous data points need to be deleted during the inspection process. Next, the timestamp is converted to a unified format, such as a UNIX timestamp or datetime object, for subsequent processing. In order to obtain a regular time series, the data needs to be resampled. For example, if the sampling interval of the original data is uneven, a fixed time interval (such as 1 minute) can be selected for resampling. This can be achieved by linear interpolation or more complex methods (such as cubic spline interpolation). Assuming that there are original data points (t1, x1, y1) and (t2, x2, y2), to interpolate at t', the linear interpolation formula can be used: , and the calculation of y' is similar. Finally, the processed data is organized into a time series format, that is, {(t1,x1,y1),(t2,x2,y2),...,(tn,xn,yn)}, where ti represents time and (xi,yi) represents the corresponding position coordinates. This series should be equally spaced, time-ordered, and have no missing values ​​or outliers.

[0132] Next, the Augmented Dickey-Fuller Test method is used to verify the stationarity of the position coordinate time series. The purpose of this step is to determine whether the time series has a unit root, that is, whether it is a non-stationary series. The Augmented Dickey-Fuller (ADF) test is a widely used unit root test method, which determines the stationarity of the series by testing whether there is a unit root in the autoregressive model of the series. The basic idea of ​​the ADF test is to model the time series as an autoregressive process and then test whether there is a unit root in the model. Specifically, for the time series yt, the ADF test considers the following regression model:

[0133] Δyt=α+βt+γyt-1+δ1Δyt-1+δ2Δyt-2+...+δpΔyt-p+εt

[0134] Among them, Δ represents the difference operation, α is a constant term, βt is the time trend term, γ is the parameter we are interested in, δi is the coefficient of the lagged difference term, and εt is white noise.

[0135] The null hypothesis H0 of the ADF test is that the sequence has a unit root (non-stationary), and the alternative hypothesis H1 is that the sequence does not have a unit root (stationary). The test statistic is the t-statistic of γ, whose distribution does not follow the standard t-distribution, but follows a special DF distribution. For the position coordinate time series, it is necessary to perform ADF tests on the x-coordinate sequence and the y-coordinate sequence respectively. If both sequences are stationary, the entire position coordinate time series can be considered to be stationary. The effect of implementing this step is to determine the stationary characteristics of the position coordinate time series. This is crucial for subsequent model selection and parameter estimation. If the sequence is stationary, the ARMA model can be used directly; if it is non-stationary, it is necessary to perform differential processing or consider using the ARIMA model. Through the ADF test, it is possible to avoid the incorrect application of statistical methods that are only applicable to stationary sequences on non-stationary sequences, thereby improving the accuracy of subsequent predictions.

[0136] If the position coordinate time series is not stationary, the position coordinate time series is subjected to difference processing, and the difference order of the difference processing is counted. Difference is a common method to transform a non-stationary time series into a stationary series. The basic idea of ​​difference is to calculate the difference between adjacent observations in the series, thereby eliminating the trend or periodic components in the series.

[0137] For the time series {yt}, the first-order difference is defined as: Δyt=yt-yt-1

[0138] The second-order difference is defined as: Δ²yt=Δ(Δyt)=(yt-yt-1)-(yt-1-yt-2)=yt-2yt-1+yt-2

[0139] In general, the d-order difference can be expressed as: Δ^dyt=(1-B)^dyt, where B is the lag operator, defined as Byt=yt-1.

[0140] The steps of differential processing are as follows:

[0141] 1. Starting from the first order difference, calculate the first order difference of the series.

[0142] 2. Perform ADF test on the differenced sequence again.

[0143] 3. If the series is still not stationary, perform second-order differences and repeat step 2 above.

[0144] 4. Repeat this process until a stationary series is obtained or the preset maximum difference order is reached.

[0145] For the position coordinate time series, the x-coordinate sequence and the y-coordinate sequence need to be differentiated respectively. The final difference order should take the maximum difference order required for the two sequences. The effect of implementing this step is to convert the non-stationary sequence into a stationary sequence, and obtain the difference order required to make the sequence stationary. Through difference processing, the trend and periodic components in the sequence can be eliminated, making the sequence more suitable for statistical modeling and prediction. At the same time, recording the difference order also provides necessary information for reverse differentiation (i.e., restoring to the original scale) of the subsequent prediction results.

[0146] If the position coordinate time series is stationary, the difference order corresponding to the position coordinate time series is set to 0. This step is a supplement and summary of the previous two steps. If the sequence itself is found to be stationary through the augmented Dickey-Fuller test, then there is no need for difference processing. In this case, the difference order d is set to 0, indicating that the original sequence has met the stationarity requirements and can be directly used for subsequent model construction. Specifically, for the position coordinate time series, the stationarity of the x-coordinate sequence and the y-coordinate sequence needs to be considered separately. If both sequences are stationary, the entire position coordinate time series is considered to be stationary. The significance of setting the difference order to 0 is that it indicates that the difference operation does not need to be included in the subsequent model construction process. In the ARIMA (p, d, q) model, d = 0 means that the model degenerates into the ARMA (p, q) model. This simplifies the model structure, reduces the number of parameters, and generally improves the stability and prediction accuracy of the model.

[0147] Next, the autocorrelation function is used to determine the autoregressive model parameters of the location coordinate time series. The purpose of this step is to determine the order p of the autoregressive (AR) part in the ARIMA model. The autocorrelation function (ACF) and partial autocorrelation function (PACF) are important tools for this purpose.

[0148] The autocorrelation function measures the correlation of a time series with itself at different time lags. For lag k, the autocorrelation coefficient is defined as:

[0149] ,

[0150] Where Cov represents covariance and Var represents variance.

[0151] The partial autocorrelation function measures the direct relationship between a time series and its lagged values, excluding the effects of intermediate lags. The PACF can be obtained by solving the Yule-Walker system of equations. Observe the PACF plot. If the PACF is "truncated" (i.e., becomes insignificant) after lag p, this indicates that the order of the AR model may be p. "Significant" usually means outside the 95% confidence interval. For the ARIMA model, the order q of the moving average (MA) part also needs to be considered. This can be determined by observing the ACF plot. If the ACF is "truncated" after lag q, this indicates that the order of the MA model may be q. It should be noted that for the location coordinate time series, this analysis needs to be performed separately for the x-coordinate series and the y-coordinate series. Different values ​​of p and q may be obtained. In this case, larger values ​​can be selected, or separate models can be built for the x- and y-coordinates. The effect of implementing this step is to determine the initial parameters of the ARIMA model. This provides important guidance for the construction of the model. By analyzing the ACF and PACF, you can understand the autocorrelation structure of the series, which helps to choose the appropriate model order, thereby improving the model's fit and prediction accuracy. At the same time, this process also helps to understand the intrinsic structure and dynamic characteristics of the sequence, providing a basis for subsequent model diagnosis and optimization.

[0152] Combining the autoregressive model parameters and the difference order fitting, the autoregressive integrated moving average model of the location coordinate time series is constructed. This step integrates the previously determined parameters (autoregressive order p, difference order d, moving average order q) to construct the ARIMA (p, d, q) model. The ARIMA model is a widely used time series forecasting model that combines three components: autoregression (AR), difference (I), and moving average (MA).

[0153] The general form of the ARIMA (p, d, q) model can be expressed as:

[0154] (1-φ1B-...-φpB^p)(1-B)^dYt=(1+θ1B+...+θqB^q)εt

[0155] Among them, B is the lag operator, φi is the AR parameter, θi is the MA parameter, and εt is the white noise.

[0156] In the actual fitting process, it is necessary to build ARIMA models for the x-coordinate and the y-coordinate separately. This means that there may be two different ARIMA models, one for predicting the x-coordinate and the other for predicting the y-coordinate.

[0157] During the model fitting process, the following aspects need to be considered:

[0158] Model selection: Although the p and q values ​​have been preliminarily determined by ACF and PACF, you may need to try different parameter combinations. You can use information criteria (such as AIC or BIC) to compare the performance of different models and select the optimal model.

[0159] Residual analysis: After fitting the model, you need to check whether the residuals meet the white noise assumption. This can be done by plotting the ACF plot of the residuals and performing the Ljung-Box test. If the residuals are not white noise, you may need to adjust the model parameters or consider a more complex model.

[0160] Overfitting check: You need to ensure that the model is not overfitting the data. You can use cross-validation or simply split the dataset into training and test sets to evaluate the generalization ability of the model.

[0161] Seasonality considerations: If your time series has a significant seasonal pattern, you may want to consider using a SARIMA (Seasonal ARIMA) model.

[0162] Exogenous variables: If there are other factors that may affect the location (such as weather conditions, time of day, etc.), you can consider using an ARIMAX model to include these factors as exogenous variables in the model.

[0163] The effect of implementing this step is to obtain a mathematical model capable of describing and predicting the time series of location coordinates. This model captures the autocorrelation structure, trend and possible seasonal patterns of the series. With this model, the inherent dynamics of the movement patterns of construction workers can be understood and future location forecasts can be made. In addition, the advantage of the ARIMA model is that it can handle non-stationary time series, transforming them into stationary series through a difference operation. This allows the model to adapt to various types of location data, whether it is data that shows a clear trend or data with greater volatility.

[0164] Next, the multiple future moving coordinates of unknown construction workers are predicted based on the preset prediction step and the autoregressive integrated moving average model. This step uses the ARIMA model constructed previously to predict the location of construction workers in the future. The prediction step refers to the number of future time points you want to predict. The prediction process is based on the recursive nature of the ARIMA model. For the ARIMA (p, d, q) model, the predicted value at time t+h can be expressed as:

[0165] Ŷt+h=E(Yt+h|Yt,Yt-1,...,Y1)

[0166] Where E represents conditional expectation. The specific calculation process depends on the specific parameters of the model and historical data.

[0167] In practice, prediction is usually performed through the following steps:

[0168] Determine the prediction step size: Set the prediction step size h according to actual needs. For example, if you need to predict the position in the next 2 hours, and the data sampling interval is 5 minutes, then h=24.

[0169] Step-by-step prediction: Starting from time t+1, the prediction value is calculated step by step until time t+h. Each step of the prediction is based on the previous prediction results and the actual observation value.

[0170] Calculate prediction intervals: In addition to point forecasts, you can also calculate prediction intervals, typically 95% confidence intervals. This provides an estimate of the uncertainty in the forecast.

[0171] Back-difference: If differencing is performed during the modeling process, the prediction results need to be back-differentiated to obtain the predicted values ​​on the original scale.

[0172] For the position coordinate time series, the x-coordinate and y-coordinate need to be predicted separately. This means that two sets of prediction values ​​will be obtained: {Xt+1, Xt+2, ..., Xt+h} and {Yt+1, Yt+2, ..., Yt+h}.

[0173] In addition, if the SARIMA model (taking seasonality into account) is used, the forecasting process will be slightly more complicated because the seasonal cycle needs to be taken into account. The effect of implementing this step is to obtain a series of predicted future position coordinates. Connect the series of predicted coordinate points obtained previously to form a continuous predicted movement path. This step not only considers the predicted coordinate points, but also the connection between these points, and how to visualize and apply the forecast results to actual scenarios.

[0174] The specific integration process may include the following aspects:

[0175] Coordinate point connection: The simplest method is to directly connect adjacent predicted coordinate points with straight lines. However, this may not accurately reflect the actual movement path, especially when the prediction interval is large. More complex methods may include the use of spline interpolation or other curve fitting techniques to generate a smoother and more natural path.

[0176] Path smoothing: The predicted coordinate points may have some random fluctuations, and connecting them directly may cause the path to look unnatural. You can use moving average or other smoothing techniques to reduce this fluctuation and generate a smoother path.

[0177] Physical constraints consideration: The predicted path should take into account actual physical constraints, such as obstacles like walls, equipment, etc. Map information of the construction site can be integrated into the path generation process to ensure that the predicted path is physically feasible.

[0178] Uncertainty representation: ARIMA models often provide prediction intervals. This uncertainty can be represented on the forecast path, for example using a gradient or transparency to indicate how the confidence in the forecast changes over time.

[0179] Consider multiple possibilities: If the prediction interval is wide, you can consider generating multiple possible paths to reflect different movement scenarios.

[0180] Time dimension representation: In addition to spatial location, you also need to consider how to represent the time dimension. You can mark time points on the path, or use animation to show the movement process over time.

[0181] Integration with the actual environment: Overlay the predicted movement path onto a map or 3D model of the construction site to provide more intuitive understanding and application value.

[0182] In one embodiment, the method further comprises the steps of:

[0183] If the time difference between the latest location update time of any target construction worker and the current system time exceeds the preset second time threshold, the corresponding target construction worker will be marked as an emergency construction worker, and a location update voice message will be sent to the smart safety helmet worn by the emergency construction worker. Based on the construction worker list, an emergency management strategy for the emergency construction worker is generated, and the second time threshold is greater than the first time threshold.

[0184] In this embodiment, generating an emergency management strategy for emergency construction workers based on the construction worker list includes the following steps:

[0185] Extracting a real-time location update list and location update time list of emergency construction workers from the construction worker list;

[0186] Pre-process the real-time location update list and location update time list of the emergency construction personnel to obtain the historical movement path of the emergency construction personnel;

[0187] Obtain a construction site map of the target construction site, and extract map topology features from the construction site map;

[0188] Extracting a movement path data set from a preset information database based on the historical movement paths, wherein the information database pre-stores the historical comprehensive movement paths of all construction workers in the target construction site, and the path similarity between all the historical comprehensive movement paths in the movement path data set and the historical movement paths exceeds a preset similarity threshold;

[0189] Build a precise path prediction model based on the convolutional neural network model, and train the precise path prediction model by combining map topology features and mobile path datasets;

[0190] Inputting the historical moving path into the trained path precision prediction model to obtain a second predicted moving path output by the path precision prediction model;

[0191] Based on the second predicted moving path and according to the time difference, a second predicted moving destination of the emergency construction personnel is calculated;

[0192] Based on the second predicted moving end point, traverse the real-time location update list of all other safety managers in the construction personnel list except the emergency construction personnel, select multiple safety managers as emergency safety managers, and the distance between the latest real-time location in the real-time location update list of the emergency safety managers and the second predicted moving end point is less than a preset distance threshold;

[0193] A second emergency management strategy is generated for emergency construction personnel. The second emergency management strategy is configured as follows:

[0194] Combined with the basic information of the emergency construction personnel in the construction personnel list, the time difference corresponding to the emergency construction personnel and the second predicted moving end point, a second safety alarm voice message is generated, and the second safety alarm voice message is pushed to the smart safety helmet of the emergency safety manager.

[0195] In this embodiment, the path precision prediction model includes an environmental feature extraction layer for extracting environmental semantic information from map topological features, a convolutional network layer for extracting spatial dimension features of the moving path, and a long short-term memory network layer for extracting temporal dimension features of the moving path.

[0196] In this embodiment, when the time difference between the latest location update time of any target construction worker and the current system time exceeds the preset second time threshold, the corresponding target construction worker is marked as an emergency construction worker. After being marked as an emergency construction worker, the system will immediately take two actions: first, send a location update voice message to the smart helmet worn by the emergency construction worker to remind him to update the location information or confirm the safety status; second, generate an emergency management strategy for the emergency construction worker based on the list of construction workers. This hierarchical response mechanism can effectively balance the strictness of safety supervision and the flexibility of operation, avoid frequent triggering of emergency response due to short-term communication interruption or equipment failure, and ensure that timely action can be taken when it is really needed. The transmission of voice information can be transmitted through a wireless network or a dedicated communication channel using the built-in communication module of the smart helmet. The effect of this method is that it can quickly identify potential dangerous situations, remind relevant personnel to pay attention to safety, and provide a basis for subsequent emergency management. At the same time, this mechanism can also help managers quickly locate construction personnel who need special attention and improve the overall efficiency of construction site safety management.

[0197] Next, extract the real-time location update list and location update time list of the emergency construction workers from the list of construction workers. Specifically, once the emergency construction workers are found, two key data lists need to be extracted from their records: the real-time location update list and the location update time list. The real-time location update list contains all the historical location records of the construction worker, each record may be a two-dimensional vector containing latitude and longitude or xy coordinates. The location update time list contains the timestamp corresponding to each location record. These two lists should correspond one to one, that is, each location record has a corresponding timestamp. During the extraction process, some data cleaning and formatting operations may also be required, such as handling possible missing values, outliers, or converting timestamps to a unified format. In addition, considering the needs of subsequent analysis, it may be necessary to perform preliminary sorting of the extracted data, usually in ascending order of timestamps, to facilitate subsequent path reconstruction and analysis.

[0198] The real-time location update list and location update time list of emergency construction workers are preprocessed to obtain the historical movement path of the personnel. This step involves multiple links such as data cleaning, interpolation, and path reconstruction. First, the original data needs to be cleaned to remove possible outliers or erroneous data. This can be achieved by setting reasonable thresholds. For example, if the distance between two consecutive location points exceeds the maximum possible movement speed of the personnel, it may be abnormal data. Secondly, it is necessary to deal with possible missing values ​​in the data. This can be achieved through interpolation methods. Commonly used interpolation methods include linear interpolation, spline interpolation, etc. For example, if linear interpolation is used, for the missing positions between time points t1 and t2, the formula can be used To calculate, where P represents the position and t represents the time. Then, the discrete location points need to be connected into a continuous path. This can be achieved by connecting adjacent location points in chronological order. In this process, it may be necessary to consider the actual conditions of the construction site, such as the location of obstacles such as buildings and equipment, to ensure that the generated path is physically feasible. In addition, the path may need to be smoothed to reduce path irregularities caused by positioning errors or data noise. This can be achieved through moving averages or more complex smoothing algorithms. Finally, the path may need to be resampled to ensure that the points on the path are evenly distributed in time for subsequent analysis and prediction. For example, a fixed time interval (such as 1 minute) can be selected to resample the path. The effect of this preprocessing step is to obtain a clear, continuous, and physically compliant historical movement path that accurately reflects the movement of emergency construction personnel over a period of time.

[0199] Obtaining the site map of the target site and extracting the map topological features are key steps in building an accurate prediction model. First, the site map may exist in many forms, such as CAD drawings, GIS data, or three-dimensional building information models (BIM). These raw map data need to be converted into a format suitable for computer processing, usually a rasterized image or a vectorized data structure. In the rasterized representation, the site is divided into grids, each of which may contain information such as obstacles, passages, and dangerous areas. The vectorized representation may contain geometric elements such as points, lines, and surfaces, which are more suitable for representing complex spatial relationships. Extracting topological features from these maps involves multiple steps. The first is edge detection, which can use algorithms such as Sobel or Canny to identify boundaries and contours in the map. The second is region segmentation, which divides the site into different functional areas, such as construction areas, material storage areas, office areas, etc. This can be achieved through image segmentation algorithms such as the watershed algorithm or more advanced deep learning methods. Then there is path identification, which identifies the main passages and walkable paths within the site. This can be done through skeletonization algorithms or graph-based shortest path algorithms. In addition, it is necessary to identify key points, such as entrances, exits, intersections, etc., which play an important role in subsequent path prediction. Finally, it is necessary to construct a connectivity graph of the entire construction site to represent the connection relationship between different areas. This connectivity graph can be represented by an adjacency matrix or an adjacency list. For example, if it is represented by an adjacency matrix A, then A[i][j]=1 means that area i is connected to area j, otherwise A[i][j]=0.

[0200] The effect of this step is to obtain a detailed and structured representation of the topological features of the construction site, which contains not only static spatial information, but also some dynamic information, such as the degree of congestion and danger level of certain areas. This rich environmental representation provides important contextual information for subsequent path prediction, which can significantly improve the accuracy and practicality of the prediction. For example, the prediction algorithm can use these topological features to avoid predicting impossible paths (such as crossing obstacles) or prioritize more likely paths (such as moving along the main corridor).

[0201] Extracting the mobile path dataset from the preset information database based on the historical mobile path is a key data preparation step. First, the preset information database stores the historical comprehensive mobile paths of all construction workers in the target construction site. This database may be a structured relational database (such as MySQL) or a non-relational database (such as MongoDB), in which each record represents a complete historical mobile path, containing information such as path ID, construction worker ID, timestamp sequence, location coordinate sequence, etc. The extraction process first needs to define the calculation method of path similarity. Commonly used path similarity metrics include Fréchet distance, Dynamic Time Warping (DTW) distance, etc. Taking DTW as an example, it can handle paths of different lengths and changing speeds. The calculation formula is: DTW(A,B)=min(sqrt(sum((ai-bj)^2))+DTW(Ai-1,Bj-1),DTW(Ai-1,Bj),DTW(Ai,Bj-1)), where A and B are two paths, and ai and bj are points on the path. Then, you need to set a similarity threshold, for example, you can set the threshold to 10% of the historical path length. Next, for a given historical mobile path, the system will traverse all paths in the database and calculate their similarity with the given path. This process may be computationally intensive, so you can consider using a spatial index (such as an R-tree) or performing a rough screening first (such as comparing the start and end points of the path) to improve efficiency. For each path, if the calculated similarity exceeds the preset threshold, it will be added to the mobile path dataset.

[0202] This dataset not only contains the path itself, but may also contain some metadata, such as the time when the path occurred, weather conditions, construction stage, etc., which may be helpful for subsequent predictions. During the extraction process, attention should also be paid to the representativeness and diversity of the data. For example, it may be necessary to ensure that the extracted paths cover different time periods (early shift, late shift), different working days (weekdays, weekends), etc. If there are too few samples of certain types of paths, data enhancement techniques may be required, such as making small changes to existing paths to generate new samples. The effect of implementing this step is to obtain a high-quality mobile path dataset that is relevant to the current situation. This dataset not only contains historical paths similar to the current situation, but may also reflect the general movement patterns and specific laws of the construction site. This provides rich training data for subsequent machine learning models, which helps to improve the accuracy and generalization of predictions.

[0203] Building a path prediction model based on a convolutional neural network model is a complex process, involving multiple links such as model design, feature engineering, and model training. First of all, the overall architecture of the path prediction model consists of three main parts: an environmental feature extraction layer, a spatial dimension feature extraction layer, and a temporal dimension feature extraction layer. The environmental feature extraction layer extracts environmental semantic information such as building distribution, road layout, and construction area boundaries from the map topological features. The convolutional network layer is responsible for extracting the spatial dimension features of the mobile path, such as turning frequency, straight segment length, path curvature, etc. The long short-term memory network layer is specifically used to capture the temporal dimension features of the mobile path, such as speed changes, dwell time, and repetitive patterns. Specifically, the environmental feature extraction layer is mainly used to process map topological features. This layer can use multi-layer convolutional neural networks (CNNs) to extract local and global features of the map. For example, a 3x3 or 5x5 convolution kernel can be used to capture local texture information, and then the receptive field is gradually increased through the pooling layer, and finally a global understanding of the entire map is obtained. Specifically, the network structure is as follows:

[0204] Conv2D(32,(3,3),activation='relu')->MaxPooling2D((2,2))->Conv2D(64,(3,3),activation='relu')->MaxPooling2D((2,2))->Conv2D(64,(3,3),activation='relu')

[0205] The spatial dimension feature extraction layer also uses CNN to process the spatial information of the mobile path. The input of this layer is the coordinate sequence on the path, which can be regarded as a two-dimensional image, where one dimension is the serial number of the path point and the other dimension is the coordinate value (x and y). Through the convolution operation, the local shape and global trend of the path can be captured. The time dimension feature extraction layer uses a long short-term memory network (LSTM) to process the time series characteristics of the path. The advantage of LSTM is that it can capture long-term dependencies, which is very important for understanding and predicting complex movement patterns. The input of the LSTM layer is the timestamp and corresponding coordinates of each point on the path. The outputs of these three feature extraction layers are merged (for example, through splicing operations), and then further feature fusion and mapping are performed through several fully connected layers, and finally the predicted next position coordinates are output.

[0206] During the training process, the mobile path dataset needs to be divided into a training set and a validation set. For each path, a sliding window method can be used to generate multiple training samples, each of which contains a historical path and the corresponding next position. The loss function of the model can be selected as mean square error (MSE) or mean absolute error (MAE). During training, optimizers such as Adam can be used, and batch gradient descent can be used to update model parameters. In order to prevent overfitting, a variety of regularization techniques can be used, such as L2 regularization, Dropout, etc. In addition, data enhancement techniques such as adding noise, path rotation and scaling can be used to increase the diversity of training data. The implementation effect of this step is to obtain a path prediction model that can comprehensively consider environmental characteristics, historical movement patterns, and time series information. This model can not only capture complex spatial-temporal dependencies, but also adapt to different environmental conditions, thereby providing high-precision path prediction results.

[0207] The trained deep learning model is used to infer the new input data to obtain the possible future movement path. In the specific implementation, the historical movement path data needs to be converted into an input format acceptable to the model, which usually involves data standardization and reshape operations. Then, the processed data is input into the model for forward propagation calculation. For example, assuming that the historical movement path is a coordinate sequence of the last 10 time points [(x1, y1), (x2, y2), ..., (x10, y10)], it needs to be converted into a numpy array with a shape of (1, 10, 2). The output of the model can be the predicted coordinates of several future time points, such as [(x11', y11'), (x12', y12'), ..., (x20', y20')], which is the so-called second predicted movement path. In order to improve the reliability of the prediction, an integrated learning method can be used, such as using multiple models for prediction and taking the average, or using Monte Carlo dropout technology for multiple sampling predictions. The effect of implementing this step is to obtain a high-precision prediction path based on historical data and environmental characteristics. This path not only takes into account the movement habits of construction workers, but also integrates the influence of the construction site environment. Therefore, it can more accurately reflect the possible future movement trajectory of emergency construction workers and provide key decision-making basis for subsequent safety management measures.

[0208] Next, the second predicted moving destination of the emergency construction personnel is calculated based on the second predicted moving path and according to the time difference. The implementation principle of this step is to determine the possible location of the construction personnel at a specific time point by extrapolating the predicted path and combining the time information. In the specific implementation, it is first necessary to determine the time difference, that is, the time interval Δt from the last position update to the current time. Then, based on the speed and direction information of the second predicted moving path, calculate the possible distance moved within this time interval. The calculation method can use simple linear extrapolation or more complex curve fitting methods. For example, assuming that the last two points of the second predicted moving path are (x_n-1, y_n-1) and (x_n, y_n), and the time interval is t, then the average speed v=sqrt((x_n-x_n-1)^2+(y_n-y_n-1)^2) / t can be calculated. Then, the second predicted moving destination can be estimated as: , where θ is the moving direction angle. The effect of this step is to obtain a specific spatial coordinate point, which represents the most likely location of the emergency construction personnel at the current moment. This predicted endpoint not only takes into account historical movement patterns and environmental factors, but also combines real-time time information, so it can provide safety managers with more accurate search targets, greatly improving the efficiency and pertinence of emergency response.

[0209] Then, based on the second predicted moving end point, the real-time location update list of all other safety managers in the construction personnel list except the emergency construction personnel is traversed, and multiple safety managers are selected as emergency safety managers. The distance between the latest real-time position in the real-time location update list of the emergency safety manager and the second predicted moving end point is less than the preset distance threshold. The purpose of this step is to find the safety manager who is most suitable for handling emergency situations. The specific implementation process is as follows: First, obtain the real-time location update list of all safety managers. This list usually contains the ID of each safety manager and their latest location coordinates. Then, for each safety manager, calculate the distance between its latest position and the second predicted moving end point of the emergency construction personnel. The calculated distance is compared with the preset distance threshold. This threshold may be determined according to the specific situation of the construction site, for example, it may be set to 100 meters. If the calculated distance is less than this threshold, the safety manager is added to the list of emergency safety managers. This process will traverse all safety managers and finally obtain a list of emergency safety managers that meets the distance requirements. This method can ensure that the selected safety manager can quickly reach the predicted location of the emergency construction personnel, thereby improving the efficiency of emergency response.

[0210] Finally, a second emergency management strategy for emergency construction workers is generated. The second emergency management strategy is configured as follows: combining the basic information of the emergency construction workers in the construction worker list, the time difference corresponding to the emergency construction workers, and the second predicted moving end point, generates a second safety alarm voice message, and pushes the second safety alarm voice message to the smart helmet of the emergency safety manager. The implementation of this step involves multiple aspects: First, extract the basic information of the emergency construction workers from the construction worker list, which may include name, work number, responsibilities, special skills or precautions, etc. Then, combine the previously calculated time difference and the second predicted moving end point to generate a detailed safety alarm message. This message may include the following: the identity information of the emergency construction workers, the expected arrival location, the expected arrival time, the potential dangers that may be faced (inferred based on the environmental information of the predicted location), and the recommended safety measures. Next, convert this text message into a voice message. This process may use text-to-speech (TTS) technology to convert the text message into a clear and understandable voice message. Finally, push this voice message to the selected smart helmet of the emergency safety manager through wireless communication technology (such as Wi-Fi, 4G or 5G network). The smart helmet is equipped with a voice broadcast module, which can play the received voice information in time.

[0211] The present invention also discloses an intelligent safety helmet management system based on the Internet of Things, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the intelligent safety helmet management method based on the Internet of Things as described in any one of the above embodiments is implemented.

[0212] Among them, the processor can adopt a central processing unit (CPU). Of course, according to actual usage, other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. can also be adopted. The general-purpose processor can adopt a microprocessor or any conventional processor, etc., and this application does not impose any restrictions on this.

[0213] Among them, the memory can be an internal storage unit of a computer device, such as a hard disk or memory of a computer device, or an external storage device of a computer device, such as a plug-in hard disk, a smart memory card (SMC), a secure digital card (SD) or a flash memory card (FC) equipped on the computer device, and the memory can also be a combination of an internal storage unit and an external storage device of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or is to be output, and this application does not impose any restrictions on this.

[0214] The embodiment of the present application also discloses a computer-readable storage medium, and the computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the smart helmet management method based on the Internet of Things in the above embodiment is adopted.

[0215] Among them, the computer program can be stored in a machine-readable medium, the computer program includes computer program code, the computer program code can be in the form of source code, object code, executable file or certain middleware, etc. The machine-readable medium includes any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the machine-readable medium includes but is not limited to the above-mentioned components.

[0216] Among them, through this computer-readable storage medium, the comprehensive fault detection method for transmission lines in the above embodiment is stored in a computer-readable storage medium, and is loaded and executed on a processor to facilitate the storage and application of the above method.

[0217] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0218] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of protection of the present application is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present application as above, which are not provided in detail for the sake of simplicity.

[0219] One or more embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the present application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of one or more embodiments of the present application should be included in the protection scope of the present application.

Claims

1. A smart helmet management method based on the Internet of Things, characterized in that: A smart construction site management system is applied to a target construction site, the smart construction site management system comprising three positioning communication base stations, multiple radio frequency identification devices, and smart safety helmets worn by construction workers at the target construction site; The three positioning communication base stations are arranged around the midpoint of the core construction area of ​​the target construction site, and the multiple radio frequency identification devices are respectively arranged at the edge of the construction area of ​​the target construction site and at the entrance of the construction site, and the device setting positions of all the radio frequency identification devices are located outside the communication range of the positioning communication base station; the smart safety helmet is equipped with a radio frequency identification chip, a positioning communication module and a voice broadcast module; The method comprises the following steps: Acquire the personnel entry information of the target construction personnel entering the target construction site through the entrance radio frequency identification device at the construction site entrance, and generate a construction personnel list according to the personnel entry information, the construction personnel list including the real-time location update list and location update time list of all the target construction personnel, the target construction personnel including the safety management personnel; When the first personnel location information of the target construction personnel is obtained through any one of the radio frequency identification devices, the construction personnel list is updated using the first personnel location information; When an activation signal sent by the positioning communication module is obtained through any one of the positioning communication base stations, the second personnel positioning information of the positioning communication module corresponding to the target construction personnel is obtained by using all the positioning communication modules and by using a signal arrival time difference calculation method; Continuously updating the construction personnel list using the second personnel location information; Real-time monitoring of the latest location update time of all the target construction workers in the construction worker list; If the time difference between the latest location update time of any target construction worker and the current system time exceeds the preset first time threshold, the corresponding target construction worker is marked as an unknown construction worker, a location update voice message is sent to the smart safety helmet worn by the unknown construction worker, and an emergency management strategy for the unknown construction worker is generated based on the construction worker list; The generating of the emergency management strategy for the unknown construction workers based on the construction workers list comprises the following steps: Extracting the real-time location update list and the location update time list of the unknown construction workers from the construction worker list; Combining the real-time location update list and the location update time list of the unknown construction worker, and predicting a first predicted moving path of the unknown construction worker by constructing an autoregressive prediction model; Obtaining a first predicted moving destination of the unknown construction worker based on the first predicted moving path and according to the time difference; Based on the first predicted moving destination, traverse the real-time location update list of all the safety managers in the construction personnel list except the unknown construction personnel, and select the safety manager whose latest real-time location in the real-time location update list is closest to the first predicted moving destination as the target safety manager; Generate a first emergency management strategy for the unknown construction worker, wherein the first emergency management strategy is configured as follows: Combined with the basic construction worker information corresponding to the unknown construction worker in the construction worker list, as well as the time difference corresponding to the unknown construction worker and the first predicted moving end point, a first safety alarm voice message is generated, and the first safety alarm voice message is pushed to the smart safety helmet of the target safety manager.

2. The method for managing smart helmets based on the Internet of Things according to claim 1, characterized in that: When the first personnel location information of the target construction personnel is obtained through any one of the radio frequency identification devices, updating the construction personnel list using the first personnel location information includes the following steps: When the first personnel location information of the target construction personnel is obtained by the entrance radio frequency identification device, it is determined whether the first personnel location information obtained by the entrance radio frequency identification device has duplicate information with the construction personnel list; If the first personnel location information acquired by the entrance radio frequency identification device has duplicate information with the construction personnel list, deleting the duplicate construction personnel information from the construction personnel list; If there is no duplication between the first personnel location information acquired by the entrance radio frequency identification device and the construction personnel list, adding the first personnel location information acquired by the entrance radio frequency identification device to the construction personnel list; When the first personnel positioning information of the target construction personnel is acquired by all other edge RFID devices except the entrance RFID device, the construction personnel list is updated according to the first personnel positioning information acquired by the edge RFID device.

3. The method for managing smart helmets based on the Internet of Things according to claim 1, characterized in that: The method of using all the positioning communication modules and using a signal arrival time difference calculation method to obtain the second personnel positioning information of the positioning communication module corresponding to the target construction personnel comprises the following steps: Acquire the positioning signal of the positioning communication module through each of the positioning communication base stations respectively; Based on the signal arrival time difference of the positioning signal arriving at each of the positioning communication base stations, the initial distance difference between the positioning communication module and each of the positioning communication base stations is calculated; Using a non-line-of-sight error identification algorithm to identify whether the initial distance difference is in a non-line-of-sight state; If the initial distance difference is not in a non-line-of-sight state, taking the initial distance difference as the final distance difference; If the initial distance difference is in the non-line-of-sight state, a Kalman filter method is used to correct the non-line-of-sight error of the initial distance difference to obtain the final distance difference; The final distance difference of all the positioning communication base stations is combined and calculated through the CHAN positioning algorithm to obtain the second personnel positioning information of the positioning communication module corresponding to the target construction personnel.

4. The method for managing smart helmets based on the Internet of Things according to claim 1, characterized in that: The step of combining the real-time location update list and the location update time list of the unknown construction worker and predicting the first predicted moving path of the unknown construction worker by constructing an autoregressive prediction model comprises the following steps: Preprocessing the real-time location update list and the location update time list of the unknown construction worker to obtain a location coordinate time series of the unknown construction worker; The augmented Dickey-Fuller test method is used to verify the sequence stationarity of the position coordinate time series; If the position coordinate time series is not stationary, performing a differential process on the position coordinate time series, and counting the differential order of the differential process; If the position coordinate time series is stationary, the difference order corresponding to the position coordinate time series is set to 0; Determining the autoregressive model parameters of the position coordinate time series using the autocorrelation function; Combining the autoregressive model parameters and the difference order fitting to construct the autoregressive integrated moving average model of the position coordinate time series; Predicting multiple future movement coordinates of the unknown construction worker based on a preset prediction step and through the autoregressive integrated moving average model; All of the future movement coordinates are integrated to generate a first predicted movement path of the unknown construction worker.

5. The method for managing smart helmets based on the Internet of Things according to claim 1, characterized in that: The method further comprises the steps of: If the time difference between the latest location update time of any one of the target construction workers and the current system time exceeds a preset second time threshold, the corresponding target construction worker will be marked as an emergency construction worker, the location update voice information will be sent to the smart safety helmet worn by the emergency construction worker, and an emergency management strategy for the emergency construction worker will be generated based on the construction worker list, and the second time threshold is greater than the first time threshold.

6. The method for managing smart helmets based on the Internet of Things according to claim 5, characterized in that: The generating of the emergency management strategy for the emergency construction personnel based on the construction personnel list comprises the following steps: Extracting the real-time location update list and the location update time list of the emergency construction workers from the construction worker list; Preprocessing the real-time location update list and the location update time list of the emergency construction personnel to obtain the historical movement path of the emergency construction personnel; Acquire a construction site map of the target construction site, and extract map topology features from the construction site map; Extracting a movement path data set from a preset information database based on the historical movement path, wherein the information database pre-stores the historical comprehensive movement paths of all construction workers in the target construction site, and the path similarity between all the historical comprehensive movement paths in the movement path data set and the historical movement paths exceeds a preset similarity threshold; Building a path precision prediction model based on a convolutional neural network model, and training the path precision prediction model in combination with the map topology features and the mobile path dataset; Inputting the historical moving path into the trained path precision prediction model to obtain a second predicted moving path output by the path precision prediction model; Based on the second predicted moving path and according to the time difference, a second predicted moving destination of the emergency construction worker is calculated; Based on the second predicted moving end point, traverse the real-time location update list of all the safety managers except the emergency construction personnel in the construction personnel list, select multiple safety managers as emergency safety managers, and the distance between the latest real-time location in the real-time location update list of the emergency safety managers and the second predicted moving end point is less than a preset distance threshold; Generate a second emergency management strategy for the emergency construction personnel, wherein the second emergency management strategy is configured as follows: Combined with the basic information of the emergency construction worker corresponding to the construction worker in the construction worker list, as well as the time difference corresponding to the emergency construction worker and the second predicted moving end point, a second safety alarm voice message is generated, and the second safety alarm voice message is pushed to the smart safety helmet of the emergency safety manager.

7. The method for managing smart helmets based on the Internet of Things according to claim 6, characterized in that: The path precision prediction model includes an environmental feature extraction layer for extracting environmental semantic information from the map topological features, a convolutional network layer for extracting spatial dimension features of the moving path, and a long short-term memory network layer for extracting temporal dimension features of the moving path.

8. An intelligent helmet management system based on the Internet of Things, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the smart helmet management method based on the Internet of Things as described in any one of claims 1 to 7 is implemented.

9. A computer-readable storage medium having a computer program stored therein, characterized in that: When the computer program is loaded and executed by the processor, the smart helmet management method based on the Internet of Things described in any one of claims 1 to 7 is adopted.

Images