Limited space operation safety intelligent monitoring method and system
By using a deep hierarchical self-organizing network and an improved Gaussian process regression model, the problems of unstable data transmission, inaccurate trajectory prediction, and high node switching delay in confined space operations were solved, and full-process safety monitoring was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING INST OF RADIO METROLOGY & MEASUREMENT
- Filing Date
- 2025-12-08
- Publication Date
- 2026-04-21
AI Technical Summary
Existing technologies suffer from unstable data transmission, inaccurate trajectory prediction, and high node switching delays in confined space operations, making it impossible to effectively conduct comprehensive collaborative supervision of the environment, personnel, and equipment.
By dividing a limited space into deep hierarchical layers and dynamically adjusting the signal transmission power, an adaptive hierarchical self-organizing network is constructed. An improved Gaussian process regression model is used for trajectory prediction, and the optimal candidate signal nodes are selected for pre-binding and switching to transmit environmental monitoring data, operator status data, and on-site video data.
It achieved stable signal transmission, accurate trajectory prediction, and timely node switching within a limited space, completed safety monitoring of the entire operation process, and solved the problems of unstable data transmission, inaccurate trajectory prediction, and high node switching delay.
Smart Images
Figure CN121908219A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of intelligent monitoring, specifically to an intelligent monitoring method and system for safety monitoring in confined space operations. Background Technology
[0002] As a crucial component of gas pipeline networks, gate wells are enclosed spaces with complex structures, making them prone to safety hazards such as poor ventilation, insufficient oxygen, and excessive concentrations of toxic and harmful gases during operations, seriously threatening the lives of workers. Currently, existing technologies employ gas detection instruments, ventilation systems, and drones to monitor gate well operations, but several shortcomings remain: Firstly, traditional unified self-organizing network technologies struggle to adapt to the signal attenuation differences within the vertical space of gate wells, leading to data disconnections when workers move while wearing equipment. Secondly, trajectory prediction technologies largely rely on linear motion assumptions, failing to accommodate non-linear movements such as bending or turning, and node switching is often passively triggered, resulting in high response delays. Furthermore, existing monitoring methods fail to achieve comprehensive, collaborative supervision of the environment, personnel, and equipment, resulting in fragmented monitoring processes. Summary of the Invention
[0003] To address the problems of unstable data transmission, inaccurate trajectory prediction, and high node switching delay in the current confined space monitoring process in practical applications, this application provides an intelligent monitoring method and system for confined space operation safety.
[0004] The first aspect of this application provides an intelligent safety monitoring method for confined space operations, the method comprising: A limited space is divided into deep layers, and the signal transmission power is dynamically adjusted according to the depth layers and environmental parameters to construct an adaptive hierarchical self-organizing network; Collect real-time location data of the equipment worn by the workers, and predict the movement trajectory based on the real-time location data to obtain the predicted position and prediction confidence level at the next moment. Optimal candidate signal nodes are selected based on the predicted location and the predicted confidence level; Based on the optimal candidate signal node, pre-binding switching is performed to transmit environmental monitoring data, worker status data, and on-site video data within a limited space.
[0005] In a possible implementation, the dynamic adjustment of signal transmission power based on the depth layering and environmental parameters includes: The confined space is divided into three layers according to depth: upper layer, middle layer and lower layer. Real-time depth data of the wearable device is collected, and the well wall material parameters and device density parameters of the confined space are obtained at the same time. The layer in which the wearable device is located is determined based on the real-time depth data, and the signal path loss is calculated by combining the well wall material parameters and the device density parameters. The signal transmission power is adjusted according to the signal path loss and the preset sensitivity requirements.
[0006] In a possible implementation, the step of predicting the movement trajectory based on the real-time location data to obtain the predicted position and prediction confidence level at the next moment includes: The real-time location data is used to form a historical location sequence within a preset time period; Based on the historical location sequence and the corresponding timestamp, the movement trajectory is predicted to obtain the predicted position at the next moment and the prediction confidence level.
[0007] In a possible implementation, the movement trajectory prediction based on the historical location sequence and the corresponding timestamps includes: Extract the timestamp corresponding to each location data in the historical location sequence and establish the association between location and time; Spatiotemporal constraints are added during trajectory prediction to correct the initial prediction results; Based on the Bayesian posterior inference method, and combined with the constructed combined kernel function and the correction results of the spatiotemporal constraints, the predicted position at the next moment and the prediction confidence are obtained.
[0008] In a possible implementation, constructing the combined kernel function includes: Construct a squared exponential kernel function for spatial correlation of location data based on the distance relationships between location data; Construct an exponential kernel function for the temporal continuity of location data based on the interval relationship between timestamps; The combined kernel function is generated by fusing the squared exponential type kernel function with the exponential type kernel function and adding a noise correction term.
[0009] In a possible implementation, the step of adding spatiotemporal constraints during trajectory prediction to correct the initial prediction results includes: Obtain the physical boundary parameters of the finite space, construct the spatial boundary constraint term, and penalize and correct the initial prediction result when the position in the initial prediction result exceeds the physical boundary parameters. Set a maximum step size threshold for the movement of the operator, construct a time continuity constraint term, and penalize and correct the initial prediction result when the position change at adjacent time points in the initial prediction result exceeds the maximum step size threshold; The spatial boundary constraint term and the temporal continuity constraint term are fused together to make an overall correction to the initial prediction result, resulting in a corrected prediction position.
[0010] In a possible implementation, the step of selecting the optimal candidate signal node based on the predicted location and the predicted confidence level includes: Based on the predicted location and each pre-deployed signal node, generate the node distance; Obtain the link quality metrics for each pre-deployed signal node; When a signal node's distance to another signal node is less than a preset distance threshold, its link quality index is higher than a preset quality threshold, and its prediction confidence meets the confidence threshold, the current signal node is determined as a candidate signal node. If multiple candidate signal nodes exist, the optimal candidate signal node is selected based on the comprehensive evaluation result of the prediction confidence and node distance.
[0011] In a possible implementation, the pre-binding switch based on the optimal candidate signal node includes: After receiving a switching request, the signal node currently connected to the wearable device sends the communication parameters of the optimal candidate signal node back to the wearable device. The communication parameters include the node address and frequency band parameters. The wearable device sends a pre-binding update message to the optimal candidate signal node, and the optimal candidate signal node repeatedly checks the address of the wearable device and records the address of the wearable device as pending activation. Based on the real-time location data, it is determined whether the real-time location has entered the coverage area of the optimal candidate signal node; When the real-time location enters the coverage area of the optimal candidate signal node, the wearable device sends a formal binding update message to the optimal candidate signal node, the optimal candidate signal node activates the main connection with the wearable device, and the currently connected signal node stops data forwarding and releases resources.
[0012] In a possible implementation, the method further includes: When the prediction confidence does not meet the confidence threshold, multiple signal nodes with the best comprehensive evaluation results are selected as candidate signal nodes. The wearable device establishes a pre-binding relationship with each of the multiple candidate signal nodes; Based on the actual location of the wearable device, one of the candidate signal nodes is dynamically selected to activate the main connection.
[0013] A second aspect of this application provides an intelligent safety monitoring system for confined space operations, the system comprising: A construction module is used to divide a limited space into deep layers, dynamically adjust the signal transmission power according to the depth layers and environmental parameters, and construct an adaptive hierarchical self-organizing network. The prediction module is used to collect real-time location data of the equipment worn by the operator, and to predict the movement trajectory based on the real-time location data to obtain the predicted position and prediction confidence level at the next moment. The filtering module is used to filter the optimal candidate signal nodes based on the predicted location and the predicted confidence level; The transmission module is used to perform pre-binding switching based on the optimal candidate signal node, and to transmit environmental monitoring data, worker status data and on-site video data within a limited space.
[0014] As can be seen from the above technical solution, this application divides the space into upper, middle, and lower layers according to signal attenuation characteristics through deep layering, clarifying the signal transmission requirements of different areas. Based on the logarithmic distance path loss model and combined with environmental parameters, the signal attenuation at different depths is calculated, and the optimal transmission power is derived in reverse, ensuring that the signal strength at the receiving end remains within a stable range. This avoids interference caused by excessively weak or strong signals due to depth changes, eliminating the risk of disconnection caused by uneven coverage of fixed power in the spatial dimension. Furthermore, this application employs an improved Gaussian process regression model, which uses a combination of kernel functions to characterize the spatial correlation and temporal continuity of the location. It also incorporates spatial boundary constraints and temporal continuity constraints to output a high-confidence predicted location for the next moment. This predicted location ensures that subsequent handover preparation actions precede the movement of personnel, eliminating the risk of disconnection caused by handover delay in the temporal dimension. Finally, this application transmits environmental monitoring data, personnel status data, and on-site video data within a limited space through the aforementioned adaptive layered self-organizing network, completing the entire process of safety monitoring and effectively solving the problems of unstable data transmission, inaccurate trajectory prediction, and high node handover delay in current limited space monitoring processes. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a schematic diagram illustrating the specific process of performing confined space operations in a gate well, as described in this application embodiment.
[0017] Figure 2 This is a flowchart illustrating the intelligent monitoring method for confined space operation safety in the embodiments of this application.
[0018] Figure 3 This is a schematic diagram of the structure of the intelligent monitoring system for confined space operation safety in the embodiments of this application. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0020] It should be noted that the specific regulatory process for the intelligent safety supervision method for confined space operations in gate wells is as follows: Figure 1As shown, the regulatory process of this application is divided into pre-operation and in-operation supervision, with the in-operation supervision further divided into supervision above and below the gate well. Regarding the safety requirements for gate well operations, it is essential to ensure that all safety protection preparations are complete before operation. This application uses a mobile video tree erected outside the work warning line to collect video footage of the work site. The video tree connects to an integrated device to transmit video data. The integrated device has a built-in safety production element recognition algorithm that automatically identifies whether safety protection equipment within the work area is complete, and promptly alerts if any equipment is missing. Regarding the issue of environmental monitoring before entering the well, this application uses gas detection equipment to collect gas concentration information at three locations—upper, middle, and lower—before entering the well. This information is transmitted via network to the integrated device for analysis to determine if it meets the Level 3 operation standard. If it does, the integrated device sends a voice command to the video tree to remind on-site personnel to begin work. If the gas environment inside the well is unsuitable for personnel entry, the integrated device sends a voice command to the video tree to remind personnel to ventilate, and then repeats the detection until the gas environment inside the well is suitable for personnel entry. To address safety issues in well operations, this application implements intelligent safety supervision from aspects such as the identity of workers, the location of supervisors, and the safe behavior of workers. Specifically, through video captured by a mobile video tree, the integrated machine uses a built-in AI algorithm to identify whether workers are acquainted with those working underground. It also issues warnings to non-workers entering the contaminated area within the warning line, prompting them to stay away from the work area to avoid injury or creating unsafe conditions for the operation. According to safety production requirements, safety supervisors must remain within the work area at all times to ensure timely handling of potential safety hazards. This application also employs identity authentication to automatically track the supervisors' movements and confirm the presence of proactive safety supervisors during operations. Simultaneously, this invention automatically identifies and analyzes potential safety hazards within the warning line in real time, detecting behaviors strictly prohibited in confined spaces like wells, such as smoking, making phone calls, or gatherings, and promptly alerts and addresses any identified hazards. For underground workers, this invention achieves intelligent safety supervision from the perspectives of both worker safety and the safety of the gas environment inside the well. Workers wear wearable gas environment detectors to monitor the surrounding gas environment in real time and upload the collected data to an integrated device via a self-organizing network. The integrated device has built-in algorithms to analyze the underground gas environment and confirm whether the concentrations of gases such as oxygen, hydrogen sulfide, carbon monoxide, and methane meet personnel safety standards. When the concentration of a gas that could endanger personnel exceeds the standard, an automatic voice message is sent to the video system to remind safety supervisors to evacuate the workers from the mine and ventilate the mine until the working conditions are met before re-entering to resume work.Workers wear smart safety helmets, which collect real-time health indicators such as heart rate and blood oxygen levels from the workers underground. The collected data is then transmitted to an integrated device, which analyzes the data using built-in algorithms to determine whether the workers' health status meets the requirements for underground operations. If any abnormality is detected, an audio message is sent to the video feed to alert safety supervisors that a person's health condition is abnormal and to promptly evacuate the person to prevent safety issues during operations.
[0021] The following will provide a detailed description of this application. First, this application provides an implementation method for an intelligent safety monitoring method for confined space operations, such as... Figure 2 As shown, it includes: S101, Divide the limited space into deep layers, dynamically adjust the signal transmission power according to the depth layers and environmental parameters, and construct an adaptive layered self-organizing network; S102, collect real-time location data of the equipment worn by the operator, predict the movement trajectory based on the real-time location data, and obtain the predicted position and prediction confidence level at the next moment; S103, Filter the optimal candidate signal node based on the predicted location and the predicted confidence level; S104, based on the optimal candidate signal node, perform pre-binding switching to transmit environmental monitoring data, worker status data and on-site video data within the limited space.
[0022] It should be noted that the "limited space" in this application specifically refers to enclosed or semi-enclosed spaces with restricted access and complex operating environments, such as gate wells, underground pipelines, and sealed containers, with gate wells being a typical application scenario. Adaptive hierarchical self-organizing networks refer to self-organizing networks that dynamically adjust signal transmission power based on the depth of the limited space and environmental parameters to achieve full-space network coverage. Pre-binding handover operation refers to a method of establishing a communication association between the device and candidate signal nodes before the operator moves to the coverage area of the target node, and quickly activating the connection when the handover conditions are met.
[0023] The inventors of this application discovered that in confined spaces, especially manholes, there are significant differences in signal attenuation due to vertical depth. Typically, the signal is strong in the upper layers near the wellhead and weak in the lower layers further away. Traditional fixed-power ad hoc networks lead to insufficient signal strength in the lower layers and excessive signal fluctuations during movement, resulting in connection drops. This application addresses this by dividing the space into upper, middle, and lower layers based on signal attenuation characteristics, clearly defining the signal transmission requirements of different areas. Based on a logarithmic distance path loss model and combined with environmental parameters, the signal attenuation at different depths is calculated, and the optimal transmit power is derived in reverse, ensuring that the signal strength at the receiving end remains within a stable range. This avoids interference caused by excessively weak or strong signals due to depth variations, eliminating the risk of connection drops caused by uneven coverage of fixed power in the spatial dimension. Furthermore, this application employs an improved Gaussian process regression model, using a combination of kernel functions to characterize the spatial correlation and temporal continuity of the location. It also incorporates spatial boundary constraints and temporal continuity constraints, outputting a high-confidence predicted location for the next moment. This predicted location ensures that subsequent handover preparation actions precede the movement of personnel, eliminating the risk of connection drops caused by handover delays in the temporal dimension. Finally, this application uses the adaptive hierarchical self-organizing network to transmit environmental monitoring data, worker status data, and on-site video data within a limited space, thereby completing the safety monitoring of the entire operation process.
[0024] In one embodiment of this application, the dynamic adjustment of signal transmission power based on the depth layering and environmental parameters includes: S201, the limited space is divided into three layers according to depth: upper layer, middle layer and lower layer, and the real-time depth data of the wearable device is collected. At the same time, the well wall material parameters and device density parameters of the limited space are obtained. S202, determine the layer in which the wearable device is located based on the real-time depth data, and calculate the signal path loss by combining the well wall material parameters and device density parameters; S203, adjust the signal transmission power according to the signal path loss and the preset sensitivity requirements.
[0025] This application achieves deep layering and dynamic power adjustment within a limited space by constructing an adaptive hierarchical self-organizing network, thereby offsetting the signal attenuation differences in different regions within the limited space.
[0026] First, this application divides the confined space into depth layers. For example, in a gate well scenario, it is divided into three layers—upper, middle, and lower—based on signal attenuation characteristics. The upper layer is close to the gate well opening, where signal attenuation is relatively small; the middle layer is the intermediate area, where signal attenuation is moderate due to reflections from the well wall and equipment blockage; and the lower layer is far from the well opening, where signal attenuation is severe. It should be noted that the number of layers in this application is not limited to this and can be flexibly adjusted according to the actual depth and structural complexity of the confined space.
[0027] Secondly, based on depth layering and dynamic adjustment of signal transmission power according to environmental parameters, in this embodiment of the application, a logarithmic distance path loss model is used to calculate signal attenuation at different depths. The specific calculation formula is as follows: ,in, To represent the path loss at depth h, which is the amount of power attenuation of the signal as it travels from the transmitter to the receiver; To indicate the reference distance The path loss at that point is a preset baseline attenuation value; The path loss index is related to the environment of a confined space, such as well wall material and equipment density, and varies with different layers. Different values; To represent the finite spatial depth in which the current device is located; This is a preset fixed value used to represent a reference distance; To represent shadow fading, it follows a Gaussian distribution with a mean of 0, and is used to describe signal attenuation fluctuations caused by random occlusion; This represents an environmental correction factor, used to correct for the influence of wellbore materials such as concrete and metal, or other fixed environmental factors, on the signal.
[0028] Based on the path loss model described above, the dynamic adjustment formula for transmit power is as follows: ,in, To represent the dynamic transmission power at depth h; This represents the minimum sensitivity of the receiver, which is the lowest power at which the receiver can normally receive signals. This indicates the link margin, used to cope with sudden signal attenuation and ensure link reliability; To indicate the gain of the transmitting antenna; This represents the receiving antenna gain.
[0029] In this embodiment, real-time depth data of the device is collected by a positioning device, and combined with preset environmental parameters, such as well wall material and device density, the target transmission power is calculated by substituting these parameters into the above formula. This power is then transmitted to the wireless communication unit via a control module to achieve dynamic power adjustment. It should be noted that environmental parameters can be pre-stored in the integrated device or collected in real-time by sensors; this application is not limited to this.
[0030] In one embodiment of this application, the step of predicting the movement trajectory based on the real-time location data to obtain the predicted position and prediction confidence level at the next moment includes: S301, using the real-time location data, a historical location sequence within a preset time period is formed; S302, based on the historical location sequence and the corresponding timestamp, predict the movement trajectory to obtain the predicted position at the next moment and the prediction confidence.
[0031] It should be noted that in this embodiment, a positioning device deployed on a worker's worn device, such as a smart safety helmet or wearable detector, continuously collects the device's location data. For example, the positioning device may employ TOF positioning technology (Time-of-Flight positioning), triangulation technology, etc., but this application is not limited to these. The positioning device collects location data at a preset frequency, forming a historical location sequence within a preset time period. This historical location sequence includes multiple location data points and a timestamp corresponding to each location data point, used for subsequent trajectory prediction.
[0032] In one embodiment of this application, the movement trajectory prediction based on the historical location sequence and the corresponding timestamps includes: S401, Extract the timestamp corresponding to each location data in the historical location sequence and establish the association between location and time; S402, in the trajectory prediction process, spatiotemporal constraints are added to correct the initial prediction results; S403, based on the Bayesian posterior inference method, combined with the constructed combined kernel function and the correction results of the spatiotemporal constraints, the predicted position at the next time moment and the prediction confidence are obtained.
[0033] Extract the timestamp corresponding to each location data in the historical location sequence, and establish the association between location and time. For example, if the historical location sequence contains... For each location data, then each location data Each corresponds to a timestamp ,form The associated dataset. It should be noted that the data for each location... These are three-dimensional coordinates, corresponding to the x, y, and h directions in a finite space.
[0034] It should be noted that the posterior mean is the predicted position, and the posterior variance is the confidence level of the prediction. The smaller the posterior variance, the higher the confidence level of the prediction.
[0035] The method for obtaining the predicted location and prediction confidence at the next time step based on Bayesian posterior inference is as follows: The posterior distribution of the predicted location follows a Gaussian distribution, and the formulas for the posterior mean and posterior variance are as follows: , in, Let represent the posterior mean of the predicted position at the next time step; Let this be the covariance vector between the historical timestamp and the predicted timestamp; Let this represent the covariance matrix between historical timestamps; To represent the identity matrix; The output vector represents the sequence of historical positions. The gradient of the constraint penalty term is used to correct the predicted position; Let this represent the posterior variance of the predicted position at the next time step; This represents the autocovariance corresponding to the predicted timestamp.
[0036] In one embodiment that can be implemented in this application, the construction of the combined kernel function includes: S501, construct a squared exponential kernel function for spatial correlation of location data based on the distance relationship between location data; S502, construct an exponential kernel function for the temporal continuity of location data based on the interval relationship between timestamps; S503, the squared exponential type kernel function is fused with the exponential type kernel function, and a noise correction term is added to generate the combined kernel function.
[0037] In this embodiment of the application, the formula for combining kernel functions is as follows: ,in, To represent timestamps and Covariance between corresponding location data; The amplitude of the kernel function is used to adjust the signal variance. To represent the squared exponential kernel function, used to characterize the spatial correlation of location data, the formula is: ,in, Indicates the first Location data in Dimension coordinates express The length scale parameter of the dimension; To represent the exponential kernel function, used to characterize the temporal continuity of location data, the formula is: ,in, Indicates the time scale parameter; To represent the noise variance, used to correct for the effects of positioning errors; To represent the Kronecker function, when hour ,otherwise .
[0038] It should be noted that the types of combined kernel functions in this application are not limited to this. For example, the combined kernel function used can be a combination of square exponential kernel and Markov kernel, or a combination of exponential kernel and rational quadratic kernel, etc. Any combined kernel function that can simultaneously characterize spatial correlation and temporal continuity is acceptable.
[0039] In one embodiment that can be implemented in this application, the step of adding spatiotemporal constraints during trajectory prediction to correct the initial prediction result includes: S601, Obtain the physical boundary parameters of the finite space, construct the spatial boundary constraint term, and when the position in the initial prediction result exceeds the physical boundary parameters, penalize and correct the initial prediction result; S602, set the maximum step size threshold for the operator's movement, construct a time continuity constraint term, and when the position change at adjacent times in the initial prediction result exceeds the maximum step size threshold, penalize and correct the initial prediction result; S603, the spatial boundary constraint term and the temporal continuity constraint term are fused together to make an overall correction to the initial prediction result, thereby obtaining the corrected prediction position.
[0040] This application incorporates spatiotemporal constraints to correct the initial prediction results. Since the finite space has physical boundaries and the movement of workers is continuous, this embodiment corrects the prediction results by constructing a constraint penalty term. The optimized objective function is as follows. ,in, To represent the optimization objective function; Let represent the log-likelihood function of a Gaussian process regression; To represent spatial constraint weights; To represent the spatial boundary constraint term, the formula is: ,in , They represent The minimum and maximum boundaries of the dimension are used to impose a penalty value on the constraint term when the predicted position exceeds the boundary, forcing the model to correct the prediction result. To represent the weights of the time continuity constraint; To represent the time continuity constraint term, the formula is: ,in This represents the maximum allowed step size threshold. When the step size of the predicted trajectory exceeds the threshold, the constraint term generates a penalty value to ensure the smoothness of the trajectory. To indicate the predicted position at the next moment; This indicates the actual location at the current moment.
[0041] It should be noted that, in the embodiments of this application, the spatial boundary constraint term and the temporal continuity constraint term are fused by weighted summation.
[0042] In one embodiment of this application, the step of selecting the optimal candidate signal node based on the predicted location and the predicted confidence level includes: S701, Generate node distance based on the predicted location and each pre-deployed signal node; S702, obtain the link quality indicators of each pre-deployed signal node; S703, when a signal node has a node distance less than a preset distance threshold, a link quality index higher than a preset quality threshold, and a prediction confidence level that meets the confidence level threshold, the current signal node is determined as a candidate signal node. S704, if there are multiple candidate signal nodes, the optimal candidate signal node is selected based on the comprehensive evaluation result of the prediction confidence and node distance.
[0043] First, the location information of all pre-deployed signal nodes within the limited space is obtained. For example, the signal nodes are evenly deployed at different depths and orientations within the limited space to form a fully covered node network, and the node location information is pre-stored in the all-in-one machine.
[0044] Secondly, the distance between the predicted location and each signal node is calculated, and the link quality index of each signal node is detected. It should be noted that the link quality index is used to evaluate the communication quality of the signal node. The calculation method is not limited in this application. For example, the link index can be calculated by comprehensively considering parameters such as the normalized value of the received signal strength, the data packet error rate, and the communication delay.
[0045] Then, it is determined whether the prediction confidence level meets the preset confidence level threshold. If the prediction confidence level meets the threshold, it means that the prediction result is reliable, and candidate nodes are selected based on distance and link quality indicators. If the threshold is not met, it means that the prediction result is less reliable, and a backup plan is adopted.
[0046] Finally, signal nodes whose distance to other nodes is less than a preset distance threshold, whose link quality index is higher than a preset quality threshold, and whose prediction confidence meets the confidence threshold are selected as candidate signal nodes. If there are multiple candidate signal nodes, the optimal candidate signal node is selected based on the comprehensive evaluation result of prediction confidence and node distance. For example, the comprehensive score is calculated by multiplying the confidence weight by the inverse of the distance, and the node with the highest score is selected as the optimal candidate signal node.
[0047] In one embodiment that can be implemented in this application, the pre-binding switching based on the optimal candidate signal node includes: S801, after receiving the switching request, the signal node currently connected to the wearable device feeds back the communication parameters of the optimal candidate signal node to the wearable device. The communication parameters include the node address and frequency band parameters. S802, the wearable device sends a pre-binding update message to the optimal candidate signal node, and the optimal candidate signal node repeatedly checks the address of the wearable device and records the address of the wearable device as a pending activation state; S803, using the real-time location data, determine whether the real-time location has entered the coverage area of the optimal candidate signal node; S804, when the real-time location enters the coverage area of the optimal candidate signal node, the wearable device sends a formal binding update message to the optimal candidate signal node, the optimal candidate signal node activates the main connection with the wearable device, and the currently connected signal node stops data forwarding and releases resources.
[0048] First, after receiving the switching request, the signal node currently connected to the wearable device sends the communication parameters of the candidate signal node back to the wearable device. The communication parameters include node address, frequency band parameters, communication protocol, etc., to ensure communication compatibility between the wearable device and the candidate node.
[0049] Secondly, the wearable device sends a pre-binding update message to the optimal candidate signal node. The optimal candidate signal node repeatedly checks the address of the wearable device to avoid address conflicts and records the address of the wearable device as pending activation. At the same time, a temporary data forwarding channel is established between the current node and the optimal candidate node to ensure that data is not lost during the pre-binding period.
[0050] Then, the actual location of the wearable device is collected in real time by the positioning device to determine whether the actual location is within the coverage range of the optimal candidate signal node. For example, the coverage range of the signal node is determined according to the communication radius of the signal node, but this application is not limited thereto.
[0051] Finally, when the actual location enters the coverage area of the optimal candidate signal node, the wearable device sends a formal binding update message to the optimal candidate node. The candidate node activates the main connection with the wearable device, the current node stops data forwarding and releases resources, and the signal node switch is completed.
[0052] In one embodiment that can be implemented in this application, the method further includes: S901, when the prediction confidence does not meet the confidence threshold, select the signal node with the best comprehensive evaluation result as the candidate signal node; S902, the wearable device establishes a pre-binding relationship with each of the multiple candidate signal nodes; S903, dynamically select one of the candidate signal nodes to activate the main connection based on the actual location of the wearable device.
[0053] It should be noted that when the prediction confidence level does not meet the confidence level threshold, the embodiments of this application adopt a backup scheme, which selects multiple signal nodes with the best comprehensive evaluation results as alternative candidate signal nodes. The wearable device establishes a pre-binding relationship with multiple alternative candidate nodes respectively, and then dynamically selects one of the alternative nodes to activate the main connection according to the actual location of the wearable device, so as to avoid signal node switching failure due to inaccurate prediction.
[0054] Furthermore, this application provides an intelligent monitoring system for safety in confined space operations, such as... Figure 3 As shown, the system includes: The construction module 1001 is used to divide a limited space into deep layers, dynamically adjust the signal transmission power according to the depth layers and environmental parameters, and construct an adaptive hierarchical self-organizing network. Prediction module 1002 is used to collect real-time location data of the equipment worn by the operator, and to predict the movement trajectory based on the real-time location data to obtain the predicted position and prediction confidence level at the next moment. The filtering module 1003 is used to filter the optimal candidate signal node based on the predicted position and the predicted confidence level; The transmission module 1004 is used to perform pre-binding switching based on the optimal candidate signal node, and to transmit environmental monitoring data, worker status data and on-site video data within a limited space.
[0055] It is understood that the technical effects of the intelligent monitoring system for safety of confined space operations provided in this disclosure are consistent with the technical effects of the method embodiments in the foregoing embodiments, and this disclosure will not elaborate on them.
[0056] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.
Claims
1. A method for intelligent monitoring of safety in confined space operations, characterized in that, The method includes: A limited space is divided into deep layers, and the signal transmission power is dynamically adjusted according to the depth layers and environmental parameters to construct an adaptive hierarchical self-organizing network; Collect real-time location data of the equipment worn by the workers, and predict the movement trajectory based on the real-time location data to obtain the predicted position and prediction confidence level at the next moment. Optimal candidate signal nodes are selected based on the predicted location and the predicted confidence level; Based on the optimal candidate signal node, pre-binding switching is performed to transmit environmental monitoring data, worker status data, and on-site video data within a limited space.
2. The intelligent monitoring method for safety in confined space operations according to claim 1, characterized in that, The dynamic adjustment of signal transmission power based on the depth layering and environmental parameters includes: The confined space is divided into three layers according to depth: upper layer, middle layer and lower layer. Real-time depth data of the wearable device is collected, and the well wall material parameters and device density parameters of the confined space are obtained at the same time. The layer in which the wearable device is located is determined based on the real-time depth data, and the signal path loss is calculated by combining the well wall material parameters and the device density parameters. The signal transmission power is adjusted according to the signal path loss and the preset sensitivity requirements.
3. The intelligent monitoring method for safety in confined space operations according to claim 1, characterized in that, The process of predicting the movement trajectory based on the real-time location data to obtain the predicted position and prediction confidence level at the next moment includes: The real-time location data is used to form a historical location sequence within a preset time period; Based on the historical location sequence and the corresponding timestamp, the movement trajectory is predicted to obtain the predicted position at the next moment and the prediction confidence level.
4. The intelligent monitoring method for safety in confined space operations according to claim 3, characterized in that, The movement trajectory prediction based on the historical location sequence and corresponding timestamps includes: Extract the timestamp corresponding to each location data in the historical location sequence and establish the association between location and time; Spatiotemporal constraints are added during trajectory prediction to correct the initial prediction results; Based on the Bayesian posterior inference method, and combined with the constructed combined kernel function and the correction results of the spatiotemporal constraints, the predicted position at the next moment and the prediction confidence are obtained.
5. The intelligent safety monitoring method for confined space operations according to claim 4, characterized in that, Constructing a combined kernel function includes: Construct a squared exponential kernel function for spatial correlation of location data based on the distance relationships between location data; Construct an exponential kernel function for the temporal continuity of location data based on the interval relationship between timestamps; The combined kernel function is generated by fusing the squared exponential type kernel function with the exponential type kernel function and adding a noise correction term.
6. The intelligent monitoring method for safety in confined space operations according to claim 4, characterized in that, The step of adding spatiotemporal constraints during trajectory prediction to correct the initial prediction results includes: Obtain the physical boundary parameters of the finite space, construct the spatial boundary constraint term, and penalize and correct the initial prediction result when the position in the initial prediction result exceeds the physical boundary parameters. Set a maximum step size threshold for the movement of the operator and construct a time continuity constraint term. When the position change at adjacent time points in the initial prediction result exceeds the maximum step size threshold, the initial prediction result is penalized and corrected. The spatial boundary constraint term and the temporal continuity constraint term are fused together to make an overall correction to the initial prediction result, resulting in a corrected prediction position.
7. The intelligent monitoring method for safety in confined space operations according to claim 1, characterized in that, The step of selecting the optimal candidate signal node based on the predicted location and the predicted confidence level includes: Based on the predicted location and each pre-deployed signal node, generate the node distance; Obtain the link quality metrics for each pre-deployed signal node; When a signal node's distance to another signal node is less than a preset distance threshold, its link quality index is higher than a preset quality threshold, and its prediction confidence level meets the confidence threshold, the current signal node is determined as a candidate signal node. If multiple candidate signal nodes exist, the optimal candidate signal node is selected based on the comprehensive evaluation result of the prediction confidence and node distance.
8. The intelligent monitoring method for safety in confined space operations according to claim 4, characterized in that, The pre-binding switching based on the optimal candidate signal node includes: After receiving a switching request, the signal node currently connected to the wearable device sends the communication parameters of the optimal candidate signal node back to the wearable device. The communication parameters include the node address and frequency band parameters. The wearable device sends a pre-binding update message to the optimal candidate signal node, and the optimal candidate signal node repeatedly checks the address of the wearable device and records the address of the wearable device as pending activation. Based on the real-time location data, it is determined whether the real-time location has entered the coverage area of the optimal candidate signal node; When the real-time location enters the coverage area of the optimal candidate signal node, the wearable device sends a formal binding update message to the optimal candidate signal node, the optimal candidate signal node activates the main connection with the wearable device, and the currently connected signal node stops data forwarding and releases resources.
9. The intelligent monitoring method for safety in confined space operations according to claim 8, characterized in that, The method further includes: When the prediction confidence level does not meet the confidence threshold, multiple signal nodes with the best comprehensive evaluation results are selected as candidate signal nodes. The wearable device establishes a pre-binding relationship with each of the multiple candidate signal nodes; Based on the actual location of the wearable device, one of the candidate signal nodes is dynamically selected to activate the main connection.
10. A confined space operation safety intelligent monitoring system, characterized in that, The system includes: A construction module is used to divide a limited space into deep layers, dynamically adjust the signal transmission power according to the depth layers and environmental parameters, and construct an adaptive hierarchical self-organizing network. The prediction module is used to collect real-time location data of the equipment worn by the operator, and to predict the movement trajectory based on the real-time location data to obtain the predicted position and prediction confidence level at the next moment. The filtering module is used to filter the optimal candidate signal nodes based on the predicted location and the predicted confidence level; The transmission module is used to perform pre-binding switching based on the optimal candidate signal node, and to transmit environmental monitoring data, worker status data and on-site video data within a limited space.