CN117496431A - Outdoor operation safety monitoring method based on indoor and outdoor positioning system - Google Patents
Outdoor operation safety monitoring method based on indoor and outdoor positioning system Download PDFInfo
- Publication number
- CN117496431A CN117496431A CN202311456064.8A CN202311456064A CN117496431A CN 117496431 A CN117496431 A CN 117496431A CN 202311456064 A CN202311456064 A CN 202311456064A CN 117496431 A CN117496431 A CN 117496431A
- Authority
- CN
- China
- Prior art keywords
- staff
- data
- worker
- indoor
- risk
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000012544 monitoring process Methods 0.000 title claims abstract description 43
- 230000000694 effects Effects 0.000 claims abstract description 78
- 238000005516 engineering process Methods 0.000 claims abstract description 46
- 238000013468 resource allocation Methods 0.000 claims abstract description 43
- 238000012502 risk assessment Methods 0.000 claims abstract description 36
- 238000009826 distribution Methods 0.000 claims abstract description 23
- 238000001514 detection method Methods 0.000 claims abstract description 22
- 238000012300 Sequence Analysis Methods 0.000 claims abstract description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 58
- 238000011156 evaluation Methods 0.000 claims description 35
- 230000003993 interaction Effects 0.000 claims description 29
- 238000012549 training Methods 0.000 claims description 14
- 230000006399 behavior Effects 0.000 claims description 13
- 230000003245 working effect Effects 0.000 claims description 13
- 238000004458 analytical method Methods 0.000 claims description 12
- 230000003287 optical effect Effects 0.000 claims description 12
- 230000009471 action Effects 0.000 claims description 11
- 230000001133 acceleration Effects 0.000 claims description 9
- 230000007613 environmental effect Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 8
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 claims description 8
- 238000012706 support-vector machine Methods 0.000 claims description 8
- 238000004140 cleaning Methods 0.000 claims description 7
- 238000004891 communication Methods 0.000 claims description 7
- 230000007774 longterm Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 238000009499 grossing Methods 0.000 claims description 5
- 230000007246 mechanism Effects 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 230000004888 barrier function Effects 0.000 claims description 4
- 125000004122 cyclic group Chemical group 0.000 claims description 4
- 230000007787 long-term memory Effects 0.000 claims description 4
- 230000006403 short-term memory Effects 0.000 claims description 4
- 238000000547 structure data Methods 0.000 claims description 4
- 230000000007 visual effect Effects 0.000 claims description 4
- 230000002452 interceptive effect Effects 0.000 claims description 3
- 230000015654 memory Effects 0.000 claims description 3
- 238000012731 temporal analysis Methods 0.000 claims description 3
- 238000000700 time series analysis Methods 0.000 claims description 3
- 238000004088 simulation Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 208000030251 communication disease Diseases 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000012913 prioritisation Methods 0.000 description 1
- 230000001681 protective effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S11/00—Systems for determining distance or velocity not using reflection or reradiation
- G01S11/12—Systems for determining distance or velocity not using reflection or reradiation using electromagnetic waves other than radio waves
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/08—Construction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/269—Analysis of motion using gradient-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/277—Analysis of motion involving stochastic approaches, e.g. using Kalman filters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/62—Extraction of image or video features relating to a temporal dimension, e.g. time-based feature extraction; Pattern tracking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30232—Surveillance
Abstract
The application provides an outdoor operation safety monitoring method based on an indoor and outdoor positioning system, which comprises the following steps: according to real-time monitoring data of a working area, combining indoor and outdoor positioning technologies, accurately judging the position distribution of workers; the output data of the indoor and outdoor positioning technology is utilized, and the movement track of the staff is dynamically analyzed by combining time sequence analysis, so that the activity range of the staff in a short time is predicted; according to the predicted activity range and indoor and outdoor positioning data of the staff, constructing a working area risk assessment model, and analyzing whether an object falling risk exists or not; when the falling risk of the object is detected, immediately sending safety warning information to related staff; according to future working contents, safety influence and indoor and outdoor positioning data of the staff, resource allocation and staff scheduling are carried out on the working area; and adjusting the resource allocation of the working area and optimizing the working flow by combining the obstacle detection result, the movable range of the staff and the indoor and outdoor positioning data.
Description
Technical Field
The invention relates to the technical field of information, in particular to an outdoor operation safety monitoring method based on an indoor and outdoor positioning system.
Background
With the rapid development of modern engineering and construction industry, the safety problem of outdoor operation is increasingly emphasized. Especially in construction sites, bridge repairs, large equipment installations, etc., workers may be faced with a number of potential risks, such as equipment collisions, object drops, and personnel injuries. In order to improve the working efficiency and ensure the working safety, the risk factors must be monitored and pre-warned in real time. In the past, the safety of outdoor work has relied primarily on manual inspection and periodic training. However, as technology advances, the use of digital and automated technology in outdoor work safety has become increasingly popular. Traditional security monitoring methods, such as CCTV camera systems, while capable of providing real-time video monitoring, often have limitations in prediction and early warning. They often cannot accurately predict the next action of the staff, let alone pre-warning of impending danger. In recent years, the development of indoor and outdoor positioning technology brings new opportunities for outdoor operation safety monitoring. By accurately tracking the positions of the staff and equipment, the activity on the site can be known in real time. However, depending on the positioning data alone, prediction of dynamic blind areas, which are areas that a worker may not see or predict in some cases, such as line of sight occlusion by a mobile device or temporary building, still cannot be achieved. Furthermore, conventional work distribution is often based on the experience and responsibility of the staff member, rather than its actual skill or current work requirements. This may lead to mismatch of work content and worker skills, thereby affecting work efficiency and safety. Therefore, in combination with the latest technical development, the creation of a comprehensive and intelligent safety monitoring system for outdoor operation has become a primary task of the industry.
Disclosure of Invention
The invention provides an outdoor operation safety monitoring method based on an indoor and outdoor positioning system, which mainly comprises the following steps:
according to real-time monitoring data of a working area, combining indoor and outdoor positioning technologies, accurately judging the position distribution of workers; the output data of the indoor and outdoor positioning technology is utilized, and the movement track of the staff is dynamically analyzed by combining time sequence analysis, so that the activity range of the staff in a short time is predicted; according to the predicted activity range and indoor and outdoor positioning data of the staff, constructing a working area risk assessment model, and analyzing whether an object falling risk exists or not; using an analysis result of the falling risk of the object, combining indoor and outdoor positioning data, judging whether an up-down operation relationship exists between workers, and monitoring the falling possibility of the object in real time; when the falling risk of the object is detected, immediately sending safety warning information to related staff; adopting time sequence analysis, predicting future working contents of each worker by using data of indoor and outdoor positioning technology and the aforementioned worker activity range prediction, and judging the safety influence of each worker on other people; according to future working contents, safety influence and indoor and outdoor positioning data of the staff, resource allocation and staff scheduling are carried out on the working area; based on resource allocation, personnel scheduling output and indoor and outdoor positioning technologies, obstacle detection is carried out on the working area of each worker, so that working interference and object collision risks are avoided; and adjusting the resource allocation of the working area and optimizing the working flow by combining the obstacle detection result, the movable range of the staff and the indoor and outdoor positioning data.
Further, the method for accurately judging the position distribution of the staff according to the real-time monitoring data of the working area and combining the indoor and outdoor positioning technology comprises the following steps:
acquiring real-time monitoring data through a high-resolution camera; identifying a humanoid outline in the video by utilizing a YOLOv3 algorithm, filtering out the characteristics of workers and distinguishing the characteristics from the background; acquiring the position information of a person by adopting an indoor and outdoor positioning technology, and matching an object detection result with the person positioning data; correcting the position information through a Kalman filter; combining the human shape recognition result and the corrected positioning data to confirm the accurate coordinates of each worker; classifying the positions of all the workers according to the labels of the areas, floors or working groups, and storing classification data; calculating personnel density of each region by using a nuclear density estimation method, and marking the region with the density higher than a preset threshold value; aiming at a high-density area, a Lucas-Kanade optical flow method is applied to obtain the moving path of the staff, and the mutual distance and potential conflict points of the staff are judged; acquiring a moving track, a stay point and an interaction point of each worker in a working area; using an Apriori algorithm to determine a common behavior mode of a worker under similar environments and tasks; predicting future positions and potential behaviors of the staff according to the result of association rule learning; and generating a final staff position distribution map through matplotlib, and displaying real-time and historical distribution of staff.
Further, the output data using the indoor and outdoor positioning technology, combined with time sequence analysis, dynamically analyzes the movement track of the staff, predicts the movement range in a short time, and includes:
indoor and outdoor positioning technology is utilized to obtain indoor movement data and outdoor movement data of a worker; combining indoor and outdoor movement data to obtain a complete movement track; cleaning the data, removing abnormal values or repeated data, and performing time standardization processing; determining a long-term movement trend of a worker by using an exponential smoothing method, and identifying a repeated movement mode in a specified time period; according to the extracted activity modes and trends, performing model training by using a K-means algorithm, and identifying a main activity area; determining main activity points of workers in a designated time period according to the stay time and the access frequency of each activity area; according to the long-term movement trend of the staff, predicting the movement trend of the staff in a short time by utilizing an ARIMA algorithm; and determining the short-term activity range of the staff by combining the prediction result and the known movement track.
Further, the constructing a working area risk assessment model according to the predicted activity range and the indoor and outdoor positioning data of the staff, and analyzing whether the object falling risk exists includes:
The complete positioning data of the staff is obtained through the indoor positioning data and the outdoor positioning data; according to the positioning data of the staff, a long-term and short-term memory network algorithm is applied to predict the action route of the staff at the next time; performing intersection analysis on the predicted action route and a high risk area to determine a possible high risk area, wherein the high risk area comprises a landing entrance, a window edge, an edge and a high altitude area; real-time tracking the activity state of the staff in the high risk area by using a sensor or video monitoring; when a worker enters a high risk area, the probability of dropping an object is evaluated in real time; three-dimensional model data of a building site are acquired, and dangerous points in the three-dimensional model data are identified, wherein the dangerous points comprise barriers and suspended areas; evaluating object falling or other risks by combining the moving track of the worker with the dangerous points; according to the space structure data of the dangerous points, the worker activities and the historical accident data, a support vector machine is used for constructing a working area risk assessment model, and risk assessment is carried out, wherein the risk assessment comprises high risk, medium risk and low risk; if the high risk is predicted, alarming in real time and recording an event; the criteria and parameters of risk assessment are adjusted and optimized periodically based on the actual events that occur.
Further, the analysis result of the falling risk of the object is combined with indoor and outdoor positioning data to judge whether an upper operation relationship and a lower operation relationship exist between staff, and the falling possibility of the object is monitored in real time, and the method comprises the following steps:
using a working area risk assessment model to obtain risk ratings of each working area; acquiring the real-time position of a worker through indoor and outdoor positioning data; constructing an input data set according to the risk rating and the positions of the staff, dividing a working area into an upper layer and a lower layer, and marking the hierarchy of each staff according to the real-time position data; judging whether a worker between the upper layer and the lower layer has an upper-lower operation relationship by using a KNN algorithm; adopting a real-time monitoring mechanism to monitor the continuous object movement condition of a working area, adopting an optical flow method to identify the movement of an object in real time, and predicting the falling point of the object according to the current position, speed and acceleration of the object; judging whether the possibility of falling of the object and the staff possibly influenced are present according to the predicted falling point and the position of the staff; if the position distance between the predicted falling point of the object and the worker is smaller than a preset threshold value, marking the object as high risk; according to the predicted falling point and the position data of the staff, the safety state of the staff is updated by combining the risk rating; if a high risk state is detected, generating a complete risk assessment report by using an automatic summarization technology of NLP according to all information related to the detected high risk state; further comprises: predicting a dynamic blind area to be formed and giving a warning to a worker.
The prediction is about to form a dynamic blind area and gives a warning to staff, and the method specifically comprises the following steps:
real-time data streams in the work environment are acquired, including the location, speed, and trajectory of equipment and personnel. And detecting the activity or the object which possibly generates the dynamic blind area in real time by using the SORT tracking algorithm, and obtaining the real-time record of the dynamic blind area generation factors. And according to the real-time record of the dynamic blind area generating factors, predicting the dynamic blind area possibly formed in the future based on the past data by utilizing a long-short-period memory network algorithm to obtain a prediction graph of the dynamic blind area. And reconfiguring task allocation of the workflow or personnel according to the prediction graph of the dynamic blind area and the current workflow to obtain the optimized workflow. And according to the prediction graph of the dynamic blind area, if a worker or equipment enters the predicted dynamic blind area or the distance between the worker or equipment and the predicted dynamic blind area is smaller than a preset threshold value, immediately giving out a warning.
Further, when the risk of falling the object is detected, immediately sending a safety warning message to the relevant staff, including:
acquiring past object drop event data, including environmental parameters, object characteristics, cleaning, normalization and labeling data; according to the processed data, training an object falling risk detection model by using a support vector machine algorithm; acquiring environmental data in real time by using a high-precision sensor, and obtaining an object falling risk level according to an object falling risk detection model; triggering a safety warning generation flow when the falling risk level of the object exceeds a preset threshold value; acquiring contact information of a possibly affected worker by accessing a worker database; using a push service to transmit a safety warning to a target staff based on the acquired contact information; if communication disorder occurs, the system monitors the communication state and automatically retries sending the warning until success or the preset number of attempts is reached.
Further, the method for predicting future working contents of each worker and judging the safety influence of the worker on other people by using time series analysis and using the data of the indoor and outdoor positioning technology and the prediction of the working range of the worker comprises the following steps:
acquiring historical activity range data of a worker, processing the historical data by using an ARIMA algorithm, and extracting key features including activity duration, activity range size and frequency; acquiring real-time position information of a worker by using an indoor and outdoor positioning technology; fusing key features of the real-time position data and the historical activity data, constructing a worker behavior model by using a cyclic neural network algorithm, and predicting future work content of workers; drawing a future activity track of each worker according to the prediction result; determining space crossing points among the staff members according to the future activity track of each staff member, acquiring interaction relation of the staff members and evaluating interaction complexity, wherein the crossing points are commonly used tools or equipment areas; according to interaction complexity and safety evaluation factors of the staff, calculating the safety influence index of each staff on other people; classifying the staff as safe or potential risk according to the safety impact index; when the state of the staff is a potential risk, automatically generating a safety suggestion, and issuing the safety suggestion to related staff through a communication channel; further comprises: and calculating the safety influence index of each worker on other people according to the positions of the workers, the interaction records and the working property data.
According to the position, interaction record and working property data of the staff, calculating the safety influence index of each staff to other people, specifically comprising:
determining interaction complexity according to the work dependence among the staff and the frequency of mutual contact; acquiring safety evaluation factors of workers, including spatial proximity, working properties and historical safety records, wherein the spatial proximity is determined through the physical distance between the workers; weight is distributed according to the influence degree of each factor on the safety; acquiring the position, interactive record and working property data of a worker; for each worker, normalizing the obtained value to be in the range of 0-1 according to the formula safety impact index = Σ (evaluation factor value x weight), and obtaining the safety impact index of each worker.
Further, the resource allocation and personnel scheduling for the working area according to the future working content, the security influence and the indoor and outdoor positioning data of the staff comprises:
acquiring all the working contents, distributing weights for each working content according to the importance and the emergency degree of the working contents, and sequencing according to the importance and the emergency degree; according to indoor and outdoor positioning data and safety influence data, a K-means algorithm is used for distributing a safety level label for each working area, wherein the safety level label comprises high risk, medium risk and low risk; according to the weight of the working content and the security level label of the working area, distributing a priority to each resource; according to the classification of the working contents and the weight thereof and the security level label of the working area, the priority of the resources is ordered, and the priority of the resources is determined, wherein the priority comprises high priority, medium priority and low priority; allocating high-priority resources to corresponding working areas and ensuring matching with security level labels; if the resources of a certain high-priority area are insufficient, allocating the medium-priority area or the low-priority area; determining the initial position of a worker through indoor and outdoor positioning data; according to the initial position of the staff and the weight of the working content, a specific task is allocated to each staff; scheduling staff according to the task allocation result to ensure the matching between staff and resources; further comprises: and evaluating the matching degree of the working content and the skill of the staff, and carrying out fine adjustment according to the emergency degree and the importance of the work.
The evaluation of the matching degree of the working content and the skill of the staff, and the fine adjustment is carried out according to the emergency degree and the importance of the work, specifically comprises the following steps:
a skill assessment table for each worker is obtained and its integrity is verified. Each skill in the skill assessment list is scored according to the actual application frequency and importance of each skill. The overall skill score S (i) for the worker is calculated from the score for each skill and its importance using a weighted average method. The type and level of skill required is determined based on each job. Skill demand indicators, including years of experience, certificates, and weights are set for them. And calculating the skill demand degree T (j) of the working content according to the skill index and the weight required by the working content. The skill score of each worker and the skill requirement level of each work content are extracted from the database. The efficiency of the staff member to complete similar work content in the past is evaluated, and the historical efficiency E (i, j) is calculated. Using the formula M (i, j) =s (i) ×t (j) ×e (i, j), a matching degree is calculated for each worker and work content. The calculated matching results are stored in a matrix and visualized in the system. And distributing the working content which is most matched with the skill of each worker to each worker based on the matching degree matrix. And fine tuning is performed according to the emergency degree and importance of the work, a work task list of each worker is generated in the system, and a work task notification is sent to the worker.
Further, the method for detecting the obstacle in the working area of each worker based on the resource allocation, the personnel scheduling output and the indoor and outdoor positioning technology, avoiding the working interference and the object collision risk, comprises the following steps:
acquiring resource allocation data, and allocating tasks and working areas for each worker according to the quantity and the types of resources required by each working content; acquiring real-time position information of a worker by using a sensor of an indoor and outdoor positioning technology; pre-training a model by using a YOLO algorithm according to the image in the working area, and identifying obstacles in the working area; scanning a working area by using a pre-trained YOLO model, detecting obstacles in the working area, and marking the positions of the obstacles; setting an initial state for the Kalman filter, including a current position and velocity estimate; using a Kalman filter, predicting the positions of the staff and the obstacle in the future based on the estimated positions and speeds of the staff and the obstacle at the last moment and the estimated positions and speeds of the staff and the obstacle at the current moment, and obtaining a predicted track or path, and displaying the expected moving direction and speed of the staff and the obstacle; if the distance between the staff and the obstacle in the predicted track is smaller than a preset safety threshold value, judging that the collision risk is possible; if the risk of a certain working area exceeds a safety threshold, re-allocating resources of the working area; if the resource redistribution is not feasible, the task positions and the task time of the staff are rearranged according to the risk assessment result, so that high risk areas are avoided; when the system predicts a collision risk, warning notification is sent to the relevant staff immediately by means of sound, vibration or visual notification.
Further, the step of adjusting the resource allocation of the working area and optimizing the workflow by combining the obstacle detection result, the movable range of the staff and the indoor and outdoor positioning data comprises the following steps:
marking and classifying obstacles in the video or the image by using a pre-trained YOLO model according to indoor and outdoor video streams or image data to obtain indoor and outdoor obstacle positions and types; according to the indoor and outdoor sensor data, determining the real-time position and the activity range of each worker; calculating available space and resources in a working area according to the positions of the obstacles and the moving ranges of the staff, comparing the available space and resources with the moving ranges of the obstacles and the staff, and determining the utilization rate of the resources; determining whether the current resource allocation is reasonable or not according to the resource utilization rate data and the preset threshold value comparison; if the current resource allocation is unreasonable, a linear programming algorithm is used according to the current resource allocation data, and the resource allocation is adjusted according to the current obstacle distribution and the working personnel activity range, so that a new resource allocation scheme is obtained; simulating a workflow under new resource allocation, and determining whether an efficiency problem exists; if the workflow has efficiency problems, adjusting the workflow according to the positioning of the staff and the distribution of the obstacles; according to the working range of the staff and the new workflow scheme, using a workflow simulation tool Arena to simulate the workflow and calculate efficiency indexes to obtain the efficiency evaluation of the new workflow; and comparing the efficiency and the risk of the two workflows according to the old workflow, the new workflow and the efficiency evaluation result to obtain the effect evaluation of workflow optimization.
The technical scheme provided by the embodiment of the invention can have the following beneficial effects:
the invention provides an outdoor operation safety monitoring method based on an indoor and outdoor positioning system, which can accurately capture the position and the movement track of a worker in real time and predict a dynamic blind area and early warn risks according to the data. More advanced, the system can also predict the future work content of each worker according to the prediction data, so as to ensure that the future work content is matched with the skills and experience of the worker, thereby reducing the potential risk. In addition, to optimize workflow and improve efficiency, the method can also perform intelligent resource allocation and personnel scheduling while avoiding potential work disturbance and collision risk through object detection. In general, the invention provides an omnibearing and multidimensional outdoor operation safety solution, which remarkably improves the working efficiency and the safety and brings revolutionary innovation to the safety management of modern engineering and construction industry.
Drawings
Fig. 1 is a flowchart of an outdoor operation safety monitoring method based on an indoor and outdoor positioning system.
Fig. 2 is a schematic diagram of an outdoor operation safety monitoring method based on an indoor and outdoor positioning system according to the present invention.
Fig. 3 is a schematic diagram of an outdoor operation safety monitoring method based on an indoor and outdoor positioning system according to the present invention.
Detailed Description
In order to make the technical solutions in the present specification better understood by those skilled in the art, the technical solutions in the embodiments of the present specification will be clearly and completely described below with reference to the drawings in the embodiments of the present specification, and it is obvious that the described embodiments are only some embodiments of the present specification, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.
The outdoor operation safety monitoring method based on the indoor and outdoor positioning system of the embodiment specifically comprises the following steps:
s101, accurately judging the position distribution of the staff according to real-time monitoring data of the working area and combining indoor and outdoor positioning technologies.
And acquiring real-time monitoring data through the high-resolution camera. And identifying the humanoid outline in the video by utilizing the YOLOv3 algorithm, filtering out the characteristics of the workers and distinguishing the workers from the background. And (3) acquiring the position information of the personnel by adopting an indoor and outdoor positioning technology, and matching the object detection result with the personnel positioning data. The position information is corrected by a kalman filter. And combining the human shape recognition result and the corrected positioning data to confirm the accurate coordinates of each worker. And classifying the positions of all the workers according to the labels of the areas, floors or working groups, and storing classification data. And calculating the personnel density of each region by using a nuclear density estimation method, and marking the region with the density higher than a preset threshold value. Aiming at a high-density area, a Lucas-Kanade optical flow method is applied to obtain the moving path of the staff, and the mutual distance and potential conflict points of the staff are judged. Acquiring a moving track, a stay point and an interaction point of each worker in a working area; using an Apriori algorithm to determine a common behavior mode of a worker under similar environments and tasks; and predicting the future position and potential behavior of the staff according to the result of the association rule learning. And generating a final staff position distribution map through matplotlib, and displaying real-time and historical distribution of staff. For example, real-time monitoring data is acquired by a 4K camera. And identifying the humanoid outline in the video by utilizing the YOLOv3 algorithm, filtering out the characteristics of the workers and distinguishing the workers from the background. And (3) acquiring the position information of the personnel by adopting an indoor and outdoor positioning technology, and matching the object detection result with the preliminary positioning data. The sensor data and the predicted position are fused by using a kalman filter. And combining the human shape recognition result and the corrected positioning data, confirming the accurate coordinates of each worker, and determining the coordinates of a certain worker as (5, 8) meters. The position of each worker is classified according to the labels of the areas, floors or working groups, and classification data is stored, such as classifying a worker into an area B, a building 2 and a production working group. The nuclear density estimation method is used to calculate the personnel density of each area, such as the "B area, the 2 building" personnel density of 1 person/square meter. Areas with densities above a preset threshold are marked, such as "B area, floor 2" as high density areas. Aiming at a high-density area, a Lucas-Kanade optical flow method is applied to obtain the moving path of the staff, the mutual distance and the potential conflict point of each staff are judged, the distance between the two staff is 2 meters according to the analysis of the optical flow method, and the potential conflict point exists. And obtaining the moving track, the stay point and the interaction point of each worker in the working area to obtain the record that the moving track of the worker A in the 'B area and the moving track of the worker B building 2' are [ (5, 8), (7, 9), (9, 0) ], the stay point is [ (7, 9) ], and the interaction point with the worker B is [ (7, 9) ]. The method comprises the steps of determining a common behavior pattern of a worker under similar environments and tasks by using an Apriori algorithm, and analyzing the common behavior pattern of the worker in a ' B area and a ' 2 building ' according to the Apriori algorithm to move to (7, 9) to stay and then move to other positions. And predicting the future position and potential behavior of the staff according to the result of the association rule learning, obtaining the association rule obtained according to the learning, predicting the position where the staff A is likely to move to (0, 1) next, and possibly interacting with the staff C. Generating a final position distribution diagram of the staff through matplotlib, displaying real-time and historical distribution of the staff, generating a diagram, wherein the horizontal axis represents time, the vertical axis represents position coordinates, and displaying the movement track and the stay points of the staff through connecting lines and mark points.
S102, dynamically analyzing the movement track of the staff by utilizing the output data of the indoor and outdoor positioning technology and combining time sequence analysis, and predicting the movement range of the staff in a short time.
And acquiring indoor movement data and outdoor movement data of the staff by using an indoor and outdoor positioning technology. And combining the indoor and outdoor movement data to obtain a complete movement track. And cleaning the data, removing abnormal values or repeating the data, and performing time normalization processing. Determining a long-term movement trend of a worker by using an exponential smoothing method, and identifying a repeated movement mode in a specified time period; according to the extracted activity modes and trends, performing model training by using a K-means algorithm, and identifying a main activity area; the main activity points of the staff in the appointed time period are determined according to the stay time and the access frequency of each activity area. According to the long-term movement trend of the staff, the ARIMA algorithm is utilized to predict the movement trend of the staff in a short time. And determining the short-term activity range of the staff by combining the prediction result and the known movement track. For example, there is a large office building in which it is desired to analyze the movement track and the movement range of a worker in and out of the room. The position data of each worker can be acquired by using indoor and outdoor positioning technology. Indoor movement data is acquired, and a worker moves to a position A and a position B in an office. Outdoor mobile data are acquired, and workers move to the position X and the position Y outside the building. The indoor and outdoor movement data are combined to obtain a complete movement track, so that the staff member moves to the position a in the office first in the time period 1 and then moves to the position X outdoors in the time period 2. And cleaning abnormal values or repeated data in the mobile data, so as to ensure the accuracy and the integrity of the data. The movement data of different time periods are subjected to normalization processing, such as converting the time of the time period 1 and the time period 2 into a unified time unit, such as minutes. And determining the long-term movement trend of the staff by using an exponential smoothing method, for example, performing exponential smoothing on movement data of the staff in different time periods to obtain the movement trend of each time period. By analyzing the movement data of the staff member, the staff member is found to move around the same position every day within a certain period of time, and can be identified as a repeated movement pattern. And (3) performing model training by using a K-means algorithm, and identifying main active areas of the staff, such as performing cluster analysis on the movement data of the staff to obtain different active areas. And determining the main activity points of the staff in a specified time period according to the stay time and the access frequency of each activity area, so that the main activity points of the staff in a certain time period are determined to be the position A and the position X by analyzing the stay time and the access frequency of the staff in different activity areas. And analyzing the long-term movement trend of the staff by using an ARIMA model, and predicting the following short-term movement trend. In combination with the predicted result and the known movement trajectory, a short-term movement range of the worker is determined, and it is possible to determine that the worker may move near the position B and the position Y in the next period of time, for example, based on the predicted movement tendency and the known movement trajectory.
S103, constructing a working area risk assessment model according to the predicted activity range and indoor and outdoor positioning data of the staff, and analyzing whether the object falling risk exists.
And obtaining complete positioning data of the staff through the indoor positioning data and the outdoor positioning data. And according to the positioning data of the staff, a long-term and short-term memory network algorithm is applied to predict the action route of the staff at the next time. Intersection analysis is performed on the predicted action route and the high risk area is determined, wherein the high risk area possibly enters the high risk area, and the high risk area comprises a landing entrance, a window edge, an edge and a high altitude area. The activity state of the staff in the high-risk area is tracked in real time by using a sensor or video monitoring. When a worker enters a high risk area, the probability of dropping an object is evaluated in real time. Three-dimensional model data of a building site are acquired, and dangerous points including barriers and suspended areas are identified. And evaluating the falling or other risks of the object by combining the moving track of the worker with the dangerous points. And constructing a working area risk assessment model by using a support vector machine according to the spatial structure data of the dangerous points, the worker activity and the historical accident data, and carrying out risk assessment, wherein the risk assessment comprises high risk, medium risk and low risk. If a high risk is predicted, an alarm is given in real time and an event is recorded. The criteria and parameters of risk assessment are adjusted and optimized periodically based on the actual events that occur. For example, the next walking route of the worker is predicted from the movement locus data of the worker. Based on historical data analysis, workers tend to move from office area to work area and then to storage area during a certain period of time. And determining the current position of the staff according to the indoor positioning data and the outdoor positioning data. And determining that the staff is currently in a certain area through Wi-Fi signal intensity and Bluetooth signal intensity analysis. And predicting the action route of the staff in the next time by using a long-term and short-term memory network algorithm. Based on the course of action and time in the past hour, it is predicted that the staff will go to the office area in the next half hour. A likely-entry high-risk region is determined based on an intersection analysis of the predicted course of action and the high-risk region. And identifying the elevator entrance, window edges, edges and high-altitude areas as high-risk areas according to the three-dimensional model data of the building site. The activity state of the staff in the high-risk area is tracked in real time by using a sensor or video monitoring. And detecting whether the staff enters a high risk area through video monitoring, and recording the activity state. When a worker enters a high risk area, the probability of dropping an object is evaluated in real time. Whether the worker approaches the suspended area or not is detected through the sensor, and the falling probability of the object is estimated by combining the weight and the height of the object. And evaluating the falling or other risks of the object by combining the moving track of the worker with the dangerous points. And evaluating the possibility of falling of the object according to the moving track of the worker and the position of the suspended area. And constructing a working area risk assessment model by using a support vector machine, and carrying out risk assessment. And establishing a risk assessment model according to the spatial structure data of the dangerous points, the worker activities and the historical accident data, and dividing the working area into high-risk, medium-risk and low-risk areas. If a high risk is predicted, an alarm is given in real time and an event is recorded. When it is predicted that a worker will enter a high risk area, an alarm is sent to notify the relevant person and record an event. The criteria and parameters of risk assessment are adjusted and optimized periodically based on the actual events that occur. And according to the actually-occurring object falling event, adjusting and optimizing the object falling probability and the dangerous point position in the risk assessment model.
S104, judging whether an up-down operation relationship exists between the staff by using an analysis result of the object falling risk and combining indoor and outdoor positioning data, and monitoring the possibility of object falling in real time.
And obtaining the risk rating of each working area by using the working area risk assessment model. And acquiring the real-time position of the staff through indoor and outdoor positioning data. And constructing an input data set according to the risk rating and the positions of the staff, dividing the working area into an upper layer and a lower layer, and marking the hierarchy of each staff according to the real-time position data. And judging whether the working personnel between the upper layer and the lower layer have an upper-lower operation relationship by using a KNN algorithm. And continuously monitoring the movement condition of the object in the working area by adopting a real-time monitoring mechanism, identifying the movement of the object in real time by adopting an optical flow method, and predicting the falling point of the object by the current position, the speed and the acceleration of the object. And judging whether the possibility of falling the object and the staff possibly influenced are present according to the predicted falling point and the position of the staff. If the position distance between the predicted falling point of the object and the worker is smaller than a preset threshold value, marking the object as high risk. And according to the predicted falling point and the position data of the staff, updating the safety state of the staff by combining the risk rating. If a high risk state is detected, generating a complete risk assessment report according to all information related to the detected high risk state by using an automatic summarization technology of NLP. For example, a working area risk assessment model is used to assess a working area, so that the risk rating of the working area is 3, and the risk rating is from 1 to 5, and 5 is the highest risk. And obtaining real-time positions (20, 30) of the staff A according to the indoor and outdoor positioning data. And constructing an input data set according to the risk rating and the position of the staff, and dividing the working area into an upper layer and a lower layer. If the working area of the upper layer is the coordinate ranges (0, 0) to (50, 50), the working area of the lower layer is the coordinate ranges (0, 0) to (100 ); and judging the level of the staff A by using a KNN algorithm. And calculating the distance between the worker A and the upper and lower layers, wherein if the distance between the upper layer and the worker A is 10, the distance between the lower layer and the worker A is 5. According to the nearest neighbor algorithm, staff member a is closer to the lower floor and is therefore marked as the lower floor staff member. And a real-time monitoring mechanism is adopted, and an optical flow method is used for monitoring the movement condition of the object. If the optical flow vector of one object is detected to be (2, -1), the speed of the object in the X-axis direction is 2, the speed in the Y-axis direction is-1, and the falling point of the object is predicted according to the current position, the speed and the acceleration of the object. If the current position of the object is (50, 40), the speed is (2, -1), the acceleration is (0, 0), and the position of the object at the next moment is predicted to be (52,39) according to the speed and the acceleration of the object. And judging whether the possibility of falling the object and the staff possibly influencing the falling exist according to the predicted falling point and the position of the staff A, and if the distance between the predicted falling point and the position of the staff A is 8 and is smaller than a preset threshold value 10, considering that the risk of falling the object possibly exists. And updating the safety state of the staff member A according to the predicted falling point and the position data of the staff member A and combining the risk rating, so that the safety state of the staff member A is marked as high risk. If a high risk state is detected, generating a complete risk assessment report according to all information related to the detected high risk state by using an automatic summarization technology of NLP. The generated report comprises information such as risk rating of 3 in the working area, risk of falling of objects and the like, and the staff A is located at the lower layer.
Predicting a dynamic blind area to be formed and giving a warning to a worker.
Real-time data streams in the work environment are acquired, including the location, speed, and trajectory of equipment and personnel. And detecting the activity or the object which possibly generates the dynamic blind area in real time by using the SORT tracking algorithm, and obtaining the real-time record of the dynamic blind area generation factors. And according to the real-time record of the dynamic blind area generating factors, predicting the dynamic blind area possibly formed in the future based on the past data by utilizing a long-short-period memory network algorithm to obtain a prediction graph of the dynamic blind area. And reconfiguring task allocation of the workflow or personnel according to the prediction graph of the dynamic blind area and the current workflow to obtain the optimized workflow. And according to the prediction graph of the dynamic blind area, if a worker or equipment enters the predicted dynamic blind area or the distance between the worker or equipment and the predicted dynamic blind area is smaller than a preset threshold value, immediately giving out a warning. For example, a working area risk assessment model is used to assess a working area, so that the risk rating of the working area is 3, and the risk rating is from 1 to 5, and 5 is the highest risk. And obtaining real-time positions (20, 30) of the staff A according to the indoor and outdoor positioning data. And constructing an input data set according to the risk rating and the position of the staff, and dividing the working area into an upper layer and a lower layer. If the working area of the upper layer is the coordinate ranges (0, 0) to (50, 50), the working area of the lower layer is the coordinate ranges (0, 0) to (100 ); and judging the level of the staff A by using a KNN algorithm. And calculating the distance between the worker A and the upper and lower layers, wherein if the distance between the upper layer and the worker A is 10, the distance between the lower layer and the worker A is 5. According to the nearest neighbor algorithm, staff member a is closer to the lower floor and is therefore marked as the lower floor staff member. And a real-time monitoring mechanism is adopted, and an optical flow method is used for monitoring the movement condition of the object. If the optical flow vector of one object is detected to be (2, -1), the speed of the object in the X-axis direction is 2, the speed in the Y-axis direction is-1, and the falling point of the object is predicted according to the current position, the speed and the acceleration of the object. If the current position of the object is (50, 40), the speed is (2, -1), the acceleration is (0, 0), and the position of the object at the next moment is predicted to be (52,39) according to the speed and the acceleration of the object. And judging whether the possibility of falling the object and the staff possibly influencing the falling exist according to the predicted falling point and the position of the staff A, and if the distance between the predicted falling point and the position of the staff A is 8 and is smaller than a preset threshold value 10, considering that the risk of falling the object possibly exists. And updating the safety state of the staff member A according to the predicted falling point and the position data of the staff member A and combining the risk rating, so that the safety state of the staff member A is marked as high risk. If a high risk state is detected, generating a complete risk assessment report according to all information related to the detected high risk state by using an automatic summarization technology of NLP. The generated report comprises information such as risk rating of 3 in the working area, risk of falling of objects and the like, and the staff A is located at the lower layer.
And S105, immediately sending a safety warning message to relevant staff when the object falling risk is detected.
And acquiring past object drop event data, including environmental parameters, object characteristics, cleaning, normalization and labeling data. And training an object falling risk detection model by using a support vector machine algorithm according to the processed data. And acquiring environmental data in real time by using a high-precision sensor, and obtaining the object falling risk level according to the object falling risk detection model. And triggering a safety warning generation flow when the object falling risk level exceeds the preset threshold value. By accessing the staff database, the contact information of the staff possibly affected is obtained. Using a push service, a security alert is delivered to a target staff member based on the acquired contact details. If communication disorder occurs, the system monitors the communication state and automatically retries sending the warning until success or the preset number of attempts is reached. For example, high-precision sensors are installed in a working area to monitor environmental parameters such as wind speed, temperature, humidity, illumination in real time. Cameras or other sensors are mounted to monitor the position and characteristics of the object, such as size, weight. Using the data recording device, the past object falling event is recorded, including information such as time, place, object characteristics, environmental parameters and the like of the event. Cleaning data, deleting abnormal values or incomplete records, and normalizing the data to ensure that the data of different sensors have consistent scales and units. Each event is labeled with a risk level, which can be classified into low, medium, and high levels. And training an object falling risk detection model by using a support vector machine algorithm, wherein the input comprises environmental parameters and object characteristics, and the output is a risk level. The data set is divided into a training set and a test set for evaluating the model performance. And deploying a real-time monitoring system, continuously collecting environment data, and inputting the environment data into a trained SVM model to obtain the falling risk level of the object. And setting a preset threshold value, and triggering a safety warning generation flow if the risk level exceeds the threshold value. A staff database is accessed and based on the monitored risk level, which staff members are likely to be affected is determined. The contact information of the target staff, such as an email address or a mobile phone number, is obtained. Using the push service, a safety alert is sent to the target staff, including event information and suggested actions. If communication disorder occurs, the system automatically retries sending the warning until success or the preset number of attempts is reached, so as to ensure that the staff can obtain the warning information in time.
S106, adopting time sequence analysis, predicting future working contents of each worker by using the data of the indoor and outdoor positioning technology and the working range prediction of the workers, and judging the safety influence of the workers on other people.
Acquiring historical activity range data of a worker, processing the historical data by using an ARIMA algorithm, and extracting key features including activity duration, activity range size and frequency; and acquiring real-time position information of the staff by using an indoor and outdoor positioning technology. And fusing key features of the real-time position data and the historical activity data, constructing a worker behavior model by using a cyclic neural network algorithm, and predicting future work content of workers. And drawing the future activity track of each worker according to the prediction result. And determining a space crossing point between the workers according to the future activity track of each worker, acquiring the interaction relation of the workers and evaluating the complexity of the interaction, wherein the crossing point is a commonly used tool or equipment area. And calculating the safety influence index of each worker on other people according to the interaction complexity and the safety evaluation factors of the workers. The staff is classified as safe or potentially dangerous according to the safety impact index. When the staff status is a potential risk, a safety suggestion is automatically generated, and the safety suggestion is issued to related staff through a communication channel, for example, historical activity range data of the staff A is obtained, and the historical activity range data is processed by using an ARIMA algorithm. In the analysis, the key characteristic activity duration was extracted, and from the history data, the average activity duration of the worker a was calculated to be 5 hours. From the historical data, it was determined that the range of motion of staff member A was on average 500 square meters. From the historical data, it was calculated that worker a performed 10 activities on average per day. Meanwhile, real-time position information of the staff A is acquired by using an indoor and outdoor positioning technology, and is fused with key features of historical activity data. Through a cyclic neural network algorithm, a behavior model of the staff A is constructed to predict future working contents. From the prediction result, the future activity trajectory of the worker a may be plotted. If predicted from the model, staff member A will work in office A during 2 to 4 pm and then go to office B during 30 minutes from 4 to 6. From the future activity trajectory of the staff member a, a spatial intersection between the staff members can be determined, and thus, the a office area and the B office area will be intersections of the staff members a and B, because they will work in these areas during the same period of time, which may be represented as commonly used tools or equipment areas. In order to evaluate the interaction complexity, the interaction complexity of the staff and other safety evaluation factors are considered, the safety influence index of the staff A on other people is calculated, and if the safety influence index of the staff A is 0.8. The staff is classified as safe or potentially dangerous according to the safety impact index. If the safety impact index of staff A is above a certain threshold value of 0.7, it is classified as a potential risk. When a worker is classified as a potential risk, the system automatically generates a corresponding safety suggestion and issues the safety suggestion to the relevant worker through a communication channel. For staff a, the system may generate advice asking him to use protective equipment in the intersection area to reduce the potential risk.
And calculating the safety influence index of each worker on other people according to the positions of the workers, the interaction records and the working property data.
Determining interaction complexity according to the work dependence among the staff and the frequency of mutual contact; acquiring safety evaluation factors of workers, including spatial proximity, working properties and historical safety records, wherein the spatial proximity is determined through the physical distance between the workers; weight is distributed according to the influence degree of each factor on the safety; acquiring the position, interactive record and working property data of a worker; for each worker, normalizing the obtained value to be in the range of 0-1 according to the formula safety impact index = Σ (evaluation factor value x weight), and obtaining the safety impact index of each worker. For example, there is a large plant where the dependency of work and the frequency of contact between workers is such that each day, worker a and worker B, need to work closely together, their interaction complexity assessment factor is 5. Staff a only occasionally comes into contact with staff C, and their interaction complexity assessment factor value is 2. Staff B and staff C need to be exposed daily and their interaction complexity assessment factor value is 4. According to the evaluation factor values, the interaction complexity between each pair of staff can be determined, and weights are distributed, wherein the interaction complexity evaluation factor value is 5, the weight is 4, the interaction complexity evaluation factor value is 4, the weight is 3, and the interaction complexity evaluation factor value is 2, and the weight is 2; next, security assessment factors of the staff, including spatial proximity, performance properties, historical security records, need to be obtained. If the spatial proximity evaluation factor value of the worker A is 3, the working property evaluation factor value of the worker A is 4, and the historical security record evaluation factor value of the worker A is 5; according to the influence degree of each factor on the safety, the weight is that the spatial proximity evaluation factor value is 3, the weight of the working property evaluation factor value is 4, the weight of the historical safety record evaluation factor value is 5 and the weight of the historical safety record evaluation factor value is 3; the spatial proximity can be determined according to the physical distance between the staff members, and if the physical distance between the staff member a and the staff member B is 10 meters, the physical distance between the staff member a and the staff member C is 20 meters. Next, the safety impact index of each worker is calculated from the formula safety impact index= Σ (evaluation factor value×weight). Finally, normalizing the obtained safety impact index to be in the range of 0-1. If the maximum safety impact index is 10 and the minimum safety impact index is 2, the normalized safety impact index is 0.571 for the worker A; through the calculation, the safety influence index of each worker can be obtained, and the influence degree of the worker on the safety in work can be known.
And S107, performing resource allocation and personnel scheduling on the working area according to future working contents, security influence and indoor and outdoor positioning data of the staff.
And acquiring all the working contents, distributing weights for each working content according to the importance and the emergency degree of the working contents, and sequencing according to the importance and the emergency degree. According to indoor and outdoor positioning data and safety influence data, a K-means algorithm is used for distributing a safety level label for each working area, wherein the safety level label comprises high risk, medium risk and low risk; and allocating a priority to each resource according to the working content weight and the security level label of the working area. And according to the classification of the working contents and the weight thereof and the security level label of the working area, the priority of the resources is ordered, and the priority of the resources is determined, wherein the priority comprises high priority, medium priority and low priority. High priority resources are allocated to the corresponding work areas and a match with the security level tag is ensured. If the resources of a certain high priority area are insufficient, the medium priority area or the low priority area is allocated. And determining the initial position of the staff through indoor and outdoor positioning data. And distributing specific tasks to each worker according to the initial positions of the workers and the weights of the work contents. And scheduling the staff according to the task allocation result to ensure the matching between the staff and the resources. For example, the evaluation results of the work content, the importance thereof and the degree of emergency are that the importance of the maintenance equipment is 8, and the degree of emergency is 9; the importance of data analysis is 7, the importance of training new staff is 6, the importance of financial statement establishment is 4, the importance of financial statement establishment is 9, and the emergency degree is 5; each work content may be assigned a weight based on the assessment of importance and urgency. And according to the indoor and outdoor positioning data and the security influence data, a security level label is allocated to each working area by using a K-means algorithm. If there are 3 working areas, their security level labels are high risk, medium risk and low risk, respectively. Next, each resource is assigned a priority based on the work content weight and the security level label of the work area. If the resource A is in the working area 1, the security level is high risk, the resource B is in the working area 2, the security level is medium risk, the resource C is in the working area 3, and the security level is low risk; the resources can be prioritized according to the classification of the work content, the weight of the work content and the security level label of the work area. The result of the sorting is that the resource A is high priority, is a high risk area, the resource B is medium priority, is a medium risk area, the resource C is low priority, and is a low risk area; according to the prioritization, high priority resources can be allocated to the corresponding work areas and ensure matching with the security level tag. If the resources of a certain high priority region are insufficient, the resources can be allocated from the middle priority or low priority region. And finally, determining the initial position of the staff according to the indoor and outdoor positioning data. And distributing specific tasks to each worker according to the initial positions of the workers and the weights of the work contents. If the staff members and the initial positions thereof are that the initial position of the staff member X is a working area 1, the initial position of the staff member Y is a working area 2, and the initial position of the staff member Z is a working area 3; according to the task distribution result, the staff can be scheduled, and the matching between the staff and the resources is ensured. Therefore, the worker X can perform maintenance equipment tasks, the worker Y can perform data analysis tasks, and the worker Z can perform training new staff tasks.
And evaluating the matching degree of the working content and the skill of the staff, and carrying out fine adjustment according to the emergency degree and the importance of the work.
Specifically, a skill assessment table for each worker is obtained and its integrity is verified. The three staff members A, B and C are skill assessment tables for staff member a, skill 1:7/10, skills 2:8/10, skills 3:6/10; skill assessment table, skill 1, for staff B: 9/10, skills 2:5/10, skills 3:8/10; skill assessment table, skill 1 for staff C: 6/10, skills 2:7/10, skills 3:9/10; and scoring each skill in the skill assessment table according to the actual application frequency and importance of each skill, if the importance of the skill 1 is 4, the importance of the skill 2 is 3, and the importance of the skill 3 is 3, calculating the comprehensive skill score of each worker, and determining the required skill type and level according to each working content. Setting skill requirement indexes including experience years and certificates, and setting weights for the skill requirement indexes, if the work content 1 needs a skill type A and a skill type B, wherein the importance of the skill type A is 6, the importance of the skill type B is 4, and the experience years and the certificates are respectively set with weights of 7 and 3. According to skill indexes and weights required by the working contents, calculating the skill requirement degree of the working contents, and if the requirement degree of the working contents 1 on the skill type A is 8 and the requirement degree on the skill type B is 6, the skill requirement degree of the working contents 1 is (8*6) + (6*4) =48+24=72; the skill score of each worker and the skill requirement of each work content are extracted from the database, and if the skill score of the worker A is 7, the skill score of the worker B is 75, and the skill score of the worker C is 72. The efficiency of the staff completing similar work content in the past is evaluated, the history efficiency is calculated, if the history efficiency of the staff A to the work content 1 is 9, the history efficiency of the staff B is 8, and the history efficiency of the staff C is 85. Using the formula M (i, j) =s (i) ×t (j) ×e (i, j), a matching degree is calculated for each worker and work content. Therefore, the matching degree between the worker a and the work content 1 is 7×72×9=4536, and the calculated matching degree result is stored in a matrix and visualized in the system. And distributing the working content which is most matched with the skill of each worker to each worker based on the matching degree matrix. And fine tuning is performed according to the emergency degree and importance of the work, a work task list of each worker is generated in the system, and a work task notification is sent to the worker. A skill assessment table for each worker is obtained and its integrity is verified. The three staff members A, B and C are skill assessment tables for staff member a, skill 1:7/10, skills 2:8/10, skills 3:6/10; skill assessment table, skill 1, for staff B: 9/10, skills 2:5/10, skills 3:8/10; skill assessment table, skill 1 for staff C: 6/10, skills 2:7/10, skills 3:9/10; and scoring each skill in the skill assessment table according to the actual application frequency and importance of each skill, if the importance of the skill 1 is 4, the importance of the skill 2 is 3, and the importance of the skill 3 is 3, calculating the comprehensive skill score of each worker, and determining the required skill type and level according to each working content. Setting skill requirement indexes including experience years and certificates, and setting weights for the skill requirement indexes, if the work content 1 needs a skill type A and a skill type B, wherein the importance of the skill type A is 6, the importance of the skill type B is 4, and the experience years and the certificates are respectively set with weights of 7 and 3. According to skill indexes and weights required by the working contents, calculating the skill requirement degree of the working contents, and if the requirement degree of the working contents 1 on the skill type A is 8 and the requirement degree on the skill type B is 6, the skill requirement degree of the working contents 1 is (8*6) + (6*4) =48+24=72; the skill score of each worker and the skill requirement of each work content are extracted from the database, and if the skill score of the worker A is 7, the skill score of the worker B is 75, and the skill score of the worker C is 72. The efficiency of the staff completing similar work content in the past is evaluated, the history efficiency is calculated, if the history efficiency of the staff A to the work content 1 is 9, the history efficiency of the staff B is 8, and the history efficiency of the staff C is 85. Using the formula M (i, j) =s (i) ×t (j) ×e (i, j), a matching degree is calculated for each worker and work content. Therefore, the matching degree between the worker a and the work content 1 is 7×72×9=4536, and the calculated matching degree result is stored in a matrix and visualized in the system. And distributing the working content which is most matched with the skill of each worker to each worker based on the matching degree matrix. And fine tuning is performed according to the emergency degree and importance of the work, a work task list of each worker is generated in the system, and a work task notification is sent to the worker.
S108, performing obstacle detection on the working area of each worker based on resource allocation, personnel scheduling output and indoor and outdoor positioning technologies, and avoiding working interference and object collision risks.
And acquiring resource allocation data, and allocating tasks and working areas for each worker according to the quantity and the types of resources required by each working content. Acquiring real-time position information of a worker by using a sensor of an indoor and outdoor positioning technology; pre-training a model by using a YOLO algorithm according to the image in the working area, and identifying obstacles in the working area; the working area is scanned using a pre-trained YOLO model, obstacles within the working area are detected, and their locations are marked. Setting an initial state for the Kalman filter, including a current position and velocity estimate; using a kalman filter, based on the estimated position and speed of the worker or the obstacle at the last time and the estimated position and speed of the worker or the obstacle at the current time, predicting the positions of the worker and the obstacle in the future, and obtaining a predicted trajectory or path, displaying the direction and speed of the intended movement of the worker and the obstacle. If the distance between the staff and the obstacle in the predicted track is smaller than a preset safety threshold value, the possible collision risk is judged. And if the risk of a certain working area exceeds the safety threshold, re-allocating the resources of the working area. If the resource redistribution is not feasible, the task positions and the task time of the staff are rearranged according to the risk assessment result, so that the high risk area is avoided. When the system predicts a collision risk, warning notification is sent to the relevant staff immediately by means of sound, vibration or visual notification. For example, there is a workplace with 10 workers and 5 work areas, each of which requires a different type and amount of resources to complete a work task. The working area 1 requires 2 laser ranging sensors and 3 camera sensors. The working area 2 requires 1 temperature sensor and 2 humidity sensors. The working area 3 requires 4 light sensors and 2 sound sensors. The working area 4 requires 3 pressure sensors and 1 gas sensor. The working area 5 requires 5 current sensors and 4 voltage sensors. The position information of each staff can be acquired in real time by using the sensor of the indoor and outdoor positioning technology. If the worker 1 is currently located in the work area 2, the worker 2 is located in the work area 4, the worker 3 is located in the work area 1, and so on. Images of each work area are scanned and analyzed using a pre-trained YOLO model to identify obstacles. In the working area 1, the YOLO algorithm detects an obstacle and marks its position. An initial state is set for the Kalman filter, including a current position of the worker and a velocity estimate of 2 m/s, resulting in a current position of the worker 1 of (1, 1). The future position of the staff member is predicted based on the position and velocity estimation at the previous time and the position and velocity estimation at the current time using a kalman filter, and therefore, the staff member 1 is predicted to be located at the next time (2, 3) based on the above-described initial state and calculation of the kalman filter. By calculating the predicted positions of the staff and the obstacle, a predicted trajectory or path can be obtained and the expected movement direction and speed of the staff and the obstacle are displayed, and if the distance between the staff 1 and the obstacle in the predicted trajectory is smaller than the preset safety threshold, namely 5 meters, the possible collision risk can be determined. If the risk of a certain working area exceeds the safety threshold, if the collision risk of the working area 1 exceeds the threshold, the resource allocation of the working area can be carried out again. If resource reallocation is not feasible, the task locations and times of the staff members may be rearranged based on the risk assessment results to ensure that high risk areas are avoided, and staff member 1 is rearranged from work area 2 to work area 5. When the system predicts a collision risk, a warning notice is sent to the relevant staff immediately through sound, vibration or visual notice, and then the warning notice is sent through the mobile phone application program of the staff 1 to remind the staff to avoid the collision risk.
S109, adjusting the resource allocation of the working area and optimizing the workflow by combining the obstacle detection result, the movable range of the staff and the indoor and outdoor positioning data.
And marking and classifying the obstacles in the video or the image by using a pre-trained YOLO model according to the indoor and outdoor video streams or image data to obtain the indoor and outdoor obstacle positions and types. And determining the real-time position and the activity range of each worker according to the indoor and outdoor sensor data. And calculating available space and resources in the working area according to the positions of the obstacles and the moving ranges of the staff, comparing the available space and resources with the moving ranges of the obstacles and the staff, and determining the utilization rate of the resources. And according to the resource utilization data, comparing the preset threshold values, and determining whether the current resource allocation is reasonable. If the current resource allocation is unreasonable, a linear programming algorithm is used according to the current resource allocation data, and the resource allocation is adjusted according to the current obstacle distribution and the working personnel activity range, so that a new resource allocation scheme is obtained. And simulating a workflow under new resource allocation to determine whether the efficiency problem exists. If the workflow has efficiency problems, the workflow is adjusted according to the positioning of the staff and the distribution of the obstacles. And according to the working range of the staff and the new workflow scheme, simulating the workflow by using a workflow simulation tool Arena, and calculating efficiency indexes to obtain the efficiency evaluation of the new workflow. And comparing the efficiency and the risk of the two workflows according to the old workflow, the new workflow and the efficiency evaluation result to obtain the effect evaluation of workflow optimization. For example, there is an indoor work area with a plurality of workers and obstacles inside. And analyzing the video of the working area by using a pre-trained YOLO model to obtain the position and type of the obstacle. If there are 3 obstacles in the video, their positions are (1, 1), (2, 2) and (3, 3), respectively. According to the indoor sensor data, the real-time position and the moving range of each worker can be determined, so that two workers are obtained, the positions of the workers are respectively (0, 0) and (2, 2), and the moving range is a circular area with the radius of 2. According to the position of the obstacle and the activity range of the staff, the available space and resources in the working area can be calculated, if the size of the working area is 5x5, and the working area comprises 3 obstacles and the activity range of 2 staff, the available space is 16 units, the resource utilization rate is (25-16)/25=36%, whether the current resource allocation is reasonable is judged according to the comparison of the resource utilization rate and a preset threshold value, and if the threshold value is 70%, the current resource allocation is unreasonable. And according to the current resource allocation data and the obstacle distribution and the working range of the staff in the working area, using a linear programming algorithm to adjust the resource allocation to obtain a new resource allocation scheme, and if the adjusted resource allocation scheme is to finely adjust the working range of the staff, so that the adjusted resource allocation scheme is not overlapped with the obstacle. And according to the new resource allocation scheme, simulating a workflow, determining whether an efficiency problem exists, and if the simulation result shows that the staff has no efficiency problem under the new resource allocation. And according to the positioning of the staff and the distribution of the obstacles, carrying out workflow adjustment. The adjusted workflow is to rearrange the order of certain tasks to avoid collisions between workers and obstacles. According to the working range of the staff and the new workflow scheme, the workflow simulation tool Arena is used for simulation, and the efficiency index is calculated, if the efficiency index of the new workflow is 10 tasks completed per hour. And comparing the efficiency and risk of the two workflows according to the old workflow, the new workflow and the efficiency evaluation result to obtain the effect evaluation of workflow optimization. If the new workflow is 20% improved in efficiency compared with the old workflow, the efficiency problem is solved and the risk is reduced.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (10)
1. An outdoor operation safety monitoring method based on an indoor and outdoor positioning system is characterized by comprising the following steps:
according to real-time monitoring data of a working area, combining indoor and outdoor positioning technologies, accurately judging the position distribution of workers; the output data of the indoor and outdoor positioning technology is utilized, and the movement track of the staff is dynamically analyzed by combining time sequence analysis, so that the activity range of the staff in a short time is predicted; according to the predicted activity range and indoor and outdoor positioning data of the staff, constructing a working area risk assessment model, and analyzing whether an object falling risk exists or not; using an analysis result of the falling risk of the object, combining indoor and outdoor positioning data, judging whether an up-down operation relationship exists between workers, and monitoring the falling possibility of the object in real time; when the falling risk of the object is detected, immediately sending safety warning information to related staff; adopting time sequence analysis, predicting future working contents of each worker by using data of indoor and outdoor positioning technology and the aforementioned worker activity range prediction, and judging the safety influence of each worker on other people; according to future working contents, safety influence and indoor and outdoor positioning data of the staff, resource allocation and staff scheduling are carried out on the working area; based on resource allocation, personnel scheduling output and indoor and outdoor positioning technologies, obstacle detection is carried out on the working area of each worker, so that working interference and object collision risks are avoided; and adjusting the resource allocation of the working area and optimizing the working flow by combining the obstacle detection result, the movable range of the staff and the indoor and outdoor positioning data.
2. The method of claim 1, wherein the accurately determining the position distribution of the staff member according to the real-time monitoring data of the working area and the indoor and outdoor positioning technology comprises:
acquiring the real-time monitoring data by adopting a high-resolution camera; the identification of the human outline in the video is obtained through a YOLOv3 algorithm, and the characteristics of the working personnel are filtered, so that the distinction between the personnel and the background is realized; acquiring personnel position information produced by an indoor and outdoor positioning technology, and matching with an object detection result; correcting the position information through a Kalman filter, and determining the accurate coordinates of the staff; judging the specific position of each worker according to the human shape recognition result and the corrected positioning data, classifying according to the region, the floor or the working group label, and storing the classification data; calculating personnel density of each region by using a nuclear density estimation method, and marking the region exceeding a preset threshold value; obtaining a moving path of the staff by adopting a Lucas-Kanade optical flow method, and judging the mutual distance and potential conflict points between the staff; recording a moving track, a stay point and an interaction point of the staff in a working area; determining a common behavior pattern of the staff under similar environments and tasks by using an Apriori algorithm; predicting future positions and potential behaviors of the staff according to the result of association rule learning; finally, the position distribution map of the staff is displayed through a matplotlib tool, wherein the position distribution map comprises real-time and historical data.
3. The method of claim 1, wherein the outputting data using the indoor and outdoor positioning technology, in combination with time series analysis, dynamically analyzes the movement track of the staff member, predicts the movement range thereof in a short time, comprises:
acquiring indoor and outdoor movement data of the staff by adopting an indoor and outdoor positioning technology; the complete moving track of the staff is obtained by combining indoor and outdoor moving data; cleaning the data, removing abnormal or repeated data, and performing time standardization treatment; determining the movement trend of the staff by using an exponential smoothing method, and repeating the movement mode in a designated time period; according to the extracted activity modes and trends, training a model by using a K-means algorithm, and identifying a main activity area; judging main activity points of staff in a specified time period according to the residence time and the access frequency of each activity area; predicting the movement trend of the staff in a short time by utilizing an ARIMA algorithm and combining the long-term movement trend; and determining the short-term activity range of the staff by combining the predicted result and the known movement track.
4. The method of claim 1, wherein the constructing a work area risk assessment model according to the predicted activity range of the staff and the indoor and outdoor positioning data, and analyzing whether there is a risk of dropping the object, comprises:
The complete positioning data of the staff is obtained through the indoor positioning data and the outdoor positioning data; according to the positioning data of the staff, a long-term and short-term memory network algorithm is applied to predict the action route of the staff at the next time; performing intersection analysis on the predicted action route and a high risk area to determine a possible high risk area, wherein the high risk area comprises a landing entrance, a window edge, an edge and a high altitude area; real-time tracking the activity state of the staff in the high risk area by using a sensor or video monitoring; when a worker enters a high risk area, the probability of dropping an object is evaluated in real time; three-dimensional model data of a building site are acquired, and dangerous points in the three-dimensional model data are identified, wherein the dangerous points comprise barriers and suspended areas; evaluating object falling or other risks by combining the moving track of the worker with the dangerous points; according to the space structure data of the dangerous points, the worker activities and the historical accident data, a support vector machine is used for constructing a working area risk assessment model, and risk assessment is carried out, wherein the risk assessment comprises high risk, medium risk and low risk; if the high risk is predicted, alarming in real time and recording an event; the criteria and parameters of risk assessment are adjusted and optimized periodically based on the actual events that occur.
5. The method of claim 1, wherein the determining whether the staff has a relationship of up-down operation by using the analysis result of the object falling risk and combining the indoor and outdoor positioning data, and monitoring the possibility of the object falling in real time comprises:
using a working area risk assessment model to obtain risk ratings of each working area; acquiring the real-time position of a worker through indoor and outdoor positioning data; constructing an input data set according to the risk rating and the positions of the staff, dividing a working area into an upper layer and a lower layer, and marking the hierarchy of each staff according to the real-time position data; judging whether a worker between the upper layer and the lower layer has an upper-lower operation relationship by using a KNN algorithm; adopting a real-time monitoring mechanism to monitor the continuous object movement condition of a working area, adopting an optical flow method to identify the movement of an object in real time, and predicting the falling point of the object according to the current position, speed and acceleration of the object; judging whether the possibility of falling of the object and the staff possibly influenced are present according to the predicted falling point and the position of the staff; if the position distance between the predicted falling point of the object and the worker is smaller than a preset threshold value, marking the object as high risk; according to the predicted falling point and the position data of the staff, the safety state of the staff is updated by combining the risk rating; if a high risk state is detected, generating a complete risk assessment report by using an automatic summarization technology of NLP according to all information related to the detected high risk state; further comprises: predicting a dynamic blind area to be formed, and giving a warning to a worker;
The prediction is about to form a dynamic blind area and gives a warning to staff, and the method specifically comprises the following steps: acquiring real-time data streams in a working environment, including positions, speeds and tracks of equipment and personnel; detecting the activity or the object which possibly generates a dynamic blind area in real time by using an SORT tracking algorithm to obtain a real-time record of the generation factors of the dynamic blind area; according to the real-time record of the dynamic blind area generating factors, predicting the dynamic blind area possibly formed in the future based on past data by utilizing a long-short-period memory network algorithm to obtain a prediction graph of the dynamic blind area; according to the prediction graph of the dynamic blind area and the current workflow, reconfiguring the task allocation of the workflow or personnel to obtain an optimized workflow; and according to the prediction graph of the dynamic blind area, if a worker or equipment enters the predicted dynamic blind area or the distance between the worker or equipment and the predicted dynamic blind area is smaller than a preset threshold value, immediately giving out a warning.
6. The method of claim 1, wherein the immediately sending the safety warning message to the relevant staff member when the risk of dropping the object is detected comprises:
acquiring and processing past object drop event data, wherein the data comprises environmental parameters and object characteristics; training an object falling risk detection model by using the data and a support vector machine algorithm; acquiring environmental data in real time by adopting a high-precision sensor, and determining the falling risk level of the object according to the risk detection model; judging whether the obtained risk level exceeds a preset threshold value, if so, determining to trigger the generation of a safety warning; obtaining the contact way of the affected staff by accessing a staff database; transmitting a safety warning to a target staff based on the acquired contact information by using a push service; in the event of a communication failure, the system will monitor the communication status and automatically attempt to resend the alert until the delivery is successful or a predetermined number of attempts is reached.
7. The method of claim 1, wherein said predicting future work content of each worker and determining its safety impact on others using time series analysis using data of indoor and outdoor location technology and the aforementioned worker's range of motion prediction comprises:
acquiring historical activity range data of the staff, and processing the data by utilizing an ARIMA algorithm to extract key characteristics including duration, range size and frequency of the activity; acquiring real-time position information of a worker through an indoor and outdoor positioning technology; fusing the key features of the real-time position data and the historical activity data, and constructing a worker behavior model by using a cyclic neural network algorithm so as to predict future work content; drawing a future activity track of each worker according to the prediction result; determining space crossing points among the staff members, evaluating the complexity of interaction, and calculating the safety influence index of each staff member on other people according to the space crossing points; classifying the staff as safe or potential risk according to the safety impact index; further comprises: according to the positions of the staff, the interaction records and the working property data, calculating the safety influence index of each staff on other people;
According to the position, interaction record and working property data of the staff, calculating the safety influence index of each staff to other people, specifically comprising: determining interaction complexity according to the work dependence among the staff and the frequency of mutual contact; acquiring safety evaluation factors of workers, including spatial proximity, working properties and historical safety records, wherein the spatial proximity is determined through the physical distance between the workers; weight is distributed according to the influence degree of each factor on the safety; acquiring the position, interactive record and working property data of a worker; for each worker, normalizing the obtained value to be in the range of 0-1 according to the formula safety impact index = Σ (evaluation factor value x weight), and obtaining the safety impact index of each worker.
8. The method of claim 1, wherein the allocating resources and scheduling personnel for the work area according to future work content, security impact and indoor and outdoor positioning data of the staff comprises:
acquiring working content, and distributing weight for the working content by adopting a weight distribution method; sorting the working contents according to the weights through a sorting algorithm; adopting a K-means algorithm to allocate a security level label for the working area based on the positioning data and the security influence data; combining the working content weight and the regional security level label to obtain a resource priority; the high-priority resources are allocated to the corresponding working areas, and allocation is carried out when the resources are insufficient; determining an initial position of the staff member, and distributing tasks to the staff member based on the initial position; further comprises: evaluating the matching degree of the working content and the skill of the staff, and carrying out fine adjustment according to the emergency degree and the importance of the work;
The evaluation of the matching degree of the working content and the skill of the staff, and the fine adjustment is carried out according to the emergency degree and the importance of the work, specifically comprises the following steps: acquiring a skill evaluation list of each worker and verifying the integrity of the skill evaluation list; scoring each skill in the skill assessment list according to the actual application frequency and importance of each skill; calculating the comprehensive skill score S (i) of the worker according to the score of each skill and the importance thereof by using a weighted average method; determining a required skill type and level according to each working content; setting skill requirement indexes including experience years and certificates, and setting weights for the skill requirement indexes; according to skill indexes and weights required by the working contents, calculating the skill demand degree T (j) of the working contents; extracting skill scores of each worker and skill demand degrees of each work content from a database; evaluating the efficiency of the staff to finish similar work contents in the past, and calculating the historical efficiency E (i, j) of the staff; calculating a matching degree for each worker and work content using the formula M (i, j) =s (i) ×t (j) ×e (i, j); storing the calculated matching degree result in a matrix and visualizing the matching degree result in a system; distributing work content which is most matched with the skill of each worker on the basis of the matching degree matrix; and fine tuning is performed according to the emergency degree and importance of the work, a work task list of each worker is generated in the system, and a work task notification is sent to the worker.
9. The method of claim 1, wherein the performing obstacle detection on the work area of each worker based on the resource allocation, the output of the personnel scheduling, and the indoor and outdoor positioning technology, avoiding work interference and risk of object collision, comprises:
acquiring task and region allocation data; acquiring the position information of the staff by adopting an indoor and outdoor positioning technology; identifying obstacles within the work area by using a pre-trained YOLO algorithm; obtaining the prediction of the future positions of the staff and the obstacle through a Kalman filter; judging collision risk in the predicted track, and when the collision risk is detected, reallocating resources or adjusting the task position and time of the staff to obtain an adjusted workflow; and sends a warning to the staff member by means of an audible, vibratory or visual notification.
10. The method of claim 1, wherein the adjusting the resource allocation of the work area and optimizing the workflow in combination with the obstacle detection result, the movement range of the worker, and the indoor and outdoor positioning data comprises:
obtaining the marks and classifications of indoor and outdoor barriers by using a pre-trained YOLO model; determining the real-time position and the activity range of the staff; calculating available space and resources in the working area, and comparing the available space and resources with the obstacle and the activity range of the staff to obtain the utilization rate of the resources; the current resource allocation is adjusted by using a linear programming algorithm, so that a new resource allocation scheme is obtained; simulating a new workflow, calculating an efficiency index, and performing efficiency evaluation; the efficiency of the old workflow and the new workflow is compared.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311456064.8A CN117496431A (en) | 2023-11-03 | 2023-11-03 | Outdoor operation safety monitoring method based on indoor and outdoor positioning system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311456064.8A CN117496431A (en) | 2023-11-03 | 2023-11-03 | Outdoor operation safety monitoring method based on indoor and outdoor positioning system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117496431A true CN117496431A (en) | 2024-02-02 |
Family
ID=89675793
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311456064.8A Pending CN117496431A (en) | 2023-11-03 | 2023-11-03 | Outdoor operation safety monitoring method based on indoor and outdoor positioning system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117496431A (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8548838B1 (en) * | 2005-09-28 | 2013-10-01 | Kyyba, Inc. | Method, software and/or process for optimizing the distribution of labor in a manufacturing environment |
| CN107609835A (en) * | 2017-07-28 | 2018-01-19 | 国网辽宁省电力有限公司 | A kind of power network manapower allocation application system and method |
| CN109819208A (en) * | 2019-01-02 | 2019-05-28 | 江苏警官学院 | A kind of dense population security monitoring management method based on artificial intelligence dynamic monitoring |
| CN113902237A (en) * | 2021-06-01 | 2022-01-07 | 贵州电网有限责任公司 | On-site operation risk management and control system |
| US20220051807A1 (en) * | 2020-07-29 | 2022-02-17 | Accenture Global Solutions Limited | Dynamic risk assessment |
| CN115661755A (en) * | 2022-11-08 | 2023-01-31 | 大连理工大学 | Worker safety dynamic evaluation method based on computer vision and trajectory prediction |
| KR20230079680A (en) * | 2021-11-29 | 2023-06-07 | 한국생산기술연구원 | A service system that supports safety managers in an integrated way based on industrial safety intelligence and a safety management method using the same |
| KR20230087383A (en) * | 2021-12-09 | 2023-06-16 | 한국가스안전공사 | Risk Gradation System for Worker Safety Distance based on Risk Grade |
| WO2023131095A1 (en) * | 2022-01-04 | 2023-07-13 | 武汉路特斯汽车有限公司 | Traffic environmental risk assessment method and apparatus, and risk assessment method and apparatus |
| CN116485021A (en) * | 2023-04-20 | 2023-07-25 | 内蒙古科技大学 | Coal enterprise technical skill person sentry matching prediction method and system |
| CN116798272A (en) * | 2023-08-23 | 2023-09-22 | 威海爱思特传感技术有限公司 | Road crossroad blind area vehicle early warning system and method based on vehicle communication |
| CN116822715A (en) * | 2023-05-30 | 2023-09-29 | 北京东方通网信科技有限公司 | Safety production monitoring and early warning system based on artificial intelligence |
| CN116882735A (en) * | 2023-06-28 | 2023-10-13 | 刘汉波 | Chemical risk correction early warning method based on process safety index |
-
2023
- 2023-11-03 CN CN202311456064.8A patent/CN117496431A/en active Pending
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8548838B1 (en) * | 2005-09-28 | 2013-10-01 | Kyyba, Inc. | Method, software and/or process for optimizing the distribution of labor in a manufacturing environment |
| CN107609835A (en) * | 2017-07-28 | 2018-01-19 | 国网辽宁省电力有限公司 | A kind of power network manapower allocation application system and method |
| CN109819208A (en) * | 2019-01-02 | 2019-05-28 | 江苏警官学院 | A kind of dense population security monitoring management method based on artificial intelligence dynamic monitoring |
| US20220051807A1 (en) * | 2020-07-29 | 2022-02-17 | Accenture Global Solutions Limited | Dynamic risk assessment |
| CN113902237A (en) * | 2021-06-01 | 2022-01-07 | 贵州电网有限责任公司 | On-site operation risk management and control system |
| KR20230079680A (en) * | 2021-11-29 | 2023-06-07 | 한국생산기술연구원 | A service system that supports safety managers in an integrated way based on industrial safety intelligence and a safety management method using the same |
| KR20230087383A (en) * | 2021-12-09 | 2023-06-16 | 한국가스안전공사 | Risk Gradation System for Worker Safety Distance based on Risk Grade |
| WO2023131095A1 (en) * | 2022-01-04 | 2023-07-13 | 武汉路特斯汽车有限公司 | Traffic environmental risk assessment method and apparatus, and risk assessment method and apparatus |
| CN115661755A (en) * | 2022-11-08 | 2023-01-31 | 大连理工大学 | Worker safety dynamic evaluation method based on computer vision and trajectory prediction |
| CN116485021A (en) * | 2023-04-20 | 2023-07-25 | 内蒙古科技大学 | Coal enterprise technical skill person sentry matching prediction method and system |
| CN116822715A (en) * | 2023-05-30 | 2023-09-29 | 北京东方通网信科技有限公司 | Safety production monitoring and early warning system based on artificial intelligence |
| CN116882735A (en) * | 2023-06-28 | 2023-10-13 | 刘汉波 | Chemical risk correction early warning method based on process safety index |
| CN116798272A (en) * | 2023-08-23 | 2023-09-22 | 威海爱思特传感技术有限公司 | Road crossroad blind area vehicle early warning system and method based on vehicle communication |
Non-Patent Citations (4)
| Title |
|---|
| 何承坤;赵虹;王东;韩文栋;: "基于层次分析法在船厂管加工车间生产绩效考核中的应用研究", 中国水运(下半月), no. 09, 15 September 2017 (2017-09-15) * |
| 孙兆亮;: "泛在无线信号辅助的室内外无缝定位方法与关键技术", 电子制作, no. 10, 15 May 2015 (2015-05-15) * |
| 王恩伟;曾惜;龙思璇;杨凤生;汪林;: "智能化电力巡检安全隐患监管技术应用研究", 电脑知识与技术, no. 22, 5 August 2020 (2020-08-05) * |
| 穆波;刘超;王廷春;慕?;王辉;王云龙;李千登;: "危化品企业人员作业安全与风险智能管控平台建设研究", 安全与环境工程, no. 05, 25 September 2020 (2020-09-25) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102005188B1 (en) | Industrial site safety management system based on artificial intelligence using real-time location tracking and Geographic Information System, and method thereof | |
| CN117172414A (en) | Building curtain construction management system based on BIM technology | |
| CN107610282A (en) | A kind of bus passenger flow statistical system | |
| JP2017091008A (en) | Data processing method and data processing system | |
| CN116739245B (en) | Intelligent fire-fighting city alarm receiving and scheduling system | |
| CN116308960B (en) | Intelligent park property prevention and control management system based on data analysis and implementation method thereof | |
| CN111311133B (en) | Monitoring system applied to power grid production equipment | |
| JP5487588B2 (en) | Information processing apparatus, program, and method for maintaining a remote machine | |
| CN110989042A (en) | Intelligent prediction method for highway fog-clustering risk | |
| US20210271217A1 (en) | Using Real Time Data For Facilities Control Systems | |
| CN111709597B (en) | Power grid production domain operation monitoring system | |
| CN117496431A (en) | Outdoor operation safety monitoring method based on indoor and outdoor positioning system | |
| CN116629726A (en) | Environment monitoring management method and system for precise copper pipe packaging workshop | |
| Sun et al. | Dynamic human systems risk prognosis and control of lifting operations during prefabricated building construction | |
| JP6445175B2 (en) | Congestion prediction system and congestion prediction method | |
| Fang et al. | Design and Development of Industrial Safety APPs | |
| WO2021111558A1 (en) | Information processing device, control method, and storage medium | |
| Duan et al. | Safety Risk Diagnosis Based on Motion Trajectory for Construction Workers: An Integrated Approach | |
| CN116862244B (en) | Industrial field vision AI analysis and safety pre-warning system and method | |
| JP2020166307A (en) | Apparatus control support device, program and control support method | |
| CN116633990B (en) | Remote information digital supervision system and method based on cloud computing | |
| CN115334452B (en) | Centralized building intelligent alarm method | |
| US20230112611A1 (en) | Emulated facility safety with embedded enhanced interface management | |
| US20230100635A1 (en) | Emulated facility safety with correlated sound frequency modeling | |
| CN115293458A (en) | Construction and use method of crowd safety risk analysis model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |