40 results about "Risk zone" patented technology

The invention provides an intelligent inspection risk assessment method and system based on multi-sensor fusion, and the method comprises the steps: obtaining original multi-source data of a transformer substation, the original multi-source data comprises a binocular vision image, infrared thermal image data and environment sensor data, and carrying out the time-space calibration and preprocessing of the original multi-source data, obtaining a multi-source sensor data stream; carrying out feature extraction on the multi-source sensor data stream to obtain a multi-modal feature set, generating a refined semantic mask based on the multi-modal feature set, constructing an initial scene relation graph, calculating a risk level based on a multi-layer perceptron classifier, and generating a risk level evaluation result and a risk distribution graph; outputting a safety distance violation warning and a risk area identifier; and generating comprehensive risk early warning information based on the risk level assessment result, the risk distribution diagram, the safety distance violation warning and the risk area identifier.

The invention relates to the technical field of mine safety monitoring and early warning, and provides a non-coal mine risk situation deduction system based on digital twinning and reinforcement learning, which comprises at least two twinning spaces; wherein the first twinborn space is used for building a mine digital twinborn body mapped by the current mine entity, and the second twinborn space is used for mapping a target dynamic event containing at least one risk factor in dynamic update of the mine digital twinborn body; the at least one reinforcement learning agent generates a third twinborn deduction model used for predicting the risk evolution trend of the mine; the risk early warning module is used for receiving the real-time multi-source data, determining a target dynamic event in the real-time multi-source data through the first twinborn space, judging whether the target dynamic event has a risk factor through the second twinborn space, and if the risk factor exists in the target dynamic event, performing early warning on the target dynamic event; and generating deduction data of the target dynamic event through a third deduction model, and determining a risk area and a risk probability.

The invention discloses an intelligent contract risk perception fuzzy testing method based on a large language model, and the method comprises the steps: introducing a risk perception mechanism based on the large language model, testing a resource scheduling strategy and semantic knowledge modeling, precisely guiding a fuzzy tester to focus on a high-risk region in an intelligent contract, and achieving the intelligent contract risk perception fuzzy testing. Therefore, the efficiency and accuracy of vulnerability detection are remarkably improved. According to the method, a risk strategy is generated by using a large language model, and the method comprises the steps of identifying a high-risk function and outputting a structured variation parameter suggestion; test resource allocation is optimized for the three key stages, and a fuzzy tester can test a high-risk code path preferentially; a knowledge graph is constructed by analyzing a function call relationship and a data dependency relationship in an intelligent contract, and the knowledge graph is used as a context semantic basis to assist a large language model in more accurately identifying a potential risk function and positioning a key mutation point.

This invention relates to the technical field of industrial machine vision and collaborative control of swarm robots, and particularly to a quality inspection method for swarm vision robots based on multi-agent reinforcement learning. The method includes: acquiring images of the workpiece to be inspected at a preset sampling period, and writing robot pose identifiers and camera imaging parameter identifiers into each frame of the image to form a quality inspection observation set; performing N forward inferences on the quality inspection observation set under a defect detection network with a random deactivation layer, generating uncertainty prediction values ​​using the defect confidence variance, and calculating a risk score matrix to determine a risk region set; inputting the risk score matrix, the risk region set, and the robot pose identifiers into a multi-agent reinforcement learning model to output a candidate observation action set; acquiring supplementary images based on the candidate observation action set and updating the risk score matrix using weighted voting based on the uncertainty prediction values; writing risk regions that meet the threshold into a review task set and rolling them into the next sampling period.

This invention discloses a method for identifying geological hazard risk zones based on a local search strategy, belonging to the field of geological hazard monitoring and early warning technology. The method includes the following steps: locating seed points within a target area; finding similar developmental condition r grid cells within the local search window Ω of the seed point s, forming a similar developmental condition region; then, using the raster area where the seed point s is located as the initial region, iteratively performing morphological dilation, expanding the current region by a certain number of pixels during each iteration and intersecting it with the similar developmental condition region, updating the current region with the intersecting region; after iteration terminates, obtaining the similar neighborhood of the seed point; converting the similar neighborhood region raster of the seed point into vector polygons, and optimizing the boundary smoothing to obtain the risk zone identification range. This invention focuses on core area calculation, avoiding the low efficiency problem of full-domain analysis and identification, and is suitable for rapid screening and dynamic updating of large-scale risk zones.

The application relates to a danger judgment and active guidance method and system for power construction safety supervision, and belongs to the technical field of power construction safety. The method comprises the following steps: collecting multi-modal data of a construction scene through intelligent glasses; analyzing the data at an edge computing end to obtain a scene semantic category and an interactive behavior of a worker; according to the scene semantic category, dynamically scheduling and hot loading a matched risk identification model set from a model library to identify a risk factor and a three-dimensional position thereof; acquiring a three-dimensional voxelized static danger level base map; mapping the risk factor into a dynamic risk observation point cloud in a voxel space; taking the static base map as a priori and taking the dynamic point cloud as observation evidence, performing recursive Bayesian filtering through a dynamic Bayesian network to generate a dynamic danger level three-dimensional distribution; generating and rendering graded AR early warning information; when a violation trend occurs, generating and rendering a three-dimensional safety path of a risk avoidance area. The method realizes adaptive, accurate and active construction safety supervision.

The invention provides a multi-party cooperative progress visualization monitoring method for intelligent building facilities, and relates to the technical field of intelligent building construction monitoring, and the method comprises the steps: collecting data, and building a three-dimensional model; establishing a causal relationship identifier based on the time sequence and the associated features to obtain a causal map, calculating an abnormal propagation probability and labeling a risk region; performing information mapping to generate a monitoring data stream and extracting features; calculating feature correlation, and mapping the feature correlation into a scene semantic segmentation map; and rendering in combination with the risk area and the progress information to obtain a visual three-dimensional dynamic construction progress. According to the invention, risk early warning and progress visual monitoring of the construction process are realized.

The invention relates to the technical field of target detection, in particular to an adaptive detection method based on regional dynamic reasoning optimization, and the method comprises the steps: 1, presetting an abnormal target; step 2, carrying out risk level division on each abnormal target; step 3, setting a detection result without an abnormal target and contribution weights of the abnormal target of each risk level to risk scores, and determining risk score ranges and detection frequencies corresponding to risk areas of different levels; 4, starting a camera device, and dividing a detection picture obtained by the camera device into AxB grid regions; 5, performing risk area grade division on each grid area according to the corresponding risk score; 6, performing risk detection on each grid region based on the corresponding detection frequency according to the corresponding risk region grade, and outputting a risk detection result; and step 7, updating the risk score and the risk area grade of each grid area according to the risk detection result.

The invention discloses a PET facing intelligent flat pasting system based on deep learning defect detection, and relates to the technical field of data analysis. By constructing a region fragment sequence which is continuously arranged along a fitting path and calculating a coupling risk value between adjacent regions in combination with image information and sensor feedback information, the system can identify a potential interaction risk region caused by material physical property difference before defects are not explicitly exposed, so that the defect detection accuracy is improved. A risk hot area map is generated, and local process intervention operation is executed in real time in a linkage manner, so that the high-order structural defects which are originally difficult to perceive visually and are easy to evolve into cracks, edge warping or stripping in the fitting process are effectively avoided. Compared with a conventional independent discrimination mode based on local visual features, the method achieves the dynamic modeling and risk pre-control of the regional coupling behavior, and remarkably improves the defect prediction capability of the system in a complex material matching scene and the stability of the fitting quality.

An information processing apparatus includes a risk area identification unit configured to identify a risk area outside a moving object and a transmission control unit configured to perform control for transmitting risk area information representing the risk area identified by the risk area identification unit to a server configured to retain information related to the risk area, in which the risk area identification unit is configured to identify an area defined by a plurality of points as the risk area, and the risk area identification unit is configured to identify, based on a boundary between a first area in which the moving object is prohibited from moving and a second area in which the moving object is allowed to move, a shape of the risk area on a side of the first area.

The invention discloses a port storage tank risk prediction method and system based on machine learning, and the method comprises the steps: obtaining the storage tank data and regional data of a preset port, carrying out the logistics transportation analysis and energy density simulation according to the regional data, carrying out the statistics of historical risk factors of a risk region based on the simulation energy density fluctuation range, and obtaining a risk prediction result; and constructing a criticality matrix according to the historical risk factors and the logistics grid point locations, performing risk assessment on the risk region by using the criticality matrix, inputting a risk assessment result and risk factor items of the storage tank data into a random forest for training, and obtaining a storage tank risk prediction model. And outputting a risk prediction result of the storage area according to the storage tank risk prediction model. According to the method, risk prediction is carried out by utilizing logistics grid division, risk area irregular splitting, energy density simulation and quality detection data of the storage tank, a multi-dimensional decision basis is provided for safe operation of a port, and meanwhile, the method has relatively good interpretability.

The present application relates to the technical field of mine safety monitoring and early warning, and provides a non-coal mine risk situation deduction system based on digital twinning and reinforcement learning, comprising: at least two twin spaces; wherein the first twin space is used to build a mine digital twin of the current mine entity mapping, and the second twin space is used to map a target dynamic event containing at least one risk factor in the dynamic update of the mine digital twin; at least one reinforcement learning intelligent agent generates a third twin deduction model for predicting the evolution trend of mine risks; a risk warning module is used to receive real-time multi-source data, determine the target dynamic event in the real-time multi-source data through the first twin space, determine whether the target dynamic event has a risk factor through the second twin space, and when the target dynamic event has a risk factor, generate deduction data of the target dynamic event through the third deduction model, and determine the risk area and the risk probability.

The application discloses a laboratory motion line safety early warning method and device, which realizes personnel space-time trajectory reconstruction and accurate identification of abnormal motion line segments by real-time collection of relevant personnel positioning and behavior monitoring data streams, formation of motion line basic data with space-time stamps and identity tags, and combination with standard templates, and provides reliable support for risk assessment. By combining personnel characteristics and basic risk coefficients corresponding to the identity, the personal risk contribution value is calculated and the personal risk field is generated, realizing accurate quantification and differentiated assessment of personnel risk. Superimposing the personal risk field generates a dynamic group risk heat map, which can intuitively present the global risk distribution; building a risk coupling network can predict the conduction effect of high-risk areas, dynamically determine the early warning threshold according to the regional personnel and activity types, effectively improve the early warning accuracy, and reduce false positives and false negatives. In summary, the application realizes accurate control and risk early disposal of the whole process of laboratory motion line safety, and significantly improves the fine and intelligent level of safety management.

The invention relates to the technical field of risk assessment, in particular to a method for providing risk assessment level map information, which comprises the following steps of: acquiring real-time data from a plurality of sources, and performing unified standardization processing on the real-time data to form a fusible data set; analyzing historical event data based on the fusible data set, and generating a historical event mode and a corresponding initialization parameter; and performing uncertainty fusion on the fusible data set in combination with a historical event mode, allocating various data weights, and generating a regional risk level with a confidence index. According to the method, a historical event association graph is constructed, time and space association rules between events are mined, and a historical event mode is used for initializing event propagation model parameters, so that prospective space prediction of potential risk events between regions is realized; therefore, the system can predict the propagation trend of the event between the areas and the potential high-risk area.

This invention provides a construction AI-based risk analysis and decision-making method for safety management. Through a knowledge base construction mechanism involving intelligent parsing, vectorized storage, and incremental updates, it transforms scattered safety knowledge from projects into digital assets, enabling the continuous accumulation of enterprise safety knowledge assets. A RAG-enhanced large language model question-answering system equips each on-site personnel with a personal AI expert, providing near-instantaneous responses to on-site issues and improving efficiency. By employing a performance quantitative analysis model, an Attention-LSTM time-series risk prediction model, and a hazard clustering analysis algorithm, it uncovers management shortcomings and predicts risk trends from massive amounts of data, making safety management decisions more data-driven and moving away from experience-based, extensive management. AI time-series prediction identifies high-risk areas and hazard trends seven days in advance, generating actionable control recommendations, shifting safety management from post-event remediation to pre-event prevention, significantly reducing the probability of accidents.

This invention discloses a method, system, and electronic device for personnel operation assessment based on video temporal joint analysis, relating to the field of power safety technology. This application involves acquiring video from the work site and extracting frames; extracting the sequence of key human points from each frame; dynamically defining risk areas for the head, torso, and hands based on these key points; calculating hand movement speed, distance from the risk area, and movement consistency characteristics; triggering refined behavior recognition when hand movement characteristics indicate a tendency to enter the risk area; extracting multi-scale temporal features before and after the trigger; and judging specific risk behaviors using a pre-trained action recognition model. This invention, through a two-level mechanism of "trend triggering + refined recognition," achieves early warning and accurate identification of risk behaviors such as removing helmets and gloves, effectively improving the real-time performance and preventative capabilities of work safety monitoring. This application realizes a shift from post-event identification to pre-event warning, resulting in better safety monitoring performance.

The invention belongs to the technical field of intelligent water affair and urban disaster prevention, and provides a drainage pipeline risk early warning method, system and device and a medium, and the method comprises the following steps: dividing risk areas through combining pipeline GIS data and historical faults, and carrying out the differentiated arrangement of temperature and strain sensing optical fibers to form a double-parameter network; synchronously acquiring data by using a demodulator, processing the data by using a dynamic self-adaptive sliding window of a preset threshold value, and identifying abnormal segments; after multi-source fusion correction, inputting a pre-trained CNN-LSTM model to preliminarily evaluate the event type and risk; carrying out digital twinning simulation on consequences of moderate and above risks by using pipelines, and judging fourth-level risks; early warning is pushed according to grades, response is triggered, disposal feedback is collected, and model parameters and risk thresholds are dynamically adjusted through incremental learning; according to the invention, accurate capture of micro-anomaly of the drainage pipeline and real-time quantitative assessment of risk are realized, the life cycle of the system is prolonged, the operation and maintenance cost is reduced, and intelligent and safe operation and maintenance of the drainage pipeline are achieved.

This invention discloses a method and system for dynamic management of mine risk areas based on the Industrial Internet of Things (IIoT), relating to the technical field of mine management. The method includes: constructing a network of sensing devices to collect multi-source mine area data streams and transmitting them to a cloud processing center; performing data fusion modeling and situation analysis to construct a three-dimensional real-time mine situation map, while simultaneously pre-setting a risk rule base; predicting and marking risk areas to obtain a set of potential mine risk areas, while simultaneously retrieving multi-physics monitoring data; verifying and reviewing the set of potential mine risk areas to determine the target mine risk area set and perform risk quantification assessment and collaborative dynamic control. This invention solves the technical problems of lagging mine risk identification, severe information silos, and a lack of collaborative risk control methods in existing technologies, achieving comprehensive real-time perception of mine risk areas, dynamic and proactive prediction of risk situations, and improving the accuracy and reliability of risk identification and control.

The invention relates to the technical field of industrial machine vision and group robot cooperative control, in particular to a group vision robot quality inspection method based on multi-agent reinforcement learning. A robot pose identifier and a camera imaging parameter identifier are written into each frame of image to form a quality inspection observation set; carrying out N times of forward reasoning on the quality inspection observation set under a defect detection network containing a random inactivation layer, generating an uncertainty prediction value by a defect confidence coefficient variance, and calculating a risk scoring matrix to determine a risk region set; inputting the risk score matrix, the risk region set and the robot pose identifier into a multi-agent reinforcement learning model to output a candidate observation action set; and acquiring supplementary images according to the candidate observation action set, performing weighted voting by using an uncertainty predicted value to update the risk scoring matrix, writing a risk region meeting a threshold value into a recheck task set, and rolling to bring the risk region into a next sampling period.

The invention relates to the control of a vehicle (1) based on risk area information (13.1, 13.2). In this context, a method is proposed, among other things, which may comprise the following steps: - Recording of risk area information (13.1, 13.2), - Transmitting the risk area information (13.1, 13.2) to a vehicle (1) and - Controlling the vehicle (1) based on the risk area information (13.1, 13.2), wherein a local arrangement, geometry and type of risk area (12) are determined based on the risk area information (13.1, 13.2).

The application discloses a community pension service management system and method based on risk analysis, and belongs to the technical field of pension service management. The application uses a YOLO algorithm to locate risk positions in a community, forms a community risk position set, constructs a community three-dimensional map, forms a risk area in the community three-dimensional map from the community risk position set, and labels risk information; exploratory data analysis is performed, a statistical model is constructed, and a risk area expansion factor is obtained; based on the risk area expansion factor and current weather data, the risk area of the community three-dimensional map and the corresponding risk information are adjusted; health data of a user is acquired, a scoring standard is set, and a health score is calculated; combined with individual conditions of the user, whether the user is currently suitable for going downstairs for exercise is evaluated based on the total health score, and a health evaluation result is obtained; when the health evaluation result of the user is suitable for exercise or cautious exercise, a planned exercise route of the user is adjusted, and risk information is prompted.

The invention belongs to the technical field of natural resource inspection, and particularly relates to a natural resource multi-sensing integrated monitoring method and platform. According to the method, through fusion correction of time synchronization, coordinate unification and historical tracks, the influence of time and coordinate abnormity and accidental positioning deviation on border crossing judgment can be effectively eliminated, so that misinformation and misscheduling are reduced, the self-learning responsibility range is automatically expanded and shrunk according to the space density and the inspection frequency, inspection resources can be concentrated in a high-demand or high-risk area, and the inspection efficiency is improved. Unnecessary repeated inspection on a low-risk area is avoided, manpower and operation cost is reduced, time-space matching and continuity inspection of events and tracks can restore inspection behavior paths and trends before and after the events occur, managers are assisted in judging event properties and responsibility ownership, the treatment pertinence and efficiency are improved, and the management cost is reduced. Through fusion evaluation of three types of risk indexes of space, border crossing and events, grading early warning of the natural resource state is realized.

The invention discloses an inspection scheduling method and system based on MEMS piezoelectric intelligent sensing and reinforcement learning, and the method comprises the following steps: carrying out the active ultrasonic monitoring through an MEMS piezoelectric intelligent sensing element, and carrying out the nonlinear feature extraction and preprocessing of a collected piezoelectric signal; risk weights of the road sections are calculated based on multi-source data fusion, and high-risk road sections are identified; finely dividing the high-risk road sections, and determining key monitoring areas; establishing a probability distribution model of a transverse wheel track band on the high-risk road section; the probability distribution model is used for guiding the piezoelectric sensors to be optimally arranged at candidate positions of the cross section two-dimensional space; a multi-level sensing and data processing system is constructed to realize data acquisition, transmission, processing and decision making; and based on a reinforcement learning algorithm, generating an inspection scheduling scheme in combination with risk grading and time window constraint. According to the invention, early warning of early damage of the road structure in the high-risk area and optimal configuration of resources are realized, the inspection efficiency is improved, and the operation and maintenance cost is reduced.

The application discloses a personnel positioning frequency self-adaptive adjustment method and device based on a risk field, and belongs to the field of personnel safety positioning. The method comprises the following steps: a continuous environmental risk field model covering a monitoring area is constructed, and a continuous risk intensity value is given to each coordinate point; a real-time position of a positioning terminal is acquired, and a risk field model is inquired to obtain a current risk intensity value; the risk intensity value is mapped to a target positioning working frequency through a continuously monotonically increasing frequency mapping function; the terminal is controlled to perform positioning data collection and reporting according to the target frequency; meanwhile, risk prediction is performed according to the change trend of the risk intensity value, and a warning is given when the risk continuously rises and the rising rate exceeds a threshold value. The application realizes fine description of risk through continuous risk field modeling, adjusts the positioning frequency steplessly according to the risk level, reduces energy consumption in a low-risk area, guarantees real-time performance in a high-risk area, and actively gives a warning in combination with the risk trend, so that the energy consumption and safety guarantee demand of the positioning system are effectively balanced.

The invention discloses a construction site safety intelligent early warning system and method based on the Internet of Things, and relates to the technical field of safety early warning. According to the method, the historical accident evolution models are established according to historical construction accident records, a plurality of accident correlation factors are marked in the historical accident evolution models, the accident correlation factors in different historical accident evolution models are matched with each other, and the historical accident evolution models are correlated according to the matching result. The method comprises the following steps of: establishing an accident evolution factor network, collecting real-time construction scene data of a construction site, establishing a dynamic construction model based on the real-time construction scene data, identifying a plurality of risk areas from the dynamic construction model through the accident association factors, and establishing an accident evolution factor network according to the risk areas. And carrying out risk simulation on the risk area in the dynamic construction model, and sending a safety early warning to construction personnel according to a risk simulation result.

The present application relates to the technical field of deep learning, in particular to a silicon carbide component stress distribution monitoring and crack risk prediction method, which realizes comprehensive perception of the stress state of the silicon carbide component, accurate positioning of the risk area and early warning of the crack fault. Multimodal data is synchronously collected by multiple sensors, and cross-modal timing synchronization is realized through feature alignment; a crack risk multi-branch feature extraction module is designed, general deep features and risk features facing thermal stress mismatch, micro-crack evolution and structural instability are respectively extracted through a shared backbone network and a special branch network; a stress cloud map generation and risk area positioning module based on a graph neural network is constructed, visual reasoning from discrete features to full-field stress distribution and risk area marking are realized; a crack risk comprehensive prediction module based on multi-dimensional risk feature fusion is designed, instantaneous state and evolution trend are fused, a multi-risk confidence vector is output and a hierarchical warning is triggered.

The application provides a method for drawing and using a direct current bias magnet risk map, which comprises the following steps: firstly, establishing a risk grade division representation; then, positioning the geographic positions of each direct current pole and substation; dividing the risk area through a direct current pole circle drawing method and a triangular area division method; then, processing the risk area such as a risk overlapping area and an uncovered area; and finally, completing the drawing of the direct current bias magnet risk map. The serious or severe or general risk area is selected from the direct current bias magnet risk map, and a treatment scheme is selected according to the grounding condition of the substation, the distance from the grounding pole and the current size obtained through modeling calculation. The method of the application establishes a predicted risk map of the direct current bias magnet risk, proposes a direct current bias magnet treatment scheme under different conditions of a power system, can provide a theoretical reference and technical guidance for the power grid operation and maintenance department to carry out treatment work for the direct current bias magnet problem, and helps to maintain the safe and reliable operation of the power substation.

The invention discloses a bridge anti-icing control system based on multi-source sensing data analysis and a control method thereof, and relates to the technical field of bridge anti-icing. The method comprises the following steps: marking a position where an icing event easily occurs in a bridge as a risk area, collecting risk factor data of the risk area, constructing a risk model used for evaluating the occurrence probability of the icing event, and calculating the probability of the icing event in the risk area in a current risk factor data state by using the risk model. According to a probability calculation result, the risk area type is divided into a high risk area, a medium risk area or a low risk area, corresponding anti-icing control strategies are made for areas of different risk levels, and icing detection, early warning and automatic control measures are organically combined to form a complete anti-icing solution, so that the anti-icing efficiency is improved. Therefore, bridge icing can be effectively prevented and treated, the safety and reliability of the bridge running in winter are improved, and the influence of the bridge icing on traffic is reduced.

The embodiment of the invention provides a data processing method, a model training method, system and device, equipment, a storage medium and a program product. According to the scheme provided by the embodiment of the invention, after the input data (including the input text and / or the input image) is determined according to the target modal data to be subjected to risk detection, the risk detection of the input data is realized by utilizing the risk detection model, so that the corresponding risk detection result is obtained. Specifically, when it is detected that the input data has a risk, the risk detection result includes risk content extracted from the input data. The risk content comprises at least one of the following items: a risk word extracted from the input text and a risk area marked in the input image. The risk word comprises an adversarial variant word. The adversarial variant words are variant expressions which are formed by transforming known risk words and have unchanged semantics.

The invention relates to a process island and a safety control method, and the method comprises the steps: collecting the position information of a person in the process island, and enabling the process island to be divided into a plurality of risk regions; matching an associated target risk area according to the position information of the personnel; determining a target risk level of the personnel according to a pre-established corresponding relationship between a risk area and a risk level; and if the target risk level is a danger level, controlling an intelligent transport trolley and automatic equipment in the process island to stop running, and triggering a danger alarm signal. When the intelligent transport trolley enters and exits the process island and the safety grating loses efficacy, personnel safety problems caused by the fact that personnel follow to enter the process island can be effectively avoided.