Blasting operation field behavior compliance identification method, system, equipment and medium

Abstract

The invention discloses a blasting operation field behavior compliance identification method, system and device and a medium, and relates to the technical field of computers. The method comprises the following steps: acquiring a multi-source video stream and environment monitoring data of a blasting operation site; identifying a current operation stage according to the multi-source video stream; identifying a scene risk factor according to the multi-source video stream and the environment monitoring data; inputting the multi-source video stream into a pre-trained deep learning model, and analyzing identity information, a behavior sequence, an equipment wearing state and position information of each operator; calling a compliance rule set matched with the current operation stage and the scene risk factor in a pre-constructed security rule library; based on the compliance rule set, performing compliance verification on the identity information, the behavior sequence, the equipment wearing state and the position information; and if the violation behavior exists through verification, generating a violation judgment result containing the violation type, the risk level, the associated personnel and the spatial region identifier. According to the invention, intelligent monitoring and violation identification of the whole process of blasting operation can be realized.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, system, equipment, and medium for identifying compliance in blasting operations. Background Technology

[0002] Blasting operations are indispensable in mining, tunneling, and building demolition projects. However, due to the high-risk nature of these operations, involving the transportation, storage, loading, and detonation of explosives and other hazardous materials, and often conducted in complex and high-risk environments, safety and compliance are paramount. National and industry standards have established strict regulations governing the behavior of personnel, the wearing of equipment, and area control. Any minor violation can lead to serious accidents, causing loss of life and property, and environmental damage.

[0003] However, existing regulatory methods still have significant shortcomings. Key aspects such as the separation of responsibilities (e.g., the same person must not be responsible for requisitioning, issuing, or blasting equipment) and regulations on identifiable attire are often difficult to implement due to management oversights or difficulties in resource allocation. Violations such as personnel gathering, unlicensed work, and incompatible equipment are frequent. Traditional manual inspections or video surveillance are inefficient, subjective, and unable to achieve 24 / 7 comprehensive monitoring. In addition, static safety rules cannot dynamically adjust protection requirements in the face of sudden changes in scenarios (such as rainstorms or leaks), resulting in delayed risk response. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method, system, equipment, and medium for identifying compliance in blasting operations, which can realize intelligent monitoring and violation identification throughout the entire blasting operation process.

[0005] To achieve the above objectives, the present invention provides a method for identifying compliance in blasting operations, comprising: Acquire multi-source video streams and environmental monitoring data from the blasting operation site; the environmental monitoring data includes temperature, humidity, wind speed, and hazardous gas concentration. The current operation stage is identified based on the multi-source video stream; Based on the multi-source video streams and environmental monitoring data, identify scene risk factors; The multi-source video streams are input into a pre-trained deep learning model to parse out the identity information, behavior sequence, equipment wearing status and location information of each operator. Call the set of compliance rules in the pre-built security rule base that matches the current operation stage and scenario risk factors; Based on the aforementioned compliance rule set, compliance verification is performed on the identity information, behavior sequence, equipment wearing status, and location information. If a violation is found during the verification, a violation determination result will be generated, which includes the violation type, risk level, associated personnel, and spatial area identifier.

[0006] Optionally, identifying the current operation stage based on the multi-source video stream includes: The multi-source video streams are spatiotemporally synchronized to generate a calibrated multi-view video sequence; Extract equipment operation features and personnel collaboration features from the multi-view video sequence; Based on preset work phase transition rules, a time-series correlation analysis is performed on the equipment operation characteristics and personnel collaboration characteristics; Based on the correlation analysis results, the current operation stage is determined.

[0007] Optionally, based on the multi-source video streams and environmental monitoring data, scene risk factors are identified, including: The environmental monitoring data is compared with a preset safety threshold to generate an environmental risk level. Analyze the multi-source video streams to extract environmental interference factors such as smoke and dust concentration, rain and fog interference intensity, and the coverage area of ​​muddy areas on the ground; Based on pre-constructed risk mapping rules, the environmental risk level and environmental disturbance factor characteristics are integrated to generate scenario risk factors.

[0008] Optionally, the compliance rule set includes: qualification rules, behavioral rules, equipment rules, and spatial rules; based on the compliance rule set, compliance verification is performed on the identity information, behavioral sequence, equipment wearing status, and location information, including: The identity information is verified according to the qualification rules. When the operator's qualification is invalid or the job authority is exceeded, it is determined that there is a violation. According to the behavior rules, the behavior sequence is compared. When a key action is missing, the operation is out of order, or a high-risk redundant action is identified, it is determined that there is a violation. The equipment wearing status is verified according to the equipment rules. When the protective equipment is missing, malfunctioning, or does not match the requirements of the current operation stage, it is determined that there is a violation. Based on the spatial rules and the location information, if the workers are in a prohibited area, exceed the gathering limit, or fail to evacuate as required, it is determined that there is a violation.

[0009] Optionally, the method further includes: When the violation determination result is generated, real-time video segments corresponding to the violation type, associated personnel, and spatial area identifier are extracted from the multi-source video stream; Add timestamps and spatial location markers to the real-time video clips, and associate and bind them with the violation judgment results, environmental monitoring data and the current operation stage to generate an electronic evidence data package; Add a digital signature and hash value to the electronic evidence data packet to generate an immutable electronic evidence record; The electronic evidence records are stored in an anti-tampering storage system to form a traceable chain of electronic evidence.

[0010] Optionally, the multi-source video stream is input into a pre-trained deep learning model, prior to which the method further includes: Acquire point cloud data and on-site audio stream from the blasting operation site; The intensity of environmental interference is determined based on the environmental monitoring data and the multi-source video stream; When the intensity of the environmental interference exceeds a preset threshold, the multi-source video stream is input into a pre-trained deep learning model. Specifically, the multi-source video stream, point cloud data, and on-site audio stream are input into the pre-trained deep learning model. The point cloud data is used to supplement spatial positioning information, and the on-site audio stream is used to extract acoustic event features associated with standard operating procedures.

[0011] This invention also provides a compliance identification system for blasting operation site behavior, comprising: The acquisition unit is used to acquire multi-source video streams and environmental monitoring data from the blasting operation site; the environmental monitoring data includes temperature and humidity, wind speed, and hazardous gas concentration. A job phase identification unit is used to identify the current job phase based on the multi-source video stream; The scene risk analysis unit is used to identify scene risk factors based on the multi-source video streams and environmental monitoring data; The personnel behavior analysis unit is used to input the multi-source video stream into a pre-trained deep learning model to analyze the identity information, behavior sequence, equipment wearing status and location information of each operator. The rule dynamic scheduling unit is used to call the set of compliance rules in the pre-built security rule library that matches the current operation stage and scenario risk factors; The compliance verification unit is used to perform compliance verification on the identity information, behavior sequence, equipment wearing status and location information based on the compliance rule set. The violation decision generation unit is used to generate a violation judgment result that includes the violation type, risk level, related personnel, and spatial area identifier if a violation is found during verification.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described identification method.

[0013] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described identification method.

[0014] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects: The blasting operation site behavior compliance identification method provided by this invention achieves real-time, multi-dimensional perception of the operation scene by acquiring multi-source video streams and environmental monitoring data (including temperature, humidity, wind speed, and hazardous gas concentration) at the blasting operation site. This completely solves the problems of low efficiency in traditional manual inspections and the inability of video surveillance to automatically identify risks. By using multi-source video streams to identify the current operation stage and combining environmental monitoring data to dynamically identify scene risk factors, it significantly improves the situational analysis capability in complex environments (such as rain, fog, and smoke), overcoming the shortcomings of traditional methods such as low identification accuracy and poor robustness caused by environmental interference.

[0015] By inputting multi-source video streams into a pre-trained deep learning model, the system accurately analyzes personnel identity information, behavioral sequences, equipment wearing status, and location information. This enables the automated conversion from raw video to structured compliance elements, effectively capturing instantaneous violations such as unlicensed work, missing equipment, and disordered operations. It solves the problems of unclear personnel responsibilities and difficulty in timely detection of hidden violations in existing technologies. Dynamically invoking compliance rule sets matched to the current work stage and scenario risk factors allows safety supervision requirements to adjust in real time according to work progress and environmental threats (e.g., automatically strengthening anti-slip equipment verification during sudden rainstorms), overcoming the limitation of static rules that cannot dynamically respond to changes in the scenario. Attached Figure Description

[0016] The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same parts.

[0017] Figure 1 This is a schematic diagram of the method flow for the compliance identification method for blasting operation site behavior as shown in an embodiment of the present invention; Figure 2 This is a schematic diagram of the module structure of a blasting operation site behavior compliance identification system according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] Please see Figure 1 , Figure 1 A flowchart illustrating the method for identifying compliance in blasting operations.

[0020] Methods for identifying compliance issues at blasting operation sites include: S101: Acquire multi-source video streams and environmental monitoring data from the blasting operation site.

[0021] The environmental monitoring data includes temperature, humidity, wind speed, and concentration of hazardous gases.

[0022] In application, to establish a comprehensive understanding of the safety status at blasting operation sites, the first step is to systematically acquire multi-dimensional real-time data from the site. Specifically, multiple visible light and infrared video monitoring devices deployed in key areas of the operation site (such as temporary explosive storage points, loading areas, wiring areas, detonation stations, warning boundaries, and personnel passageways) continuously collect dynamic video streams covering different perspectives and operational stages, forming multi-source video streams. These video streams complement each other, aiming to eliminate blind spots from a single perspective and provide basic visual information for subsequent identification of personnel behavior, equipment status, operational stages, and some environmental interference factors.

[0023] Simultaneously, to accurately capture environmental physical conditions that may affect operational safety or exacerbate the risk of violations, environmental monitoring data must be acquired concurrently. This is achieved by deploying corresponding sensor networks at key locations on the work site. The collected data specifically includes: temperature and humidity data reflecting on-site comfort and potential electrostatic risks; wind speed data affecting the accumulation and diffusion of harmful gases and the prediction of the trajectory of blast debris; and concentration data of hazardous gases directly related to explosion risks and personnel health (such as flammable, explosive, or toxic gases like methane, carbon monoxide, and hydrogen sulfide).

[0024] The simultaneous acquisition and fusion of multi-source video streams with environmental monitoring data (temperature, humidity, wind speed, and hazardous gas concentrations) forms the data foundation of the entire compliance identification method. The video streams provide direct visual observation of personnel activities, the status of goods, and work processes, while the environmental monitoring data quantifies the physical and chemical risk factors present on site.

[0025] S102: Identify the current operation stage based on multi-source video streams.

[0026] In applications, after acquiring multi-source video streams from a blasting operation site, it is crucial to accurately identify the current stage of the operation. This is a key prerequisite for achieving dynamic and compliant supervision. To achieve this goal, the acquired multi-channel video streams are first processed for spatiotemporal synchronization. Through timestamp alignment and spatial coordinate calibration, perspective misalignment caused by differences in equipment deployment locations or clock deviations is eliminated, generating a multi-view video sequence that is consistent in time and space, providing a coherent and consistent visual input for subsequent feature analysis.

[0027] Based on the calibrated video sequences, deep analysis of the video content is used to extract key operational status features. The focus is on extracting equipment operation features reflecting equipment usage status (e.g., the action patterns of specific equipment such as drilling machine operation, activation of explosive loading equipment, and detonator readiness status), as well as personnel collaboration features reflecting personnel organization and cooperation patterns (e.g., safety officers demarcating warning zones, blasters loading explosives and connecting lines, evacuation of clearing personnel, and multi-person collaborative handling). These features are the core visual identifiers representing different operational stages.

[0028] The extracted equipment operation features and personnel collaboration features are input into a pre-defined work stage transition rule base for time-series correlation analysis. This rule base, based on standards such as the "Blasting Safety Regulations," clearly defines the logical transition conditions and typical feature combinations between each work stage (such as preparation stage, charging stage, connection stage, warning stage, detonation stage, and all-clear and inspection stage). The analysis process dynamically tracks the order of appearance, duration, and combination relationships of features on the timeline to determine the specific stage node of the current work process.

[0029] Specifically, the above-mentioned identification of the current operation stage based on multi-source video streams includes the following steps: Spatiotemporal synchronization processing is performed on multi-source video streams to generate calibrated multi-view video sequences; Extract equipment operation features and personnel collaboration features from multi-view video sequences; Based on preset work phase transition rules, a time-series correlation analysis is performed on equipment operation characteristics and personnel collaboration characteristics; Based on the correlation analysis results, the current operation stage is determined.

[0030] The automatic identification of the current operation stage achieved through the above steps replaces the inefficient traditional method that relies on manual experience, significantly improving the objectivity and real-time nature of stage judgment. On the one hand, it provides an accurate basis for subsequent dynamic scheduling and matching of the safety rule set for this stage (such as strictly checking anti-static equipment during the charging stage and ensuring site clearance during the detonation stage); on the other hand, by monitoring the operation progress in real time, the system can predict potential risk points in the next stage (such as strengthening personnel clearance verification in advance before entering the detonation stage), thereby achieving proactive safety supervision and effectively avoiding compliance risks caused by misjudgment of stages or delayed response.

[0031] S103: Identify scene risk factors based on multi-source video streams and environmental monitoring data.

[0032] In practice, to dynamically assess the overall safety situation at blasting operation sites, it is necessary to comprehensively analyze video and environmental data to identify potential risk factors. First, real-time environmental monitoring data (including temperature, humidity, wind speed, and hazardous gas concentration) is automatically compared with preset safety thresholds. For example, when methane concentration exceeds 20% of the lower explosive limit, wind speed is higher than the safe operating limit, or humidity is too low, causing static electricity risk, a quantitative environmental risk level is generated based on preset classification standards (such as low, medium, high, and extremely high), directly reflecting the degree of threat posed by the physical environment to operational safety.

[0033] Simultaneously, computer vision technology is used to deeply analyze multi-source video streams and extract visual environmental interference features. Three key factors are identified: first, smoke concentration is calculated by analyzing smoke diffusion patterns and visibility changes to assess its interference with workers' vision and equipment recognition; second, raindrop trajectories, fog density, and light attenuation are detected to quantify the intensity of rain and fog interference; and third, the coverage area of ​​muddy areas is identified and calculated based on ground reflectivity, water accumulation areas, and personnel walking postures. These features collectively constitute the environmental interference factor feature set.

[0034] By employing pre-constructed risk mapping rules, the aforementioned generated environmental risk levels are fused with a multi-dimensional set of environmental disturbance factor features. This rule base defines the coupling effects of risk factors under different environmental conditions (e.g., high wind speeds combined with high dust concentrations significantly increase the risk of flying rocks, and heavy rain leading to the expansion of muddy areas may cause people to slip or equipment to malfunction). The fusion process uses weighted aggregation or decision tree models, ultimately outputting dynamic scene risk factors. These factors not only quantify the inherent danger of the environment but also assess its impact on the accuracy of the monitoring system.

[0035] Specifically, the above-mentioned identification of scene risk factors based on multi-source video streams and environmental monitoring data includes the following steps: Environmental monitoring data is compared with preset safety thresholds to generate environmental risk levels; Analyze multi-source video streams to extract environmental interference factors such as smoke and dust concentration, rain and fog interference intensity, and the coverage area of ​​muddy areas on the ground. Based on pre-built risk mapping rules, environmental risk levels and environmental disturbance factor characteristics are integrated to generate scenario risk factors.

[0036] The above steps overcome the limitations of traditional static safety assessments. On one hand, by integrating multi-source heterogeneous data (sensor data and video features), it is possible to perceive complex risks that are difficult for the human eye to detect (such as flammable gas leaks accompanied by low visibility). On the other hand, dynamically generated scenario risk factors provide a scientific basis for subsequent steps (such as adaptively adjusting the strictness of compliance rules and triggering specific early warning mechanisms). For example, when "high rain and fog interference + excessive mud on the ground" is identified, the verification intensity of personnel anti-slip equipment can be automatically increased, or the fault tolerance threshold of the spatial positioning algorithm can be improved, thereby maintaining the effectiveness and robustness of supervision in harsh environments and significantly reducing the risk of missed detections due to sudden environmental changes.

[0037] S104: Input the multi-source video streams into the pre-trained deep learning model to parse out the identity information, behavior sequence, equipment wearing status and location information of each operator.

[0038] In applications, to accurately extract individualized compliance elements for operators, deep semantic parsing is required on the synchronously processed multi-source video streams. This process is accomplished by inputting the video streams into a pre-trained deep learning model. This model is trained on a large-scale blasting operation scenario dataset and adopts a multi-task learning architecture. It can output four types of key information in parallel: identity information (by integrating facial recognition, work uniform color coding recognition, and badge detection technology, it distinguishes roles such as blasters, safety officers, and guards, and verifies their qualifications and validity), behavior sequence (through a spatiotemporal action segmentation model, it continuously captures and labels standard actions such as "carrying explosives," "loading drill holes," "connecting detonators," and "leaving the warning area" and their temporal relationships), equipment wearing status (based on target detection and attribute classification models, it identifies the integrity and validity of wearing protective equipment such as safety helmets, anti-static clothing, and goggles, such as detecting whether the helmet straps are fastened and whether the protective clothing is damaged), and location information (by combining multi-view geometric reconstruction and real-time positioning algorithms, it accurately outputs the coordinates of personnel in the three-dimensional work space and associates them with pre-divided functional areas such as "loading area," "outside the warning line," and "blasting avoidance point."

[0039] The model design is optimized for the specific challenges of blasting sites. On the one hand, it utilizes the spatial complementarity of multi-source video streams. When a certain perspective becomes difficult to identify due to occlusion (such as equipment obstruction) or environmental interference (such as smoke and dust), it automatically switches or weights and fuses reliable features from other perspectives to ensure the robustness of the analysis results. On the other hand, the model uses a transfer learning mechanism to fine-tune specific blasting scenarios (such as low light underground or strong reflective surfaces in the open air) based on pre-trained general features. This significantly improves the recognition accuracy of identifying elements (such as reflective vest markings and warning tape numbers required by the "General Rules for Warning and Registration Markings of Civil Explosives"), overcoming the problem of high false detection rates in complex environments caused by traditional methods.

[0040] This invention utilizes a deep learning model to perform comprehensive analysis of personnel identity, behavior, equipment, and location, providing data support for subsequent rule verification and completely replacing the inefficient manual frame-by-frame review method. Secondly, the analysis process is real-time and highly accurate, instantly capturing momentary violations such as "unlicensed personnel entering the loading area" and "failure to wear anti-static gloves during the connection phase." Finally, the multi-view fusion mechanism effectively solves the blind spot problem of single cameras, ensuring full coverage of supervision over hidden corners or densely populated areas, and fundamentally avoiding compliance risks caused by visual omissions.

[0041] Regarding the deep learning model, it adopts an ensemble architecture based on multi-task learning. The core of this architecture is a shared visual feature extraction backbone network; for example, this backbone network can use an I3D structure pre-trained on large-scale video datasets such as Kinetics, which contains multiple 3D convolutional layers, 3D pooling layers, and non-linear activation layers, effectively extracting spatiotemporal features from the video. To process multi-source video streams, the system equips each video source with the aforementioned backbone network of identical structure for independent feature extraction, and then integrates the multi-view features through a feature fusion layer; this fusion layer may include channel concatenation operations and one or more subsequent 3D convolutional layers for feature dimensionality reduction and enhancement, ultimately generating a unified multi-view joint feature representation.

[0042] This joint feature representation integrates visual information from different perspectives, providing a common and rich data foundation for multiple subsequent recognition tasks. Based on this, the model constructs four specific task-specific parsing branches in parallel: identity parsing branch, behavior parsing branch, equipment detection branch, and location parsing branch.

[0043] Specifically, the identity resolution branch may include a Region Proposal Network (RPN), an ROI alignment layer, and two fully connected layers for performing personnel detection, bounding box regression, and identity classification, and can fuse the output of a dedicated face recognition sub-network. The behavior resolution branch includes a temporal modeling module (such as a Long Short-Term Memory network layer or a spatiotemporal graph convolutional layer), followed by a fully connected classification layer, for temporal segmentation and recognition of continuous behaviors. The equipment detection branch employs an anchor-based target detection structure (such as Faster R-CNN or YOLO variants), containing several 2D convolutional layers, upsampling layers, and prediction layers, for locating and classifying various security equipment and determining its wearing status. The location resolution branch is constructed as an encoder-decoder semantic segmentation network (such as U-Net); the encoder extracts features through multiple convolutional and pooling layers, while the decoder restores spatial resolution through deconvolutional layers or upsampling operations, finally outputting spatial region labels for personnel through a pixel-by-pixel classification layer.

[0044] The training of the deep learning model adopts a phased strategy. First, it is pre-trained on large-scale general datasets such as Kinetics and COCO to initialize network parameters and build basic feature extraction capabilities. Then, it is fine-tuned using data specific to blasting operation scenarios. The dedicated dataset required for fine-tuning comes from multi-view videos of real blasting sites, covering different lighting conditions, weather interference, and operation stages, and is labeled with four types of ground truth data by professionals: identity, behavior, equipment, and location.

[0045] Training employed a multi-task joint loss function, where cross-entropy loss was used for identity recognition, temporal focus loss was used for behavior sequences to focus on key actions, equipment detection combined bounding box regression and classification loss, and location parsing used Euclidean distance loss and region classification loss. The optimization process used an adaptive moment estimation optimizer, supplemented by a dynamic learning rate adjustment strategy. Data augmentation techniques such as random occlusion simulation and illumination perturbation were introduced during training to effectively improve the model's generalization ability to complex on-site disturbances.

[0046] Furthermore, to meet deployment requirements, the model can be compressed using knowledge distillation technology to generate a lightweight version suitable for edge computing nodes, ensuring real-time processing performance. During the model validation phase, an independent test set is used for evaluation, requiring key indicators to reach set thresholds (e.g., identity recognition accuracy > 95%, behavior sequence F1 score > 0.92, equipment detection mAP > 0.89, positioning error < 0.3 meters). Continuous incremental learning is performed based on misjudged samples from real-world scenarios, thereby ensuring the model's adaptability and accuracy in different demolition environments and providing reliable data support for compliance verification.

[0047] S105: Call the set of compliance rules in the pre-built security rule base that matches the current operation stage and scenario risk factors.

[0048] In application, to specifically verify the compliance of blasting operations, safety rules need to be dynamically scheduled and adapted to the current situation. This is based on a pre-built safety rule library—formed through structured parsing of standards such as the "Blasting Safety Regulations" and the "General Rules for Warning Signs and Registration Marks of Civil Explosives"—containing hundreds of atomic rules covering the entire process, and stored with multi-level indexes according to operation stages (preparation, charging, connection, warning, detonation, and disarming) and risk dimensions (environment, behavior, equipment, and space).

[0049] Once the current operation stage and scenario risk factors are obtained, the rule base is quickly retrieved through a dual matching mechanism. Specifically, the baseline rule set that must be enforced in the operation stage is first selected based on the operation stage (for example, the loading stage must include rules such as "anti-static equipment verification" and "single-person operation time limit"). Then, incremental rules are dynamically loaded based on scenario risk factors (for example, when the "high wind speed" risk factor is identified, rules such as "heavy object reinforcement inspection" and "personnel prohibited from entering the flying rock protection zone" are added). Finally, a compliance rule set that takes effect in real time is generated.

[0050] The matching process employs a hybrid retrieval strategy based on hash tables and decision trees. The job stage serves as the primary key to quickly locate a subset of rules, while scenario risk factors activate associated rules through a weight matrix (e.g., the "methane concentration exceeding the standard" factor activates high-risk rules such as "interrupt operation" and "evacuation order," while "medium level of rain and fog interference" only triggers adaptive rules such as "enhanced location verification" and "reduced behavior recognition confidence threshold"). To ensure the accuracy of rule scheduling, each rule has a pre-defined activation condition expression (e.g., "wind speed > 8 m / s, stage = alert"), which is parsed and executed in real time by the rule engine.

[0051] This invention dynamically loads rules based on the current operational stage and scenario risk factors, adapting to actual operational progress and environmental threats (e.g., automatically increasing the verification intensity of anti-slip equipment when encountering sudden rainstorms during the connection stage). This overcomes the deficiency of static rules in existing technologies, which cannot dynamically adjust protection requirements. Furthermore, the refined pruning of the rule set avoids redundant verification, significantly improving processing efficiency (e.g., retaining only core rules in low-risk stages). More importantly, this mechanism provides flexible adaptability to compliance requirements under different risk scenarios—for example, when the scenario risk factor identifies "high gas concentration + low visibility," the compliance rule set will simultaneously strengthen the priority of "personnel positioning accuracy verification" and "automatic shutdown command," building an adaptive safety protection network at the rule level to effectively prevent the occurrence of complex risk events.

[0052] S106: Based on the compliance rule set, perform compliance verification on identity information, behavior sequence, equipment wearing status and location information.

[0053] The compliance rule set includes: qualification rules, behavior rules, equipment rules, and space rules.

[0054] In application, the structured information parsed from S104 is validated based on the dynamically generated compliance rule set. The validation process is executed in four categories according to the rule type, as detailed below.

[0055] Qualification rule verification: The identity information (such as personnel ID, role category, certificate number) is compared with the preset license list and permission matrix in the rule set. If an expired qualification certificate is detected, the operator's role conflicts with the current stage's prohibited access permissions (such as a non-safety officer appearing in the detonation zone during the alert stage), or unlicensed personnel are detected (such as unregistered faces or illegal uniform codes), then it is determined as a qualification violation.

[0056] Behavioral rule verification: Match the behavioral sequence (e.g., "handling explosives → loading boreholes → connecting detonators → evacuation") with the standard operating procedure (SOP) timing template for the current stage within the rule set. Detect missing critical actions (e.g., "electrostatic release operation not detected after loading"), out-of-order operations (e.g., "trigger detection signal triggered before connection completion"), or high-risk redundant actions (e.g., "returning to the loading area during detonation") using a dynamic time warping (DTW) algorithm, and mark behavioral deviations in real time.

[0057] Equipment rule verification: Based on target detection results, the system verifies whether the equipment wearing status meets the stage requirements. For example, during the explosive loading stage, the integrity of the antistatic clothing (including no damage to the clothing and a connected grounding strap) and the wearing of goggles are forcibly verified; during the detonation stage, the focus is on the disabled status of wireless communication equipment. The rule engine analyzes the equipment attributes and rule conditions in real time (such as "protective clothing type = level A, current stage = explosive loading"). When it detects missing, malfunctioning (such as helmet straps not fastened) or mismatched types (such as ordinary work clothes replacing antistatic clothing), a violation judgment is triggered.

[0058] Spatial rule verification: The location information (3D coordinates and associated area labels) is compared with the electronic fence database. Verification includes intrusion into prohibited areas (e.g., unauthorized personnel entering a temporary explosives storage area), exceeding personnel gathering limits (e.g., the number of people in the restricted area exceeds the preset capacity), and lingering during clearing operations (e.g., personnel still emitting heat signals outside the designated avoidance point after the detonation countdown has ended). Spatial verification integrates real-time positioning and historical trajectory prediction to ensure accurate detection of transient violations (e.g., rapid entry into a restricted area).

[0059] The verification process employs a multi-level confidence mechanism: for ambiguous data caused by environmental interference (such as a 60% confidence level for behavior recognition in rain and fog), the judgment threshold is dynamically adjusted by automatically associating with scenario risk factors (such as reducing the confidence threshold to 50% in high-risk scenarios). Simultaneously, the rule engine supports triggering composite conditions (such as "identity = demolition worker, location = connected area, no anti-static gloves detected"), enabling multi-dimensional violation correlation analysis.

[0060] Specifically, the above-mentioned compliance verification of identity information, behavior sequence, equipment wearing status, and location information based on the compliance rule set includes the following steps: According to the qualification rules, the identity information is verified. When the operator's qualification is invalid or the job authority is exceeded, it is determined that there is a violation. By comparing the behavior sequence with the behavior rules, when a key action is missing, the operation is out of order, or a high-risk redundant action is identified, it is determined that there is a violation. The equipment wearing status is checked according to the equipment rules. When protective equipment is missing, ineffective, or does not match the requirements of the current operation stage, it is determined that there is a violation. Based on spatial rules and location information comparison, when workers are in prohibited areas, exceed gathering limits, or fail to evacuate as required, it is determined that there is a violation.

[0061] S107: If a violation is found during the verification, a violation judgment result containing the violation type, risk level, associated personnel, and spatial area identifier will be generated.

[0062] In the application, when a violation is detected during the compliance verification process, a structured violation judgment result will be generated. This result first precisely identifies the violation type, and its classification strictly corresponds to the four-dimensional architecture of the rule base: qualification violation (such as unlicensed operation or overreach of authority), behavioral violation (such as missing key actions or high-risk redundant operations), equipment violation (such as malfunctioning or incompatible protective equipment), and spatial violation (such as trespassing into restricted areas or exceeding gathering limits). Each type is associated with a preset risk level assessment model—this model is trained based on the nature of the violation (such as whether it directly triggered an explosion), scenario risk factors (such as not wearing anti-static clothing in a high-concentration flammable gas environment), and historical accident data, and outputs four levels of quantitative risk labels: low, medium, high, and emergency; for example, "personnel were not evacuated during the detonation phase" is automatically labeled as "emergency".

[0063] The judgment result is simultaneously linked to the information of related personnel. By extracting the identity identifiers (such as facial ID, employee number) and role attributes (blaster / safety officer) parsed in S104, the responsible person is accurately identified, and their behavioral sequence fragments are linked as corroborating evidence. Spatial area identification is based on the electronic fence system and real-time positioning data, marking the specific functional area where the violation occurred (such as "northeast corner of the explosive loading area" or "5 meters outside the warning line"), supplemented by three-dimensional coordinates and regional risk attributes (such as "high-risk gas accumulation area"). Finally, the four elements (violation type, risk level, related personnel, and spatial area identification) are encapsulated into standardized data objects for use in early warning and evidence chain construction.

[0064] This invention enables precise output and rapid response in risk decision-making. First, structured judgment results replace traditional vague textual descriptions, allowing the command center to instantly understand the nature of violations (e.g., "Spatial violation - Emergency level: Zhang San is lingering in the detonation zone"), significantly improving emergency decision-making efficiency. Second, automated risk level assessment (e.g., combining "high wind speed" scenario factors to upgrade "personnel gathering in the fly rock protection zone" to high risk) solves the problem of strong subjectivity in manual judgment, ensuring that resources are prioritized for truly high-risk events. Finally, binding personnel and spatial information provides core elements for subsequent electronic evidence chains, fundamentally avoiding disputes during liability tracing and significantly strengthening the authority and enforceability of safety supervision.

[0065] In one embodiment, before inputting the multi-source video streams into the pre-trained deep learning model, the method further includes: Acquire point cloud data and on-site audio stream from the blasting operation site; The intensity of environmental interference was determined based on environmental monitoring data and multi-source video streams. When the intensity of environmental interference exceeds a preset threshold, the multi-source video stream is input into the pre-trained deep learning model. Specifically, the multi-source video stream, point cloud data, and on-site audio stream are input into the pre-trained deep learning model. Point cloud data is used to supplement spatial positioning information, while on-site audio streams are used to extract acoustic event features associated with standard operating procedures.

[0066] In the application, before inputting multi-source video streams into a pre-trained deep learning model for parsing, a multimodal data augmentation acquisition and adaptive triggering process can be executed. Specifically, firstly, source sensing devices are added to key areas of the blasting operation site to simultaneously acquire point cloud data generated by millimeter-wave radar and on-site audio streams captured by a high-sensitivity acoustic sensor array. The point cloud data provides spatial depth information unconstrained by visible light conditions and can penetrate a certain degree of smoke and water mist interference; the on-site audio stream fully records various sound events in the operation environment, including human voices, mechanical operation sounds, and blasting equipment assembly sounds.

[0067] Continuous joint analysis of environmental monitoring data and multi-source video streams is conducted to comprehensively assess the current intensity of environmental interference. Specifically, this involves comparing real-time wind speed, rainfall, visual smoke and dust concentration, and visibility estimates with corresponding preset values, and introducing a machine learning-based fusion evaluation model to output a quantitative environmental interference intensity score. This score reflects the degree of perception challenges currently faced by visible light visual sensors.

[0068] When the environmental interference intensity is determined to exceed a preset threshold based on the score, meaning that the visual perception conditions have significantly deteriorated to the point that may affect the recognition accuracy, the multimodal fusion analysis mechanism is automatically triggered. Under this mechanism, the step of inputting multi-source video streams into the pre-trained deep learning model is specifically extended to simultaneously inputting multi-source video streams, point cloud data, and on-site audio streams into the deep learning model. The point cloud data is mainly used to supplement and provide more robust 3D spatial positioning and motion trajectory information of personnel and large equipment in situations of visual occlusion or low visibility. The on-site audio stream is processed through the acoustic event recognition branch of the model, aiming to extract acoustic event features closely related to the standard operating procedures of blasting operations, such as the knocking sound of specific tools, the repetition of safety commands, and the specific friction sound during explosive loading. These acoustic features can serve as an effective supplement and cross-validation of visual behavior sequences.

[0069] By introducing the aforementioned multimodal data adaptive fusion mechanism, the accuracy of behavior analysis in harsh environments is significantly improved, ensuring the continuity and reliability of the compliance monitoring process.

[0070] Furthermore, the deep learning model analyzes the identity information, behavior sequences, equipment wearing status, and location information of each worker based on multi-source video streams, point cloud data, and on-site audio streams. This process deeply integrates multi-source heterogeneous data to improve the robustness and accuracy of the analysis. The model first processes the point cloud data, extracting and tracking the dynamic contours of each worker through point cloud clustering and 3D target tracking algorithms, thereby generating their precise 3D spatial coordinates and motion trajectory information. This effectively compensates for the potential failure of purely visual positioning in situations of visual occlusion or extremely low visibility.

[0071] Simultaneously, the model performs real-time analysis of the on-site audio stream, using acoustic event detection and sound source localization technology to separate and identify specific acoustic events strongly correlated with standard blasting operations from complex background noise, such as the crisp knocking sound when detonators are connected, the rumbling sound of a specific type of drilling rig, or the standardized safety commands between personnel, and associates these acoustic events with their approximate location.

[0072] Subsequently, the model's built-in cross-modal feature alignment module begins operation. This module is responsible for aligning and mapping spatial positioning information from point cloud data, acoustic event features from audio streams, and multi-view joint feature representations generated by feature fusion of multi-source video streams (which include visual information such as personnel appearance, action posture, and equipment shape) onto a unified spatiotemporal reference. Based on this, the module employs an attention-based weighted fusion strategy to adaptively fuse the features from the three modalities, ultimately generating a unified and enhanced multimodal feature representation.

[0073] It should be noted that the weighting coefficients in this weighted fusion process are not fixed, but dynamically adjusted by the model based on the real-time assessment of environmental interference intensity. For example, when environmental interference intensity is extremely high (such as dense smoke) causing a severe deterioration in visual feature quality, the model automatically increases the fusion weight of point cloud data and audio stream features, relying on the spatial perception capability of radar and the temporal characteristics of acoustic events to maintain the ability to extract key information; conversely, when visual conditions are good, it relies more on rich visual features. Ultimately, based on this dynamically fused multimodal feature representation, the model, through its parallel multiple task parsing branches, stably and accurately outputs the identity information of each operator, continuous behavioral sequences, complete equipment wearing status, and precise location information.

[0074] In one embodiment, the above method further includes: When generating violation determination results, extract real-time video segments from multi-source video streams that correspond to the violation type, associated personnel, and spatial area identifiers; Add timestamps and spatial location markers to real-time video clips, and associate and bind them with violation judgment results, environmental monitoring data and the current operation stage to generate electronic evidence data packages; Add digital signatures and hash values ​​to electronic evidence data packets to generate tamper-proof electronic evidence records; Electronic evidence records are stored in an tamper-proof storage system to form a traceable chain of electronic evidence.

[0075] In the application, once a violation determination result is generated, the process of constructing an electronic evidence chain will be automatically initiated. Specifically, based on the violation type, the identity of the associated personnel, and the spatial area where the violation occurred contained in the violation determination result, the original real-time video clips that precisely correspond to these criteria from multiple source video streams are retrieved and extracted. This extraction process ensures that the video evidence is completely synchronized with the moment the violation occurred in time and covers the specific area view where the violation occurred in space, thereby realistically and completely reconstructing the scene of the violation.

[0076] Subsequently, the extracted raw real-time video clips undergo enhanced tagging processing. High-precision timestamp information is overlaid; simultaneously, spatial location markers based on on-site 2D or 3D electronic maps are overlaid, clearly indicating the specific geographical location or functional area where the violation occurred. After tagging, the video clip is structurally associated and bound with the currently generated violation determination result, the environmental monitoring data snapshot at the time of the violation, and the identified current operation stage information, encapsulating them into a complete and interconnected electronic evidence data package. This data package has a unified internal logic, allowing retrieval and tracing of all other related information through any element.

[0077] To ensure the authenticity and integrity of evidence and prevent tampering during transmission, storage, or retrieval, cryptographic techniques can be used to strengthen electronic evidence data packets. An asymmetric encryption algorithm is used to generate a digital signature for the data packet, and the packet content is then hashed to generate a unique hash value. The digital signature verifies the legitimacy of the evidence's source and the integrity of its content, while the hash value serves as the data packet's "digital fingerprint"—any minor alteration will invalidate the hash value. This generates an immutable electronic evidence record.

[0078] Finally, the generated electronic evidence records are instantly transmitted and stored in a dedicated tamper-proof storage system. This storage system can employ Write Once Read Many (WORM) technology or blockchain-based distributed storage technology to eliminate the possibility of records being modified or deleted afterward, both physically and in terms of system architecture. All records are indexed and stored according to time sequence and event type, forming a complete, clear, and traceable chain of electronic evidence.

[0079] Corresponding to the aforementioned application function implementation method embodiments, the present invention also provides a blasting operation site behavior compliance identification system and corresponding embodiments.

[0080] Please see Figure 2 , Figure 2 This is a schematic diagram of the module structure of a compliance identification system for blasting operations.

[0081] The blasting operation site behavior compliance identification system includes: Acquisition unit 21 is used to acquire multi-source video streams and environmental monitoring data from the blasting operation site; the environmental monitoring data includes temperature and humidity, wind speed, and hazardous gas concentration. The job phase identification unit 22 is used to identify the current job phase based on the multi-source video stream; Scene risk analysis unit 23 is used to identify scene risk factors based on multi-source video streams and environmental monitoring data; The personnel behavior analysis unit 24 is used to input multi-source video streams into a pre-trained deep learning model to analyze the identity information, behavior sequence, equipment wearing status and location information of each operator. The rule dynamic scheduling unit 25 is used to call the set of compliance rules in the pre-built security rule library that matches the current operation stage and scenario risk factors; The compliance verification unit 26 is used to perform compliance verification on identity information, behavior sequence, equipment wearing status and location information based on the compliance rule set; The violation decision generation unit 27 is used to generate a violation judgment result containing the violation type, risk level, related personnel, and spatial area identifier if a violation is found during verification.

[0082] In one embodiment, the job stage identification unit 22 is specifically used for identifying the current job stage based on multi-source video streams: Spatiotemporal synchronization processing is performed on multi-source video streams to generate calibrated multi-view video sequences; Extract equipment operation features and personnel collaboration features from multi-view video sequences; Based on preset work phase transition rules, a time-series correlation analysis is performed on equipment operation characteristics and personnel collaboration characteristics; Based on the correlation analysis results, the current operation stage is determined.

[0083] In one embodiment, the scenario risk analysis unit 23 is specifically used for identifying scenario risk factors based on multi-source video streams and environmental monitoring data: Environmental monitoring data is compared with preset safety thresholds to generate environmental risk levels; Analyze multi-source video streams to extract environmental interference factors such as smoke and dust concentration, rain and fog interference intensity, and the coverage area of ​​muddy areas on the ground. Based on pre-built risk mapping rules, environmental risk levels and environmental disturbance factor characteristics are integrated to generate scenario risk factors.

[0084] In one embodiment, regarding compliance verification of identity information, behavior sequence, equipment wearing status, and location information based on a set of compliance rules, the aforementioned compliance verification unit 26 is specifically used for: According to the qualification rules, the identity information is verified. When the operator's qualification is invalid or the job authority is exceeded, it is determined that there is a violation. By comparing the behavior sequence with the behavior rules, when a key action is missing, the operation is out of order, or a high-risk redundant action is identified, it is determined that there is a violation. The equipment wearing status is checked according to the equipment rules. When protective equipment is missing, ineffective, or does not match the requirements of the current operation stage, it is determined that there is a violation. Based on spatial rules and location information comparison, when workers are in prohibited areas, exceed gathering limits, or fail to evacuate as required, it is determined that there is a violation.

[0085] In one embodiment, the acquisition unit 21 is further configured to: Acquire point cloud data and on-site audio stream from the blasting operation site; The above system also includes: The environmental interference determination unit is used to determine the intensity of environmental interference based on environmental monitoring data and multi-source video streams. In terms of inputting multi-source video streams into a pre-trained deep learning model, the aforementioned personnel behavior parsing unit 24 is specifically used for: When the intensity of environmental interference exceeds a preset threshold, multi-source video streams, point cloud data, and on-site audio streams are input into a pre-trained deep learning model. Point cloud data is used to supplement spatial positioning information, while on-site audio streams are used to extract acoustic event features associated with standard operating procedures.

[0086] In one embodiment, the system further includes: The evidence chain generation unit is used for: When generating violation determination results, extract real-time video segments from multi-source video streams that correspond to the violation type, associated personnel, and spatial area identifiers; Add timestamps and spatial location markers to real-time video clips, and associate and bind them with violation judgment results, environmental monitoring data and the current operation stage to generate electronic evidence data packages; Add digital signatures and hash values ​​to electronic evidence data packets to generate tamper-proof electronic evidence records; Electronic evidence records are stored in an tamper-proof storage system to form a traceable chain of electronic evidence.

[0087] Regarding the system in the above embodiments, the specific manner in which each unit module performs operations has been described in detail in the embodiments related to the method, and will not be elaborated further here.

[0088] This invention also provides an electronic device, including a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described positioning method.

[0089] Please see Figure 3 The electronic device 3000 includes a memory 3010 and a processor 3020.

[0090] The processor 3020 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0091] Memory 3010 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. ROM may store static data or instructions required by processor 3020 or other modules of the computer. Permanent storage devices may be read-write storage devices. Permanent storage devices may be non-volatile storage devices that retain stored instructions and data even when the computer is powered off. In some embodiments, permanent storage devices use mass storage devices (e.g., magnetic or optical disks, flash memory) as permanent storage devices. In other embodiments, permanent storage devices may be removable storage devices (e.g., floppy disks, optical drives). System memory may be a read-write storage device or a volatile read-write storage device, such as dynamic random access memory. System memory may store some or all of the instructions and data required by the processor during operation. Furthermore, memory 3010 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (e.g., DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory 3010 may include a removable storage device that is readable and / or writable, such as a laser disc (CD), a read-only digital multifunction optical disc (e.g., DVD-ROM, dual-layer DVD-ROM), a read-only Blu-ray disc, an ultra-high density optical disc, a flash memory card (e.g., SD card, minSD card, Micro-SD card, etc.), a magnetic floppy disk, etc. Computer-readable storage media do not contain carrier waves or transient electronic signals transmitted wirelessly or via wired connections.

[0092] The memory 3010 stores executable code, which, when processed by the processor 3020, can cause the processor 3020 to execute part or all of the methods described above.

[0093] Furthermore, the method according to the present invention can also be implemented as a computer program or computer program product, which includes computer program code instructions for performing some or all of the steps in the above-described method of the present invention.

[0094] Alternatively, the present invention may also be implemented as a computer-readable storage medium (or a non-transitory machine-readable storage medium or a machine-readable storage medium) storing executable code (or computer program or computer instruction code) thereon, which, when executed by a processor of an electronic device (or server, etc.), causes the processor to perform part or all of the steps of the methods described above according to the present application.

[0095] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for identifying compliance in blasting operations, characterized in that, include: Acquire multi-source video streams and environmental monitoring data from the blasting operation site; The environmental monitoring data includes temperature, humidity, wind speed, and concentration of hazardous gases; The current operation stage is identified based on the multi-source video stream; Based on the multi-source video streams and environmental monitoring data, identify scene risk factors; The multi-source video streams are input into a pre-trained deep learning model to parse out the identity information, behavior sequence, equipment wearing status and location information of each operator. Call the set of compliance rules in the pre-built security rule base that matches the current operation stage and scenario risk factors; Based on the aforementioned compliance rule set, compliance verification is performed on the identity information, behavior sequence, equipment wearing status, and location information. If a violation is found during the verification, a violation determination result will be generated, which includes the violation type, risk level, associated personnel, and spatial area identifier.

2. The method for identifying compliance of blasting operation site behavior according to claim 1, characterized in that, Identifying the current operation stage based on the multi-source video streams includes: The multi-source video streams are spatiotemporally synchronized to generate a calibrated multi-view video sequence; Extract equipment operation features and personnel collaboration features from the multi-view video sequence; Based on preset work phase transition rules, a time-series correlation analysis is performed on the equipment operation characteristics and personnel collaboration characteristics; Based on the correlation analysis results, the current operation stage is determined.

3. The method for identifying compliance of blasting operation site behavior according to claim 1, characterized in that, Based on the multi-source video streams and environmental monitoring data, scene risk factors are identified, including: The environmental monitoring data is compared with a preset safety threshold to generate an environmental risk level. Analyze the multi-source video streams to extract environmental interference factors such as smoke and dust concentration, rain and fog interference intensity, and the coverage area of ​​muddy areas on the ground; Based on pre-constructed risk mapping rules, the environmental risk level and environmental disturbance factor characteristics are integrated to generate scenario risk factors.

4. The method for identifying compliance of blasting operation site behavior according to claim 1, characterized in that, The compliance rule set includes: qualification rules, behavioral rules, equipment rules, and spatial rules; based on the compliance rule set, compliance verification is performed on the identity information, behavioral sequence, equipment wearing status, and location information, including: The identity information is verified according to the qualification rules. When the operator's qualification is invalid or the job authority is exceeded, it is determined that there is a violation. According to the behavior rules, the behavior sequence is compared. When a key action is missing, the operation is out of order, or a high-risk redundant action is identified, it is determined that there is a violation. The equipment wearing status is verified according to the equipment rules. When the protective equipment is missing, malfunctioning, or does not match the requirements of the current operation stage, it is determined that there is a violation. Based on the spatial rules and the location information, if the workers are in a prohibited area, exceed the gathering limit, or fail to evacuate as required, it is determined that there is a violation.

5. The method for identifying compliance of blasting operation site behavior according to claim 1, characterized in that, The method further includes: When the violation determination result is generated, real-time video segments corresponding to the violation type, associated personnel, and spatial area identifier are extracted from the multi-source video stream; Add timestamps and spatial location markers to the real-time video clips, and associate and bind them with the violation judgment results, environmental monitoring data and the current operation stage to generate an electronic evidence data package; Add a digital signature and hash value to the electronic evidence data packet to generate an immutable electronic evidence record; The electronic evidence records are stored in an anti-tampering storage system to form a traceable chain of electronic evidence.

6. The method for identifying compliance of blasting operation site behavior according to claim 1, characterized in that, Before inputting the multi-source video stream into a pre-trained deep learning model, the method further includes: Acquire point cloud data and on-site audio stream from the blasting operation site; The intensity of environmental interference is determined based on the environmental monitoring data and the multi-source video stream; When the intensity of the environmental interference exceeds a preset threshold, the multi-source video stream is input into a pre-trained deep learning model. Specifically, the multi-source video stream, point cloud data, and on-site audio stream are input into the pre-trained deep learning model. The point cloud data is used to supplement spatial positioning information, and the on-site audio stream is used to extract acoustic event features associated with standard operating procedures.

7. A compliance identification system for blasting operation site behavior, characterized in that, include: The acquisition unit is used to acquire multi-source video streams and environmental monitoring data from the blasting operation site. The environmental monitoring data includes temperature, humidity, wind speed, and concentration of hazardous gases; A job phase identification unit is used to identify the current job phase based on the multi-source video stream; The scene risk analysis unit is used to identify scene risk factors based on the multi-source video streams and environmental monitoring data; The personnel behavior analysis unit is used to input the multi-source video stream into a pre-trained deep learning model to analyze the identity information, behavior sequence, equipment wearing status and location information of each operator. The rule dynamic scheduling unit is used to call the set of compliance rules in the pre-built security rule library that matches the current operation stage and scenario risk factors; The compliance verification unit is used to perform compliance verification on the identity information, behavior sequence, equipment wearing status and location information based on the compliance rule set. The violation decision generation unit is used to generate a violation judgment result that includes the violation type, risk level, related personnel, and spatial area identifier if a violation is found during verification.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the identification method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When a computer program is executed by a processor, it implements the identification method as described in any one of claims 1-6.

Images