A production safety inspection and monitoring integrated simulation modeling management and control system

Abstract

The application provides a production safety inspection and monitoring integrated simulation modeling management and control system, relates to the technical field of production safety monitoring and simulation modeling, and has the technical scheme as follows: multi-source input data of a pipe gallery area are acquired, inspection and monitoring candidate observation relations are generated for each hidden danger point, visible boundary checking containing sight line continuity and observation angle conditions is performed on each candidate observation relation, effective observation relations are determined, the checking state of the hidden danger point is determined and coverage state data are constructed according to the effective observation relations, therefore, the actual observability of the hidden danger point in a multi-view intersection scene can be accurately identified, coverage determination deviation caused by a view angle fault is avoided, and the production safety coverage determination accuracy and reliability are improved.

Description

Technical Field

[0001] This application relates to the field of production safety monitoring and simulation modeling technology, specifically to an integrated simulation modeling and control system for production safety inspection and monitoring. Background Technology

[0002] In industrial production safety management, to achieve integrated simulation modeling and control of production safety inspections and monitoring, the conventional approach is to create a factory area model in a virtual 3D space, plan the movement paths and stopping points of inspection personnel, and mark the installation locations and monitoring ranges of fixed monitoring cameras. The system evaluates the overall safety coverage in the simulation platform by statically overlaying the inspection routes with the camera coverage. However, in densely populated pipe corridor areas, pipelines are intricate, equipment is arranged in high and low layers, and there are numerous corners, supports, and partial obstructions. In such complex scenarios, the dynamic observation perspective of inspection personnel moving on ground passages, combined with the static observation perspective of fixed monitoring equipment installed at high or side positions, constitutes the core logic of multi-object spatial correlation in determining the coverage of potential hazards such as valves and flange connection points within the pipe corridor. When the inspection personnel move along the main channel, they can observe the back side of the equipment or low-level blind spots that the fixed monitoring equipment cannot reach due to height difference and angle limitations. However, although some high-level pipeline connections are within the field of view of the fixed monitoring equipment, the inspection personnel cannot form an effective observation angle when inspecting from the ground due to the distance and the existence of viewing angle deviation.

[0003] If, during simulation modeling, all potential hazards located near the inspection path and within the camera's field of view are simply marked as covered, the overlap, compensation, and gaps between the two perspectives at specific hazard points will be ignored. This discontinuity in perspective coverage causes the simulation system to deviate in determining whether hazard points have been effectively verified. Areas that appear to be doubly covered may actually contain unobserved gaps. Existing simulation modeling and control methods typically use simplified bounding boxes or distance thresholds to determine coverage relationships, failing to accurately depict visible boundaries and occlusion logic from different perspectives. This easily equates path passage or inclusion in the view with effective verification, thus failing to accurately identify this perspective discontinuity. Consequently, inspection tasks and monitoring strategies output by the control end contain security loopholes.

[0004] Therefore, in densely equipped utility tunnel areas, when high and low-level equipment and corner obstructions coexist, the visible boundaries of the same hidden danger point are inconsistent between the close-range dynamic perspective of the inspection personnel and the long-range static perspective of the fixed monitoring equipment. This leads to a deviation in the determination of the coverage relationship of the hidden danger point, where the surface appears to be covered but the actual situation has not been verified. This, in turn, affects the integrated simulation modeling and control of production safety inspection and monitoring.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] The purpose of this application is to provide an integrated simulation modeling and control system for production safety inspection and monitoring. This system has the advantages of being able to accurately identify potential hazards in pipe gallery areas with high and low-level equipment and obstructing structures, avoiding coverage omissions caused by perspective breaks, and improving the accuracy and reliability of production safety coverage determination.

[0007] This application provides an integrated simulation modeling and control system for production safety inspection and monitoring, the technical solution of which is as follows:

[0008] include:

[0009] The acquisition unit is used to acquire input data, which includes the location and local observation direction information of each hidden danger point in the utility tunnel area, the pose information of the inspection path, the pose and field of view information of the fixed monitoring equipment, and the geometric information of the obstruction structure.

[0010] The generation unit is used to generate inspection candidate observation relationships for each potential hazard point based on the location of the potential hazard point and the pose information of the inspection path, and to generate monitoring candidate observation relationships based on the location of the potential hazard point and the pose and field of view information of the fixed monitoring equipment.

[0011] The verification unit is used to perform visible boundary verification on each candidate observation relationship. The visible boundary verification includes: a line-of-sight check to determine whether the line of sight is blocked based on the geometric information of the occlusion structure, and an observation angle check to determine whether the observation angle meets the identification conditions based on the local observation direction information of the hidden danger point. The candidate observation relationship that passes the visible boundary verification is determined as a valid observation relationship.

[0012] The verification unit is used to determine the verification status of each potential hazard point based on the effective observation relationship of that hazard point. The verification status is at least divided into patrol inspection effectively undertaken, monitoring effectively undertaken, or both patrol inspection and monitoring are invalid.

[0013] The construction unit is used to build coverage status data of the simulation modeling and control system based on the verification status of each hidden danger point, so that the coverage conclusion in the coverage status data corresponds to the valid verification results at the hidden danger point level.

[0014] Furthermore, this application also proposes performing a visible boundary check on each candidate observation relationship, including:

[0015] Obtain the type of potential hazards at the detection point;

[0016] Identify the key areas corresponding to the hazard points based on their types;

[0017] In the line-of-sight inspection, the geometric information of the obstruction structure is used to determine whether the key area of ​​the potential hazard point is obstructed. If the key area of ​​the potential hazard point is obstructed, the line-of-sight inspection of the candidate observation relationship is deemed to be unsuccessful.

[0018] Obtain the observation subject type corresponding to the candidate observation relationship. The observation subject type is either patrol personnel or fixed monitoring equipment.

[0019] Based on the type of the subject being observed and the type of potential hazard, determine the corresponding local identifiable conditions. The local identifiable conditions should include at least the distance condition, the unobstructed ratio condition, and the inspection dwell time condition or the monitoring image area ratio condition.

[0020] In the observation angle check, the candidate observation relationship is judged to meet the identification requirements based on the local identifiable conditions.

[0021] A candidate observation relationship is deemed to have passed the visible boundary check only if it simultaneously meets the conditions that the key area of ​​the hidden danger point is not obstructed in the line-of-sight continuity check and the condition that it is locally identifiable in the observation angle check.

[0022] Furthermore, this application proposes to determine candidate observation relationships that pass the visible boundary check as valid observation relationships, including:

[0023] For the inspection candidate observation relationship that has passed the visible boundary verification, the inspection observation interval is determined based on the pose information of the inspection path.

[0024] Obtain at least one of the following: continuous information, persistent information, orientation naturalness information, speed adaptation information, and historical repetitive information of the inspection observation area;

[0025] At least based on the information obtained, determine whether the inspection and observation range is a stable window;

[0026] Only the inspection candidate observation relationships identified as stable windows are determined as valid observation relationships.

[0027] Furthermore, this application also proposes methods for determining whether an inspection observation interval is a stable window, including:

[0028] Obtain the continuity indicators, persistence indicators, orientation naturalness indicators, speed adaptation indicators, and historical repeatability indicators corresponding to the inspection and observation interval;

[0029] Based on the continuity index, persistence index, orientation naturalness index, speed adaptation index, historical repeatability index, and their respective weighting coefficients, the comprehensive stability index of the inspection observation range is calculated.

[0030] The overall stability index is compared with the preset stability threshold and the preset instantaneous threshold. The stability threshold is greater than the instantaneous threshold.

[0031] When the comprehensive stability index is greater than or equal to the stability threshold, the inspection observation interval is determined as the stability window;

[0032] When the overall stability index is less than the instantaneous threshold, the inspection observation interval is determined as the instantaneous window;

[0033] When the overall stability index is between the stability threshold and the instantaneous threshold, the inspection observation interval is determined as the window to be reviewed or the condition window.

[0034] Instantaneous windows, windows awaiting review, and conditional windows are not considered stable windows.

[0035] Furthermore, this application also proposes, and includes:

[0036] When the inspection observation range is determined to be a condition window that requires additional actions to form an effective observation based on at least one of continuous information, persistent information, orientation natural information, speed adaptation information and historical repeatability information, action constraint requirements are generated, and the action constraint requirements are associated with the condition window and the corresponding hidden danger point.

[0037] Write the action constraints into the inspection task;

[0038] Obtain the actual inspection trajectory, verify whether the action constraint requirements have been executed based on the actual inspection trajectory, and whether the inspection observation interval corresponding to the condition window has passed the visible boundary verification again after the action is executed;

[0039] If the validation result is successful, the condition window is updated to show valid observations.

[0040] Furthermore, this application also proposes determining the verification status of the potential hazard point, including:

[0041] Based on the effective observation relationship of the potential hazard points, the verification status is determined as one of the following: effective undertaking of inspection, effective undertaking of monitoring, joint undertaking of inspection and monitoring, inspection candidate exists but no effective observation relationship is formed, monitoring candidate exists but no effective observation relationship is formed, or both inspection and monitoring are ineffective undertakings.

[0042] Generate failure cause identifiers and compensation direction identifiers for potential hazards. The failure cause identifiers indicate the reasons why the potential hazard was not effectively verified, and the compensation direction identifiers indicate the suggested directions for supplementary verification or blind spot filling for the potential hazard.

[0043] Furthermore, this application also proposes, and includes:

[0044] Acquire actual inspection trajectory and pose information and actual online status information of monitoring equipment;

[0045] The actual inspection trajectory pose information is used as the pose information of the inspection path, and the fixed monitoring equipment currently in online status is determined based on the actual online status information of the monitoring equipment.

[0046] For each potential hazard point, the inspection candidate observation relationship is regenerated based on the location of the potential hazard point and the actual inspection trajectory pose information, and the monitoring candidate observation relationship is regenerated based on the location of the potential hazard point and the pose and field of view information of the fixed monitoring equipment currently online.

[0047] Perform visible boundary verification on each newly generated candidate observation relationship, and redetermine the verification status of the hidden danger points to obtain the updated verification status;

[0048] Update the coverage status data based on the updated verification status of each potential hazard point;

[0049] For potential hazards that are identified as ineffective for both inspection and monitoring after the update, a supplementary inspection task instruction or early warning message will be generated.

[0050] Furthermore, this application also proposes, and includes:

[0051] Obtain verification feedback information, including on-site verification results, historical omission records, or safety supervision spot check data;

[0052] Based on the verification feedback information, determine the deviation between the verification status in the coverage status data of at least one potential hazard point and the actual verification result;

[0053] Determine the type of visible boundary verification rule corresponding to the deviation;

[0054] Based on the rule type, the rule parameters used for visible boundary verification are modified;

[0055] For the potential vulnerabilities affected by the correction, the steps of generating candidate observation relationships, performing visible boundary verification, and determining the verification status are re-executed to update the coverage status data.

[0056] Furthermore, this application also proposes, and includes:

[0057] Before generating inspection candidate observation relationships based on the location of the hazard point and the pose information of the inspection path, and before generating monitoring candidate observation relationships based on the location of the hazard point and the pose and field of view information of the fixed monitoring equipment, for each hazard point, it is determined whether the input information of the hazard point meets the preset reliability conditions based on the input data.

[0058] If the condition is not met, the potential hazard point is marked as pending confirmation, and when constructing the coverage status data, potential hazard points in the pending confirmation state are prohibited from being included in the coverage conclusion.

[0059] Furthermore, this application also proposes that the coverage status data includes the hazard point identifier, verification status, effective observation relationship source type, observation subject identifier, observation location or observation range, and visible boundary verification results corresponding to each hazard point;

[0060] The effective observation relationship source type is used to indicate that the verification status of the corresponding hidden danger point is formed by the inspection candidate observation relationship, the monitoring candidate observation relationship, or both. The visible boundary verification result is used to record the verification information of the corresponding effective observation relationship through line-of-sight access check and observation angle check.

[0061] As can be seen from the above, the integrated simulation modeling and control system for production safety inspection and monitoring provided in this application generates inspection and monitoring candidate observation relationships for each hidden danger point by acquiring multi-source input data of the pipe gallery area. It then performs visible boundary verification, including line-of-sight connectivity and observation angle conditions, on each candidate observation relationship to determine the effective observation relationship. Based on the effective observation relationship, it determines the verification status of the hidden danger point and constructs coverage status data. Therefore, it can accurately identify the actual observability of hidden danger points in multi-view intersection scenarios, avoid coverage judgment deviations caused by viewpoint breaks, and has the advantage of improving the accuracy and reliability of production safety coverage judgment. Attached Figure Description

[0062] Figure 1 A schematic diagram of the system provided in this application. Detailed Implementation

[0063] The following description, in conjunction with the technical solution of this application, provides a clearer and more complete explanation of the relevant content. It should be noted that the embodiments described herein are only a part of the implementation methods of this application, and not all of them. Other implementation methods obtained by those skilled in the art based on the embodiments of this application without creative effort should also fall within the protection scope of this application. Furthermore, in the description of this application, the terms "first," "second," etc., are mainly for distinction and should not be construed as indicating relative importance.

[0064] Reference Figure 1 This application proposes an integrated simulation modeling and control system for production safety inspection and monitoring, which is applied to pipe gallery areas with high and low-level equipment and obstruction structures. The system includes an acquisition unit, a generation unit, a verification unit, a check unit, and a construction unit.

[0065] The acquisition unit is used to acquire input data, which specifically includes the location of the hidden danger point and local observation direction information, the pose information of the inspection path, the pose and field of view information of the fixed monitoring equipment, and the geometric information of the obstruction structure.

[0066] The acquisition unit transmits the input data to the generation unit. For each potential hazard point, the generation unit generates a candidate observation relationship for inspection based on the location of the potential hazard point and the pose information of the inspection path. At the same time, it generates a candidate observation relationship for monitoring based on the location of the potential hazard point and the pose and field of view information of the fixed monitoring equipment.

[0067] The verification unit receives these candidate observation relationships and performs visible boundary verification. The verification process includes line-of-sight continuity check and observation angle check. The line-of-sight continuity check determines whether the line of sight is blocked based on the geometric information of the obstructing structure. The observation angle check determines whether the observation angle meets the identifiable condition based on the local observation direction information of the hidden danger point. The verification unit determines the candidate observation relationships that pass the visible boundary verification as valid observation relationships.

[0068] The verification unit receives valid observation relationships. For each potential hazard point, it determines the verification status of the hazard point based on the valid observation relationships of that hazard point. The verification status includes valid patrol and inspection, valid monitoring, and invalid patrol and monitoring.

[0069] Finally, the construction unit constructs and outputs coverage status data based on the verification status of each hazard point. This coverage status data includes the valid verification results at the hazard point level, i.e., the coverage conclusion.

[0070] In densely populated utility tunnel areas, pipelines, supports, columns, beams, and corner structures are intricately distributed, with high and low-level equipment arranged in layers. Many potential hazards, such as valves, flange connections, drain outlets, and instrument panels, are located behind equipment, at low joints, or in recessed areas, often subject to complex obstructions and spatial angle limitations. In such scenarios, the dynamic observation perspective formed by inspection personnel moving along the corridor differs from the static observation perspective formed by fixed monitoring equipment in a fixed position, resulting in inconsistent visibility of the same hazard. If the simulation modeling and control system directly marks all hazard points falling within the vicinity of the inspection path or within the monitoring field of view as covered, it is easy to mistake false coverage caused by proximity or inclusion in the view as genuine verification. This leads to the output of conclusions in the coverage status data that do not match the actual safety verification status, causing subsequent inspection task scheduling and monitoring blind spot analysis to be based on distorted coverage information.

[0071] After receiving spatial layout data and related operational information of the utility tunnel area, the system of this application uses each potential hazard point as an independent minimum judgment unit. Through acquisition units, it organizes multi-source input data into an information foundation for verification and judgment. The location and local observation direction information of the potential hazard point can be extracted from existing 3D models of the plant area, laser scanning point clouds, as-built drawings, or on-site calibration results. Local observation direction information is used to characterize the orientation in which the potential hazard point can be effectively observed, such as the local normal direction of the surface where the potential hazard point is located or the recommended observation side direction. The pose information of the inspection path can be derived from the draft inspection plan path or provided by the personnel positioning trajectory during actual execution, including spatial coordinates and orientation at different positions along the path. The pose and field of view information of the fixed monitoring equipment includes the installation coordinates, installation height, pitch and yaw orientation, field of view boundaries, and effective observation distance range of each camera. The geometric information of the obstructing structure includes the spatial geometric description of components such as supports, pipelines, equipment bodies, beams, platforms, and enclosures within the utility tunnel, used to determine obstruction relationships during subsequent line-of-sight checks.

[0072] For each potential hazard point, the generation unit first generates preliminary candidate observation relationships based on the spatial location of the hazard point. When generating inspection candidate observation relationships, an observable neighborhood can be determined on the inspection path based on the location of the hazard point. A certain density of path poses is extracted from this neighborhood as candidate observation positions, and their corresponding position and orientation information are recorded. When generating monitoring candidate observation relationships, it can be checked whether the hazard point falls within the theoretical field of view cone of each fixed camera, and cameras that are significantly too far away or facing away are filtered out based on distance constraints. In this way, the candidate observation relationships serve as a broad-screening result, providing input for the subsequent verification unit, while avoiding indiscriminate fine-grained occlusion and angle verification at the level of all path points and all cameras, thus maintaining engineering feasibility.

[0073] The verification unit performs visible boundary verification on each candidate observation relationship. Its purpose is to identify and filter relationships that appear valid due to proximity or seemingly valid field of view angles, but which do not actually meet the effective observation conditions. In the line-of-sight connectivity check, the system establishes a main line of sight between the observation subject and the potential hazard point based on the geometric information of the obstructing structure, and determines whether this line of sight is blocked by supports, pipelines, equipment bodies, or other components. For some potential hazard points that are not single mathematical points but localized areas with a certain area or volume, the system can also generate auxiliary sampling points around the observable area of ​​the hazard point to determine whether at least a portion of the key area remains unobstructed, thus avoiding the complete elimination of partially visible observation windows simply because the main line of sight is blocked. In the observation angle check, the system calculates the angle between the observation direction and the local observation direction information of the hazard point, and determines whether this angle is within the preset allowable range of identifiable conditions. For example, if an inspector's observation direction is almost parallel to the flange face, even though the distance is short and the line of sight is unobstructed, it cannot effectively identify the connection status of the flange face. This relationship will be identified as insufficient to support the verification during the observation angle check and thus excluded from the valid observation relationship. Only candidate observation relationships that pass both the line-of-sight continuity check and the observation angle check are determined as valid observation relationships by the verification unit.

[0074] After obtaining the valid observation relationships for each potential hazard point, the verification unit determines the verification status of the hazard point accordingly. If a hazard point only has a valid observation relationship on the inspection side, it is determined to be validly undertaken by the inspection team; if it only has a valid observation relationship on the monitoring side, it is determined to be validly undertaken by the monitoring team; if neither the inspection nor the monitoring side has a valid observation relationship, it is determined to be invalid for both inspection and monitoring. For hazard points that have both valid inspection and monitoring observation relationships, they can be further classified into a shared responsibility status to retain the flexibility of primary and secondary responsibility allocation in subsequent task scheduling. The determination of the verification status no longer depends on whether the entire device has been passed or whether the screen has ever covered it, but is directly based on the results of point-by-point visible boundary verification, thereby enabling each hazard point to obtain an independent judgment with physical visibility support in the coverage status data.

[0075] The construction unit aggregates the verification status of all potential hazards, forming the coverage status data of the simulation modeling and control system. In this coverage status data, the coverage conclusion is not a general coverage percentage output by region, but rather corresponds to the effective verification results at each hazard point level. For example, for a certain pipe gallery section, the coverage status data can indicate which flange connection points are stably handled by fixed monitoring, which low-level joints require effective observation by inspection personnel along specific paths, and which valve backsides lack effective observation relationships under the existing inspection paths and monitoring layout, representing true blind spots requiring supplementary inspection or blind spot filling. In this way, even if the hazard point-level verification status is subsequently rolled up level by level to generate equipment-level or section-level aggregated statistics, these aggregated results can always be traced back to the actual verification source and visible boundary conditions of the specific hazard point, thus avoiding the situation where coarse-grained conclusions for the entire equipment or region mask local uncovered risks.

[0076] Through the collaborative process of acquisition, generation, verification, inspection, and construction described above, this system can, in complex utility tunnel environments where high and low-level equipment and obstructing structures coexist, correct coverage determination from a mere formality of proximity and field of view inclusion to a more effective hazard-point-level verification determination based on line-of-sight accessibility and identifiable observation angles. Some hazard points, although adjacent to the inspection passage, may be obstructed by supports or parallel pipelines, preventing effective observation from a normal walking perspective; others, although within the camera's field of view, may only be imaged as blurry outlines due to excessive angle or distance, failing to meet the identification requirements for safety inspection. These points, easily mismarked as covered in traditional coarse-grained coverage determination, will be classified as invalid or pending inspection in this system because they fail visible boundary verification, thus preventing false safety conclusions from being output based on coverage status data. When reading the coverage status data, the inspection task scheduling system can accurately distinguish which potential hazards can indeed be addressed through actual inspection, which require additional stops or detours, which need to rely on monitoring to fill gaps, and which are currently unserved areas in the network layout. This allows it to generate work orders and strategies supported by real verification conditions. Similarly, the monitoring gap-filling analysis can directly reuse the failure causes and spatial identifiers at the hazard level to determine the direction, location, and priority of gap-filling, avoiding vague suggestions to add more cameras.

[0077] It is understandable that the candidate observation relationship generation method used by the generation unit in this system, the specific geometric algorithm for line-of-sight connectivity checks in the verification unit, the identifiable condition threshold for observation angle checks, and the specific classification detail of the verification status can all be configured according to the scale, accuracy requirements, and downstream system capacity of the utility tunnel in actual engineering projects, without limiting the scope of system protection. For example, the local observation direction information of potential hazards can be organized and stored in the form of a potential hazard location mapping table, where each potential hazard is bound to its associated equipment, spatial coordinates, local normal direction, and optional key observation surface definition. However, this is only one possible form of information organization. Furthermore, the generation of candidate observation relationships can employ a rapid elimination method based on spatial distance or field cone angle, but this does not preclude other technical means that can achieve coarse screening centered on potential hazard points.

[0078] In the aforementioned basic visible boundary verification, if the overall bounding box or center point of a potential hazard is not blocked by an obstructing structure, and the angle between the observation direction and the local normal of the hazard is within the general threshold range, the candidate observation relationship may be considered to have passed the verification. However, in the actual environment of utility tunnels, although the main body of a potential hazard may be visible, its key verifiable parts may be obstructed by pipelines or supports, making it impossible to effectively verify the hazard. Furthermore, although inspection personnel or monitoring equipment may be able to see the hazard, their actual identification capability may not meet the requirements for effective verification due to factors such as excessive distance, insufficient image area ratio, or lack of personnel intervention. Therefore, based on the basic visible boundary verification, it is necessary to further introduce fine-grained verification based on the obstruction of key areas of the hazard and verification of local identifiability conditions based on differences in the observed subject.

[0079] Furthermore, in some preferred methods, a visible boundary check is performed on each candidate observation relationship, including: obtaining the hazard point type of the detection point; determining the key area of ​​the hazard point corresponding to the hazard point based on the hazard point type; in the line-of-sight check, determining whether the key area of ​​the hazard point is obstructed based on the geometric information of the obstruction structure; if the key area of ​​the hazard point is obstructed, the line-of-sight check of the candidate observation relationship is deemed to have failed; obtaining the observation subject type corresponding to the candidate observation relationship, where the observation subject type is either patrol personnel or fixed monitoring equipment; determining the corresponding local identifiability conditions based on the observation subject type and the hazard point type, where the local identifiability conditions include at least distance conditions, unobstructed ratio conditions, and patrol dwell time conditions or monitoring image area ratio conditions; in the observation angle check, determining whether the candidate observation relationship meets the identifiability requirements based on the local identifiability conditions; only when the candidate observation relationship simultaneously satisfies that the key area of ​​the hazard point is not obstructed in the line-of-sight check and meets the local identifiability conditions in the observation angle check is the candidate observation relationship determined to have passed the visible boundary check.

[0080] After identifying the types of potential hazards at the detection points, the corresponding critical areas are determined based on these types. The critical areas are typically defined by observable structures directly related to leaks, loosening, or abnormal conditions. For example, the critical areas for flange-related hazards may include the flange connection gap and the bolt distribution area on both sides; for valve-related hazards, the critical areas may include the valve stem sealing gland area, the welded area connecting the valve body and the pipeline, and the actuator status indicator window; and for instrument-related hazards, the critical areas may include the dial readings and pressure tap interface area. The definition of critical areas can be expanded or adjusted according to on-site safety requirements.

[0081] In the line-of-sight inspection, the focus is on the critical areas of potential hazards. When determining whether a critical area is obstructed based on the geometric information of the obstructing structure, a finite number of auxiliary visual sampling points can be generated around the critical area. If all sampling points in the critical area are obstructed, the line-of-sight inspection is deemed a failure. If some sampling points in the critical area are visible, and the visible portion is sufficient to cover the core structure of the critical area, it can be considered a pass. This auxiliary mechanism can enhance the robustness of the judgment to a certain extent, avoiding excessive rejection due to the obstruction of individual sampling points.

[0082] After obtaining the observation subject type corresponding to the candidate observation relationship, the corresponding local identifiability conditions are determined based on the observation subject type and the hazard point type. The parameters for the local identifiability conditions can be set based on the typical identifiability capability of the observation subject. Distance conditions can be set based on the typical identifiability distance of the observation subject; for example, the effective identification distance of an inspector for a low-level flange is usually less than the effective identification distance for a high-level pipeline connection point. The unobstructed ratio condition requires that the visibility ratio of the critical area of ​​the hazard point be higher than a preset threshold. The inspection dwell time condition can be derived from the minimum dwell time in the inspection task specifications. The monitoring image area ratio condition can be calculated based on the camera resolution and the scale of typical hazard points. Different combinations of hazard point types and different observation subject types can correspond to different condition configurations. These configurations can be maintained through preset tables or configuration files and can be adjusted based on historical review data.

[0083] During the observation angle inspection, the candidate observation relationship is judged to meet the identifiability requirements based on the local identifiability condition. Only when the candidate observation relationship simultaneously meets the conditions that the key area of ​​the hazard point is not obstructed in the line-of-sight inspection and meets the local identifiability condition in the observation angle inspection, is the candidate observation relationship determined to have passed the visible boundary check.

[0084] When the type of a potential hazard is not predefined, or the definition of the key area of ​​the potential hazard is missing, the candidate observation relationship of the potential hazard can be marked as undefined, or the entire potential hazard can be used for occlusion judgment, while reducing the credibility level of its verification status.

[0085] Through the above method, candidate observation relationships whose key areas are obscured, or whose unobstructed areas do not meet the identifiable conditions such as distance, unobstructed ratio, dwell time, or image area ratio, will change from passing the initial basic verification to failing the visible boundary verification and be excluded from the valid observation relationships. Only when the key areas are not obscured and all identifiable conditions are met is the candidate observation relationship officially confirmed as having passed the visible boundary verification and entered the set of valid observation relationships. This, to a certain extent, eliminates the problem of blind spots in the verification caused by the obscuration of key areas, prevents false identifiable judgments due to excessive distance, insufficient image area, or lack of dwell time, and ensures that subsequent status determination is based on a higher-quality set of valid observation relationships. At the same time, the valid observation relationships selected through refined verification provide a cleaner input basis for subsequent judgments on the stability of the inspection observation window.

[0086] Furthermore, in some preferred methods, for inspection candidate observation relationships that have passed the visible boundary verification, the inspection observation interval is determined based on the pose information of the inspection path; at least one of the following is obtained: continuity information, persistent information, orientation naturalness information, speed adaptation information, and historical repeatability information of the inspection observation interval; at least based on the obtained information, it is determined whether the inspection observation interval is a stable window; only inspection candidate observation relationships that are determined to be stable windows are determined as valid observation relationships.

[0087] In densely populated utility tunnel areas, even if the candidate inspection observation relationships have passed basic checks on line-of-sight connectivity and observation angles, their reliability may still be lacking in actual implementation. Due to the interweaving of high and low-level equipment and the dense arrangement of support pipelines within the tunnel, many potential hazards often only become apparent when the inspector reaches a specific corner or briefly turns their head. These geometrically visible observation opportunities are typically extremely short-lived and highly dependent on instantaneous posture changes. Directly recognizing such observation relationships that have passed basic checks as valid verification can easily lead to misjudging momentarily visible or intermittently visible windows as stable verification windows, resulting in the inspection path appearing to pass through and be visible, but in reality, no effective verification has been achieved. To address this, a layered stability verification logic for inspection observation windows is introduced to distinguish between geometrically visible and behaviorally stable verification.

[0088] When determining the inspection observation interval, pose segments that continuously satisfy the visibility condition are extracted from the inspection path state sequence and combined into inspection observation intervals. For each inspection observation interval, at least one of the following is acquired: continuity information, persistence information, orientation naturalness information, velocity adaptation information, and historical repeatability information. Continuity information can be obtained by comparing the pass rate of adjacent poses to reflect whether the window is frequently interrupted. Persistence information can be calculated from the timestamp difference and path length to distinguish between fleeting visibility and continuous visibility. Orientation naturalness information can be evaluated by the change in the angle between the travel direction vector and the observation direction vector to distinguish between natural observation and observation that depends on significant pose deflection. Velocity adaptation information can be determined by the degree of conformity between the actual velocity and the normal inspection velocity range. Historical repeatability information can be obtained by statistically analyzing the frequency and stability of the interval in multiple inspection records. Based on the acquired information, it is determined whether the inspection observation interval constitutes a stable window. If it is determined to be a stable window, the candidate inspection observation relationship is marked as a valid observation relationship. If it is determined to be an unstable window, it is not considered a valid observation relationship. Instable windows can be further subdivided into instantaneous windows, conditional windows, and windows awaiting review. These unstable windows reflect the fragility of the observation opportunity to some extent and do not directly assume the responsibility of verification.

[0089] For example, a low-level flange connection point on the main passage of a utility tunnel will only be briefly exposed when an inspector reaches a specific corner edge and is facing sideways. This section is short-lived and requires a noticeable turning motion; its continuity and orientation information typically do not meet the preset conditions for a stable window. After introducing stability stratification, this section will be judged as an unstable window and thus will not be considered a valid observation relationship. This stability stratification verification can prevent the erroneous transmission of the assumption that a brief glance indicates verification has been completed. Different types of potential hazards can be preset with different stability emphases. For example, for high-risk flange connection points, the requirements for continuity and orientation can be increased to a certain extent. This ensures that inspection observation opportunities entering a valid observation relationship must be observation windows that are stably established and repeatable under normal inspection behavior, thereby improving the reliability of inspection judgments and reducing missed judgments caused by formal modeling of the dynamic observation process.

[0090] After obtaining the inspection observation range and acquiring basic visible continuity information, although it is possible to initially identify observation ranges with continuity, in the context of utility tunnels, if the stability level of the range is not further subdivided, windows with extremely short durations or those dependent on accidental postures are easily regarded as valid verification sources, leading to the problem of instantaneous window false rise. In order to determine the stability level of the inspection observation range and avoid misjudging instantaneous visible windows as stable verification windows, it is necessary to introduce quantitative evaluation.

[0091] Furthermore, in some preferred methods, determining whether an inspection observation interval is a stable window includes: obtaining the continuity index, persistence index, orientation naturalness index, speed adaptation index, and historical repeatability index corresponding to the inspection observation interval; calculating the comprehensive stability index of the inspection observation interval based on the continuity index, persistence index, orientation naturalness index, speed adaptation index, historical repeatability index, and their respective weighting coefficients; comparing the comprehensive stability index with a preset stability threshold and a preset instantaneous threshold, where the stability threshold is greater than the instantaneous threshold; when the comprehensive stability index is greater than or equal to the stability threshold, the inspection observation interval is determined as a stable window; when the comprehensive stability index is less than the instantaneous threshold, the inspection observation interval is determined as an instantaneous window; when the comprehensive stability index is between the stability threshold and the instantaneous threshold, the inspection observation interval is determined as a window to be reviewed or a conditional window; instantaneous windows, windows to be reviewed, and conditional windows are not considered stable windows.

[0092] In one implementation, the acquired continuity information, persistence information, orientation naturalness information, speed adaptation information, and historical repeatability information are converted into quantitative indicators. The continuity indicator can be derived from the proportion of poses that continuously pass the basic visibility check within the inspection observation interval to the total number of sampled poses in the interval. The persistence indicator can be obtained by normalizing the continuous length or duration along the path within the interval. The orientation naturalness indicator can be characterized by the cosine value or deviation of the angle between the average observation orientation of the inspectors within the interval and the direction of travel. The speed adaptation indicator can be calculated by the ratio or difference between the average travel speed of the inspectors within the interval and the preset normal inspection speed. The historical repeatability indicator can be determined based on the proportion of times the window was identified as continuously visible in multiple past inspections along the same route.

[0093] After obtaining the above indicators, a comprehensive stability index is obtained by weighting each indicator using weighting coefficients. The weighting coefficients can be configured based on the type of hazard point, risk level, or location characteristics. For hazard points requiring a stable viewpoint and sufficient time to observe, higher weights can be set for persistence and speed adaptation. For hazard points requiring observation during natural movement, higher weights can be set for orientation and naturalness.

[0094] After calculating the comprehensive stability index, it is compared with the preset stability threshold and instantaneous threshold. The stability threshold and instantaneous threshold can be determined based on on-site verification data and model validation statistics, ensuring that most known stable windows score higher than the stability threshold and known instantaneous windows score lower than the instantaneous threshold.

[0095] When the overall stability index is greater than or equal to the stability threshold, the inspection observation interval is designated as a stable window. This stable window can be included in the effective inspection observation relationship table as a basis for effective inspection. When the overall stability index is less than the instantaneous threshold, the inspection observation interval is designated as an instantaneous window, which is excluded from the effective verification sources.

[0096] When the overall stability index falls between the stability threshold and the instantaneous threshold, it needs further classification as a conditional window or a window awaiting review. If the speed fit index or the orientation naturalness index is low within the interval, indicating a need for deceleration or a turn, it is likely to be classified as a conditional window. If it falls into the middle zone mainly due to insufficient historical repeatability samples, it can be classified as a window awaiting review. Window conditions and windows awaiting review are in an intermediate state and are not directly counted as valid checks; they await further action binding, review, or supplementary checks.

[0097] By employing the aforementioned weighted dual-threshold mechanism, the method of determining stability based solely on geometric continuity or simple counting is modified to a certain extent. It can identify and filter out windows formed by instantaneous passing or occasional attitude changes, preventing them from being erroneously escalated into valid inspection tasks. Simultaneously, by identifying conditional windows and windows awaiting verification, the possibility of achieving effective verification through additional actions or further confirmation is preserved, preventing windows from being directly discarded. This reduces the phenomenon of surface coverage substituting for substantive verification, improving the authenticity and robustness of effective verification results at the hazard point level in coverage status data.

[0098] In the aforementioned stability stratification process, the conditional window state has been identified. The conditional window theoretically meets the basic visibility requirements, but its effective observation depends on the inspector performing additional actions. If we only stop at the stratification identification stage without subsequent action binding and closed-loop verification, the conditional window may be marked as checked by default during actual inspections because personnel have not performed the corresponding actions. Therefore, this application further proposes an action constraint and closed-loop verification mechanism for the conditional window.

[0099] Furthermore, in some preferred methods, when the inspection observation interval is determined to be a conditional window that requires additional actions to form a valid observation based on at least one of continuous information, persistent information, orientation naturalness information, speed adaptation information, and historical repetition information, action constraint requirements are generated. These action constraint requirements are associated with the conditional window and the corresponding hidden danger point. The action constraint requirements are written into the inspection task. The actual inspection trajectory is obtained, and the execution of the action constraint requirements is verified based on the actual inspection trajectory. It is also verified whether the inspection observation interval corresponding to the conditional window passes the visible boundary check again after the action is executed. If the verification result is passed, the conditional window is updated to a valid observation relationship.

[0100] When a specific inspection observation range is designated as a condition window, action constraints are generated based on the characteristics of that window. For example, if the speed adaptation information indicates that the current inspection speed is too fast, resulting in insufficient observation time, a requirement to decelerate to the observation speed can be generated; if the orientation information shows a significant deviation between the observation direction and the travel direction, a requirement to turn around and face the potential hazard can be generated; if the continuity information indicates that a longer observation time is needed, a requirement for the duration of stay within that range can be generated. These action constraints are associated with the condition window and the corresponding potential hazard, thus clarifying the behavioral prerequisites for the effective verification of that potential hazard.

[0101] The generated motion constraints are written into the inspection task, providing guidance for inspectors during task execution. The actual inspection trajectory is acquired during or after the task. This trajectory typically includes pose sequences, velocity curves, and orientation change data recorded by the positioning device. Based on this data, it can be verified whether the motion constraints have been executed. For example, analyzing the velocity curve determines if a deceleration phase meeting the constraints occurred; analyzing the pose sequence determines if a pause occurred; and analyzing the orientation change determines if a turning motion was performed.

[0102] After confirming that the action constraints have been executed, a visible boundary check will be performed again on the inspection observation area corresponding to the condition window, based on the state after the action is executed. This step aims to confirm whether the area meets the identifiable conditions such as line-of-sight connectivity and observation angle after the additional action is completed. If the actual trajectory verification passes and the second check also passes, the condition window will be updated to a valid observation relationship, and the corresponding hidden danger point can be included in the scope of the effective inspection responsibility.

[0103] If the verification result is unsuccessful, meaning the inspector did not perform the additional actions or the visible boundary verification was still not met after execution, the condition window will remain in an invalid observation state. To a certain extent, this can downgrade the status or transfer it to the supplementary inspection branch, avoiding misjudging it as having been verified. In addition, the strictness of action constraints can be set differently for potential hazards of different risk levels or different types of locations, so as to ensure both verification quality and inspection efficiency.

[0104] By binding the above actions with the actual execution closed loop, it is possible to avoid misjudging inspection observation windows that only exist in theory and have not been confirmed by the execution actions as valid inspections to a certain extent. This ensures that the hidden danger points marked as valid inspections have actual execution records to support the actual observation actions, thereby improving the verification authenticity of the integrated simulation modeling and control system in dense utility tunnel scenarios.

[0105] In densely populated utility tunnels, simply categorizing the status of potential hazards as covered or uncovered is often insufficient to guide subsequent inspection scheduling and monitoring gap-filling operations. For example, when a low-level flange is determined to be uncovered, the control center cannot directly determine whether it's due to support obstruction or excessive distance, making it difficult to decide whether to increase manual inspection or install additional cameras. Therefore, fine-grained status categorization and supplementary information are needed to support the closed-loop execution of integrated management.

[0106] Furthermore, in some priority methods, the verification status of the hazard point is determined, including: based on the effective observation relationship of the hazard point, determining the verification status as one of the following: effective undertaking of inspection, effective undertaking of monitoring, joint undertaking of inspection and monitoring, inspection candidate exists but no effective observation relationship is formed, monitoring candidate exists but no effective observation relationship is formed, or both inspection and monitoring are ineffective undertakings; generating a failure cause identifier and a compensation direction identifier for the hazard point, the failure cause identifier is used to indicate the reason why the hazard point was not effectively verified, and the compensation direction identifier is used to indicate the suggested direction for supplementary inspection or blind spot filling of the hazard point.

[0107] Specifically, when both effective observation relationships from patrol and monitoring are established simultaneously, it can be determined that patrol and monitoring share the responsibility. In this case, although both can observe, it is still necessary to clearly define primary and secondary responsibilities in the actual task arrangement to prevent a responsibility vacuum when one party is offline or its path is blocked. When only effective observation relationships from patrol are established, it is determined that patrol is responsible. This means that fixed cameras cannot reach the potential hazard point due to height differences or angle limitations, requiring manual close-range verification. Conversely, when only effective observation relationships from monitoring are established, it is determined that monitoring is responsible. In this case, the frequency of manual patrols can be appropriately reduced to alleviate the burden on patrol personnel.

[0108] It's worth noting that the "inspection candidate exists but no effective observation relationship has been formed" status is primarily used to identify potential hazards near the inspection path that haven't been checked due to corner obstructions or insufficient dwell time. This status can, to some extent, prevent a mere passing through the path from being mistaken for an inspection. Similarly, the "monitoring candidate exists but no effective observation relationship has been formed" status is used to identify potential hazards that fall within the camera's field of view but are insignificant due to distance or obstruction by pipelines in key areas. When neither condition is met, the inspection and monitoring are deemed invalid, triggering the rectification process.

[0109] When generating failure cause identifiers, they can be extracted and categorized from the rejection reasons of visible boundary verification. For example, failure cause identifiers may include structural obstruction, corner obstruction, elevation difference causing the observation surface to face away, excessive distance, inclusion in the image but insufficient image area, path passing through but without effective orientation, or an unreliable inspection window that is too short. These identifiers can transform verification failures into physical space constraints, helping managers understand why potential hazards have not been effectively verified.

[0110] The compensation direction indicators are generated based on the cause of failure and the spatial location of potential hazards. For example, if a low-level flange faces away from a fixed camera and is partially obscured by a bracket, but a manual observation window can be formed within a certain range along a lateral branch passage, the compensation direction indicator can indicate supplementary inspections and suggest observing from the west branch passage. Similarly, if a high-level connection point is difficult to inspect manually, but a blind spot exists in the vicinity of an existing high-level camera, the compensation direction indicator can indicate blind spot coverage and suggest adding a downward observation position at the south crossbeam. In this way, the output results can serve subsequent engineering design or task adjustments, reducing the workload of secondary manual analysis to some extent.

[0111] This status classification and identifier generation transforms coverage status data from indicators into actionable instructions. Downstream task modules can parse these identifiers, binding hazard points that are only effectively inspected to observation intervals, generating supplementary inspection requirements for hazard points that exist as inspection candidates but have not yet formed an effective observation relationship, and outputting blind spot suggestions for hazard points that exist as monitoring candidates but have not yet formed an effective observation relationship. This, to a certain extent, avoids oversights that may occur from manual interpretation of status, improves the automation level of control, and fundamentally reduces the risk of surface coverage replacing verification.

[0112] Furthermore, in some preferred methods, the actual inspection trajectory pose information and the actual monitoring equipment online status information are obtained; the actual inspection trajectory pose information is used as the pose information of the inspection path, and the fixed monitoring equipment currently in the online status is determined according to the actual monitoring equipment online status information; for each hidden danger point, the inspection candidate observation relationship is regenerated according to the location of the hidden danger point and the actual inspection trajectory pose information, and the monitoring candidate observation relationship is regenerated according to the location of the hidden danger point and the pose and field of view information of the fixed monitoring equipment currently in the online status; the visible boundary verification is performed on each regenerated candidate observation relationship, and the verification status of the hidden danger point is re-determined to obtain the updated verification status; the overlay status data is updated according to the updated verification status of each hidden danger point; for hidden danger points that are determined to be invalid for both inspection and monitoring after the update, a supplementary inspection task instruction or early warning prompt information is generated.

[0113] During actual inspections, situations may arise such as personnel not following the planned path, moving too fast and failing to reach the observation window, personnel not stopping in designated areas, or certain fixed monitoring devices going offline due to malfunction. These situations may lead to discrepancies between the actual verification status and the planned coverage data. To address this, the actual inspection trajectory and pose information can be obtained, which can come from the orientation data uploaded by positioning devices or inspection terminals. Simultaneously, the online status information of the actual monitoring equipment can be obtained, which can be acquired through queries on the device management platform or through heartbeat detection mechanisms.

[0114] After obtaining the actual data, the pose information of the original planned inspection path is replaced with the actual inspection trajectory pose information, and the fixed monitoring devices currently online are selected based on their actual online status. For each potential hazard point, new inspection candidate observation relationships and monitoring candidate observation relationships are generated, and visible boundary verification is performed on the candidate relationships. During this process, the verification rules established earlier can be used. The verification status of each potential hazard point is redefined, and the overlay status data is updated accordingly. If some of the actual execution data is missing, such as a segment of trajectory positioning being lost, the relevant potential hazard points can be kept in a pending confirmation state to avoid directly determining them as invalid.

[0115] Through the above processing, the coverage status data can be transformed from a static snapshot based on planning assumptions to a dynamic reflection based on actual execution. Hazard points that were considered effectively undertaken in the planning but not effectively verified in practice can have their verification status updated to "not effectively undertaken." For hazard points identified as ineffective in both inspection and monitoring after the update, supplementary inspection task instructions or early warning messages can be generated. Supplementary inspection task instructions can include hazard point identification, suggested observation locations, and failure cause labels, while early warning messages can be highlighted on the control interface or pushed to management personnel terminals.

[0116] By implementing dynamic recalculation and closed-loop processing after execution, loopholes in the verification process that are difficult to foresee in static planning can be automatically identified and exposed to a certain extent, preventing conclusions already covered in the planning from being carried over to subsequent management stages. By automatically generating supplementary inspection task instructions or early warning prompts, unchecked potential hazards can be promptly reinstated into inspection tasks or attract the attention of management personnel, thereby preventing safety oversights from entering the shift handover or responsibility determination stages. Furthermore, the deviation records and actual execution data generated during dynamic recalculation can provide input sources for revising subsequent judgment rules.

[0117] During long-term operation, the initially set visible boundary verification rules may differ from the actual situation of the utility tunnel scenario, leading to discrepancies between the judgment results of potential hazards and the actual verification results. To address the problem of judgment distortion caused by rigid rules or gradual environmental changes, this application further proposes, in some preferred approaches: obtaining verification feedback information, including on-site review results, historical omission records, or safety supervision spot check data; determining, based on the verification feedback information, the deviation between the verification status in the coverage status data of at least one potential hazard and the actual verification result; determining the visible boundary verification rule type corresponding to the deviation; correcting the rule parameters on which the visible boundary verification is based according to the rule type; and for potential hazards affected by the correction, re-executing the steps of generating candidate observation relationships, performing visible boundary verification, and determining the verification status to update the coverage status data.

[0118] Verification feedback information can come from multiple channels. On-site verification results can include the visible status recorded by inspection personnel or safety supervisors after conducting on-site inspections of potential hazards. Historical omission records can reflect potential hazards marked as covered in past periods but where incidents actually occurred. Safety supervision spot check data can show discrepancies between the coverage status and the actual situation discovered by regulatory authorities during inspections. By collecting this information, it can be compared with the verification status in the coverage status data.

[0119] When determining deviations, inconsistencies in the presentation of the same hazard point across multiple feedback reports can be analyzed. If coverage status data indicates that a hazard point has been effectively verified, but on-site verification results repeatedly show that the hazard point cannot be effectively observed in practice, a positive deviation can be identified for that hazard point. Conversely, if coverage status data indicates that a hazard point has not been effectively verified, but on-site verification repeatedly demonstrates that it meets the observation conditions, a negative deviation can be identified for that hazard point. This comparison can identify the degree of deviation between the initial rule settings and the actual on-site situation.

[0120] After identifying the deviation, it is necessary to analyze its causes and categorize it into the corresponding visible boundary verification rule type. Visible boundary verification rule types can include angle threshold rules, key area definition rules, occlusion ratio threshold rules, local identifiability condition parameter rules, and path representation density rules. If the deviation manifests as the angle between the observation direction and the local normal of the hazard point being within the allowable range but actually unidentifiable, it can be categorized into angle threshold rules. If the deviation manifests as the key observation surface of the hazard point being partially occluded but still deemed valid, it can be categorized into occlusion ratio threshold rules or key area definition rules.

[0121] Based on the determined rule type, the rule parameters used for visible boundary verification can be modified. For angle threshold rules, modification can be manifested by adjusting the allowable glimpse angle value corresponding to the hazard point type. For key area definition rules, modification can be manifested by updating the mapping table of key areas for hazard points. For occlusion ratio threshold rules, modification can be manifested by adjusting the upper limit of the allowable occlusion ratio of key areas. For locally identifiable condition parameter rules, modification can be manifested by adjusting the distance threshold or dwell time parameter. Through modification, the rule parameters can be made to fit the physical conditions of the utility tunnel scene.

[0122] After the rule parameters are corrected, the steps of generating candidate observation relationships, performing visible boundary verification, and determining the verification status can be re-executed for the affected hazard points. The recalculation process ensures that the corrected rules are reflected in the coverage status data. The verification status of the affected hazard points can be changed from the deviation status to the corrected and verified status, and the coverage conclusion of the hazard point in the coverage status data can be restored to consistency with the actual verification results.

[0123] By applying the verification feedback data to adjust the parameters of the visible boundary verification rules, the judgment logic can evolve according to the actual situation on site. This approach overcomes the shortcoming of simulation systems that do not update rules after they are set, and helps to solve the problem of judgment distortion spreading due to rule setting deviations or gradual environmental changes. The system has the ability to correct deviations for utility tunnel scenarios, and can maintain the output coverage conclusions corresponding to the verification results at the hazard point level.

[0124] In actual engineering deployments, input data may be partially missing or of insufficient quality to meet judgment requirements. Directly performing judgments based on incomplete data can easily lead to false conclusions. Furthermore, in some preferred methods, before generating candidate inspection observation relationships based on the location and pose information of the inspection path for each potential hazard point, and before generating candidate monitoring observation relationships based on the location and pose and field of view information of the fixed monitoring equipment, for each potential hazard point, the input data is used to determine whether the input information meets preset reliability conditions. If it is determined that the conditions are not met, the potential hazard point is marked as pending confirmation, and when constructing coverage status data, potential hazard points in the pending confirmation state are prohibited from being included in the coverage conclusion.

[0125] Preset reliability conditions can cover evaluation criteria for the completeness, accuracy, or consistency of input data. For example, when the precise three-dimensional coordinates of a potential hazard point are missing, the orientation parameters of fixed monitoring equipment are not updated, the positioning drift of the inspection trajectory is significant, or the geometric information of obstructions is incomplete, the corresponding input information can be determined to not meet the reliability conditions. By setting such conditions, deviations in subsequent line-of-sight checks or observation angle checks can be avoided to some extent due to unreliable basic data.

[0126] When the input information for a potential hazard is deemed unreliable, the hazard will change from a state eligible for normal assessment to a state pending confirmation. During subsequent coverage data construction, hazard points in the pending confirmation state will not be forcibly included in the covered or uncovered statistics; instead, they will be excluded as uncertain items. This approach prevents false coverage conclusions based on unreliable input data, prompting on-site data supplementation or manual verification for hazard points with missing or unreliable data. This fosters a conservative safety strategy of refusing to assess uncertainty in actual production safety management, ensuring the authenticity of the overall control results.

[0127] Furthermore, in some embodiments, when performing visible boundary verification, the verification unit determines the corresponding key area of ​​the hidden danger point according to the type of hidden danger point, and divides the key area of ​​the hidden danger point into multiple key sampling points. The verification unit uses the line connecting the observation reference point of the observing subject and each key sampling point as the line segment to be detected, and determines whether the line segment to be detected intersects with the occlusion structure based on the geometric information of the occlusion structure; if the intersection point is located between the observation reference point and the key sampling point, it is determined that the key sampling point is occluded. The verification unit determines the visibility ratio of the key area according to the ratio of the number of visible key sampling points to the total number of key sampling points, and when the visibility ratio of the key area is less than the visibility ratio threshold of the corresponding hidden danger point type, it is determined that the candidate observation relationship has not passed the line connectivity check. Furthermore, the verification unit also determines whether the observation angle meets the allowable observation angle threshold for the corresponding hazard type based on the angle between the direction the observer points to the key area of ​​the hazard point and the local observation direction of the hazard point. When the observer is an inspection personnel, it also determines whether the distance and dwell time meet the conditions for human identification. When the observer is a fixed monitoring device, it also determines whether the image area ratio or pixel size of the key area of ​​the hazard point in the monitoring screen meets the conditions for monitoring identification. This avoids misjudging a hazard as valid simply because it is near the path or within the monitoring field of view.

[0128] Furthermore, in some implementations, for inspection candidate observation relationships that have passed the visible boundary verification, the system determines the inspection observation interval based on the inspection path pose sequence, and calculates a comprehensive stability index based on the continuity index, persistence index, orientation naturalness index, speed adaptation index, and historical repeatability index of the inspection observation interval. Specifically, the continuity index characterizes the degree to which the visible boundary verification is passed continuously within the observation interval; the persistence index characterizes whether the observation duration meets the minimum verification time; the orientation naturalness index characterizes the degree of matching between the inspection personnel's walking or observation orientation and the observation direction of the hidden danger point; the speed adaptation index characterizes whether the inspection personnel's movement speed within the interval is suitable for verification; and the historical repeatability index characterizes the degree to which the observation interval has repeatedly formed effective observation opportunities in historical inspections. The system calculates a comprehensive stability index based on each indicator and its corresponding weight, and compares the comprehensive stability index with a preset stability threshold and a preset instantaneous threshold. When the comprehensive stability index is greater than or equal to the stability threshold, the inspection observation interval is determined as a stable window, and the corresponding inspection candidate observation relationship is determined as a valid observation relationship. When the comprehensive stability index is less than the instantaneous threshold, it is determined as an instantaneous window. When the comprehensive stability index is between the instantaneous threshold and the stability threshold, it is determined as a conditional window or a window to be reviewed. However, instantaneous windows, conditional windows, and windows to be reviewed are not directly used as valid observation relationships.

[0129] Furthermore, in some implementations, when the inspection observation range is determined as a condition window, the system generates action constraint requirements based on the type of indicator that causes the comprehensive stability index to fail to reach the stability threshold. Specifically, when the persistence indicator is insufficient, a requirement to stay within the corresponding observation range for a preset duration is generated; when the speed adaptation indicator is insufficient, a requirement to reduce the travel speed is generated; when the orientation naturalness indicator is insufficient, a requirement to face the corresponding hidden danger point or turn to a specified orientation range is generated. The action constraint requirements are associated with the corresponding hidden danger point, the position range of the condition window, and the action judgment threshold, and written into the inspection task. After the inspection task is executed, the system obtains the actual inspection trajectory pose information, determines whether the action constraint requirements have been executed based on the position, speed, dwell time, and orientation changes in the actual trajectory, and re-executes the visible boundary verification for the hidden danger point after the action is executed; if the re-verification passes, the condition window is updated to a valid observation relationship; if the action is not executed or the re-verification fails, the invalid state of the condition window is maintained, and the corresponding hidden danger point is marked as needing supplementary inspection or pending review.

[0130] Furthermore, in some implementations, when constructing the coverage status data, the construction unit generates a corresponding coverage status record for each potential hazard point. The coverage status record includes the hazard point identifier, verification status, type of valid observation relationship source, observer identifier, observation location or observation range, and visible boundary verification result.

[0131] The system includes the following components: Hazard Point Identifier: uniquely identifies the corresponding hazard point; Verification Status: indicates whether the hazard point is under effective inspection, effective monitoring, joint inspection and monitoring, or both; Valid Observation Relationship Source Type: indicates whether the hazard point's verification status is formed by an effective inspection observation relationship, a effective monitoring observation relationship, or both; Observation Subject Identifier: records the inspection personnel, inspection path segment, or fixed monitoring equipment that formed the valid observation relationship; Observation Location or Observation Interval: records the corresponding inspection path location, inspection observation interval, fixed monitoring equipment location, or monitoring observation direction when the valid observation relationship is formed; Visible Boundary Verification Result: records at least one of the following: line-of-sight connectivity check result, observation angle check result, key area visibility ratio, and identifiable condition satisfaction. By storing the above information in the coverage status data, the coverage conclusion for each hazard point can be traced back to the corresponding observation subject, observation location, and visible boundary verification basis, facilitating subsequent supplementary checks, blind spot filling, verification, or status updates.

[0132] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. All modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A simulation modeling and control system integrating production safety inspection and monitoring, applied to pipe gallery areas with high and low-level equipment and obstruction structures, characterized in that, include: The acquisition unit is used to acquire input data, which includes the location and local observation direction information of each hidden danger point in the pipe gallery area, the pose information of the inspection path, the pose and field of view information of the fixed monitoring equipment, and the geometric information of the obstruction structure. The generation unit is used to generate inspection candidate observation relationships for each potential hazard point based on the location of the potential hazard point and the pose information of the inspection path, and to generate monitoring candidate observation relationships based on the location of the potential hazard point and the pose and field of view information of the fixed monitoring equipment. The verification unit is used to perform visible boundary verification on each candidate observation relationship. The visible boundary verification includes: a line-of-sight check to determine whether the line of sight is blocked based on the geometric information of the occlusion structure, and an observation angle check to determine whether the observation angle meets the identifiable condition based on the local observation direction information of the hidden danger point. The candidate observation relationship that passes the visible boundary verification is determined as a valid observation relationship. The verification unit is used to determine the verification status of each potential hazard point based on the effective observation relationship of that hazard point. The verification status is at least divided into effective patrol and inspection, effective monitoring, or both patrol and monitoring are invalid. The construction unit is used to construct the coverage status data of the simulation modeling and control system based on the verification status of each hidden danger point, so that the coverage conclusion in the coverage status data corresponds to the valid verification result at the level of the hidden danger point.

2. The system according to claim 1, characterized in that, Perform visible boundary checks on each candidate observation relation, including: Obtain the type of potential hazards at the detection point; Based on the type of potential hazard, determine the key area corresponding to the potential hazard; In the line-of-sight check, the geometric information of the obstruction structure is used to determine whether the key area of ​​the potential hazard point is obstructed. If the key area of ​​the potential hazard point is obstructed, the line-of-sight check of the candidate observation relationship is deemed to be unsuccessful. Obtain the observation subject type corresponding to the candidate observation relationship, wherein the observation subject type is inspection personnel or fixed monitoring equipment; Based on the observed subject type and the hidden danger point type, the corresponding local identifiable conditions are determined. The local identifiable conditions include at least the distance condition, the unobstructed ratio condition, and the inspection dwell time condition or the monitoring image area ratio condition. In the observation angle check, it is determined whether the candidate observation relationship meets the identifiability requirements based on the local identifiability conditions; The candidate observation relationship is determined to pass the visible boundary check only if it simultaneously satisfies the conditions that the key area of ​​the hidden danger point is not obstructed in the line of sight continuity check and the local identifiability condition is met in the observation angle check.

3. The system according to claim 1, characterized in that, The step of determining the candidate observation relationships that pass the visible boundary verification as valid observation relationships includes: For the inspection candidate observation relationship that passes the visible boundary verification, the inspection observation interval is determined according to the pose information of the inspection path; Obtain at least one of the following: continuity information, persistent information, orientation naturalness information, speed adaptation information, and historical repeatability information of the inspection observation range; Based at least on the information obtained, determine whether the inspection observation interval is a stable window; Only the inspection candidate observation relationships that are determined to be stable windows are identified as the valid observation relationships.

4. The system according to claim 3, characterized in that, Determining whether the inspection observation interval is a stable window includes: Obtain the continuity index, persistence index, orientation naturalness index, speed adaptation index, and historical repeatability index corresponding to the inspection observation interval; Based on the continuity index, persistence index, orientation naturalness index, speed adaptation index, historical repeatability index, and their respective weighting coefficients, the comprehensive stability index of the inspection observation interval is calculated. The stability comprehensive index is compared with a preset stability threshold and a preset instantaneous threshold, wherein the stability threshold is greater than the instantaneous threshold; When the comprehensive stability index is greater than or equal to the stability threshold, the inspection observation interval is determined as the stability window; When the comprehensive stability index is less than the instantaneous threshold, the inspection observation interval is determined as the instantaneous window; When the comprehensive stability index is between the stability threshold and the instantaneous threshold, the inspection observation interval is determined as the window to be reviewed or the condition window. The instantaneous window, the window to be reviewed, and the conditional window are not considered as the stable window.

5. The system according to claim 3, characterized in that, Also includes: When the inspection observation interval is determined to be a condition window that requires additional action to form effective observation based on at least one of the continuity information, the persistence information, the orientation naturalness information, the speed adaptation information, and the historical repeatability information, an action constraint requirement is generated, which is associated with the condition window and the corresponding hidden danger point. Write the aforementioned action constraints into the inspection task; Obtain the actual inspection trajectory, verify whether the action constraint requirement has been executed based on the actual inspection trajectory, and whether the inspection observation interval corresponding to the condition window passes the visible boundary check again after the action is executed; If the verification result is successful, the condition window is updated to reflect the valid observation relationship.

6. The system according to claim 1, characterized in that, The determination of the verification status of the potential hazard point includes: Based on the effective observation relationship of the aforementioned potential hazard points, the verification status is determined to be one of the following: effective undertaking of inspection, effective undertaking of monitoring, joint undertaking of inspection and monitoring, inspection candidate exists but no effective observation relationship is formed, monitoring candidate exists but no effective observation relationship is formed, or both inspection and monitoring are invalid undertakings. Generate a failure cause identifier and a compensation direction identifier for the hazard point. The failure cause identifier is used to indicate the reason why the hazard point was not effectively checked, and the compensation direction identifier is used to indicate the suggested direction for supplementary check or blind spot filling of the hazard point.

7. The system according to claim 1, characterized in that, Also includes: Acquire actual inspection trajectory and pose information and actual online status information of monitoring equipment; The actual inspection trajectory pose information is used as the pose information of the inspection path, and the fixed monitoring equipment currently in online status is determined according to the actual monitoring equipment online status information. For each potential hazard point, a new inspection candidate observation relationship is generated based on the location of the potential hazard point and the actual inspection trajectory pose information. A new monitoring candidate observation relationship is generated based on the location of the potential hazard point and the pose and field of view information of the currently online fixed monitoring equipment. Perform the visible boundary check on each newly generated candidate observation relationship, and redetermine the verification status of the hidden danger points to obtain the updated verification status; The coverage status data is updated based on the updated verification status of each potential hazard point; For potential hazards that are identified as ineffective for both inspection and monitoring after the update, a supplementary inspection task instruction or early warning message will be generated.

8. The system according to claim 1, characterized in that, Also includes: Obtain verification feedback information, which includes on-site review results, historical omission records, or safety supervision spot check data; Based on the verification feedback information, determine the deviation between the verification status in the coverage status data of at least one potential hazard point and the actual verification result; Determine the visible boundary verification rule type corresponding to the deviation; Based on the rule type, the rule parameters on which the visible boundary verification is based are modified; For the potential vulnerabilities affected by the correction, the steps of generating candidate observation relationships, performing visible boundary verification, and determining the verification status are re-executed to update the coverage status data.

9. The system according to claim 1, characterized in that, Also includes: Before generating inspection candidate observation relationships for each potential hazard point based on the location of the potential hazard point and the pose information of the inspection path, and before generating monitoring candidate observation relationships based on the location of the potential hazard point and the pose and field of view information of the fixed monitoring equipment, for each potential hazard point, it is determined whether the input information of the potential hazard point meets the preset reliability conditions based on the input data. If the condition is not met, the potential hazard point is marked as pending confirmation, and when constructing the coverage status data, potential hazard points in the pending confirmation state are prohibited from being included in the coverage conclusion.

10. The system according to claim 1, characterized in that, The coverage status data includes the hazard point identifier, verification status, effective observation relationship source type, observation subject identifier, observation location or observation range, and visible boundary verification results corresponding to each hazard point; The effective observation relationship source type is used to indicate that the verification status of the corresponding hidden danger point is formed by the inspection candidate observation relationship, the monitoring candidate observation relationship, or both. The visible boundary verification result is used to record the verification information of the corresponding effective observation relationship through line-of-sight access check and observation angle check.

Images