Elevator safety management method and system

Abstract

The invention relates to the technical field of intelligent building and elevator dispatching control, and particularly discloses an elevator safety management method and system. The method comprises the steps that elevator structure, electrical configuration, environmental parameters, use records and historical maintenance data are collected, and an elevator twin map in a unified format is constructed; fault evolution modeling and simulation are executed based on the atlas, and a trigger factor set is generated; a multi-elevator scheduling task graph is constructed, path optimization and task cooperation are completed, and an intelligent rescue path table is generated; abnormity is identified, a fault response trigger chain is generated, and an edge control instruction is issued; constructing an emergency response rule base and a strategy template set; and carrying out virtual drilling based on the strategy template and outputting feedback data. Compared with the prior art, intelligent scheduling based on multi-source data and a path scoring mechanism under the event that people are trapped in the elevator is achieved, and the response timeliness and the system cooperation capacity are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart buildings and elevator dispatching and control, and in particular to an elevator safety management method and system. Background Art

[0002] As urban architecture continues to grow in complexity and height, elevators, as core facilities for vertical transportation of people in high-rise buildings, are becoming increasingly critical to intelligent building management, ensuring operational safety and fault response efficiency. Existing elevator monitoring systems, which mostly rely on threshold-triggered alarm mechanisms and manual response processes, lack the ability to dynamically perceive and jointly optimize factors such as multi-elevator coordination, resource constraints, and path congestion in complex scenarios.

[0003] Especially in sudden incidents of people being trapped, traditional dispatching mechanisms usually make simple judgments based only on the shortest path or response distance, failing to fully integrate factors such as elevator operating status, maintenance resource distribution, building structure accessibility, and task conflict. This leads to irrational response path selection, low efficiency in rescue resource dispatching, and poor overall system linkage, seriously affecting the timeliness and safety of fault response.

[0004] In addition, existing systems generally lack the ability to structured modeling of elevator structural characteristics, historical usage records, and maintenance status. They are unable to build a unified fault evolution map, nor can they generate differentiated scheduling strategies for specific fault types, resulting in insufficient scheduling decision-making capabilities for high-risk scenarios. Summary of the Invention

[0005] The present invention provides an elevator safety management method and system to solve the problem of how to construct a multi-elevator scheduling task graph that integrates path scoring functions, resource matching constraints and collaborative interference evaluation mechanisms based on elevator structure maps and fault evolution factors, and realize intelligent response path optimization and task collaborative scheduling for elevator entrapment incidents in complex building scenarios.

[0006] In order to solve the above technical problems, the present invention provides an elevator safety management method, comprising:

[0007] Obtain elevator structural information, electrical system configuration, environmental parameters, usage records, and maintenance history data, complete unified format modeling, and generate an elevator twin map;

[0008] Obtain the elevator twin graph, construct a fault evolution modeling scenario, perform multi-type fault chain simulation, extract key control variables, and generate a trigger factor set;

[0009] Obtain the trigger factor set, construct a multi-elevator dispatch task graph, complete path optimization and task collaboration, and generate an intelligent rescue path table;

[0010] The priority function of the task collaboration is:

[0011]

[0012] in, is the final priority score of path i→j; Score the path fitness; k Ω i，j，k is the cumulative cooperative interference value between path i→j and other paths; E j is the current operating status score of the elevator response node j; θ1, θ2, θ3: are the priority function weight parameters; Ω i，j，k is the degree of cooperative interference between paths; E j The health status score value of the elevator response equipment;

[0013] Combine Score, select the most optimal path, and form an intelligent rescue path table;

[0014] Obtaining the intelligent rescue path table, identifying anomalies and generating a fault response trigger chain, starting edge control and generating an edge response log;

[0015] Obtain the edge response log and the intelligent rescue path table, build an emergency response rule base and generate a policy constraint parameter set and a policy template set;

[0016] The strategy template set is obtained, a virtual exercise task set is constructed and a training process is executed, operation deviations are recorded and training feedback data is generated.

[0017] Furthermore, obtaining elevator structural information, electrical system configuration, environmental parameters, usage records, and maintenance history data, completing unified format modeling, and generating an elevator twin map includes the following steps:

[0018] Obtain elevator structural information, electrical system configuration, environmental parameters, usage records, and maintenance history data to build a multi-source fusion data set;

[0019] Performing structural standardization and semantic annotation on the multi-source fusion data set to generate multi-dimensional input data in a unified format;

[0020] Graph structure modeling and parameter binding are performed on the multi-dimensional input data to generate an elevator twin graph containing structural nodes, behavioral relationships and historical attributes.

[0021] Furthermore, the construction of the fault evolution modeling scenario, execution of multiple fault chain simulations and extraction of key control variables, and generation of a trigger factor set include the following steps:

[0022] Obtain the elevator twin map, combine it with high-risk working condition simulation configuration, and build a fault evolution modeling scenario;

[0023] Performing multiple types of fault chain simulations in the fault evolution modeling scenario to generate a sequence of fault propagation paths;

[0024] Extract abnormal change key indicators and control variables from the fault propagation path sequence to generate a trigger factor set.

[0025] Furthermore, obtaining the trigger factor set and constructing a multi-elevator scheduling task graph includes the following steps:

[0026] Obtain the trigger factor set and the on-site elevator distribution status to construct a multi-elevator scheduling task graph;

[0027] Performing scheduling path optimization based on path scoring and resource constraints in the multi-elevator scheduling task graph to generate a preliminary rescue path set;

[0028] The preliminary rescue path set is subjected to collaborative task analysis and priority arrangement to generate an intelligent rescue path table.

[0029] Furthermore, in the step of completing path optimization and task collaboration to generate an intelligent rescue path table, the preliminary rescue path set is collaboratively processed based on the on-site resource status to construct the intelligent rescue path table.

[0030] Furthermore, the process of identifying anomalies, generating a fault response trigger chain, starting edge control, and generating an edge response log includes the following steps:

[0031] Obtain the intelligent rescue path table and real-time operating status data, identify local anomalies and establish a fault response trigger chain;

[0032] Performing condition matching and execution scheduling on the fault response trigger chain to start edge-level control instructions;

[0033] Collect response delay, execution status, and exception feedback data during the execution of edge-level control instructions to generate edge response logs.

[0034] Furthermore, the construction of the emergency response rule base and the generation of the policy constraint parameter set and the policy template set include the following steps:

[0035] Obtain the edge response log and the intelligent rescue path table to build an emergency response rule library;

[0036] Performing path validity analysis and control state reconstruction on the emergency response rule base to generate a policy constraint parameter set;

[0037] Combining the policy constraint parameter set with the fault evolution modeling result, a configurable policy template set is generated.

[0038] Furthermore, the construction of a virtual rehearsal task set and execution of a training process, recording operation deviations and generating training feedback data includes the following steps:

[0039] Obtaining the strategy template set and constructing a virtual drill task set corresponding to the rescue mission structure;

[0040] Executing the interactive training process in the virtual rehearsal task set and recording the operation sequence and behavioral deviations;

[0041] The operation sequence and behavioral deviations are comprehensively analyzed and structurally encoded to generate training feedback data.

[0042] Furthermore, the training feedback data is used for model optimization and structure update when subsequently constructing the elevator twin graph.

[0043] An elevator safety management system, applied to any of the above elevator safety management methods, comprising:

[0044] The modeling module is used to obtain elevator structural information, electrical system configuration, environmental parameters, usage records and maintenance history data, complete unified format modeling and generate elevator twin maps;

[0045] Fault simulation module, used to build fault evolution modeling scenarios, perform multi-type fault chain simulations, extract key control variables, and generate trigger factor sets;

[0046] A scheduling module is used to obtain the trigger factor set, build a multi-elevator scheduling task graph, complete path optimization and task coordination, and generate an intelligent rescue path table;

[0047] The edge control module is used to identify anomalies and generate a fault response trigger chain, start edge control and generate edge response logs;

[0048] A policy module, configured to obtain the edge response log and the intelligent rescue path table, build an emergency response rule base, and generate a policy constraint parameter set and a policy template set;

[0049] The training module is used to obtain the policy template set, build a virtual exercise task set and execute the training process, record operation deviations and generate training feedback data.

[0050] The key innovations of the present invention include:

[0051] (1) Five-dimensional elevator twin map construction mechanism: For the first time, structural, electrical, environmental, usage and maintenance data are unified into a model, providing a unified basis for full system modeling and scheduling calculations.

[0052] (2) Path scoring function and resource-channel joint scheduling model: The scheduling scoring function is constructed by integrating time, resource and structure information, and a multi-objective optimization function and fitness function are introduced to significantly improve the scheduling accuracy.

[0053] (3) Collaborative interference assessment and priority sorting mechanism: By calculating path overlap and delay impact, the task order is dynamically adjusted to achieve multi-path conflict avoidance and scheduling optimization.

[0054] The following are its main beneficial effects:

[0055] (1) Efficient elevator entrapment response path planning in complex building scenarios is achieved. The present invention constructs an elevator twin map to accurately restore the elevator structure, electrical configuration, maintenance records and environmental parameters, and realizes dynamic modeling of the entire life cycle of the elevator. With the help of trigger factors extracted from fault evolution simulation, high-risk nodes can be efficiently located, and a multi-elevator scheduling task graph can be constructed in combination with a scoring function. This graph embeds indicators such as response time, resource adaptability and path traffic risk, significantly improving the rationality and response speed of scheduling.

[0056] (2) Improve the intelligence level and execution stability of elevator collaborative scheduling. This invention introduces a multi-objective optimization algorithm to construct a scheduling objective function and fitness function, achieving optimal path selection and task resource allocation while meeting response time and equipment status constraints. Through a collaborative interference assessment mechanism, path conflicts and task dependencies are dynamically identified, ensuring efficient collaborative scheduling of multiple elevators in a limited space, breaking through the limitations of traditional solutions based on single-path and static scheduling.

[0057] (3) Enhance the system's drill training capabilities and achieve closed-loop optimization of strategies. The generated strategy templates can be automatically mapped into a set of virtual drill tasks. Combined with the behavioral deviation recording and feedback coding mechanism, virtual training of maintenance personnel in complex fault scenarios can be achieved. This further promotes the pre-verification and self-optimization of scheduling strategies before actual combat, and builds a complete "modeling-deduction-response-training" closed loop. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A flowchart of an elevator safety management method provided in an embodiment of the present application;

[0059] Figure 2 This is a structural block diagram of an elevator safety management system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0060] Example 1: Reference Figure 1 , is a flow chart of an elevator safety management method provided by an embodiment of the present invention, the flow chart may include at least steps S100-S600:

[0061] S100: Obtain elevator structural information, electrical system configuration, environmental parameters, usage records, and maintenance history data, complete unified format modeling, and generate an elevator twin map;

[0062] S200: Obtain the elevator twin graph, construct a fault evolution modeling scenario, perform multiple fault chain simulations, extract key control variables, and generate a trigger factor set;

[0063] S300: Obtain the trigger factor set, construct a multi-elevator dispatch task graph, complete path optimization and task collaboration, and generate an intelligent rescue path table;

[0064] S400: Obtain the intelligent rescue path table, identify anomalies and generate a fault response trigger chain, start edge control and generate an edge response log;

[0065] S500: Obtain the edge response log and the intelligent rescue path table, build an emergency response rule base, and generate a policy constraint parameter set and a policy template set;

[0066] S600: Acquire the strategy template set, construct a virtual exercise task set and execute a training process, record operation deviations and generate training feedback data.

[0067] Step S100 at least includes steps S110-S130:

[0068] S110. Obtain elevator structural information, electrical system configuration, environmental parameters, usage records, and maintenance history data to construct a multi-source fusion data set.

[0069] Specifically, this step first obtains information about key structural components such as the car frame, counterweight, traction machine, and door machine system from the elevator's sensor network, control system, and remote monitoring platform; at the same time, it collects parameter information such as the wiring diagram of the electrical system, contactor status, relay protection configuration, and emergency power supply layout; further, through the environmental perception module, it collects environmental parameters such as elevator shaft temperature and humidity, vibration intensity, and magnetic field disturbance, and simultaneously retrieves dynamic operating information such as the dispatching system log, user elevator call records, and frequency statistics in the car, as well as recent inspection, repair, and parts replacement records provided by the maintenance service platform. After the above information is transmitted through the multi-channel acquisition module, a multi-source fusion data set is constructed in the fusion layer. This set is indexed by timestamp and elevator unique number, and has structural integrity and time-series correlation, providing data support for subsequent structural modeling and attribute injection.

[0070] S120 , performing structural standardization and semantic annotation on the multi-source fusion data set to generate multi-dimensional input data in a unified format.

[0071] Specifically, after completing the construction of the multi-source fusion data set, the system executes predefined data cleaning processes for the image, structure and log data in the set, including missing field filling, outlier clipping, numerical normalization and structural rearrangement, etc., to ensure that all types of data meet the structured input specifications. Subsequently, semantic annotation is performed on different data domains through feature engineering rules to build a unified elevator attribute naming system. For example, the contactor status information in the electrical system is labeled as the switch control state node, the magnetic disturbance intensity in the environmental data is labeled as the operation stability risk factor, and event label information is added to the time series data to improve the interpretability of the data in the subsequent graph modeling process. Finally, a multi-dimensional input data set in a unified format is output, which includes node type labels, relationship candidate descriptions, time indexes and dynamic attribute dimensions.

[0072] S130. Perform graph structure modeling and parameter binding on the multi-dimensional input data to generate an elevator twin graph including structural nodes, behavioral relationships, and historical attributes.

[0073] Specifically, based on the multi-dimensional input data in the unified format, the system uses a graph modeling module to construct a heterogeneous graph structure containing structural nodes and behavioral edges. The graph structure uses key components such as cars, counterweights, door machines, and traction machines as structural nodes, and associates them with behavioral nodes such as user usage behavior, maintenance records, and alarm events. It forms a multi-type edge set through time series edges, fault cause edges, and maintenance relationship edges between structures and behaviors. After the graph structure is constructed, the system further performs parameter binding operations on all structural nodes and edge relationships, including dynamic attribute fields such as node real-time status, life prediction parameters, historical operation distribution, and abnormal trigger frequency. The above parameter binding results also constitute the embedded dimension of the graph for subsequent fault evolution modeling and path decision modeling.

[0074] Furthermore, the elevator twin graph serves as the basic data structure throughout the entire process of the present invention. Its output results are not only necessary input for constructing the fault evolution modeling scenario in S200, but also the historical operation distribution and abnormal frequency parameters therein are used in step S220 to simulate the fault propagation path under different working conditions; in addition, the maintenance record nodes bound to the twin graph are used to generate structural control constraints in the generation of the S500 module strategy template.

[0075] Through the complete operation process of the above-mentioned S100, unified modeling of the elevator's multi-dimensional data sources can be achieved, and an elevator twin graph with time series attributes, behavioral edges, historical states and structural nodes can be constructed. This provides a structured and reasonable basic semantic graph structure for subsequent fault evolution modeling and intelligent path generation, significantly improving the system's prior perception ability and response intelligence level of fault development and rescue paths.

[0076] Step S200 at least includes steps S210-S230:

[0077] S210: Obtain the elevator twin map, combine it with high-risk working condition simulation configuration, and construct a fault evolution modeling scenario.

[0078] Specifically, the structural node information, electrical system connection status and historical maintenance records contained in the twin graph will be loaded into the fault modeling engine as basic input. According to the predefined high-risk working condition simulation configuration, a parameter set containing a variety of typical fault modes such as wire rope breakage, door machine jamming, power outage, electrical protection failure, etc. is loaded, and combined with the physical boundary conditions mapped by the structural entity, a fault evolution modeling scenario with spatial structure mapping, component state constraints and time evolution rules is generated. The modeling scenario is organized in the form of a graph structure and corresponds one-to-one with the parameters bound in the previous S130 step to ensure that each risk unit maintains semantic consistency with its historical behavior data in the twin graph.

[0079] Furthermore, to adapt to the need for high-frequency emergency simulation in the elevator operating environment, the system divides the fault modeling scenario into multiple sub-areas, corresponding to key parts such as the elevator shaft, car, counterweight, control cabinet, door machine, and button panel. Simulation instances and trigger rules are configured in each sub-area to simulate the cascading effects of fault chains between different components. The structure, state, and trigger logic of the simulation instance need to be modeled in combination with the historical maintenance frequency, component replacement records, and alarm logs contained in the twin map to ensure high fidelity and temporal consistency of the modeling scenario.

[0080] S220 , executing multiple types of fault chain simulations in the fault evolution modeling scenario to generate a fault propagation path sequence.

[0081] After the fault evolution modeling scenario is loaded, the system activates the fault simulation task execution module, sequentially triggering simulation instances encompassing various fault types. These include, but are not limited to, abnormal door zone closure, traction system stall, emergency stop failure, limiter actuation failure, and elevator call system disconnection. For each fault type, propagation rules are defined based on trigger time, propagation rate, associated components, and intervention control logic, constructing a multi-level fault propagation path.

[0082] Specifically, the system organizes fault propagation paths into a graph structure based on chronological order and logical cause-and-effect, generating a path sequence with clear node weights, path depth, and propagation probability. Each path includes an initial node (e.g., a sensor anomaly or input anomaly), intermediate state transition nodes (e.g., relay activation, power switching), and a final response node (e.g., alarm execution, elevator landing failure). The system also records the response delay, control feedback status, and number of potentially affected passengers involved in the state transition.

[0083] To enhance the dynamic adaptability of the simulation path sequence, the system introduces a multi-round simulation execution mechanism. This mechanism dynamically adjusts the simulation initial conditions and fault triggering logic by combining historical anomaly cases and approximate operating condition parameters in the elevator twin map, thereby obtaining a variety of simulation results under different initial conditions. The resulting fault propagation path sequence is output as a multi-scenario structure, providing a data foundation for subsequent extraction of key control variables.

[0084] S230: Extract abnormal change key indicators and control variables from the fault propagation path sequence to generate a trigger factor set.

[0085] After the fault propagation path sequence is generated, the system calls the path parsing module to perform feature extraction on each node and path segment in the path sequence. First, all nodes in the propagation path are scored for abnormality based on multi-dimensional parameters such as fault propagation intensity, node activity, and feedback response interval. Then, based on the scoring results, several nodes that recur across multiple simulation paths and significantly influence path evolution are identified. The sensor variables, execution module states, and logical judgment conditions associated with these nodes are defined as key control variables.

[0086] Furthermore, the key control variables are bound to corresponding abnormal change indicators to form a set of candidate factors. Each factor in this set consists of a trigger source, abnormal amplitude, time window length, and response type, characterizing the initial cause and evolutionary trend of a specific fault scenario. After deduplication, normalization, and structural standardization, this set of candidate factors is transformed into a final set of trigger factors, which serves as the core input for constructing the multi-elevator scheduling task graph in S300.

[0087] Understandably, this set of trigger factors not only forms an important foundation for the subsequent task graph construction, but also serves as a criterion for initiating early warnings at edge control nodes and a reference for variable definitions during policy template generation. Thus, step S230 forms the core link between the aforementioned fault propagation and subsequent scheduling logic.

[0088] Through this step, the system implemented dynamic simulation of multiple fault chains and control variable identification based on elevator twin graphs in a high-fidelity modeling environment. This established a precise mapping pathway from structural state identification to fault trend prediction, providing key parameter support for subsequent multi-elevator intelligent scheduling and edge-response control. This process effectively improved the coverage depth and prediction granularity of elevator fault diagnosis, laying the data foundation for multi-path concurrent modeling and strategy optimization.

[0089] Step S300 at least includes steps S310-S330:

[0090] S310: Acquire the trigger factor set and the on-site elevator distribution status, and construct a multi-elevator scheduling task graph.

[0091] Specifically, S310 is based on the trigger factor set generated in step S230 and combined with the following structural input:

[0092] The spatial coordinates of the structural nodes defined in the elevator twin graph;

[0093] The spatial distribution status of all elevators in the current building;

[0094] Current elevator operation status and historical dispatch data;

[0095] Geographic location information of elevator maintenance resources;

[0096] Critical access connectivity in building structural diagrams.

[0097] On this basis, the initial scheduling task graph structure is constructed, marking the reachable paths and estimated reach time between each potential fault point and the response node.

[0098] To quantify the timeliness and resource matching between nodes in the scheduling path graph, the system defines the following path scoring function:

[0099] Π i，j =α1·T i，j +α2·R i，j +α3·C i，j

[0100] in:

[0101] Π i，j represents the comprehensive score of the path from trigger factor node i to response elevator unit j;

[0102] T i，j It represents the estimated arrival time of the fault response, which is derived from the distance between nodes in the twin graph and the current channel reachability;

[0103] R i，j Indicates the degree of resource matching required for the dispatch path, scored based on the equipment and skill tags of the maintenance team;

[0104] C i，j The congestion risk score representing the path in the building structure is generated by combining the historical dispatch density;

[0105] α1, α2, and α3 are the weight parameters of time, resource, and structure influencing factors, respectively, and are preferably 0.4, 0.3, and 0.3. Their values are derived from the optimal values obtained by fitting the system's historical path scoring database.

[0106] The scoring function above can generate a path graph matrix for all potential dispatch paths, which can be used for subsequent optimization algorithm calls. The generated results will be passed to step S320 to perform dispatch path optimization.

[0107] S320: Perform scheduling path optimization based on path scoring and resource constraints in the multi-elevator scheduling task graph to generate a preliminary rescue path set.

[0108] Based on the multi-elevator dispatching task graph and its path scoring matrix constructed in step S310, S320 further introduces resource constraints and task timeliness factors, performs dispatching path optimization through an optimization algorithm, and forms a feasible preliminary rescue path set.

[0109] 1. Construction of Scheduling Objective Function

[0110] In order to achieve minimum response time and optimal resource matching, the system sets the scheduling optimization objective function as follows:

[0111]

[0112] in:

[0113] represents the scheduling path set, i.e., a set of elevator response paths selected by the system in the graph;

[0114] T i，j : The estimated time for fault response from fault node i to elevator response node j;

[0115] R i，j : Maintenance resource scheduling matching cost;

[0116] L i，j : The current path congestion level extracted from the path scoring function;

[0117] β1, β2, β3: weighted coefficients of the scheduling objective function, which control the influence of time priority, resource optimization and channel risk respectively. The preferred values in experience are β1 = 0.5, β2 = 0.3, and β3 = 0.2.

[0118] The objective function is the original path scoring function Π i，j Further structural decomposition and multi-factor scheduling trade-offs are carried out to solve the problem that traditional path scheduling only focuses on the shortest path and ignores resource and risk allocation.

[0119] 2. Constraint Setting

[0120] The optimization problem constraints include:

[0121] Elevator response uniqueness constraint: Each fault node can only be responded to by one elevator;

[0122] Elevator load capacity constraint: The total number of response tasks for each elevator must not exceed its preset upper limit;

[0123] Path continuity constraint: The selected path must exist continuously in the task graph;

[0124] Response time threshold constraint: The response time of each path must be less than the maximum allowed time set by the system.

[0125] Combining the above objective functions and constraints, the system introduces a multi-objective scheduling algorithm based on ant colony optimization to solve the path, and measures the quality of each path with the following path fitness function:

[0126]

[0127] in:

[0128] Path fitness score, used to guide the algorithm to perform local search;

[0129] γ1, γ2, γ3: fitness function weight parameters, preferably 0.6, 0.3, 0.1;

[0130] ∈: a small constant to avoid division by zero errors;

[0131] The three parts of the fitness function encourage short response time, low resource cost and low path risk respectively.

[0132] This fitness function provides dynamic adjustment capability for scheduling path selection, enabling the algorithm to adapt to the current elevator layout and risk environment, thereby selecting several priority paths from multiple optional paths to form a preliminary rescue path set.

[0133] 3. Generation of a Preliminary Rescue Path Set

[0134] After iterative execution of the above path optimization model and algorithm, the system obtains several high-fitness paths that meet all scheduling constraints. The path set includes:

[0135] The mapping relationship between the fault point and the elevator response node;

[0136] Total response time, resource consumption, and risk estimation for each path;

[0137] The scheduling number and task priority corresponding to each path.

[0138] This preliminary rescue path set will be further used for collaborative task analysis and priority arrangement in the next step S330.

[0139] S330: Perform collaborative task analysis and priority ranking on the preliminary rescue path set to generate an intelligent rescue path table.

[0140] In step S320, the system obtains a preliminary set of rescue paths that have been optimized based on path scoring and resource constraints. Based on this, step S330 further performs task collaborative analysis and prioritization operations, ultimately generating an intelligent rescue path table that can be scheduled for execution.

[0141] 1. Construction of Task Collaboration Structure

[0142] First, the system structures the fault nodes, response elevators, scheduling time windows, and resource occupancy status corresponding to each path in the preliminary rescue path set to generate a task coordination structure. Each coordination structure records the following:

[0143] Fault node identification;

[0144] Respond to the elevator number;

[0145] Path fitness score (inherited from );

[0146] Available time periods and operating windows;

[0147] Required maintenance resources and the time period they will take up;

[0148] Path conflict measurement and synergy impact factors.

[0149] The synergy impact factor is used to describe the interdependence and scheduling competition between different paths and is defined as follows:

[0150] Ω i，j，k =δ1·Overlap(i,k)+δ2·DelayImpact(j,k)

[0151] in:

[0152] Ω i，j，k : represents the cooperative interference value between path i→j and path k;

[0153] Overlap(i,k): the degree of overlap of path time intervals;

[0154] DelayImpact(j,k): the impact of the response time delay scheduled on path k on path i→j;

[0155] δ1, δ2: Collaborative evaluation weight coefficients, empirically set to δ1 = 0.7 and δ2 = 0.3.

[0156] The above formula can automatically evaluate the degree of collaborative interference between tasks and provide a basis for subsequent priority determination.

[0157] 2. Task Prioritization Algorithm Execution

[0158] After building the task coordination structure, the system uses a multi-factor priority function to score and sort all scheduled tasks. The specific priority function is as follows:

[0159]

[0160] in:

[0161] The final priority score of path i→j;

[0162] Path fitness score (calculated by S320);

[0163] ∑ k Ω i，j，k : The cumulative cooperative interference value between path i→j and other paths;

[0164] E j : The current operating status score of the elevator response node j (taking into account factors such as operating frequency and the time of the last maintenance);

[0165] θ1, θ2, θ3: priority function weight parameters, preferably set to 0.6, 0.3, and 0.1.

[0166] This function uses path score as a positive factor and task conflict and equipment status as control factors to achieve a fine sorting of task scheduling order, thereby forming a reasonable and feasible response plan sequence in the scheduling system.

[0167]

Letter and function definition

[0168] Defined by step S320, it represents the fitness score of the response path;

[0169] Ω i，j，k : The degree of cooperative interference between paths is calculated internally in this step;

[0170] E j : Elevator response equipment health status score, derived from maintenance history and usage record data in S130;

[0171] θ1, θ2, θ3: weight parameters used to adjust the proportion of each factor in the ranking. The value can be dynamically updated according to the actual operation and maintenance strategy.

[0172] 3. Intelligent rescue path table generation

[0173] According to the calculated priority ranking results, the system sorts all preliminary paths into The values are arranged from high to low, and combined with the time window and equipment conflict constraints, the top several high-priority paths are selected to form the final intelligent rescue path table. Each path item records the following:

[0174] Path number and mission target (i.e., the faulty elevator to be rescued);

[0175] Respond to the elevator number;

[0176] Start time window and execution order;

[0177] Details of required resources and path suitability;

[0178] Path conflict records and alternative path suggestions (for use when an exception occurs on the original path).

[0179] The resulting intelligent rescue path table will be directly passed into the edge response scheduling logic in the S400 module to establish a fault response trigger chain and issue edge-level instructions.

[0180] By optimizing the logic connection and structure of the three steps S310, S320, and S330 in the above-mentioned S300 module, the present invention achieves the following technical effects:

[0181] Efficient conversion from trigger factors to path diagrams: Utilize twin graph structures and fault propagation factors to construct task scheduling diagrams, effectively bridging the interface gap between data modeling and scheduling execution.

[0182] Scheduling optimization capability integrating path scoring and resource constraints: The proposed multi-objective path optimization mechanism consisting of objective function and fitness function breaks through the traditional scheduling scheme based on the shortest path model and significantly improves response efficiency and resource utilization.

[0183] Prioritization mechanism supporting conflict detection and task collaboration: Introducing collaborative interference factors and elevator status scores to establish a complete prioritization system, effectively supporting the closed-loop control execution of subsequent edge scheduling strategies.

[0184] Step S400 at least includes steps S410-S430:

[0185] S410: Obtain the intelligent rescue path table and real-time operation status data, identify local anomalies and establish a fault response trigger chain.

[0186] Specifically, step S410 uses the intelligent rescue path table output in S330 as the primary control input, along with the current elevator operating status data, car position, transmission component operating conditions, door status, temperature and humidity environment, and state sequences generated by edge sensors, acquired by the edge acquisition module, for a joint analysis. The intelligent rescue path table, which includes fields such as dispatch priority, path number, responding elevator number, and estimated arrival time, has been explicitly modeled in the previous module.

[0187] Based on the comparison between the running status data and the path table scheduling plan, the system performs real-time abnormality identification operations. If one of the following situations is detected:

[0188] The deviation between the current position of the elevator car and the scheduled dispatch node exceeds the dynamic threshold;

[0189] Edge sensors detect abnormal equipment behavior such as car emergency stop, limiter triggering, or door opening failure;

[0190] The power consumption fluctuation of the elevator electrical system exceeds the safety limit set by the twin map;

[0191] The abnormal warning signal of the temperature, humidity or fire sensor lasts longer than the specified duration;

[0192] The system determines that the current elevator has a local failure risk or a scheduling deviation risk, and a fault response trigger chain needs to be established.

[0193] The fault response trigger chain uses the intelligent rescue path table as the skeleton, and associates the identified abnormal nodes, timestamps, risk levels, and control strategies to be dispatched. Each trigger chain structure records the following core content:

[0194] The relationship between abnormal node location and path matching;

[0195] Corresponding elevator number and dispatch number;

[0196] Trigger time point and scheduling lag;

[0197] Risk factor sources and abnormality level annotations;

[0198] Preset control response types (such as forced parking, warning activation, emergency lighting, voice alarm, etc.).

[0199] S420: Perform condition matching and execution scheduling on the fault response trigger chain, and start edge-level control instructions.

[0200] After completing the trigger chain construction in step S410, step S420 executes edge-level control operations based on the trigger chain contents. Specifically, the system matches each item in the fault response trigger chain with a corresponding control instruction template based on the fault type, risk level, scheduling node location, and pre-set control strategy rules. The corresponding control instruction template is then issued through the edge control module.

[0201] The control instructions include but are not limited to the following types:

[0202] Start the car emergency lighting system;

[0203] Control the elevator to stop at a certain floor and lock the door;

[0204] Send voice / image abnormality reports back to the monitoring center;

[0205] Activate car ventilation or audio comfort system;

[0206] Switch to backup power supply line;

[0207] Control the door machine locking device for physical intervention.

[0208] Understandably, when executing commands, the edge control module needs to confirm whether the elevator is reachable based on its current physical state. For example, if the elevator car is descending and its position is below the bottom buffer point threshold, some door lock commands will be automatically blocked by the system.

[0209] The scheduling logic of control instructions is matched by the system according to the following three conditions:

[0210] State condition matching: The prerequisite state for command execution is consistent with the real-time acquisition state, such as the transmission system is stationary, the door lock is not closed, etc.

[0211] Risk level matching: When the fault response level is high, multiple concurrent response strategies are matched first;

[0212] Path node evaluation: If the current scheduling path contains multiple intervention nodes, the execution order of the control instructions is adjusted according to the path coordination sequence priority.

[0213] After receiving the instructions, the edge-level control module controls each module component through the local interface and returns the execution signal to the central control scheduling end.

[0214] S430: Collect response delay, execution status, and abnormal feedback data during the execution of edge-level control instructions, and generate an edge response log.

[0215] Step S430 is automatically started after the edge control instruction in S420 is executed, and is used to collect multi-dimensional response data during the execution process and generate an edge response log in a standard format.

[0216] The collection content includes:

[0217] Control command issuance time and response time, and calculate command response delay;

[0218] Physical interface feedback status, including execution success, delayed response, or failure type;

[0219] Elevator module status change records, such as door control feedback open and close status, motor load data, etc.

[0220] Synchronously collected audio and video feedback, passenger interaction status, and user abnormal complaint records;

[0221] Records of instruction actions that were interrupted or modified by the system;

[0222] Event codes and judgment levels corresponding to all exception types.

[0223] The data collection process is completed through the diagnostic chip and status sampling sub-module embedded in the edge terminal. The data is periodically uploaded to the cloud main control module after being buffered locally. At the same time, if the fault level is higher than level 2, an alarm will be immediately triggered to the superior dispatch center.

[0224] The log file is organized in an event chain structure, indicating the fault event, path number, control response sequence, device status change, time node and intervention feedback, etc. The format is unified and bound to the task scheduling identifier.

[0225] The edge response log finally generated will serve as important data input for the subsequent S500 module to build the policy library and evaluate the control path.

[0226] Through the structured design of the S400 module and the coordinated execution of the three-level sub-steps, the present invention achieves the following based on existing twin modeling and scheduling path planning:

[0227] The edge trigger mechanism based on real-time status data ensures that every anomaly can be detected and matched with the corresponding response action within milliseconds.

[0228] The conditional dynamic matching and scheduling execution capabilities of edge control instructions enable the selection and sorting of control instructions based on the physical state of the elevator and task priority, avoiding invalid or dangerous operations.

[0229] The complete edge response log collection and encoding mechanism provides traceable and structured raw data support for subsequent control path evaluation and policy rule generation, enabling sustainable optimization of scheduling strategies.

[0230] Step 500 at least includes steps S510-S530:

[0231] S510: Obtain the edge response log and the intelligent rescue path table, and build an emergency response rule base.

[0232] Specifically, step S510 first performs a structural analysis operation on the edge response log generated in step S430. The edge response log includes, but is not limited to: the triggering time of the edge-level control instruction, the response delay time, the scheduling execution flag, the control completion status, the feedback exception type, the local node status snapshot, and the control completion flag field.

[0233] At the same time, the system calls the intelligent rescue path table generated in S330 and extracts the response target elevator, task node in the path, coordination number, scheduling priority and resource binding status parameters corresponding to each path.

[0234] After establishing a mapping relationship between the path number and the response identifier between the edge response log and the intelligent rescue path table, the system performs the following operations on each matching path:

[0235] Determine whether the path has completed scheduling execution;

[0236] Determine whether there are unexpected conditions such as abnormal path interruption, unreachable resources, and response failures;

[0237] Extract edge delay parameters that affect response efficiency and their offset from the scheduling time window;

[0238] Generate path status labels, including five types of labels: "normal completion", "interruption response", "redundant execution", "priority task insertion", and "path migration".

[0239] Based on the above calibration content, preliminary response rule entries are constructed. Each rule structure includes: task path identification, execution result label, edge response feature combination, scheduling trigger conditions and resource collaboration attributes.

[0240] All rule entries are eventually stored in the emergency response rule base to form a standardized set of structural rules, which can be called by subsequent policy constraint reasoning modules.

[0241] S520: Perform path validity analysis and control state reconstruction on the emergency response rule base to generate a policy constraint parameter set.

[0242] Furthermore, the step S520 performs validity evaluation and state abstraction operations on all path rule items based on the completion of the rule base construction. Specifically, this step includes the following sub-processes:

[0243] The first is the calculation of path validity identification. Based on the path execution status and exception feedback records in the S510 rule base, the system performs consistency verification and validity judgment on all response paths. The path validity judgment must meet the following three conditions at the same time:

[0244] The path execution time window does not overlap with other high-priority paths;

[0245] The path resource configuration does not cause coordination conflicts;

[0246] The path is not replaced by the dynamic insertion of subsequent paths.

[0247] The second is control state reconstruction. For paths labeled "interrupt response" and "path migration," the system re-derives the scheduling state machine transition process based on the response final state in the edge response log and the elevator structure node binding information established in S130, and maps it to generate the current elevator control state structure.

[0248] The third is parameter set extraction. Based on the completion of the validity identification and control state mapping, the system generates a policy constraint parameter set. The parameter set structure includes but is not limited to:

[0249] Path valid identification;

[0250] Control status index;

[0251] Response priority level;

[0252] State transition trajectory encoding;

[0253] path stability score;

[0254] Resource conflict coefficient;

[0255] Co-compression value;

[0256] Scheduling window interval parameters.

[0257] The above-mentioned policy constraint parameter set will be embedded in the policy template structure as constraint conditions and configuration rules in the subsequent step S530 to form an emergency policy expression framework for multiple fault scenarios.

[0258] S530 : Combining the policy constraint parameter set with the fault evolution modeling result, generate a configurable policy template set.

[0259] After completing the construction of the S520 parameter set, the S530 step further introduces the trigger factor set and fault evolution path information generated in the S230, and generates an emergency response policy template set with structured configuration capabilities using a policy template reasoning mechanism.

[0260] Specifically, the system first calls the fault node in the fault propagation path sequence that matches the current control state, and maps it to the path control interval marked in the parameter set, performs template attribution judgment based on the matching level and deviation index, and forms a preliminary strategy candidate set.

[0261] The system selects policy fragments from the candidate policy set that meet the following three types of constraints:

[0262] The scheduling priority is not lower than the original strategy task path;

[0263] There is topological consistency between the control state transition and the target elevator structure node;

[0264] The required resource is marked as available within the current time window.

[0265] Subsequently, the policy fragments that pass the screening are encapsulated. Each policy template contains: trigger factor combination identifier, priority path set number, resource collaboration label, abnormality tolerance parameter, scheduling window threshold, preset fallback policy index and path migration rule code.

[0266] All policy templates are finally stored in the policy template set structure and output to the S600 module for loading and use by the virtual exercise task structure.

[0267] Through the three-stage operation process of the S500 module, the structured expression of intelligent path execution results, the control state reconstruction of scheduling behavior logic and the rule encapsulation of policy templates are achieved. The specific technical effects are as follows:

[0268] A unified emergency response knowledge structure is formed in large-scale elevator systems to facilitate the rapid extraction of high-frequency path behavior patterns and scheduling status.

[0269] Achieve bidirectional alignment between response behavior and fault status to improve the accuracy and adaptability of policy recommendations.

[0270] It supports strategy reconstruction, template updates, and visual adjustments in complex task scheduling scenarios, enhancing the flexibility of strategy deployment and the robustness of system response.

[0271] Step S600 at least includes steps S610-S630:

[0272] S610: Acquire the strategy template set and construct a virtual drill task set corresponding to the rescue task structure.

[0273] In this step, the configurable policy template set generated in step S530 is first obtained. This policy template set includes multiple emergency response policy templates generated based on the policy constraint parameter set and the fault evolution modeling results. Each policy template corresponds to a specific fault type, control state reconstruction path, and its constraints.

[0274] Specifically, the system parses each policy template in the policy template set, extracting information such as the corresponding fault type, control state reconstruction path, and constraints. Then, combined with the elevator twin graph generated in step S130, a set of virtual drill tasks corresponding to the rescue mission structure is constructed. Each task in this virtual drill task set simulates a specific fault scenario and includes a corresponding emergency response strategy.

[0275] For example, for the "stuck elevator door" fault described in the policy template, the system will construct a virtual drill task to simulate the elevator door stuck fault scenario and include corresponding emergency response strategies, such as manual unlocking and notifying maintenance personnel. This task will include information such as fault triggering conditions, emergency response procedures, and control state changes.

[0276] Furthermore, the system will set execution conditions and evaluation criteria for virtual drill tasks based on the constraints in the strategy template. For example, if an emergency response strategy needs to be completed within a specific timeframe, the system will set a corresponding time limit in the virtual drill task and evaluate whether the operator's response time meets the requirements during the training process.

[0277] Through the above process, the system constructs a set of virtual rehearsal tasks corresponding to the rescue mission structure, providing a basis for the subsequent interactive training process.

[0278] S620: Execute the interactive training process in the virtual rehearsal task set and record the operation sequence and behavior deviation.

[0279] In this step, the system executes the interactive training process in the virtual exercise task set constructed in step S610. This training process is based on virtual reality (VR) technology, providing an immersive training environment that enables operators to conduct emergency response training in a near-realistic scenario.

[0280] Specifically, the system loads the virtual drill task into the VR training platform, and the operator enters the virtual training environment by wearing VR equipment. During the training process, the system records the operator's operation sequence in real time, including the timestamp, operation content, and operation order of each operation.

[0281] The system also records operator behavioral deviations, i.e., discrepancies between operator behavior and expected behavior. These deviations include, but are not limited to, incorrect operation sequences, time limits exceeded, and incorrect operation content. The system compares these deviations with the emergency response strategy in the policy template to identify and record relevant information.

[0282] In addition, the system will also record the operator's physiological data during training, such as heart rate, respiratory rate, etc., to assess the operator's stress level and coping ability during emergency response.

[0283] Through the above process, the system collected the operator's operation sequence and behavioral deviation data in the interactive training process, providing data support for subsequent comprehensive analysis and structural coding.

[0284] S630: Perform comprehensive analysis and structural coding on the operation sequence and behavioral deviation to generate training feedback data.

[0285] In this step, the system comprehensively analyzes and structures the operation sequence and behavior deviation data collected in step S620 to generate training feedback data. This training feedback data will be used for subsequent model optimization and structural updates of the elevator twin graph.

[0286] Specifically, the system first analyzes the operation sequence to assess whether the operator's operational procedures comply with the emergency response strategy in the policy template. The system identifies incorrect steps, missing steps, and repeated steps in the operational procedures, and calculates indicators such as the accuracy and completeness of the operational procedures.

[0287] The system then analyzes the behavioral deviation data to assess the type, frequency, and severity of operator behavioral deviations during emergency response. The system identifies common operator errors in specific failure scenarios and analyzes their possible causes, such as insufficient operational knowledge and insufficient emergency response experience.

[0288] Next, the system structures and encodes the results of the operation sequence analysis and the behavioral deviation analysis to generate training feedback data. This training feedback data will include information such as the operation process evaluation results, the behavioral deviation analysis results, and the operator's emergency response capability assessment.

[0289] In addition, the system will also make targeted training suggestions based on training feedback data, such as strengthening emergency response training for specific fault scenarios and improving operators' emergency response knowledge level.

[0290] Through the above process, the system generates structured training feedback data, providing data support for subsequent model optimization and structural updates of the elevator twin graph.

[0291] By implementing the steps of the above-mentioned S600 stage, the present invention achieves the following technical effects:

[0292] A set of virtual drill tasks corresponding to the rescue mission structure was constructed, enabling operators to conduct emergency response training in scenarios close to reality, improving the effectiveness and pertinence of the training.

[0293] Through the interactive training process, the system records the operator's operation sequence and behavioral deviations in real time, providing data support for subsequent comprehensive analysis and structural coding.

[0294] Through comprehensive analysis and structural encoding of operation sequences and behavioral deviations, the system generates structured training feedback data, providing data support for subsequent model optimization and structural updates of the elevator twin map, thereby improving the intelligence level of the system.

[0295] Through the above process, the present invention realizes closed-loop management from the strategy template set to the training feedback data, thereby improving the overall performance and reliability of the elevator safety management system.

[0296] Example 2: Figure 2 FIG. 1 shows a structural block diagram of an elevator safety management system according to an embodiment of the present invention. Figure 2 As shown, the structure may include:

[0297] Modeling module 10 is used to obtain elevator structural information, electrical system configuration, environmental parameters, usage records and maintenance history data, complete the unified format modeling of the data, and generate the elevator twin map. Specifically including:

[0298] Build a multi-source fusion data set covering mechanical, electrical, environmental, and operation and maintenance;

[0299] Perform structural standardization and semantic annotation on the above data sets;

[0300] Perform graph structure modeling and parameter binding to generate an elevator twin graph with structural nodes, behavioral relationships, and historical attributes.

[0301] The fault simulation module 20 is used to build high-risk fault modeling scenarios based on the elevator twin graph, perform multi-type fault chain simulations, extract key control variables, and output a set of trigger factors. It includes:

[0302] Construct fault evolution modeling scenarios;

[0303] Perform chain simulations for typical faults such as wire rope breakage and door crane jamming;

[0304] Extract abnormal change indicators and control variables in the fault propagation path to form a set of trigger factors with prediction and early warning capabilities.

[0305] The dispatch module 30 builds a multi-elevator dispatch task graph based on the above trigger factor set and the on-site elevator distribution status, realizes path optimization and task coordination, and generates an executable intelligent rescue path table. It mainly includes:

[0306] Build a scheduling task graph;

[0307] Execute path scoring, resource constraints, and scheduling optimization under multi-objective functions;

[0308] Perform task analysis and priority arrangement, and output an intelligent rescue path table.

[0309] The edge control module 40 is used to identify real-time anomalies based on the intelligent rescue path table, initiate edge-level control responses, perform local scheduling, and generate edge response logs. It includes:

[0310] Build a fault response trigger chain;

[0311] Match scheduling conditions and issue edge control instructions;

[0312] Collect response data, latency records, and status feedback during the execution process to form an edge response log.

[0313] The policy module 50 is used to integrate the edge response log and the intelligent rescue path table, build an emergency response rule base, and generate a policy constraint parameter set and a configurable policy template set. The main steps are as follows:

[0314] Build an emergency response rule library to map various fault response scenarios and strategic actions;

[0315] Evaluate path validity, reconstruct control state logic, and extract parameter constraint sets;

[0316] Combine the above content with the fault modeling results to output a highly adaptable policy template set.

[0317] The training module 60 is used to construct a set of virtual training tasks corresponding to the rescue mission structure based on the strategy template set, perform interactive training and generate training feedback data. It includes:

[0318] Extract fault scenario information from the policy template set to build virtual training tasks covering the entire life cycle scenario;

[0319] Perform interactive training to record operational processes, behavioral deviations, and status responses;

[0320] Analyze deviant behavior, perform structural encoding, generate structured training feedback data, and feed it back to the modeling module to support twin graph optimization.

[0321] The system structure scheme of the present invention has the following beneficial effects:

[0322] Strong global data modeling capabilities: Through modeling modules, integrated modeling of mechanical, electrical, environmental and maintenance information is achieved, supporting the construction of high-precision digital twin maps and effectively avoiding data fragmentation problems.

[0323] High fault prediction accuracy: The fault chain model built based on the simulation module can simulate more than 300 types of fault scenarios, dynamically track the propagation path, and improve the prediction and early warning capabilities of complex anomalies.

[0324] Intelligent coordination of dispatch response: The dispatch module integrates path scoring, resource allocation and multi-elevator coordinated dispatch mechanism to quickly generate rescue task sequences in high-concurrency scenarios.

[0325] Edge response with millisecond-level feedback: The edge control module processes localized commands to implement response actions such as occupant detection, lighting activation, and power-off protection, shortening emergency response time.

[0326] Flexible and adaptive policy decision-making: The policy module supports rule base self-learning and policy template reconstruction to meet personalized response requirements in multi-fault combination scenarios.

[0327] Realistic and efficient training and drills: The training module establishes a virtual drill space based on strategy templates and task structures, strengthens operators' ability to respond to high-risk events, improves training efficiency and forms a closed-loop optimization.

[0328] The system modules are clearly decoupled: the boundaries between functional modules are clear, the input and output are consistent, and they can be deployed in parallel, making them suitable for safety management in various elevator scenarios.

[0329] Supports twin graph self-optimization: Training feedback data can be used as reflux input to the modeling module to achieve dynamic evolution and parameter correction of the twin graph.

[0330] Obviously, the embodiments described above are only some of the embodiments of the present application, rather than all of the embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned specific embodiments, or to make equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the present application specification and the accompanying drawings, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present application.

Claims

1. An elevator safety management method, characterized in that: The following steps are involved: Obtain elevator structural information, electrical system configuration, environmental parameters, usage records, and maintenance history data, complete unified format modeling, and generate an elevator twin map; Obtain the elevator twin graph, construct a fault evolution modeling scenario, perform multi-type fault chain simulation, extract key control variables, and generate a trigger factor set; Obtain the trigger factor set, construct a multi-elevator dispatch task graph, complete path optimization and task collaboration, and generate an intelligent rescue path table; The priority function of the task collaboration is: in, is the final priority score of path i→j; Score the path fitness; k Ω i，j，k is the cumulative cooperative interference value between path i→j and other paths; E j is the current operating status score of the elevator response node j; θ1, θ2, θ3: are the priority function weight parameters; Ω i，j，k is the cooperative interference degree between paths; E j The health status score value of the elevator response equipment; Combine Score, select the most optimal path, and form an intelligent rescue path table; Obtaining the intelligent rescue path table, identifying anomalies and generating a fault response trigger chain, starting edge control and generating an edge response log; Obtain the edge response log and the intelligent rescue path table, build an emergency response rule base and generate a policy constraint parameter set and a policy template set; The strategy template set is obtained, a virtual exercise task set is constructed and a training process is executed, operation deviations are recorded and training feedback data is generated.

2. The elevator safety management method according to claim 1, characterized in that: The acquisition of elevator structural information, electrical system configuration, environmental parameters, usage records, and maintenance history data, the completion of unified format modeling, and the generation of an elevator twin map includes the following steps: Obtain elevator structural information, electrical system configuration, environmental parameters, usage records, and maintenance history data to build a multi-source fusion data set; Performing structural standardization and semantic annotation on the multi-source fusion data set to generate multi-dimensional input data in a unified format; Graph structure modeling and parameter binding are performed on the multi-dimensional input data to generate an elevator twin graph containing structural nodes, behavioral relationships and historical attributes.

3. The elevator safety management method according to claim 1, characterized in that: The method of constructing a fault evolution modeling scenario, performing multiple fault chain simulations, extracting key control variables, and generating a trigger factor set includes the following steps: Obtain the elevator twin map, combine it with high-risk working condition simulation configuration, and build a fault evolution modeling scenario; Performing multiple types of fault chain simulations in the fault evolution modeling scenario to generate a sequence of fault propagation paths; Extract abnormal change key indicators and control variables from the fault propagation path sequence to generate a trigger factor set.

4. The elevator safety management method according to claim 1, characterized in that: Obtaining the trigger factor set and constructing a multi-elevator scheduling task graph includes the following steps: Obtain the trigger factor set and the on-site elevator distribution status to construct a multi-elevator scheduling task graph; Performing scheduling path optimization based on path scoring and resource constraints in the multi-elevator scheduling task graph to generate a preliminary rescue path set; The preliminary rescue path set is subjected to collaborative task analysis and priority arrangement to generate an intelligent rescue path table.

5. The elevator safety management method according to claim 1, characterized in that: In the step of completing path optimization and task collaboration to generate an intelligent rescue path table, the preliminary rescue path set is collaboratively processed based on the on-site resource status to construct the intelligent rescue path table.

6. The elevator safety management method according to claim 1, characterized in that: The process of identifying anomalies, generating a fault response trigger chain, starting edge control, and generating an edge response log includes the following steps: Obtain the intelligent rescue path table and real-time operating status data, identify local anomalies and establish a fault response trigger chain; Performing condition matching and execution scheduling on the fault response trigger chain to start edge-level control instructions; Collect response delay, execution status, and exception feedback data during the execution of edge-level control instructions to generate edge response logs.

7. The elevator safety management method according to claim 1, characterized in that: The construction of the emergency response rule base and the generation of the policy constraint parameter set and the policy template set include the following steps: Obtain the edge response log and the intelligent rescue path table to build an emergency response rule library; Performing path validity analysis and control state reconstruction on the emergency response rule base to generate a policy constraint parameter set; Combining the policy constraint parameter set with the fault evolution modeling result, a configurable policy template set is generated.

8. The elevator safety management method according to claim 1, characterized in that: The process of constructing a virtual rehearsal task set and executing a training process, recording operational deviations, and generating training feedback data includes the following steps: Obtaining the strategy template set and constructing a virtual drill task set corresponding to the rescue mission structure; Executing the interactive training process in the virtual rehearsal task set and recording the operation sequence and behavioral deviations; The operation sequence and behavioral deviations are comprehensively analyzed and structurally encoded to generate training feedback data.

9. The elevator safety management method according to claim 1, characterized in that: The training feedback data is used for model optimization and structure update when subsequently constructing the elevator twin map.

10. An elevator safety management system, applied to the elevator safety management method according to any one of claims 1 to 9, characterized in that: include: The modeling module is used to obtain elevator structural information, electrical system configuration, environmental parameters, usage records and maintenance history data, complete unified format modeling and generate elevator twin maps; Fault simulation module, used to build fault evolution modeling scenarios, perform multi-type fault chain simulations, extract key control variables, and generate trigger factor sets; A scheduling module is used to obtain the trigger factor set, build a multi-elevator scheduling task graph, complete path optimization and task coordination, and generate an intelligent rescue path table; The edge control module is used to identify anomalies and generate a fault response trigger chain, start edge control and generate edge response logs; A policy module, configured to obtain the edge response log and the intelligent rescue path table, build an emergency response rule base, and generate a policy constraint parameter set and a policy template set; The training module is used to obtain the policy template set, build a virtual exercise task set and execute the training process, record operation deviations and generate training feedback data.

Images