Elevator traveling cable dynamic fault detection method, system, equipment and medium

Abstract

The invention relates to the technical field of elevator equipment detection, and discloses an elevator traveling cable dynamic fault detection method, system and device and a medium, and the method comprises the following steps: obtaining multi-mode state data of each line of a traveling cable and elevator operation state data; inputting the operation state data into a preset electromechanical coupling model to obtain a corresponding multi-dimensional health reference signal vector; performing time alignment on the multi-modal state data and the reference vector, and performing subtraction to obtain a multi-dimensional signal residual sequence; inputting the residual error sequence into a pre-training deep learning classification model to obtain a preliminary fault judgment result; a final result is confirmed through causal analysis; and if the fault exists, target fault information is obtained through the time domain reflectometer, and visual information is generated and sent to the human-computer interaction interface. According to the embodiment of the invention, the electrical characteristics of the cable and the mechanical motion state of the elevator are dynamically coupled, and high-precision and high-efficiency fault early warning and diagnosis are realized.

Description

Technical Field

[0001] This invention relates to the field of elevator equipment testing technology, and in particular to a method, system, equipment and medium for dynamic fault detection of elevator traveling cables. Background Technology

[0002] In modern society, elevators are widely used and are inseparable from people's daily lives. During elevator operation, the traveling cable needs to move up and down with the elevator car. Long-term bending motion may cause intermittent disconnection faults in the internal circuit.

[0003] Existing testing methods typically involve stopping the elevator in a fixed position and using static testing techniques (such as measuring each line with a multimeter) to check the continuity of the traveling cable. However, for intermittent faults like this, existing testing methods often require maintenance personnel to repeatedly stop the elevator at multiple locations and measure the electrical characteristics of the traveling cable line by line in order to confirm the location and time of the fault, resulting in low fault diagnosis efficiency. Summary of the Invention

[0004] In view of this, this application provides a method, system, device and medium for dynamic fault detection of elevator traveling cables to solve the above problems.

[0005] Firstly, a dynamic fault detection method for elevator traveling cables is provided, applied to an elevator control platform. This method includes: Acquire multimodal status data of each line of the elevator's traveling cable and elevator operating status data; The operating status data is input into a preset electromechanical coupling model to obtain a multidimensional health benchmark signal vector corresponding to the operating status data. The multimodal state data is time-aligned and subtracted from the multidimensional health baseline signal vector to obtain the multidimensional signal residual sequence; The multidimensional signal residual sequence is input into a pre-trained deep learning classification model to obtain preliminary fault judgment results; Perform causal analysis on the preliminary fault diagnosis results to confirm the final fault diagnosis results; When the final fault determination result is that a fault exists, the target fault information is acquired and recorded by a pre-deployed time domain reflectometer; Visual information is generated based on the target fault information and then sent to a pre-configured human-computer interaction interface.

[0006] The above technical solution achieves high-precision fault detection of elevator traveling cables through multi-dimensional data fusion and intelligent analysis. First, it simultaneously acquires multi-modal state data of the cable (such as strain, temperature, and electrical data) and elevator operating status data. Using an electromechanical coupling model, it dynamically correlates mechanical conditions with cable health benchmarks, accurately capturing abnormal signals through data alignment and residual calculation. Then, a pre-trained deep learning model quickly classifies preliminary faults, and causal analysis is combined to improve the reliability of the judgment. Once a fault is confirmed, a time-domain reflectometer accurately locates the fault information and visualizes it on the interactive interface. The entire process dynamically couples the electrical characteristics of the cable with the mechanical motion of the elevator. Through the synergy of data modeling, intelligent identification, causal verification, and precise positioning, it effectively improves the accuracy of fault warnings and the efficiency of diagnosis, achieving dynamic and efficient fault detection.

[0007] Optionally, the operating status data is input into a preset electromechanical coupling model to obtain a multidimensional health baseline signal vector corresponding to the operating status data, specifically including: Using the finite element analysis method and combining the material properties, geometric structure and physical parameters of the traveling cable, a digital twin of the traveling cable is established, and an electromechanical coupling model is constructed based on the digital twin. The electromechanical coupling model is configured to simulate the mechanical, thermal and electrical coupling relationship of the traveling cable when it is running under a preset healthy state. The position and velocity data from the operating status are input into the mechanical simulation layer of the electromechanical coupling model to obtain the theoretical mechanical parameters of the traveling cable. The theoretical mechanical parameters, elevator load data, and ambient temperature are input into the thermal simulation layer of the electromechanical coupling model to obtain the theoretical temperature distribution of the traveling cable. The theoretical mechanical parameters and theoretical temperature distribution are input into the electrical simulation layer of the electromechanical coupling model to obtain the theoretical electrical data of each circuit. By combining theoretical mechanical parameters, theoretical temperature distribution, and theoretical electrical data, a multidimensional health baseline signal vector is obtained.

[0008] The above technical solution improves the accuracy of the reference signal and strengthens the foundation of fault detection by constructing an electromechanical coupling model. First, a model is built based on the cable material, structure, and physical parameters to accurately simulate the mechanical, thermal, and electrical coupling relationships under healthy conditions. The operating position and speed are input to the mechanical simulation layer to obtain mechanical parameters such as the theoretical bending radius. Combined with load and ambient temperature, a theoretical temperature distribution is generated through a thermal simulation layer. Then, the mechanical parameters and temperature distribution are input to the electrical simulation layer to obtain theoretical electrical data, ultimately forming a multi-dimensional healthy reference signal vector. This process achieves a dynamic correlation between mechanical operating conditions, environmental factors, and electrical characteristics, making the reference signal more closely match the actual operating scenario. This provides accurate reference for subsequent residual analysis and fault diagnosis, ensuring the reliability and accuracy of fault detection from the source.

[0009] Optionally, the multimodal state data is time-aligned and subtracted from the multidimensional health baseline signal vector to obtain a multidimensional signal residual sequence, specifically including: Multimodal state data and multidimensional health baseline signal vectors are time-aligned using their respective timestamps; After time alignment, the multimodal state data at each target time point is subtracted from the corresponding multidimensional health baseline signal vector to obtain the original multidimensional residual vector; The original multidimensional residual vector is scaled using a preset normalization algorithm to obtain a normalized residual vector. Multiple normalized residual value vectors are arranged in chronological order to generate a multidimensional signal residual sequence.

[0010] The above technical solution improves the accuracy of fault feature extraction through precise time alignment and residual processing. First, it uses synchronized timestamps to align multimodal state data with multidimensional health baseline signal vectors, ensuring data matching at the same time. Then, it subtracts the aligned real-time data from the baseline vector time-by-time to obtain the original multidimensional residual vector, which intuitively reflects signal deviation. A preset normalization algorithm is used to scale the residual vector, eliminating dimensional differences between different dimensions. Finally, the normalized residual vectors are arranged in chronological order to generate a multidimensional signal residual sequence. This process effectively preserves fault feature differences, reduces the influence of interference factors, provides high-quality input for subsequent deep learning model fault judgment, and enhances the accuracy and stability of fault detection.

[0011] Optionally, the multidimensional signal residual sequence can be input into a pre-trained deep learning classification model to obtain preliminary fault judgment results, specifically including: The multidimensional signal residual sequence is sliced ​​according to the first time window to form multiple multidimensional residual segments; Each multidimensional residual fragment is input into a deep learning classification model to obtain the classification label and confidence level corresponding to the multidimensional residual fragment; If the classification labels of multiple consecutive multidimensional residual segments are the same preset fault type, and the confidence scores of multiple consecutive multidimensional residual segments are all greater than the preset threshold, a preliminary fault judgment result is generated based on the classification labels and target timestamps of multiple consecutive multidimensional residual segments. The target timestamp is the start timestamp of the first multidimensional residual segment among multiple consecutive multidimensional residual segments. The preliminary judgment result includes the preliminary fault type and the time when the preliminary fault occurred.

[0012] The above technical solution improves the accuracy of fault diagnosis by optimizing the deep learning classification process. First, the multidimensional signal residual sequence is sliced ​​into multiple residual segments according to a first time window. After inputting into the deep learning classification model, the classification labels and confidence scores of each segment are obtained. By setting a judgment condition that consecutive segments of the same fault type have confidence scores exceeding a threshold, preliminary results containing the fault type and occurrence time are generated based on the corresponding labels and start timestamps. This process captures temporal features through time window slicing, and combined with continuous segment verification and confidence score filtering, effectively reduces the impact of misjudgment of a single segment, enhances the reliability and timeliness of the preliminary fault diagnosis, and lays an accurate foundation for subsequent causal analysis.

[0013] Optionally, a causal analysis can be performed on the preliminary fault diagnosis results to confirm the final fault diagnosis results, specifically including: When the preliminary fault assessment result is received, extract the target multimodal state data and target operational state data within the historical time window corresponding to the time of the preliminary fault occurrence. The preset causal rule library is invoked to verify the target multimodal state data and target operational state data within the historical time window. The causal rule library is used to define the logical relationship between multimodal state data and operational state data under different fault types. If the target logical association between the target multimodal state data and the target operating state data within the historical time window satisfies the logical association corresponding to the preliminary fault type in the causal rule base, then the preliminary fault judgment result will be confirmed as the final fault judgment result. If the target logical association does not satisfy the logical association, the final fault judgment result is determined to be a non-fault interference or transient anomaly.

[0014] The above technical solution improves the accuracy of fault diagnosis and reduces false alarms through a causal analysis mechanism. After receiving preliminary fault results, multimodal state data and operational status data within the corresponding historical time window are extracted. A preset causal rule base is invoked to verify the data. This rule base defines the physical and logical relationships between strain, temperature, electrical data, and operational status data under different fault types. If the data trend matches the logical relationship of the corresponding fault type in the rule base, the preliminary result is confirmed as the final result; otherwise, it is determined to be a non-fault interference or transient anomaly. This process, combined with physical logic verification of deep learning results, effectively filters false alarms, strengthens the rigor and reliability of fault diagnosis, and provides accurate basis for subsequent fault handling.

[0015] Optionally, when the final fault determination result indicates the existence of a fault, the target fault information is acquired and recorded using a pre-deployed time-domain reflectometer, specifically including: Based on the final fault diagnosis results, the target line where the fault occurred is determined; An electrical pulse is injected into the target line using a time-domain reflectometer, and the reflected signal is collected. The location of the fault is determined based on the propagation time of the reflected signal, and the type of fault is determined based on the waveform characteristics of the reflected signal. The location and type of the fault are integrated with the time of the fault occurrence and the target operating status data in the final fault judgment result to form the target fault information.

[0016] The above technical solution improves fault diagnosis efficiency by accurately locating faults and integrating information using a time-domain reflectometry (TDRS). Once a fault is confirmed, the target faulty line is first identified. An electrical pulse is injected using the TDRS, and the reflected signal is collected. The fault location is determined based on the propagation time, and the fault type is determined by combining waveform characteristics. Then, the fault location, type, time of fault occurrence, and target operating status data are integrated to form the target fault information. This process utilizes the TDRS to achieve precise fault location and type identification, and integrates multi-dimensional information to improve the fault description, providing a comprehensive and accurate basis for subsequent visualization and maintenance, thus enhancing the accuracy and practicality of fault diagnosis.

[0017] Optionally, visual information is generated based on the target fault information, and the visual information is sent to a pre-configured human-machine interface, specifically including: Retrieve the preset three-dimensional digital twin model of the elevator shaft and traveling cable; Based on the location of the fault, the elevator's position and the corresponding section of the traveling cable at the time of the fault are marked on the traveling cable in the three-dimensional digital twin model. A heat map is generated on a three-dimensional digital twin model of the accompanying cable based on all historical fault information and operational data. The three-dimensional digital twin model with labels and heatmaps is displayed in association with text information containing fault type, fault occurrence time, and operational status data at the time of fault occurrence on the human-computer interaction interface.

[0018] The above technical solution enhances the efficiency and intuitiveness of fault information transmission through 3D visualization. After retrieving the 3D digital twin model of the elevator shaft and traveling cable, the real-time location of the elevator and the faulty section of the cable are marked based on the fault location. A cable heat map is generated by combining historical fault and operational data to intuitively display high-incidence areas of faults. The 3D model with markings and the heat map is then linked and displayed with text information containing fault type, time, and operational status on the human-machine interface. This process uses digital twin technology to achieve visualized reconstruction of the fault scenario, heat map to help identify risk trends, and multi-dimensional information linkage presentation, allowing maintenance personnel to quickly and accurately grasp the full picture of the fault and significantly improve fault response and handling efficiency.

[0019] A second aspect of this application provides a dynamic fault detection system for elevator traveling cables. The system includes a data acquisition module, a data processing module, a fault judgment module, a fault recording module, and a fault indication module, wherein: The data acquisition module is configured to acquire multimodal status data of each line of the elevator's traveling cable and the elevator's operating status data. The multimodal status data includes strain data, temperature data, and electrical data. The data processing module is configured to input the operating status data as operating conditions into a preset electromechanical coupling model to obtain a multi-dimensional health reference signal vector corresponding to the operating conditions. The electromechanical coupling model is used to characterize the characteristics of the traveling cable under the preset health condition. The data processing module is also configured to time-align and subtract the multimodal state data from the multidimensional health baseline signal vector to obtain a multidimensional signal residual sequence. The fault diagnosis module is configured to input the multidimensional signal residual sequence into a pre-trained deep learning classification model to obtain preliminary fault diagnosis results. The fault diagnosis module is also configured to perform causal analysis on the preliminary fault diagnosis results to confirm the final fault diagnosis results; The fault recording module is configured to acquire and record target fault information through a pre-deployed time domain reflectometer when the final fault judgment result indicates that a fault exists. The fault indication module is configured to generate visual information based on the target fault information and send the visual information to a pre-configured human-machine interface.

[0020] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the foregoing.

[0021] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the preceding descriptions.

[0022] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: By constructing a framework for "comparing actual signals with a health benchmark," dynamic, online fault detection of elevator traveling cables is achieved. A theoretical "healthy" electrical signal is generated through an electromechanical coupling model. This signal accurately reflects the electrical characteristics that the traveling cable should exhibit under the current specific operating conditions of the elevator (such as speed and position). The actual electrical signal collected on-site is compared with this "health benchmark." If the difference is small, the cable is working normally; if the difference (i.e., the residual) is significant, it indicates an anomaly. A deep learning model is used to intelligently analyze this difference to determine if a fault exists. The fundamental advantage of this method is that it is not a static, isolated detection, but rather dynamically couples the electrical characteristics of the cable with the mechanical motion state of the elevator. By eliminating the signal fluctuations caused by changes in normal operating conditions, it can more accurately and sensitively capture weak signal anomalies caused by faults, achieving high-precision fault early warning and diagnosis, and ultimately improving maintenance efficiency through visualization. Attached Figure Description

[0023] Figure 1 This is an exemplary system architecture diagram of an elevator traveling cable dynamic fault detection method or an elevator traveling cable dynamic fault detection system according to this application. Figure 2 This is a flowchart illustrating a dynamic fault detection method for elevator traveling cable according to an embodiment of this application. Figure 3 This is a schematic diagram of a dynamic fault detection system for elevator traveling cable according to an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device disclosed in the application embodiment.

[0024] Explanation of reference numerals in the attached diagram: 100, System architecture; 101, First terminal device; 102, Second terminal device; 103, Third terminal device; 104, Network; 105, Server; 301, Data acquisition module; 302, Data processing module; 303, Fault diagnosis module; 304, Fault recording module; 305, Fault indication module; 401, Processor; 402, Communication bus; 403, User interface; 404, Network interface; 405, Memory. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0026] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0027] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0028] Figure 1 This paper presents an exemplary system architecture diagram of an embodiment of a dynamic fault detection method or a dynamic fault detection system for elevator traveling cables that can be applied according to this application.

[0029] like Figure 1 As shown, the system architecture 100 may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0030] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as model training applications, video recognition applications, web browser applications, social platform software, etc.

[0031] Terminal devices 101, 102, and 103 can be either hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with displays, including but not limited to smartphones, tablets, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptops, and desktop computers, etc. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices. They can be implemented as multiple software programs or software modules (e.g., multiple software programs or software modules used to provide distributed services) or as a single software program or software module. No specific limitations are imposed here.

[0032] When terminals 101, 102, and 103 are hardware devices, video capture devices can also be installed on them. These video capture devices can be various devices capable of capturing video, such as cameras, sensors, etc. Users can use the video capture devices on terminals 101, 102, and 103 to capture video.

[0033] Server 105 can be a server that provides various services, such as a backend server for processing data displayed on terminal devices 101, 102, and 103. The backend server can analyze and process the received data and can feed back the processing results (such as recognition results) to the terminal devices.

[0034] It should be noted that a server can be either hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software programs or software modules (e.g., multiple software programs or software modules used to provide distributed services), or as a single software program or software module. No specific limitations are made here.

[0035] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included. In particular, if the target data does not need to be obtained remotely, the above system architecture may exclude the network and include only terminal devices or servers.

[0036] Figure 2 This is a flowchart illustrating a dynamic fault detection method for elevator traveling cables according to an embodiment of this application, as shown below. Figure 2 As shown, this embodiment includes: Step S201: Obtain multi-modal status data of each line of the elevator's traveling cable and the elevator's operating status data.

[0037] For example, various types of sensors are deployed on different lines of the elevator traveling cable. For instance, Hall current sensors and voltage sensors are installed on the power line and signal line, respectively, to collect the current values ​​(such as the current change of the power line when the elevator starts and the working voltage fluctuation of the signal line) and voltage signals of each line in real time. Fiber optic temperature sensors are installed in the middle and at both ends of the cable to monitor the temperature distribution of the cable during operation, such as the temperature change at the bend of the cable when the elevator is fully loaded and moving upwards. At the same time, accelerometers and strain gauges are installed at the fixed end and the movable end of the cable to obtain mechanical characteristic data such as the vibration frequency, amplitude, and mechanical stress of the cable. These cross-modal data such as current, voltage, temperature, vibration, and stress collected by different sensors together constitute multimodal state data, and each data point carries a timestamp accurate to milliseconds.

[0038] Furthermore, the elevator's operating status data is read in real time through the PLC (Programmable Logic Controller) interface of the elevator control system. This includes information such as the elevator's current physical floor (e.g., 8th floor), direction of travel (upward), real-time speed (1.2m / s), load weight in the car (650kg), door status (closed), and cumulative running time (this run has lasted 12 seconds). This operating status data also carries a timestamp synchronized with the multimodal status data, ensuring accurate correspondence in the time dimension during subsequent data processing.

[0039] Step S202: Input the operating status data into the preset electromechanical coupling model to obtain the multidimensional health benchmark signal vector corresponding to the operating status data.

[0040] For example, using the finite element analysis method, combined with the material properties, geometric structure, and physical parameters of the elevator traveling cable, a digital twin is constructed using finite element software as the physical basis of the electromechanical coupling model to simulate the coupling relationship between mechanics, heat, and electricity during operation under a preset healthy state. The position and speed data from the elevator's operating status are input into the mechanical simulation layer of the model to obtain the theoretical mechanical parameters of the traveling cable. The theoretical mechanical parameters, elevator load data, and ambient temperature are input into the thermal simulation layer to obtain the theoretical temperature distribution. The theoretical mechanical parameters and theoretical temperature distribution are input into the electrical simulation layer to obtain the theoretical electrical data of each circuit. Finally, the theoretical mechanical parameters, theoretical temperature distribution, and theoretical electrical data are combined to form a multidimensional health reference signal vector.

[0041] In one possible implementation, the operating status data is input into a preset electromechanical coupling model to obtain a multidimensional health reference signal vector corresponding to the operating status data. Specifically, this includes: using finite element analysis and combining the material properties, geometric structure, and physical parameters of the traveling cable to establish a digital twin of the traveling cable; constructing an electromechanical coupling model based on the digital twin; configuring the electromechanical coupling model to simulate the mechanical, thermal, and electrical coupling relationships of the traveling cable during operation in a preset health state; inputting the position and velocity from the operating status data into the mechanical simulation layer of the electromechanical coupling model to obtain the theoretical mechanical parameters of the traveling cable; inputting the theoretical mechanical parameters, elevator load data, and ambient temperature into the thermal simulation layer of the electromechanical coupling model to obtain the theoretical temperature distribution of the traveling cable; inputting the theoretical mechanical parameters and theoretical temperature distribution into the electrical simulation layer of the electromechanical coupling model to obtain the theoretical electrical data for each circuit; and combining the theoretical mechanical parameters, theoretical temperature distribution, and theoretical electrical data to obtain the multidimensional health reference signal vector.

[0042] Specifically, the finite element method is used, combined with the material properties, geometric structure, and physical parameters of the accompanying cable, to establish a digital twin of the cable. For example, key parameters of each component of the accompanying cable are obtained through material testing, such as the resistivity, yield strength, and thermal conductivity of the copper core conductor; the elastic modulus, Poisson's ratio, glass transition temperature, and dielectric constant of the insulation layer (e.g., neoprene rubber); and the conductivity and thickness of the shielding layer. These parameters are then categorized and entered into a material property database. The fine geometric structure of the cable is obtained through 3D laser scanning and industrial CT (Computerized Tomography) scanning, including the number of cores (e.g., 12 cores), single core diameter, stranding angle, stranding pitch, insulation layer thickness, shielding layer winding method, and outer sheath contour. This data is then used in conjunction with CAD (Computer Aided CAD) to further refine the structure. Computer-aided design (CAD) software is used to construct a 1:1 three-dimensional geometric model, which needs to accurately reproduce the spatial arrangement of the core wires, the wrapping relationship of each layer of material, and the details of the connection interfaces with the elevator car and control cabinet. The geometric model is then imported into finite element analysis software (such as ANSYS or COMSOL) for mesh generation—a fine mesh (0.1mm element size) is used for the contact interface between the core wires and the insulation layer, and for bending areas prone to stress concentration, while a transition mesh is used for the overall structure to balance computational accuracy and efficiency. Next, multiphysics coupling analysis types are defined, including mechanical fields (tension, bending, vibration) and thermal fields (Joule heating, frictional heat generation, environmental heat dissipation). The model incorporates electric fields (resistance, capacitance, signal transmission) and sets inter-field coupling relationships, such as relating mechanical deformation to conductor cross-sectional area changes (which in turn affect resistance) and temperature changes to material resistivity / dielectric constant through custom equations. Finally, boundary conditions are set, such as the fixing constraint method at both ends of the cable (rigid fixing at the connection end with the control cabinet and servo constraint at the connection end with the car), the input method of initial ambient temperature and current load, etc. The convergence of the model is verified by iterative calculation through the solver, and finally a digital twin that can dynamically respond to changes in external load is generated. This twin can output multi-dimensional data such as mechanical stress, temperature distribution, and electrical parameters at various points of the cable under different working conditions in real time.

[0043] Furthermore, based on the physical foundation of the digital twin, dynamic simulation logic is integrated to enable it to simulate the coupling relationships of multiple physics fields under operating conditions. This is achieved by dividing the system into a mechanical simulation layer, a thermal simulation layer, and an electrical simulation layer. The mechanical simulation layer utilizes the geometric structure and material mechanics parameters of the digital twin to configure stress analysis logic (e.g., calculating cable suspension length based on elevator position, and then deriving tension distribution). The thermal simulation layer, based on parameters such as the thermal conductivity and heat dissipation area of ​​the digital twin, configures heat source calculation rules (e.g., the correlation formula between Joule heat generated by current and load, and the relationship between mechanical friction heat generation and vibration frequency) and heat conduction logic (e.g., the curve of air convection heat dissipation coefficient changing with elevator speed). The electrical simulation layer utilizes parameters such as the conductivity and dielectric constant of the digital twin to configure electrical parameter calculation models (e.g., signal line voltage attenuation formula), and deeply correlates the effects of mechanical deformation and temperature on electrical parameters (e.g., changes in contact resistance caused by bending, and decreases in insulation resistance caused by temperature rise). Simultaneously, a mapping relationship is established between elevator operating status data (position, speed, load) and input parameters of the coupling model (such as speed mapping to vibration frequency and load mapping to current magnitude), ultimately forming an electromechanical coupling model that can simulate the mechanical, thermal, and electrical coupling relationship of the traveling cable under a preset healthy state.

[0044] Furthermore, the position and speed data from the operating status are input into the mechanical simulation layer of the electromechanical coupling model to obtain the theoretical mechanical parameters of the traveling cable. For example, when the elevator operating status data shows that the current position is on the 15th floor (45m from the ground floor) and the operating speed is 2m / s upward, the mechanical simulation layer calculates the force distribution of the cable under the traction of the car and outputs the theoretical mechanical parameters, including: the maximum cable tension of 850N (located at the car connection end), the minimum bending radius of 0.5m (located at the bend in the middle of the shaft), and the vibration frequency of 3.2Hz (resonance caused by the car movement).

[0045] Furthermore, the theoretical mechanical parameters, elevator load data, and ambient temperature are input into the thermal simulation layer of the electromechanical coupling model to obtain the theoretical temperature distribution of the traveling cable. Assuming that the elevator load data is 700kg (rated load 800kg) and the ambient temperature in the shaft is 26℃, the thermal simulation layer combines the tension (friction heat generation) and vibration (air convection heat dissipation) in the mechanical parameters to calculate the theoretical temperature distribution of each section of the cable: the car connection end has the highest tension and poor heat dissipation conditions, with a temperature of 32℃; the middle section has a temperature of 28℃ due to sufficient heat dissipation from vibration; and the control cabinet connection end has a temperature of 27℃, which is presented in the form of a three-dimensional temperature cloud map.

[0046] Furthermore, the theoretical mechanical parameters and theoretical temperature distribution are input into the electrical simulation layer of the electromechanical coupling model to obtain the theoretical electrical data of each line. For example, the electrical simulation layer calculates the deformation degree of the core wire (affecting the contact resistance) based on the bending radius in the mechanical parameters. Combined with the influence of temperature distribution on resistance (the temperature coefficient of copper is 0.00393 / ℃), the theoretical current of the output power line is 5.2A (driving current when the load is 700kg), the theoretical voltage of the signal line is 5.0V (allowable fluctuation ±0.1V), and the theoretical grounding resistance of the shielding layer is 0.05Ω. The theoretical mechanical parameters (tension 850N, bending radius 0.5m, vibration frequency 3.2Hz), theoretical temperature distribution (32℃, 28℃, 27℃ for each section) and theoretical electrical data (current 5.2A, voltage 5.0V, grounding resistance 0.05Ω) are combined to obtain a multidimensional health reference signal vector [850, 0.5, 3.2, 32, 28, 27, 5.2, 5.0, 0.05] containing 9 dimensions of parameters.

[0047] Step S203: Time-align the multimodal state data with the multidimensional health baseline signal vector and subtract them to obtain the multidimensional signal residual sequence.

[0048] For example, the multimodal state data and the multidimensional health baseline signal vector are time-aligned using their respective timestamps, so that the two datasets correspond precisely in the time dimension. The multimodal state data at each target time after time alignment is subtracted from the corresponding multidimensional health baseline signal vector to obtain the original multidimensional residual vector. The original multidimensional residual vector is scaled using a preset normalization algorithm to obtain a normalized residual vector. Multiple normalized residual vectors are arranged in chronological order to generate a complete multidimensional signal residual sequence that reflects the changes in residuals of each parameter at different times.

[0049] In one possible implementation, the multimodal state data and the multidimensional health baseline signal vector are time-aligned and subtracted to obtain a multidimensional signal residual sequence. Specifically, this includes: time-aligning the multimodal state data and the multidimensional health baseline signal vector using their respective timestamps; subtracting the multimodal state data at each target time after time alignment from the corresponding multidimensional health baseline signal vector to obtain the original multidimensional residual vector; scaling the original multidimensional residual vector using a preset normalization algorithm to obtain a normalized residual vector; and arranging multiple normalized residual vectors in chronological order to generate a multidimensional signal residual sequence.

[0050] Specifically, the multimodal state data and multidimensional health baseline signal vectors are time-aligned using their respective timestamps. For example, the multimodal state data includes timestamps such as current 4.8A, temperature 30℃, and vibration frequency 3.5Hz collected at 14:20:30.123 on June 25, 2025, and current 5.0A, temperature 31℃, and vibration frequency 3.6Hz collected at 14:20:30.223 on June 25, 2025. The multidimensional health baseline signal vectors are time-aligned using their respective timestamps. The Kang reference signal vector contains reference current of 5.2A, reference temperature of 29℃, and reference vibration frequency of 3.2Hz at 14:20:30.123 on June 25, 2025, and reference current of 5.3A, reference temperature of 30℃, and reference vibration frequency of 3.3Hz at 14:20:30.223 on June 25, 2025, with the same time precision. By matching the same timestamps, the two datasets are precisely correlated in the time dimension.

[0051] Furthermore, the multimodal state data at each target time after time alignment is subtracted from the corresponding multidimensional health baseline signal vector to obtain the original multidimensional residual vector. Taking 14:20:30.123 on June 25, 2025 as an example, the multimodal state data is [4.8A, 30℃, 3.5Hz], and the corresponding multidimensional health baseline signal vector is [5.2A, 29℃, 3.2Hz]. The original multidimensional residual vector obtained after subtracting the two is [-0.4A, 1℃, 0.3Hz]. Similarly, the original multidimensional residual vector at 14:20:30.223 on June 25, 2025 is [5.0A-5.3A, 31℃-30℃, 3.6Hz-3.3Hz], that is, [-0.3A, 1℃, 0.3Hz].

[0052] Furthermore, the original multidimensional residual vector is scaled using a preset normalization algorithm to obtain a normalized residual vector. For example, the preset normalization algorithm is min-max normalization. For the current residual, the historical minimum value is -1A and the maximum value is 1A. After normalization, -0.4A becomes (-0.4 - (-1)) / (1 - (-1)) = 0.3. The historical minimum value of the temperature residual is -2℃ and the maximum value is 2℃. After normalization, 1℃ becomes (1 - (-2)) / (2 - (-2)) = 0.75. The historical minimum value of the vibration frequency residual is -0.5Hz and the maximum value is 0.5Hz. After normalization, 0.3Hz becomes (0.3 - (-0.5)) / (0.5 - (-0.5)) = 0.8. Therefore, 2025-06-25 14:20:3 The normalized residual vector at time 0.123 is [0.3, 0.75, 0.8]. Similarly, the normalized residual vector at time 14:20:30.223 on June 25, 2025 is calculated to be [0.35, 0.75, 0.8]. Arranging multiple normalized residual vectors in chronological order generates a multidimensional signal residual sequence. For example, arranging the normalized residual vectors from 14:20:30.123 on June 25, 2025 to 14:20:31.123 on June 25, 2025 sequentially forms a sequence containing 10 vectors: [[0.3, 0.75, 0.8], [0.35, 0.75, 0.8], [0.32, 0.8, 0.85], ...]. This sequence fully reflects the changes in the residuals of each parameter at different times.

[0053] Step S204: Input the multidimensional signal residual sequence into the pre-trained deep learning classification model to obtain preliminary fault judgment results.

[0054] For example, the multidimensional signal residual sequence is sliced ​​according to a fixed first time window to form multiple multidimensional residual segments containing multiple sets of normalized residual vectors; a CNN-LSTM (Convolutional Neural Network-Long Short-Term Memory) hybrid architecture deep learning classification model with input layer dimension matching segment format is adopted. Local features are extracted through two 1D convolutional modules, and after global max pooling, the temporal dependencies are analyzed by the bidirectional LSTM (Long Short-Term Memory) layer. Then, the fault type probability distribution is output through a fully connected layer and a Softmax activation function. The model is trained with a large number of labeled samples and an optimization strategy is adopted to ensure performance; when multiple consecutive segments have the same classification label and the confidence level exceeds the preset threshold, a preliminary fault judgment result containing the preliminary fault type and the time of occurrence is generated based on the classification label and the start timestamp of the first segment.

[0055] It should be noted that the training process of this hybrid architecture deep learning classification model is as follows: First, a large number of labeled samples are prepared. These samples are multi-dimensional residual fragments formed by slicing at fixed time windows, with each fragment corresponding to an actual fault type label. The samples are divided into training and validation sets proportionally for model parameter learning and performance validation. During training, the Adam (Adaptive Moment Estimation) optimizer is used, with the cross-entropy loss function as the optimization objective. The model parameters are continuously adjusted through backpropagation to make the fault type probability distribution output by the model closer to the actual labels. To prevent overfitting, an early stopping strategy is adopted, i.e., training stops when the performance on the validation set shows no improvement for several consecutive rounds, ensuring that the model maintains good performance on unseen data. Finally, through training, the model achieves a high overall accuracy and specific fault category recognition performance on the test set.

[0056] In one possible implementation, the multidimensional signal residual sequence is input into a pre-trained deep learning classification model to obtain a preliminary fault judgment result. Specifically, this includes: slicing the multidimensional signal residual sequence according to a first time window to form multiple multidimensional residual segments; inputting each multidimensional residual segment into the deep learning classification model to obtain the classification label and confidence level corresponding to the multidimensional residual segment; if the classification labels of multiple consecutive multidimensional residual segments are the same preset fault type, and the confidence levels corresponding to multiple consecutive multidimensional residual segments are all greater than a preset threshold, a preliminary fault judgment result is generated based on the classification labels and target timestamps of the multiple consecutive multidimensional residual segments, wherein the target timestamp is the start timestamp of the first multidimensional residual segment among the multiple consecutive multidimensional residual segments, and the preliminary judgment result includes the preliminary fault type and the time of occurrence of the preliminary fault.

[0057] Specifically, the multidimensional signal residual sequence is input into a pre-trained deep learning classification model to obtain preliminary fault judgment results. This includes slicing the multidimensional signal residual sequence according to a first time window to form multiple multidimensional residual segments. For example, the first time window is set to 5 seconds, and each window does not overlap. For a multidimensional signal residual sequence from 15:30:00 on June 25, 2025 to 15:30:30 on June 25, 2025, after slicing according to this window, 6 multidimensional residual segments are obtained, corresponding to the time periods of 15:30:00-15:30:05, 15:30:05-15:30:10, ..., 15:30:25-15:30:30 respectively. Each segment contains 50 sets of normalized residual vectors collected at 100ms intervals within the 5 seconds, covering residual changes in 9 dimensions such as current, temperature, and vibration.

[0058] Furthermore, each multidimensional residual fragment is input into a deep learning classification model, which employs a CNN-LSTM hybrid architecture. For example, the input layer dimension is set to (50, 9) to match the fragment data format. Local features are first extracted through two 1D convolutional modules—the first module uses 32 convolutional kernels of size 3 to capture abrupt changes in the current residual within 200ms, followed by ReLU (Rectified Linear Array) processing. After activation of the first unit (corrected linear unit), the output is a (48, 32) dimension feature. The second module further compresses it to (47, 64) dimensions using 64 convolutional kernels of size 2. Then, global max pooling is used to obtain a 64-dimensional local feature vector. Subsequently, a bidirectional LSTM layer with 64 hidden units is used to analyze the temporal dependencies (such as the hysteresis change of temperature residual with vibration residual). Finally, a fully connected layer and a Softmax activation function are used to output the probability distribution of 8 fault types. For example, after inputting the first segment (15:30:00-15:30:05), the model identifies that the residual fluctuation conforms to the normal pattern and outputs the classification label "normal" with a confidence score of 0.96. In the second segment (15:30:05-15:30:10), the convolution... The first layer captured high-frequency fluctuations in the current residual within the range of 0.3-0.5. The LSTM layer found a significant temporal correlation between this fluctuation and the increase in temperature residual. The final output classification label was "poor contact" with a confidence score of 0.88. The third segment (15:30:10-15:30:15) output the classification label as "poor contact" with a confidence score of 0.92, and the fourth segment (15:30:15-15:30:20) output the classification label as "poor contact" with a confidence score of 0.91. The model was trained with 100,000 labeled samples (the training and validation sets were divided in an 8:2 ratio, using the Adam optimizer and cross-entropy loss function, and an early stopping strategy to prevent overfitting). The overall accuracy on the test set reached 97.3%, with an F1 score of 0.96 for the "poor contact" category.

[0059] Furthermore, if the classification labels of multiple consecutive multidimensional residual segments are the same preset fault type, and the confidence levels corresponding to multiple consecutive multidimensional residual segments are all greater than the preset threshold, a preliminary fault judgment result is generated based on the classification labels and target timestamps of multiple consecutive multidimensional residual segments. The preset threshold is set to 0.85. At this time, the classification labels of the second to fourth segments are all poor contact, and the confidence levels of 0.88, 0.92, and 0.91 are all greater than 0.85, which meets the condition. The target timestamp is the start timestamp of the first segment among these three segments, i.e., 15:30:05. Therefore, the preliminary judgment result is: the preliminary fault type is poor contact, and the preliminary fault occurrence time is 15:30:05 on June 25, 2025.

[0060] Step S205: Perform causal analysis on the preliminary fault judgment results to confirm the final fault judgment results.

[0061] For example, the multidimensional signal residual sequence is sliced ​​into multiple multidimensional residual segments according to a first time window. Each segment contains multiple sets of normalized residual vectors collected at fixed intervals within the corresponding time period. Each multidimensional residual segment is input into a pre-trained deep learning classification model using a CNN-LSTM hybrid architecture. This model extracts local features through convolutional modules, analyzes temporal dependencies through bidirectional LSTM layers, and outputs the probability distribution of fault types through fully connected layers and Softmax activation functions. The model is trained with a large number of labeled samples and uses optimization strategies to ensure performance. If the classification labels of multiple consecutive segments are the same preset fault type and the confidence level is greater than a preset threshold, a preliminary fault judgment result containing the preliminary fault type and the time of occurrence is generated based on the classification labels of these segments and the start timestamp of the first segment.

[0062] In one possible implementation, causal analysis is performed on the preliminary fault judgment result to confirm the final fault judgment result. Specifically, this includes: when the preliminary fault judgment result is received, extracting the target multimodal state data and target operating state data within the historical time window corresponding to the time of the preliminary fault occurrence; calling a preset causal rule library to verify the target multimodal state data and target operating state data within the historical time window, whereby the causal rule library is used to define the logical association between multimodal state data and operating state data under different fault types; if the target logical association between the target multimodal state data and target operating state data within the historical time window satisfies the logical association corresponding to the preliminary fault type in the causal rule library, then the preliminary fault judgment result is confirmed as the final fault judgment result; if the target logical association does not satisfy the logical association, then the final fault judgment result is determined to be a non-fault interference or transient anomaly.

[0063] Specifically, when the preliminary fault diagnosis result is received, the target multimodal state data and target operating state data within the historical time window corresponding to the time of the preliminary fault occurrence are extracted. For example, if the preliminary fault diagnosis result is poor contact and the time of occurrence is 15:30:05 on June 25, 2025, the historical time window is set to 10 seconds before and after this time (i.e., 15:30:00 to 15:30:10). The target multimodal state data within this period is extracted from the database, including the current residual (fluctuating from 4.8A to 5.2A, accompanied by 3 instantaneous drops to 3.5A), temperature change (rising from 29℃ to 31℃), vibration frequency (stabilizing at around 3.5Hz), and target operating state data (the elevator starts moving upward from the 7th floor, reaches the rated speed of 1.8m / s at 15:30:03, begins to decelerate at 15:30:08 to prepare to stop at the 12th floor, with a load of 600kg).

[0064] Further, the preset causal rule base is invoked to verify the target multimodal state data and target operating state data within the historical time window. The causal rule base adopts production rule representation, where the rule corresponding to poor contact fault is defined as: "If the elevator is in a dynamic operating state (starting, acceleration, deceleration process), and the multimodal data shows a sudden drop in current ≥20%, accompanied by a slight increase in temperature (1-3℃), and stable vibration frequency without violent fluctuations, then it is determined that there is a logical association of poor contact." The rule base also includes the association conditions of other fault types, such as "insulation layer aging" corresponding to "abnormal temperature rise ≥5℃ and slow current increase under continuous high load".

[0065] Furthermore, if the target logical association between the target multimodal state data and the target operating state data within the historical time window satisfies the logical association corresponding to the preliminary fault type in the causal rule base, and comparison reveals that the elevator is in a dynamic operation process from acceleration to deceleration during this period, the current experiences three instantaneous drops of ≥20% (from 5.0A to 3.5A, a drop of 30%), the temperature rises by 2℃ which meets the "small increase" characteristic, and the vibration frequency is stable and does not fluctuate by more than ±0.2Hz, fully matching the rule conditions for poor contact, then the preliminary fault judgment result is confirmed as the final fault judgment result; if the target logical association does not satisfy the logical association, for example, the preliminary fault judgment result is poor contact, but the elevator is in a stationary state within the historical window (stopped on the 5th floor and not running), the current does not experience instantaneous drops and the temperature does not change, which does not meet the necessary conditions of "dynamic operating state" and "current drop" in the rules, then the final fault judgment result is determined to be non-fault interference (possibly caused by temporary false alarms from sensors) or transient anomaly.

[0066] Step S206: When the final fault judgment result is that a fault exists, the target fault information is acquired and recorded by a pre-deployed time domain reflectometer.

[0067] For example, when the final fault diagnosis result indicates the presence of a fault, the target faulty line is determined based on this result and the line layout; an electrical pulse is injected into the target line using a pre-integrated time-domain reflectometer, and a time-domain waveform containing the incident and reflected pulses is acquired; the location of the fault is calculated based on the propagation time of the reflected signal, and the fault type is determined by combining the correspondence between the fault type and the reflection amplitude and waveform distortion in the waveform feature library; the fault location, type, occurrence time in the final fault diagnosis result, and elevator operating status data are integrated to form complete target fault information and stored in the fault database.

[0068] In one possible implementation, when the final fault determination result indicates the presence of a fault, target fault information is acquired and recorded using a pre-deployed time-domain reflectometer. Specifically, this includes: determining the target line where the fault occurred based on the final fault determination result; injecting electrical pulses into the target line using the time-domain reflectometer and acquiring the reflected signals; determining the fault location based on the propagation time of the reflected signals and determining the fault type based on the waveform characteristics of the reflected signals; and integrating the fault location and fault type with the fault occurrence time and target operating status data in the final fault determination result to form target fault information.

[0069] Specifically, based on the final fault diagnosis results, the target line where the fault occurred is determined. For example, if the final fault diagnosis result is poor contact and the initial location is abnormal signal transmission, combined with the wiring layout diagram of the elevator traveling cable (where lines 1-3 are power lines, lines 4-8 are signal lines, and lines 9-12 are grounding lines), it is determined that the fault occurred in signal line 5, which is responsible for transmitting car command signals.

[0070] Furthermore, an electrical pulse is injected into the target line using a time-domain reflectometer (TDRI) and the reflected signal is collected. The TDRI is pre-integrated in the elevator control cabinet and connected to signal line 5 via a dedicated test interface. After the test is started, a square wave electrical pulse with a peak voltage of 12V and a pulse width of 50ns is injected into the line. At the same time, the reflected signal is collected at a sampling rate of 1GHz, and the waveform changes within 200ns are continuously recorded to obtain a complete time-domain waveform diagram containing the incident pulse and the reflected pulse.

[0071] Furthermore, the location of the fault is determined based on the propagation time of the reflected signal, and the fault type is determined based on the waveform characteristics of the reflected signal. It is known that the propagation speed of the electrical pulse in signal line 5 (copper core) is 2.0 × 10⁻⁶. 8 m / s, the time difference between the incident pulse emission and the reflected pulse reception measured from the waveform diagram is 30ns. According to the formula: Fault distance = Propagation speed × Propagation time / 2, the fault occurs 3 meters away from the control cabinet (2.0 × 10 m / s). 8 m / s×30×10 -9 (s / 2=3m), and at the same time, it was observed that the amplitude of the reflected pulse was 40% of the incident pulse, and the waveform showed obvious step-like distortion. Based on the preset waveform feature library (poor contact corresponds to a reflection amplitude of 30%-50% and waveform distortion), the fault type was determined to be intermittent poor contact.

[0072] Furthermore, the fault location and fault type are integrated with the fault occurrence time and target operating status data in the final fault judgment result to form target fault information. This integrates the fault location at 3 meters, the fault type of intermittent poor contact, the occurrence time in the final fault judgment result of 15:30:05 on June 25, 2025, and the operating status data of the elevator at that time being on the 10th floor, with an upward speed of 1.5 m / s and a load of 600 kg. This forms complete target fault information containing "Fault line: Signal line No. 5; Fault location: 3 meters from the control cabinet; Fault type: Intermittent poor contact; Occurrence time: 15:30:05 on June 25, 2025; Operating status: 10th floor, upward, speed 1.5 m / s, load 600 kg", and is stored in the fault database.

[0073] Step S207: Generate visual information based on the target fault information and send the visual information to the pre-configured human-machine interface.

[0074] For example, a 3D digital twin model of the elevator shaft and traveling cable, constructed using laser scanning and CAD drawings, is retrieved. This model accurately presents the shaft layout, cable path, and segmentation information. Based on the location of the fault, the fault section and the corresponding elevator position are marked in the model, and the two are linked by lines to demonstrate spatial relationship. Combining historical faults and operating data, a heat map reflecting the frequency of faults and stress conditions is generated on the model according to preset rules. The 3D model with markings and heat map, along with text information containing fault type, occurrence time, and operating status, are displayed together on a left-right split human-computer interaction interface, supporting model operation and fault details viewing, thus integrating visualization and interactive functions.

[0075] In one possible implementation, visualization information is generated based on the target fault information, and the visualization information is sent to a pre-configured human-machine interface. Specifically, this includes: retrieving a preset three-dimensional digital twin model of the elevator shaft and the traveling cable; marking the elevator's position and the corresponding section of the traveling cable at the time of the fault on the traveling cable in the three-dimensional digital twin model based on the fault location; generating a heat map on the three-dimensional digital twin model of the traveling cable based on all historical fault information and operating data; and displaying the three-dimensional digital twin model with markings and the heat map, along with text information containing fault type, fault occurrence time, and operating status data at the time of the fault, on the human-machine interface.

[0076] Specifically, a pre-defined 3D digital twin model of the elevator shaft and traveling cable is retrieved. For example, this model is a 1:1 reconstruction model built by laser scanning the actual structure of the elevator shaft and combining it with CAD drawings. It accurately presents the spatial layout of the 20 floors in the shaft, the position of the guide rails, the suspension path of the traveling cable (extending from the top of the control cabinet on the first floor to the top of the car, with a total length of 57 meters), and the connection nodes between the cable and the car. In the model, the traveling cable is subdivided into independently identifiable segments of 1 meter each, and each segment is labeled with its corresponding distance from the control cabinet. Based on the location of the fault, the fault location is marked on the traveling cable in the 3D digital twin model. The system automatically locates the elevator's position and the corresponding section of the traveling cable in the model. For example, if the fault information indicates that the fault occurs 3 meters away from the control cabinet and the elevator is on the 10th floor (28 meters above the ground floor) at the time of the fault, the system will automatically locate the section on the traveling cable 3 meters away in the model (this section is slightly curved because the elevator is on the 10th floor and is located in the upper middle part of the shaft). A flashing red cube marker (0.1 meters on each side) will be generated at this location. At the same time, a semi-transparent car shadow will be displayed at the 10th floor car position in the model. The car shadow will be connected to the fault marker by yellow lines, which will intuitively show the spatial relationship between the fault and the elevator position.

[0077] Furthermore, based on all historical fault information and operational data, a heat map is generated on the three-dimensional digital twin model of the traveling cable. For example, the system retrieves fault records from the past year (a total of 12 faults, including 2 in the 3-meter section, 3 in the 8-meter section, 5 in the 15-meter section, and 2 in other locations). Combining the operating time of each section (e.g., the 3-meter section has accumulated 800 hours of stress due to frequent elevator starts and stops), a three-color gradient heat map rule of red-yellow-green is used (red indicates a fault frequency > 5 times / year or accumulated stress > 500 hours, yellow indicates 3-5 times / year or 300-500 hours, and green indicates < 3 times / year or < 300 hours). The heat map distribution of the traveling cable is rendered in the model, with the 15-meter section displayed as dark red, the 8-meter section as orange-yellow, the 3-meter section as yellow, and the remaining sections as green.

[0078] Furthermore, the 3D digital twin model with markers and heatmaps is displayed in conjunction with text information containing fault type, fault occurrence time, and operational status data at the time of fault occurrence on the human-machine interface. For example, the human-machine interface adopts a left-right split layout, with the 3D digital twin model displayed on the left (supporting rotation, scaling, and panning operations), and the text information panel displayed on the right. When the user clicks on the red fault marker in the left model, the right panel automatically updates with the detailed information of the fault: "Fault type: Intermittent poor contact of signal line No. 5; Occurrence time: 15:30:05 on June 25, 2025; Operation status: 10th floor upward, speed 1.5m / s, load 600kg; Historical similar faults: Poor contact has occurred once in this 3-meter section within the past 3 months." At the same time, the bottom of the panel provides operation buttons such as "View maintenance records" and "Generate repair order," achieving seamless integration of visual information and interactive functions.

[0079] Figure 3 This is a schematic diagram of a module of a dynamic fault detection system for elevator traveling cables according to an embodiment of this application, as shown below. Figure 3 As shown, the system includes: a data acquisition module 301, a data processing module 302, a fault judgment module 303, a fault recording module 304, and a fault indication module 305, wherein: The data acquisition module 301 is configured to acquire multimodal status data of each line of the elevator's traveling cable and the elevator's operating status data. The data processing module 302 is configured to input the operating status data into a preset electromechanical coupling model to obtain the multidimensional health benchmark signal vector corresponding to the operating status data. The data processing module 302 is also configured to time-align and subtract the multimodal state data from the multidimensional health baseline signal vector to obtain a multidimensional signal residual sequence. The fault judgment module 303 is configured to input the multidimensional signal residual sequence into a pre-trained deep learning classification model to obtain preliminary fault judgment results. The fault diagnosis module 303 is also configured to perform causal analysis on the preliminary fault diagnosis results to confirm the final fault diagnosis results; The fault recording module 304 is configured to acquire and record target fault information through a pre-deployed time domain reflectometer when the final fault judgment result is that a fault exists. The fault indication module 305 is configured to generate visual information based on the target fault information and send the visual information to a pre-configured human-machine interface.

[0080] Optionally, the data processing module 302 is also configured to: Using the finite element analysis method and combining the material properties, geometric structure and physical parameters of the traveling cable, a digital twin of the traveling cable is established, and an electromechanical coupling model is constructed based on the digital twin. The electromechanical coupling model is configured to simulate the mechanical, thermal and electrical coupling relationship of the traveling cable when it is running under a preset healthy state. The position and velocity data from the operating status are input into the mechanical simulation layer of the electromechanical coupling model to obtain the theoretical mechanical parameters of the traveling cable. The theoretical mechanical parameters, elevator load data, and ambient temperature are input into the thermal simulation layer of the electromechanical coupling model to obtain the theoretical temperature distribution of the traveling cable. The theoretical mechanical parameters and theoretical temperature distribution are input into the electrical simulation layer of the electromechanical coupling model to obtain the theoretical electrical data of each circuit. By combining theoretical mechanical parameters, theoretical temperature distribution, and theoretical electrical data, a multidimensional health baseline signal vector is obtained.

[0081] Optionally, the data processing module 302 is also configured to: Multimodal state data and multidimensional health baseline signal vectors are time-aligned using their respective timestamps; After time alignment, the multimodal state data at each target time point is subtracted from the corresponding multidimensional health baseline signal vector to obtain the original multidimensional residual vector; The original multidimensional residual vector is scaled using a preset normalization algorithm to obtain a normalized residual vector. Multiple normalized residual value vectors are arranged in chronological order to generate a multidimensional signal residual sequence.

[0082] Optionally, the fault diagnosis module 303 is also configured to: The multidimensional signal residual sequence is sliced ​​according to the first time window to form multiple multidimensional residual segments; Each multidimensional residual fragment is input into a deep learning classification model to obtain the classification label and confidence level corresponding to the multidimensional residual fragment; If the classification labels of multiple consecutive multidimensional residual segments are the same preset fault type, and the confidence scores of multiple consecutive multidimensional residual segments are all greater than the preset threshold, a preliminary fault judgment result is generated based on the classification labels and target timestamps of multiple consecutive multidimensional residual segments. The target timestamp is the start timestamp of the first multidimensional residual segment among multiple consecutive multidimensional residual segments. The preliminary judgment result includes the preliminary fault type and the time when the preliminary fault occurred.

[0083] Optionally, the fault diagnosis module 303 is also configured to: When the preliminary fault assessment result is received, extract the target multimodal state data and target operational state data within the historical time window corresponding to the time of the preliminary fault occurrence. The preset causal rule library is invoked to verify the target multimodal state data and target operational state data within the historical time window. The causal rule library is used to define the logical relationship between multimodal state data and operational state data under different fault types. If the target logical association between the target multimodal state data and the target operating state data within the historical time window satisfies the logical association corresponding to the preliminary fault type in the causal rule base, then the preliminary fault judgment result will be confirmed as the final fault judgment result. If the target logical association does not satisfy the logical association, the final fault judgment result is determined to be a non-fault interference or transient anomaly.

[0084] Optionally, the fault logging module 304 is also configured to: Based on the final fault diagnosis results, the target line where the fault occurred is determined; An electrical pulse is injected into the target line using a time-domain reflectometer, and the reflected signal is collected. The location of the fault is determined based on the propagation time of the reflected signal, and the type of fault is determined based on the waveform characteristics of the reflected signal. The location and type of the fault are integrated with the time of the fault occurrence and the target operating status data in the final fault judgment result to form the target fault information.

[0085] Optionally, the fault indication module 305 is also configured to: Retrieve the preset three-dimensional digital twin model of the elevator shaft and traveling cable; Based on the location of the fault, the elevator's position and the corresponding section of the traveling cable at the time of the fault are marked on the traveling cable in the three-dimensional digital twin model. A heat map is generated on a three-dimensional digital twin model of the accompanying cable based on all historical fault information and operational data. The three-dimensional digital twin model with labels and heatmaps is displayed in association with text information containing fault type, fault occurrence time, and operational status data at the time of fault occurrence on the human-computer interaction interface.

[0086] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0087] This embodiment also discloses an electronic device, as shown in the reference. Figure 4 The electronic device may include: at least one processor 401, at least one communication bus 402, user interface 403, network interface 404, and at least one memory 405.

[0088] The communication bus 402 is used to enable communication between these components.

[0089] The user interface 403 may include a display screen and a camera. Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0090] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0091] The processor 401 may include one or more processing cores. The processor 401 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 405, and by calling data stored in memory 405. Optionally, the processor 401 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 401 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 401.

[0092] The memory 405 may include random access memory (RAM) or read-only memory. Optionally, the memory 405 may include a non-transitory computer-readable storage medium. The memory 405 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 405 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 405 may also be at least one storage device located remotely from the aforementioned processor 401. Figure 4 As shown, the memory 405, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for a dynamic fault detection method for elevator traveling cables.

[0093] exist Figure 4 In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 401 can be used to call the application program of a dynamic fault detection method for elevator traveling cable stored in the memory 405. When executed by one or more processors 401, the electronic device performs one or more methods as described in the above embodiments.

[0094] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0095] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0096] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0097] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0098] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0099] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 405 and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory 405 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0100] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will be readily apparent to those skilled in the art upon consideration of the disclosure in this specification. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for dynamic fault detection of elevator traveling cables, characterized in that, Applied to an elevator control platform, the method includes: Acquire multimodal status data of each line of the elevator's traveling cable and the elevator's operating status data; The operating status data is input into a preset electromechanical coupling model to obtain a multidimensional health benchmark signal vector corresponding to the operating status data. The multimodal state data is time-aligned and subtracted from the multidimensional health baseline signal vector to obtain a multidimensional signal residual sequence; The multidimensional signal residual sequence is input into a pre-trained deep learning classification model to obtain preliminary fault judgment results; A causal analysis was performed on the preliminary fault assessment results to confirm the final fault assessment results; When the final fault determination result indicates that a fault exists, the target fault information is acquired and recorded using a pre-deployed time domain reflectometer. Visual information is generated based on the target fault information, and the visual information is sent to a pre-configured human-computer interaction interface.

2. The method according to claim 1, characterized in that, The step of inputting the operating status data into a preset electromechanical coupling model to obtain a multidimensional health benchmark signal vector corresponding to the operating status data specifically includes: Using the finite element analysis method and combining the material properties, geometric structure, and physical parameters of the accompanying cable, a digital twin of the accompanying cable is established, and an electromechanical coupling model is constructed based on the digital twin. The electromechanical coupling model is configured to simulate the mechanical, thermal, and electrical coupling relationships of the accompanying cable when it is running under a preset healthy state. The position and velocity data from the operating status data are input into the mechanical simulation layer of the electromechanical coupling model to obtain the theoretical mechanical parameters of the traveling cable. The theoretical mechanical parameters, the elevator load data, and the ambient temperature are input into the thermal simulation layer of the electromechanical coupling model to obtain the theoretical temperature distribution of the traveling cable. The theoretical mechanical parameters and the theoretical temperature distribution are input into the electrical simulation layer of the electromechanical coupling model to obtain the theoretical electrical data of each circuit. The theoretical mechanical parameters, the theoretical temperature distribution, and the theoretical electrical data are combined to obtain the multidimensional health benchmark signal vector.

3. The method according to claim 1, characterized in that, The step of time-aligning and subtracting the multimodal state data from the multidimensional health baseline signal vector to obtain the multidimensional signal residual sequence specifically includes: The multimodal state data and the multidimensional health baseline signal vector are time-aligned using their respective timestamps; After time alignment, the multimodal state data at each target time point is subtracted from the corresponding multidimensional health baseline signal vector to obtain the original multidimensional residual vector; The original multidimensional residual vector is scaled using a preset normalization algorithm to obtain a normalized residual vector. The multiple normalized residual value vectors are arranged in chronological order to generate the multidimensional signal residual sequence.

4. The method according to claim 3, characterized in that, The step of inputting the multidimensional signal residual sequence into a pre-trained deep learning classification model to obtain preliminary fault judgment results specifically includes: The multidimensional signal residual sequence is sliced ​​according to a first time window to form multiple multidimensional residual segments; Each of the multidimensional residual segments is input into the deep learning classification model to obtain the classification label and confidence level corresponding to the multidimensional residual segment; If the classification labels of multiple consecutive multidimensional residual segments are the same preset fault type, and the confidence scores corresponding to the multiple consecutive multidimensional residual segments are all greater than a preset threshold, a preliminary fault judgment result is generated based on the classification labels and target timestamps of the multiple consecutive multidimensional residual segments. The target timestamp is the start timestamp of the first multidimensional residual segment among the multiple consecutive multidimensional residual segments, and the preliminary judgment result includes the preliminary fault type and the time of occurrence of the preliminary fault.

5. The method according to claim 4, characterized in that, The step of performing causal analysis on the preliminary fault assessment results to confirm the final fault assessment results specifically includes: When the preliminary fault judgment result is received, extract the target multimodal state data and target operation state data within the historical time window corresponding to the time when the preliminary fault occurred; The preset causal rule library is invoked to verify the target multimodal state data and the target operating state data within the historical time window. The causal rule library is used to define the logical association between multimodal state data and operating state data under different fault types. If the target logical association between the target multimodal state data and the target operating state data within the historical time window satisfies the logical association in the causal rule base corresponding to the preliminary fault type, then the preliminary fault judgment result is confirmed as the final fault judgment result. If the target logical association does not satisfy the logical association, the final fault judgment result is determined to be a non-fault interference or transient anomaly.

6. The method according to claim 5, characterized in that, When the final fault determination result indicates the existence of a fault, the target fault information is acquired and recorded using a pre-deployed time-domain reflectometer, specifically including: Based on the final fault determination result, the target line where the fault occurred is identified; The time-domain reflectometer injects an electrical pulse into the target line and collects the reflected signal. The location of the fault is determined based on the propagation time of the reflected signal, and the type of fault is determined based on the waveform characteristics of the reflected signal. The location and type of the fault are integrated with the time of the fault occurrence and the target operating status data in the final fault judgment result to form the target fault information.

7. The method according to claim 6, characterized in that, The step of generating visual information based on the target fault information and sending the visual information to a pre-configured human-computer interaction interface specifically includes: Retrieve a preset three-dimensional digital twin model of the elevator shaft and the accompanying cable; Based on the location of the fault, the position of the elevator and the corresponding section of the accompanying cable at the time of the fault are marked on the accompanying cable in the three-dimensional digital twin model. A heat map is generated on the three-dimensional digital twin model of the accompanying cable based on all historical fault information and operational data. The three-dimensional digital twin model with markings and heatmaps is displayed in association with text information containing fault type, fault occurrence time, and operational status data at the time of fault occurrence on the human-computer interaction interface.

8. A dynamic fault detection system for elevator traveling cables, characterized in that, The system includes a data acquisition module, a data processing module, a fault diagnosis module, a fault recording module, and a fault indication module, among which: The data acquisition module is configured to acquire multimodal status data of each line of the elevator's traveling cable and the elevator's operating status data. The data processing module is configured to input the operating status data into a preset electromechanical coupling model to obtain a multidimensional health benchmark signal vector corresponding to the operating status data. The data processing module is further configured to time-align and subtract the multimodal state data from the multidimensional health baseline signal vector to obtain a multidimensional signal residual sequence. The fault judgment module is configured to input the multidimensional signal residual sequence into a pre-trained deep learning classification model to obtain preliminary fault judgment results. The fault judgment module is also configured to perform causal analysis on the preliminary fault judgment result to confirm the final fault judgment result; The fault recording module is configured to acquire and record target fault information through a pre-deployed time domain reflectometer when the final fault judgment result indicates that a fault exists. The fault indication module is configured to generate visual information based on the target fault information and send the visual information to a pre-configured human-computer interaction interface.

9. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-7.

Images