Electromechanical fusion intelligent operation and maintenance elevator integrated control system and method

Abstract

The invention discloses an electromechanical fusion intelligent operation and maintenance elevator integrated control system which comprises a multi-mode sensing fusion system, an edge computing control center and a cloud intelligent decision platform. The multi-mode sensing system collects elevator physical state data in real time through a high-precision electromagnetic field sensor, a vibration sensing network and an infrared-optical fiber fusion temperature measurement module. The edge computing center adopts an FPGA accelerated signal processor to realize multi-source data feature fusion, and the multi-source data feature fusion is uploaded to the cloud platform through an encrypted data link; and the cloud digital twin engine constructs a multi-body dynamics and electromagnetic field coupling model, generates an optimization control instruction in combination with a deep learning algorithm, and feeds back the optimization control instruction to the edge layer for execution. According to the system, a sensing-edge-cloud three-level collaborative architecture is innovatively provided, millisecond-level sensing, minute-level decision making and microsecond-level control closed loop of the elevator running state are achieved, and the fault early warning accuracy and the operation and maintenance response efficiency are further improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent elevator control technology, and in particular to an integrated control system and method for an electromechanical integrated intelligent operation and maintenance elevator. Background Art

[0002] Elevators are the main means of transportation for people to move vertically in buildings. The widespread use of elevator systems has brought many conveniences and benefits to people. With the development of society, buildings with elevators are becoming more and more popular. Therefore, many related issues such as elevator safety, energy saving, and environmental protection have also received more and more attention. At present, most of the elevators in use in the country do not have many measures in terms of safety management and intelligence, and there is still a gap in the management of personnel access and elevator use rights. Traditional elevator control systems have the following defects:

[0003] Mechanical monitoring and electrical control use independent subsystems with low data integration and delayed response. For example, fault diagnosis relies on threshold alarms and is unable to predict potential risks. In the daily operation and maintenance of the elevator system, there are also inefficiencies, long manual inspection cycles, and the inability to grasp hidden defects such as wire rope wear in real time. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the deficiencies in the prior art, the present invention effectively overcomes the above technical bottlenecks through electromechanical deep coupling and hierarchical collaborative computing.

[0006] (II) Technical solution

[0007] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: an integrated control system for electromechanical fusion intelligent operation and maintenance of elevators, including a multi-modal sensor fusion system, an edge computing control center and a cloud-based intelligent decision-making platform;

[0008] The multimodal sensor fusion system includes an electromagnetic field sensor array, a mechanical vibration sensor network and a temperature field monitoring module deployed in key elevator components. The data of each sensor is transmitted to the edge computing control center in real time via the EtherCAT bus;

[0009] The edge computing control center includes a hardware accelerated signal processor, a real-time control logic unit and a secure communication interface. It parses multimodal sensor data through the hardware accelerated signal processor, generates time-frequency domain fusion feature vectors, and establishes a two-way encrypted data link with the cloud intelligent decision-making platform through the secure communication interface;

[0010] The cloud-based intelligent decision-making platform includes a digital twin simulation engine, a deep learning model library, and a dynamic parameter optimization module, which receives the fused feature vector uploaded by the edge computing control center, performs device state mapping through the digital twin simulation engine, and generates control parameter optimization instructions based on the deep learning model library and transmits them back to the edge computing control center parameters;

[0011] The sensor data of the multi-physical field perception system is transmitted to the edge control center via the EtherCAT bus. A two-way data link is established between the edge control center and the cloud decision-making platform to form a perception decision-making execution closed loop.

[0012] As a preferred solution, the integrated control system composed of the multimodal sensor fusion system, edge computing control center and cloud intelligent decision-making platform includes a closed-loop control architecture of physical layer sensing, edge layer analysis, cloud decision-making and edge execution, wherein:

[0013] The electromagnetic field sensor array is embedded in the bottom of the traction machine stator slot at 120° intervals to monitor the three-phase current harmonic components in real time; the mechanical vibration sensor network includes a six-axis MEMS inertial unit installed on the car roof and an acoustic emission sensor on the side wall of the guide shoe, with a sampling rate of not less than 50kHz; the temperature field monitoring module uses an infrared thermal imager and a fiber grating array to fusion measure temperature, with a spatial resolution of 0.1℃ / pixel; the hardware acceleration signal processor integrates parallel FIR filters and fast Fourier transform kernels implemented by FPGA; the dynamic parameter optimization module uses a reinforcement learning algorithm to adjust the PID control coefficients online.

[0014] As a preferred solution, the control system operation method includes:

[0015] A1. Multimodal sensor data is transmitted to the edge computing control center via the EtherCAT bus, and the hardware accelerated signal processor completes signal denoising, time-frequency conversion and feature fusion;

[0016] A2. The fused feature vector is uploaded to the cloud-based intelligent decision-making platform through the 5G edge gateway, triggering the real-time working condition deduction of the digital twin simulation engine;

[0017] A3. The cloud-based intelligent decision-making platform calls the deep learning model library to generate a control parameter optimization matrix, which is then verified by the dynamic parameter optimization module and sent to the edge computing control center.

[0018] A4. The real-time control logic unit of the edge computing control center adjusts the inverter PWM frequency and brake opening and closing timing according to the optimized parameters to form a closed-loop control.

[0019] As a preferred solution, the electromagnetic field sensor array comprises:

[0020] Three groups of orthogonally arranged TMR magnetoresistive sensors are installed on the stator teeth of the traction machine to detect air gap magnetic field distortion;

[0021] Rogowski coil current sensor, wraps around the motor power cable, with a bandwidth covering 0-10MHz.

[0022] As a preferred solution, the mechanical vibration sensor network comprises:

[0023] The three-axis accelerometer installed on the car roof has X / Y / Z axes corresponding to the vertical lifting direction, front and rear swing direction, and left and right deflection direction of the elevator car, which is used to capture abnormal vibration modes of the car in real time;

[0024] The acoustic emission sensor on the side wall of the guide shoe has a piezoelectric ceramic probe fixed to the bottom of the V-groove on the friction surface of the guide shoe by epoxy resin glue, which is specifically used to detect the stress waves generated by the expansion of micro cracks on the contact surface of the guide rail.

[0025] As a preferred solution, the real-time control logic unit of the edge computing control center includes:

[0026] Multi-objective optimization module, balancing elevator operation efficiency and energy consumption based on NSGA-II algorithm;

[0027] Fault injection simulation module, simulating abnormal conditions such as short circuit and overload in FPGA;

[0028] Dynamically reconfigurable safety module, adopting dual CAN bus redundant architecture, with switching time ≤5ms.

[0029] As a preferred solution, the bidirectional encrypted data link adopts a layered transmission protocol. The bottom layer uses the Modbus TCP protocol encrypted by AES-256 to transmit real-time control instructions; the upper layer transmits device status data through the MQTT protocol, and the data packet is attached with a blockchain hash value check.

[0030] As a preferred solution, the digital twin simulation engine includes:

[0031] Multi-body dynamics model to accurately simulate the nonlinear vibration of wire rope-pulley contact;

[0032] Electromagnetic field finite element model to calculate electromagnetic force fluctuations caused by stator and rotor eccentricity;

[0033] The model and the physical system achieve μs-level time synchronization, and the simulation error is ≤2%.

[0034] As a preferred solution, the deep learning model library includes:

[0035] Convolutional neural network, processing local overheating area detection of infrared thermal images;

[0036] Time series prediction network, based on LSTM to predict the remaining service life of bearings;

[0037] The model uses knowledge distillation technology to compress it to 1 / 5 of its original volume and is deployed in the edge computing control center.

[0038] As a preferred solution, the control system also includes: an emergency fault-tolerant subsystem, which automatically switches to a locally cached control strategy library when a communication interruption is detected, and the strategy library stores the historical mean of the optimization parameters in the last 24 hours and a 3σ safety margin.

[0039] (III) Beneficial effects

[0040] Compared with the prior art, the present invention provides an integrated control system for electromechanical intelligent operation and maintenance elevators, which has the following beneficial effects:

[0041] The system of the present invention includes a multimodal sensor fusion system, an edge computing control center and a cloud-based intelligent decision-making platform. The multimodal sensing system collects elevator physical status data in real time through high-precision electromagnetic field sensors, vibration sensor networks and infrared-fiber fusion temperature measurement modules; the edge computing center uses an FPGA-accelerated signal processor to achieve multi-source data feature fusion, and uploads it to the cloud platform through an encrypted data link; the cloud-based digital twin engine constructs a multi-body dynamics and electromagnetic field coupling model, and combines deep learning algorithms to generate optimized control instructions that are fed back to the edge layer for execution. The system innovatively proposes a three-level collaborative architecture of "sensing-edge-cloud", which realizes millisecond-level perception, minute-level decision-making and microsecond-level control closed loop of the elevator's operating status, further improving the accuracy of fault warnings and the efficiency of operation and maintenance response. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The hierarchical architecture of the control system of the present invention is shown in FIG. Figure 1 ;

[0043] Figure 2 The hierarchical architecture of the control system of the present invention is shown in FIG. Figure 2 . DETAILED DESCRIPTION

[0044] In order to better understand the purpose, structure and function of the present invention, the following will further illustrate an integrated control system for electromechanical intelligent operation and maintenance elevators of the present invention in conjunction with the accompanying drawings and specific embodiments.

[0045] Example 1

[0046] refer to Figure 1-2 The present invention provides an integrated control system for an electromechanical fusion intelligent operation and maintenance elevator, comprising:

[0047] Multimodal sensor fusion system, edge computing control center and cloud-based intelligent decision-making platform;

[0048] The multimodal sensor fusion system includes an electromagnetic field sensor array, a mechanical vibration sensor network and a temperature field monitoring module deployed in key elevator components. The data of each sensor is transmitted to the edge computing control center in real time via the EtherCAT bus;

[0049] The edge computing control center includes a hardware accelerated signal processor, a real-time control logic unit and a secure communication interface. It parses multimodal sensor data through the hardware accelerated signal processor, generates time-frequency domain fusion feature vectors, and establishes a two-way encrypted data link with the cloud intelligent decision-making platform through the secure communication interface;

[0050] The cloud-based intelligent decision-making platform includes a digital twin simulation engine, a deep learning model library, and a dynamic parameter optimization module, which receives the fused feature vector uploaded by the edge computing control center, performs device state mapping through the digital twin simulation engine, and generates control parameter optimization instructions based on the deep learning model library and transmits them back to the edge computing control center parameters;

[0051] The sensor data of the multi-physical field perception system is transmitted to the edge control center via the EtherCAT bus. A two-way data link is established between the edge control center and the cloud decision-making platform to form a perception decision-making execution closed loop.

[0052] Specifically, the system of the present invention is composed of three core modules, among which:

[0053] Multimodal sensor fusion system

[0054] Electromagnetic field sensor array: TMR magnetoresistive sensors are embedded at 120° intervals at the bottom of the traction machine stator slot to detect air gap magnetic field distortion (accuracy ±0.5mT);

[0055] Mechanical vibration sensing network: The six-axis MEMS unit on the car top (range ±50g) and the guide shoe acoustic emission sensor (150kHz center frequency) work together to capture mechanical anomalies;

[0056] Temperature field monitoring module: Infrared thermal imager (640×480 resolution) and distributed fiber Bragg grating (0.1℃ resolution) are integrated to monitor hot spots.

[0057] Edge computing control center

[0058] Hardware accelerated signal processor: Xilinx Zynq UltraScale+FPGA implements 128-channel parallel FIR filtering (256 taps);

[0059] Real-time control logic unit: dynamically adjust PID parameters (update cycle 1ms), support online algorithm reconstruction;

[0060] Secure communication interface: integrated with national secret SM2 / SM3 encryption chip, supporting 5G and industrial Ethernet dual-mode communication.

[0061] Cloud-based intelligent decision-making platform

[0062] Digital twin simulation engine: Ansys Twin Builder builds a multi-body dynamics model of the wire rope (calculation step size 10μs);

[0063] Deep learning model library: TensorFlow framework training bearing life prediction model (test set accuracy 96.7%);

[0064] Dynamic parameter optimization module: Generates PWM frequency optimization matrix based on Q-learning algorithm.

[0065] More specifically, the integrated control system of the present invention, which is composed of a multimodal sensor fusion system, an edge computing control center, and a cloud intelligent decision-making platform, includes a closed-loop control architecture of physical layer sensing, edge layer analysis, cloud decision-making, and edge execution, wherein:

[0066] The electromagnetic field sensor array is embedded in the bottom of the traction machine stator slot at 120° intervals to monitor the three-phase current harmonic components in real time; the mechanical vibration sensor network includes a six-axis MEMS inertial unit installed on the car roof and an acoustic emission sensor on the side wall of the guide shoe, with a sampling rate of not less than 50kHz; the temperature field monitoring module uses an infrared thermal imager and a fiber grating array to fusion measure temperature, with a spatial resolution of 0.1℃ / pixel; the hardware acceleration signal processor integrates parallel FIR filters and fast Fourier transform kernels implemented by FPGA; the dynamic parameter optimization module uses a reinforcement learning algorithm to adjust the PID control coefficients online.

[0067] Specifically, the electromagnetic field sensor array of the present invention comprises:

[0068] Three groups of orthogonally arranged TMR magnetoresistive sensors are installed on the stator teeth of the traction machine to detect air gap magnetic field distortion;

[0069] Rogowski coil current sensor, wrapping around the motor power cable, with a bandwidth covering 0-10MHz;

[0070] The sensor output signal is sampled synchronously by a 24-bit Σ-Δ ADC after passing through a differential amplifier circuit.

[0071] The mechanical vibration sensing network includes:

[0072] The three-axis accelerometer installed on the car roof has X / Y / Z axes corresponding to the vertical lifting direction, front and rear swing direction, and left and right deflection direction of the elevator car, which is used to capture abnormal vibration modes of the car in real time;

[0073] The acoustic emission sensor on the side wall of the guide shoe has a piezoelectric ceramic probe fixed to the bottom of the V-groove on the friction surface of the guide shoe by epoxy resin glue, which is specifically used to detect the stress waves generated by the expansion of micro cracks on the contact surface of the guide rail.

[0074] The real-time control logic unit of the edge computing control center includes:

[0075] Multi-objective optimization module, balancing elevator operation efficiency and energy consumption based on NSGA-II algorithm;

[0076] Fault injection simulation module, simulating abnormal conditions such as short circuit and overload in FPGA;

[0077] Dynamically reconfigurable safety module, adopting dual CAN bus redundant architecture, with switching time ≤5ms.

[0078] This embodiment proposes a three-level collaborative architecture of "sensing-edge-cloud" to achieve millisecond-level perception, minute-level decision-making and microsecond-level control closed loop of the elevator's operating status.

[0079] Example 2

[0080] This embodiment proposes an integrated control system for an elevator with mechatronic intelligent operation and maintenance, and the control system operation method thereof includes:

[0081] A1. Multimodal sensor data is transmitted to the edge computing control center via the EtherCAT bus, and the hardware accelerated signal processor completes signal denoising, time-frequency conversion and feature fusion;

[0082] A2. The fused feature vector is uploaded to the cloud-based intelligent decision-making platform through the 5G edge gateway, triggering the real-time working condition deduction of the digital twin simulation engine;

[0083] A3. The cloud-based intelligent decision-making platform calls the deep learning model library to generate a control parameter optimization matrix, which is then verified by the dynamic parameter optimization module and sent to the edge computing control center.

[0084] A4. The real-time control logic unit of the edge computing control center adjusts the inverter PWM frequency and brake opening and closing timing according to the optimized parameters to form a closed-loop control.

[0085] Furthermore, the bidirectional encrypted data link in the present invention adopts a layered transmission protocol, and the bottom layer uses the Modbus TCP protocol encrypted by AES-256 to transmit real-time control instructions; the upper layer transmits device status data through the MQTT protocol, and the data packet is attached with a blockchain hash value check; the present invention adopts data interaction logic to implement control, and its uplink: sensor data is transmitted to the edge layer via the EtherCAT bus (cycle 1ms) to feature extraction (wavelet packet decomposition + FFT) to encryption and compression and then uploaded to the cloud via 5G; downlink: the cloud generates control instructions (JSON format) to the SM4 algorithm for decryption to the edge layer for parsing and execution (inverter PWM frequency adjustment error ≤ 0.1%).

[0086] In this embodiment, the digital twin simulation engine includes:

[0087] Multi-body dynamics model to accurately simulate the nonlinear vibration of wire rope-pulley contact;

[0088] Electromagnetic field finite element model to calculate electromagnetic force fluctuations caused by stator and rotor eccentricity;

[0089] The model and the physical system achieve μs-level time synchronization, and the simulation error is ≤2%.

[0090] The deep learning model library includes:

[0091] Convolutional neural network, processing local overheating area detection of infrared thermal images;

[0092] Time series prediction network, based on LSTM to predict the remaining service life of bearings;

[0093] The model uses knowledge distillation technology to compress it to 1 / 5 of its original volume and is deployed in the edge computing control center.

[0094] Take the elevator renovation of a 10-story office building as an example:

[0095] Specifically, the hardware deployment is as follows:

[0096] Install three sets of TMR sensors (model TMR2003) at equal distances in the stator slots of the traction machine, with a spacing error of <0.2mm;

[0097] An ADI ADXL356 three-axis accelerometer is placed on the car roof, with the Z axis aligned with the lifting direction;

[0098] The edge control cabinet is equipped with Xilinx ZU19EG FPGA chip, and the heat dissipation design meets the requirements for operating in an environment of 85°C.

[0099] Algorithm implementation:

[0100] Training LSTM life prediction model: input features include vibration RMS value, temperature gradient, current harmonic distortion rate, and output is the remaining life of the bearing (hours);

[0101] Digital twin model calibration: The actual measured wire rope tension data (error ±3%) is injected into the simulation model every 24 hours to update the contact stiffness parameters.

[0102] Communication configuration:

[0103] Real-time control command channel: EtherCAT bus (100Mbps), Modbus TCP protocol frame format;

[0104] Status data channel: 5G CPE equipment (Huawei MH5000 module), uplink bandwidth reserved 20Mbps.

[0105] Security Policy:

[0106] Encrypted communication: Control instructions are encrypted in SM4-CTR mode, and the key is updated every 6 hours;

[0107] Redundant switching: dual CAN bus (ZLG-CANET) sets heartbeat packet detection (cycle 10ms), and triggers switching after 3 timeouts.

[0108] The present invention adopts electromechanical deep coupling perception: through the joint analysis of magnetic field, vibration and temperature multi-physical fields, cross-domain feature correlation detection of early faults is realized, and edge-cloud collaborative decision-making is adopted, wherein the edge layer: completes emergency braking decisions within 50ms (such as detecting free fall acceleration > 0.8g); the cloud layer: executes hourly life prediction and energy efficiency optimization, reducing the edge computing load by 35%.

[0109] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. An integrated control system for electromechanical intelligent operation and maintenance of elevators, characterized in that: Including multi-modal sensor fusion system, edge computing control center and cloud-based intelligent decision-making platform; The multimodal sensor fusion system includes an electromagnetic field sensor array, a mechanical vibration sensor network and a temperature field monitoring module deployed in key elevator components. The data of each sensor is transmitted to the edge computing control center in real time via the EtherCAT bus; The edge computing control center includes a hardware accelerated signal processor, a real-time control logic unit and a secure communication interface. It parses multimodal sensor data through the hardware accelerated signal processor, generates time-frequency domain fusion feature vectors, and establishes a two-way encrypted data link with the cloud intelligent decision-making platform through the secure communication interface; The cloud-based intelligent decision-making platform includes a digital twin simulation engine, a deep learning model library, and a dynamic parameter optimization module, which receives the fused feature vector uploaded by the edge computing control center, performs device state mapping through the digital twin simulation engine, and generates control parameter optimization instructions based on the deep learning model library and transmits them back to the edge computing control center parameters; The sensor data of the multi-physical field perception system is transmitted to the edge control center via the EtherCAT bus. A two-way data link is established between the edge control center and the cloud decision-making platform to form a perception decision-making execution closed loop.

2. The electromechanical fusion intelligent operation and maintenance elevator integrated control system according to claim 1 is characterized in that: The integrated control system composed of the multimodal sensor fusion system, edge computing control center and cloud intelligent decision-making platform includes a closed-loop control architecture of physical layer sensing, edge layer analysis, cloud decision-making and edge execution, wherein: The electromagnetic field sensor array is embedded in the bottom of the traction machine stator slot at 120° intervals to monitor the three-phase current harmonic components in real time; the mechanical vibration sensor network includes a six-axis MEMS inertial unit installed on the car roof and an acoustic emission sensor on the side wall of the guide shoe, with a sampling rate of not less than 50kHz; the temperature field monitoring module uses an infrared thermal imager and a fiber grating array to fusion measure temperature, with a spatial resolution of 0.1℃ / pixel; the hardware acceleration signal processor integrates parallel FIR filters and fast Fourier transform kernels implemented by FPGA; the dynamic parameter optimization module uses a reinforcement learning algorithm to adjust the PID control coefficients online.

3. The electromechanical fusion intelligent operation and maintenance elevator integrated control system according to claim 2 is characterized in that: The control system operation method comprises: A1. Multimodal sensor data is transmitted to the edge computing control center via the EtherCAT bus, and the hardware accelerated signal processor completes signal denoising, time-frequency conversion and feature fusion; A2. The fused feature vector is uploaded to the cloud-based intelligent decision-making platform through the 5G edge gateway, triggering the real-time working condition deduction of the digital twin simulation engine; A3. The cloud-based intelligent decision-making platform calls the deep learning model library to generate a control parameter optimization matrix, which is then verified by the dynamic parameter optimization module and sent to the edge computing control center. A4. The real-time control logic unit of the edge computing control center adjusts the inverter PWM frequency and brake opening and closing timing according to the optimized parameters to form a closed-loop control.

4. The electromechanical integrated intelligent operation and maintenance elevator control system according to claim 2 is characterized in that , the electromagnetic field sensor array comprises: Three groups of orthogonally arranged TMR magnetoresistive sensors are installed on the stator teeth of the traction machine to detect air gap magnetic field distortion; Rogowski coil current sensor, wraps around the motor power cable, with a bandwidth covering 0-10MHz.

5. The electromechanical integrated intelligent operation and maintenance elevator control system according to claim 2 is characterized in that , the mechanical vibration sensing network comprises: The three-axis accelerometer installed on the car roof has X / Y / Z axes corresponding to the vertical lifting direction, front and rear swing direction, and left and right deflection direction of the elevator car, which is used to capture abnormal vibration modes of the car in real time; The acoustic emission sensor on the side wall of the guide shoe has a piezoelectric ceramic probe fixed to the bottom of the V-groove on the friction surface of the guide shoe by epoxy resin glue, which is specifically used to detect the stress waves generated by the expansion of micro cracks on the contact surface of the guide rail.

6. The electromechanical integrated intelligent operation and maintenance elevator control system according to claim 2 is characterized in that ,The real-time control logic unit of the edge computing control center includes: Multi-objective optimization module, balancing elevator operation efficiency and energy consumption based on NSGA-II algorithm; Fault injection simulation module, simulating abnormal conditions such as short circuit and overload in FPGA; Dynamically reconfigurable safety module, adopting dual CAN bus redundant architecture, with switching time ≤5ms.

7. The electromechanical integrated intelligent operation and maintenance elevator control system according to claim 1 is characterized in that ,The bidirectional encrypted data link adopts a layered transmission protocol. The bottom layer uses the Modbus TCP protocol encrypted by AES-256 to transmit real-time control instructions; the upper layer transmits device status data through the MQTT protocol, and the data packet is attached with a blockchain hash value check.

8. The electromechanical integrated intelligent operation and maintenance elevator control system according to claim 1 is characterized in that ,The digital twin simulation engine includes: Multi-body dynamics model to accurately simulate the nonlinear vibration of wire rope-pulley contact; Electromagnetic field finite element model to calculate electromagnetic force fluctuations caused by stator and rotor eccentricity; The model and the physical system achieve μs-level time synchronization, and the simulation error is ≤2%.

9. The electromechanical integrated intelligent operation and maintenance elevator control system according to claim 1 is characterized in that , the deep learning model library includes: Convolutional neural network, processing local overheating area detection of infrared thermal images; Time series prediction network, based on LSTM to predict the remaining service life of bearings; The model uses knowledge distillation technology to compress it to 1 / 5 of its original volume and is deployed in the edge computing control center.

10. The electromechanical integrated intelligent operation and maintenance elevator control system according to claim 1 is characterized in that ,The control system also includes an emergency fault tolerant subsystem, which automatically switches to the ,locally cached control strategy library when a communication interruption is detected, and its ,strategy library stores the historical mean of the optimization parameters and ,3σ safety margin in the last 24 hours.

Images