Intelligent elevator rescue and release system and rescue method

Abstract

This invention discloses an intelligent elevator rescue system and method. The system includes a fault judgment module, a safe movement judgment and power control module, a car movement control module, and a leveling detection and door opening module, all connected in sequence. The method includes a high-definition camera capturing panoramic images of the elevator car; an intelligent analysis system using AI to identify and determine if someone is trapped and outputs a signal; the safe movement judgment module collecting signals of wire rope tension and guide rail status; the car movement control module using PWM pulse signals to release the brake, driving the car to move based on the weight difference between the car and the counterweight, with a dual-redundant speed measurement system monitoring for overspeed prevention in real time; and the leveling detection module, after confirming the car has reached the leveling point, controlling the door operator to open the door and guiding the person to leave via voice prompts. This invention achieves fully intelligent rescue without human intervention, significantly shortening rescue time, with a low misjudgment rate in entrapment identification, adaptability to multiple scenarios, and significantly improving rescue efficiency and safety.

Description

Technical Field

[0001] This invention belongs to the field of elevator safety technology, specifically relating to an intelligent elevator rescue and release system and rescue method. Background Technology

[0002] Elevators, as a core vertical transportation tool in modern buildings, are widely used in residential buildings, office buildings, shopping malls, and other settings. However, due to mechanical wear, electrical faults, and environmental factors, elevator malfunctions and entrapment incidents are frequent. Traditional elevator rescue methods rely on manual fault detection and alarm activation, resulting in slow response times. Furthermore, professional personnel are required to operate the equipment on-site to move the car and open the doors to rescue people, making the rescue process complex and time-consuming. This not only may delay rescue opportunities but also increases the risk of secondary accidents due to human error, causing significant psychological stress to trapped individuals.

[0003] Existing rescue technologies, while some solutions attempt to incorporate sensors to monitor malfunctions, suffer from significant drawbacks: reliance on a single sensor (such as infrared sensing) for entrapment detection makes them susceptible to environmental interference and misjudgments; the lack of multi-dimensional speed monitoring for car movement poses a risk of overspeeding; safety movement assessments are limited to a single dimension, failing to adequately consider critical safety hazards such as uneven wire rope tension and guide rail deformation; and the brake release control method is crude, easily leading to car impact. Therefore, developing an intelligent elevator rescue system and method integrating AI-powered intelligent recognition, multi-dimensional safety assessment, precise speed control, and smooth brake release is crucial to addressing the pain points of existing technologies. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent elevator rescue system and method that enables intelligent rescue throughout the entire process, including automatic identification of trapped persons, accurate determination of safety status, smooth movement of the elevator car, and automatic door opening at the landing. This shortens rescue time, reduces rescue risks, and ensures the safety of trapped persons.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] An intelligent elevator rescue and release system includes a fault judgment module, a safe movement judgment and power control module, a car movement control module, and a leveling detection and door opening module, which are connected in sequence by signals.

[0007] The fault diagnosis module includes a high-definition camera installed inside the elevator car and an intelligent analysis system. The high-definition camera is used to collect panoramic images inside the car when the elevator stops abnormally and transmit them to the intelligent analysis system in real time. The intelligent analysis system extracts static and dynamic features of the human body, and outputs a entrapment signal after dual-dimensional recognition and result verification.

[0008] The safe movement judgment and power control module includes a safety parameter acquisition unit, a signal processing unit, a safety judgment unit, and a control output unit. The safety parameter acquisition unit acquires the wire rope tension signal and the guide rail status signal. After signal processing, the safety judgment unit performs multi-dimensional threshold judgment. After confirming that it is safe to move, it disconnects the main power supply, starts the emergency power supply, and sends an alarm message.

[0009] The car movement control module includes an emergency power control unit, a movement drive unit, a dual-redundant speed measurement unit, an overspeed protection unit, and a status feedback unit. The emergency power control unit outputs a PWM pulse signal to jog and release the elevator brake. The movement drive unit drives the car to move by relying on the weight difference between the car and the counterweight. The dual-redundant speed measurement unit monitors the elevator car's movement speed in real time. The overspeed protection unit cuts off the emergency power supply and causes the brake to engage when the speed exceeds a preset threshold.

[0010] The leveling detection and door opening module is used to continuously detect the leveling position of the car. After reaching the level, the car stops moving and the emergency power supply is used to power the door operator to open the car door and the landing door. During the door opening process, a safety prompt is sent through the voice prompt unit.

[0011] Further optimizations include a preprocessing unit, a feature extraction unit, an AI dual-dimensional recognition unit, and a result verification unit.

[0012] The preprocessing unit performs grayscale correction, Gaussian filtering, and low-light enhancement on the image to output a valid frame image; the feature extraction unit is used to extract static human features (contour, skeletal key points, human proportions) and dynamic features (movement trajectory, movement amplitude, frequency) from the valid frame image.

[0013] The AI ​​dual-dimensional recognition unit includes:

[0014] The human target detection subunit uses a lightweight target detection algorithm to analyze static features and outputs the bounding box and confidence score of the human target.

[0015] The abnormal action analysis subunit uses a temporal neural network to analyze dynamic features and identify preset abnormal actions such as patting the car wall, shouting, and repeatedly pressing buttons.

[0016] The result verification unit is connected to the AI ​​dual-dimensional recognition unit. When the confidence level of human target detection is greater than or equal to the preset threshold and the abnormal action analysis meets the characteristics of entrapment, the entrapment signal is output to the safe movement judgment module.

[0017] The AI ​​dual-dimensional recognition unit also includes a non-human target filtering subunit. If the intelligent analysis system misjudges (such as identifying luggage as a human body), the non-human target filtering subunit compares the luggage feature database (preset outlines of suitcases, strollers, etc.) to automatically correct the judgment result and avoid triggering rescue in case of accident.

[0018] If recognition fails due to insufficient light, the high-definition camera automatically switches to infrared mode, the preprocessing unit enhances image contrast, and the AI ​​model enables low-light scene weight parameters to improve recognition robustness.

[0019] Further optimization involves using the lightweight target detection algorithm, either YOLOv5 or EfficientDet, with a preset confidence threshold of ≥0.7 for the human target detection subunit, and filtering out duplicate detection boxes using the IOU threshold.

[0020] The temporal neural network is an LSTM network. The abnormal action analysis subunit determines the following for patting the car wall: the hand is in contact with the car wall for more than 3 consecutive frames and the movement amplitude is ≥5cm. It determines the following for shouting: the head swing frequency is ≥2 times / second and the duration is ≥3 seconds.

[0021] Further optimization includes a safety parameter acquisition unit comprising a wire rope tension sensor group installed in the elevator traction system and a guide rail status sensor group installed on the car. The wire rope tension sensor group is used to acquire the real-time tension signal of a single wire rope, and the guide rail status sensor group is used to acquire the guide rail flatness, verticality, and lateral clearance signals.

[0022] The signal processing unit is used to filter, temperature compensate, and perform analog-to-digital conversion on the acquired tension signal and guide rail status signal, and output a standardized digital signal.

[0023] The safety judgment unit is connected to the signal processing unit and is used to: perform single-strand threshold judgment, tension uniformity judgment, and dynamic fluctuation judgment on the wire rope tension signal; and perform flatness threshold judgment, verticality deviation judgment, and gap range judgment on the guide rail status signal. When all judgments meet the preset safety conditions, a "safe movement permitted" signal is output.

[0024] The control output unit is signal-connected to the safety judgment unit and is used to disconnect the elevator main power switch, start the emergency power supply, and send alarm information containing the elevator location to the preset department after receiving the "safe movement allowed" signal.

[0025] Further optimization includes a signal cross-verification unit in the safe movement judgment and power control module, which works in conjunction with the original elevator control system to obtain load and brake status information. When the deviation between the sensor-collected data and the theoretical value exceeds 10%, a secondary calibration is initiated.

[0026] Further optimization involves using a pin-type tension sensor group for the wire rope, with one sensor configured for each wire rope and an "N+1" redundant configuration. The sensor is installed on the wire rope idler roller shafts on both sides of the traction sheave.

[0027] The guide rail status sensor group includes a laser displacement sensor and a tilt sensor. The laser displacement sensor is installed at the four corners of the top of the car and faces the working surface of the guide rail. The tilt sensor is installed on the upper and lower beams of the car and arranged along the extension direction of the guide rail.

[0028] The wire rope tension judgment standard of the safety judgment unit is as follows: the single wire rope tension is 15-30kN for elevators with a rated load of 1000kg, the standard deviation of tension is ≤10% of the average tension, and the instantaneous change rate is ≤5kN / s. The guide rail condition judgment standard is as follows: flatness fluctuation ≤±0.5mm, verticality deviation ≤0.5°, and lateral clearance 2-10mm.

[0029] It also includes a signal cross-verification unit, which works in conjunction with the existing elevator control system to obtain information on car load and brake status. When the data collected by the sensors deviates from the theoretical calculation value by more than 10%, a secondary calibration is initiated.

[0030] The safety judgment unit can adaptively adjust the judgment threshold according to the elevator type (residential / shopping mall / hospital). For example, the tension fluctuation threshold of shopping mall elevators can be relaxed to 8kN / s, while the verticality threshold of hospital elevator guide rails can be tightened to 0.3°.

[0031] When the control output unit determines that the elevator is "unsafe to move," it sends the specific fault type (uneven wire rope tension / guide rail verticality deviation, etc.) to a preset department (such as the elevator maintenance unit or the property gatehouse) and keeps the main power supply on. For example, if the tension of a wire rope suddenly rises above a threshold, the safety judgment unit immediately determines that the elevator is "unsafe to move," the control output unit sends an "abnormal wire rope tension" alarm, maintains the main power supply on, and waits for manual inspection. Alternatively, if the system detects that the guide rail verticality deviation exceeds the threshold, it simultaneously sends an image of the guide rail position (acquired with the assistance of a shaft camera) to facilitate maintenance personnel in locating the fault point.

[0032] In a further optimization, the emergency power control unit is used to receive the "safe movement allowed" signal from the safe movement judgment module and output a pulse width modulation (PWM) signal to the elevator brake to release the brake in a jogging motion.

[0033] The dual-redundant speed measurement unit includes at least two speed measurement systems: the first is an incremental encoder installed on the end of the traction sheave shaft, and the second is a laser Doppler speedometer installed in the car, which is used to collect the car's moving speed in real time and output the speed signal.

[0034] The overspeed protection unit is signal-connected to the dual redundant speed measuring unit and is used to compare the collected speed signal with a preset safe speed threshold. When the speed exceeds the threshold and the duration is ≥100ms, the emergency power supply is cut off to cause the brake to engage.

[0035] The status feedback unit is connected to the brake and the dual redundant speed measuring unit to monitor the brake's opening and closing status and the consistency of the speed measuring signal. In case of abnormality, the movement control function is locked. The status feedback unit monitors the opening and closing angle of the brake arm through the brake limit switch (the feedback signal is a 0-5V analog quantity). When it detects that the brake has not been released as instructed, it immediately stops the pulse signal output and triggers an alarm.

[0036] In a further optimization, the dual-redundant speed measurement unit also includes a signal verification subunit, which is used to calculate the speed difference between the two speed measurement systems. If the difference exceeds 0.1 m / s, a backup judgment logic based on a preset acceleration threshold is activated.

[0037] The emergency response time of the overspeed protection unit is ≤50ms, and the time from power cut-off to full braking of the brake is ≤300ms.

[0038] The moving drive unit determines the direction of car movement by using the phase difference signal from the dual redundant speed measuring unit. If the direction does not match the expected direction of the weight difference drive, the overspeed protection unit is immediately triggered.

[0039] After the brake engages, the status feedback unit outputs a "lock signal," preventing the emergency power control unit from outputting a pulse signal again. The lock needs to be released via a manual reset command.

[0040] The preset safe speed threshold of the overspeed protection unit is 110% of the elevator's rated speed, which can be dynamically adjusted according to the car's load. The overspeed response time is ≤50ms, and the time from power cut-off to full braking is ≤300ms.

[0041] If the speed difference measured by the dual redundant speed measuring unit exceeds the preset value, the signal verification subunit starts the backup logic. If it is still determined to be overspeeding, the brake is immediately triggered to hold the brake.

[0042] If the brake is stuck, the system will attempt to retry 3 times (pulse duty cycle 50%, 60%, 70%). If it fails, it will send a "brake stuck" alarm and lock the movement function.

[0043] Further optimizations include improving the leveling position detection accuracy of the leveling detection and door opening module to ≤ ±5mm. After the door operator opens, it continuously monitors the door status and extends the power supply time and repeats safety prompts when it detects that the door is not fully open or is obstructed by personnel.

[0044] Further optimizations include the voice prompt unit supporting multilingual prompts in Chinese and English, with prompt content customizable via the property management office control terminal; and the safety movement judgment and power control module storing fault records and rescue data from the past year, supporting wired or wireless network export.

[0045] Regularly clean the high-definition camera lens and laser sensor probe, and calibrate the zero point of the tension sensor. Regularly check the emergency power supply capacity. Verify safety threshold parameters annually and adjust the tension threshold through elevator rated load testing.

[0046] The intelligent analysis system supports OTA upgrades, pushing newly trained AI models, including new scenario samples, through the elevator IoT platform. The threshold parameter table of the safety judgment unit can be updated via USB flash drive or network to adapt to the mechanical characteristics of different brands of elevators.

[0047] A method for intelligent elevator rescue and passenger release, applied to the aforementioned intelligent elevator rescue and passenger release system, includes the following steps:

[0048] Step S1: Trapped Person Fault Recognition Trigger: When the elevator stops abnormally, the high-definition camera inside the car is activated to collect a panoramic image of the car. The intelligent analysis system preprocesses the image, extracts features, performs AI dual-dimensional recognition and verifies the results. After determining that there are people trapped inside the car, a trapped person signal is output.

[0049] Step S2: Safe Movement Condition Determination and Power Control: After receiving the trapped person signal, the safety parameter acquisition unit collects the elevator wire rope tension signal and guide rail status signal. The signal processing unit performs filtering, temperature compensation, and analog-to-digital conversion on the signal. The safety judgment unit completes the safe movement condition determination based on multi-dimensional safety thresholds. If it is determined that the elevator can be moved safely, an alarm message containing the elevator's specific location and the time of the fault is sent to the preset rescue-related departments. At the same time, the elevator's main power switch is disconnected and the emergency power supply is activated. If it is determined that the elevator cannot be moved safely, specific fault type information is sent and the main power supply is kept on.

[0050] Step S3: Car safety movement control: The elevator brake is released by jogging through the output of PWM pulse signal from the emergency power supply, and the car is driven to move by the weight difference between the car and the counterweight; a dual-redundant speed measurement system is used to collect the car's moving speed in real time, and the speed signal is compared with the preset safe speed threshold. If the speed exceeds the threshold and continues for a preset time, the emergency power supply is immediately cut off to make the brake engage.

[0051] Step S4: Leveling and Door Opening Rescue: Continuously monitor the car's leveling position. When the car reaches the leveling position, stop moving the car and power the door operator via emergency power to open the car door and landing door. Simultaneously, send safety prompts to trapped personnel through the voice prompt unit to guide them to evacuate in an orderly manner, completing the rescue. For international scenarios, the voice prompt unit defaults to playing bilingual (Chinese and English) prompts and supports remote updates of multilingual audio through the hotel management system.

[0052] Further optimization involves step S2, in which if the safety judgment unit determines that the vehicle is "not safe to move", it immediately sends the specific fault type to the preset department, keeps the main power supply on, and waits for manual rescue.

[0053] Further optimization involves step S3, where the brake arm opening angle is monitored via the brake travel switch. If the brake arm is not released as instructed, the pulse output is stopped and three retries are triggered. If the retries fail, a "brake jamming" alarm message is sent.

[0054] Further optimization involves steps S2 and S3, where the safety movement judgment and power control module and the car movement control module can adaptively adjust the judgment threshold and PWM parameters according to the elevator type (residential / shopping mall / hospital).

[0055] Compared with the prior art, the beneficial effects of the present invention are:

[0056] 1) Precise and efficient entrapment detection: Through in-car cameras and intelligent analysis systems, the system can automatically detect entrapment faults and issue alarms immediately, significantly shortening rescue response time. Compared to traditional rescue methods, it can reduce rescue time by several times, giving trapped individuals valuable rescue opportunities. AI dual-dimensional recognition combined with multi-scenario adaptation technology effectively avoids misjudgments and omissions, ensuring a fast response and triggering rescue immediately.

[0057] 2) Comprehensive Safety Assurance: Multi-dimensional safety parameter monitoring, redundant design, and cross-validation mechanisms form a robust safety defense line throughout the entire process from fault identification to car movement, preventing secondary accidents. Multiple speed measurement systems are connected in series to strictly control the car's movement speed, ensuring that excessive speed does not cause danger during rescue operations. Simultaneously, multiple safety measures are implemented in areas such as safe movement judgment and power control to guarantee the safety of trapped personnel and the rescue process.

[0058] 3) Smooth and intelligent rescue process: The brake release control avoids car impact, dual redundant speed measurement and precise overspeed protection ensure safe movement, and the whole process is automated without human intervention, reducing the rescue risk caused by human operation error and improving the accuracy and reliability of rescue.

[0059] 4) Strong scenario adaptability: Parameters can be adaptively adjusted according to elevator type, load capacity, and environmental conditions; multilingual prompts and data storage export are supported to meet the needs of different application scenarios. Attached Figure Description

[0060] Figure 1 This is a block diagram of the intelligent elevator rescue and release system described in this invention;

[0061] Figure 2 This is a block diagram of the fault diagnosis module described in this invention;

[0062] Figure 3 This is a block diagram of the safe movement judgment and power control module described in this invention;

[0063] Figure 4 This is a block diagram of the car movement control module described in this invention;

[0064] Figure 5 This is a block diagram of the leveling detection and door opening module described in this invention;

[0065] In the diagram: 1-Fault diagnosis module, 11-High-definition camera, 12-Intelligent analysis system, 121-Preprocessing unit, 122-Feature extraction unit, 123-AI dual-dimensional recognition unit, 124-Result verification unit, 2-Safe movement judgment and power control module, 21-Safety parameter acquisition unit, 22-Signal processing unit, 23-Safety judgment unit, 24-Control output unit, 25-Signal cross-verification unit, 3-Car movement control module, 31-Emergency power control unit, 32-Movement drive unit, 33-Dual redundant speed measurement unit, 34-Overspeed protection unit, 35-Status feedback unit, 4-Leveling detection and door opening module, 41-Leveling position detection unit, 42-Door operator, 43-Voice prompt unit, 5-Elevator car, 6-Brake, 7-Traction sheave, 8-Guide rail. Detailed Implementation

[0066] The technical solution of the present invention will be further described in detail below through embodiments and with reference to the accompanying drawings.

[0067] Example 1: As Figure 1 As shown, an intelligent elevator rescue and release system includes a fault judgment module 1, a safe movement judgment and power control module 2, a car movement control module 3, and a leveling detection and door opening module 4, which are connected in sequence by signals. The modules work together to achieve fully automated control of the entire process from trapped person identification to door opening and rescue.

[0068] 1) such as Figure 2 As shown, the fault diagnosis module 1 includes a high-definition camera 11 and an intelligent analysis system 12.

[0069] In this embodiment, the high-definition camera 11 is a 2-megapixel infrared camera, model Hikvision DS-2CD3T25-I3, with a frame rate of 25fps, a resolution of 1920×1080P, and a minimum illumination of 0.01Lux (0Lux when infrared is on). It is installed in the center of the car's top, with the lens tilted downwards at 15° to ensure coverage of the car's interior from 0.5 to 2.5 meters in height and all corners. The camera is connected to the intelligent analysis system via a network cable, supports PoE power supply, and is suitable for operating environments from -30℃ to 60℃.

[0070] The intelligent analysis system 12 employs an edge computing terminal, specifically an NVIDIA Jetson Nano in this embodiment. It includes an integrated preprocessing unit 121, a feature extraction unit 122, an AI dual-dimensional recognition unit 123, and a result verification unit 124. The terminal has a built-in human feature library, a pre-trained YOLOv5s model, and an LSTM motion analysis model, and communicates with the safe movement judgment module via an RS485 interface. The human feature library contains over 100,000 human samples with different postures, ages, and occlusion levels.

[0071] For the fault diagnosis module, the default confidence threshold for human target detection is set to 0.7. For special scenarios, such as hospital scenarios, it can be reduced to 0.6 to adapt to people lying down.

[0072] The thresholds for abnormal actions are set as follows: slapping motions with an amplitude ≥ 5cm and a duration ≥ 3 frames; shouting motions with a head-shaking frequency ≥ 2 times / second and a duration ≥ 3 seconds. The default time for determining the presence of a human body is 30 seconds.

[0073] 2) For example Figure 3 As shown, the safe movement judgment and power control module 2 includes a safety parameter acquisition unit 21, a signal processing unit 22, a safety judgment unit 23, and a control output unit 24.

[0074] The safety parameter acquisition unit 21 includes a wire rope tension sensor guide rail status sensor. The wire rope tension sensor is a pin-type tension sensor, model YZC-528, with a range of 0-50kN and an accuracy of ±0.2%FS. Each wire rope corresponds to one sensor, and an additional redundant sensor is configured, installed on the idler roller shafts on both sides of the traction sheave, and connected to the signal processing unit via a 4-20mA analog line.

[0075] The guide rail status sensor uses a laser displacement sensor, model KEYENCE, with a resolution of 0.001mm. It is installed at the four corners of the car top, with the probe vertically aligned with the guide rail working surface (15mm away). The tilt sensor, model BNO055, has a measurement range of ±180° and an accuracy of ±0.5°. It is installed at the center of the car upper beam and arranged along the guide rail direction. Both are connected to the signal processing unit via a USB interface.

[0076] The signal processing unit 22 uses an STM32H743 microcontroller, which integrates 16 channels of 16-bit ADC, a sampling frequency of 100Hz, a Kalman filter algorithm, and a temperature compensation module. It receives sensor signals through an optocoupler isolation circuit, processes them, and outputs standardized digital signals (0-3.3V).

[0077] The safety judgment unit 23 is based on an FPGA chip for parallel computing, model XilinxArtix-7, and has a pre-stored table of safety threshold parameters for multiple scenarios, supporting threshold updates via Ethernet.

[0078] The control output unit 24 includes two relays, model OmronG2R-1, which control the main power switch and emergency power start-up respectively, and send alarm information to the rescue center via GPRS module, including latitude and longitude, fault code and timestamp.

[0079] The emergency power supply uses a 24V / 100Ah lithium battery pack with a battery life of ≥4 hours. It is equipped with a power monitoring module that triggers a low power alarm when the remaining power is <20%.

[0080] In this example, for an elevator with a rated load of 1000kg, the wire rope tension threshold is set to 15-30kN, which can be relaxed to 12-35kN for shopping mall elevators. The guide rail verticality deviation threshold is set to 0.5°, which is tightened to 0.3° for hospital elevators. The signal cross-validation deviation threshold is set to 10%, which is the maximum allowable deviation between sensor data and theoretical values.

[0081] 3) For example Figure 4 As shown, the car movement control module 3 includes an emergency power control unit 31, a movement drive unit 32, a dual-redundant speed measurement unit 33, an overspeed protection unit 34, and a status feedback unit 35.

[0082] Among them, the emergency power control unit 31 adopts an STM32F407 microcontroller, which integrates a PWM signal generator with an adjustable frequency range of 5-10Hz and a duty cycle of 30%-50%. It controls the on / off state of the brake coil through a MOS transistor drive circuit.

[0083] The dual-redundant speed measurement unit 33 includes at least two speed measurement systems: the first is an incremental encoder installed on the traction sheave shaft end, and the second is a laser Doppler tachometer installed in the car, used to acquire the car's moving speed in real time and output speed signals. The incremental encoder is model E6B2-CWZ6C, with a resolution of 1024 lines / revolution, installed on the traction sheave shaft end, and outputs speed pulses through differential signals.

[0084] The laser Doppler velocimeter, model Polytec OFV-5000, has a measurement range of 0-2 m / s and an accuracy of ±0.001 m / s. It is installed on the top of the car, and the laser beam illuminates the surface of the guide rail, outputting a 4-20 mA speed signal.

[0085] The overspeed protection unit 34 uses a dedicated comparator chip and a preset speed threshold adjustable resistor, which corresponds to 110% of the rated speed. When the speed exceeds the threshold, it outputs a high level to cut off the emergency power supply.

[0086] The status feedback unit 35 monitors the opening and closing angle of the brake arm through the brake travel switch. When it detects that the brake has not been released as instructed, it immediately stops the pulse signal output and triggers an alarm. In this embodiment, the brake travel switch is model AZ-7141, installed at the brake arm, and outputs a 0-5V analog signal, where 0V corresponds to full brake engagement and 5V corresponds to full release.

[0087] 4) For example Figure 5 As shown, the leveling detection and door opening module 4 includes a leveling position detection unit 41, a door operator 42, and a voice prompt unit 43. The leveling position detection unit 41 uses a permanent magnet sensor, installed in the leveling area at the top of the car. Magnetic shielding plates are installed at corresponding positions on each floor of the hoistway. The detection accuracy is ±3mm, and it outputs a switching signal. The door operator 42 is a variable frequency door operator, model Mitsubishi DL2-E, supporting 24V emergency power supply. The door opening time is adjustable from 0.8 to 1.5 seconds, and it receives door opening commands via a CAN bus. The voice prompt unit 43 is a 3W speaker installed inside the car, connected to an MP3 voice module. It supports SD card storage of multi-language audio, with Chinese / English as the default prompt tone, which can be customized via an RS232 interface.

[0088] Example 2: In this example, a residential elevator with a rated load of 1000kg and a speed of 1.0m / s is used as an example to illustrate the execution steps of the intelligent elevator rescue method for releasing people. The specific steps include the following:

[0089] Step S1: Trapped person fault detection trigger, specifically including:

[0090] Step S11: After the elevator stops abnormally due to a traction machine malfunction, the control cabinet outputs an "abnormal stop signal", triggering the fault judgment module to start. The response time is <100ms.

[0091] Step S12: The high-definition camera 11 is activated. If there is no light inside the car, the infrared mode is turned on to collect real-time images and transmit them to the intelligent analysis system 12.

[0092] Step S13: The preprocessing unit 121 of the intelligent analysis system 12 performs grayscale correction, Gaussian filtering, and low-light enhancement on the image to eliminate emergency light reflections, improve the contrast of dark areas, and output an effective frame image.

[0093] Step S14: Feature extraction unit 122 extracts static human features (such as head contour and shoulder key points) and dynamic features (such as hand movement trajectory) through ResNet50 network.

[0094] Step S15: AI Two-Dimensional Recognition Unit 123:

[0095] Human Target Detection Subunit: In this embodiment, the YOLOv5s model detected two human targets inside the car with confidence levels of 0.85 and 0.82.

[0096] Abnormal Action Analysis Subunit: The LSTM network identified one person who was "tapping the car wall" for 5 frames with an amplitude of 7cm.

[0097] Step S16: The result verification unit 124 determines that the "valid human body + abnormal action" condition is met, and outputs a high-level trapped person signal (lasting 100ms) to the safe movement judgment module.

[0098] Step S2: Determining safe movement conditions and controlling power supply, specifically including:

[0099] Step S21: After receiving the trapped person signal, the safety movement judgment and power control module 2 starts the safety parameter acquisition unit 21 and performs a 50ms sliding window mean filtering on the collected tension signals of the 6 steel wire ropes. The tension values ​​of the 6 steel wire ropes measured by the 6 tension sensors are 22kN, 23kN, 21kN, 24kN, 22kN and 23kN respectively, with an average tension of 22.5kN and a standard deviation of 1.02kN (≤22.5×10%=2.25kN), which meets the uniformity requirement; the instantaneous change rate of tension is ≤3kN / s, which meets the dynamic threshold.

[0100] The laser displacement sensor detected the condition of the guide rail, and the results showed that the flatness fluctuation was ±0.3mm, the verticality deviation was 0.3°, and the lateral clearance was 5mm. All indicators met the safety threshold.

[0101] Step S22: Safety judgment unit 23 outputs a "safe movement permitted" signal, controlling output unit 24 to execute:

[0102] The alarm message was sent to the property management office and the fire and rescue center via GPRS module. The message read: "People are trapped in the elevator of Unit 2, Building 3, XX Community. The location is between the 12th and 13th floors. The time is 202X-10-02, 14:30:22."

[0103] Step S23: Disconnect the main power switch and start the emergency power supply.

[0104] Step S3: Car safety movement control, specifically including:

[0105] Step S31: After receiving the "Safe movement permitted" signal, the car movement control module 3 outputs a PWM pulse signal with a frequency of 8Hz, a duty cycle of 40%, and a duration of 30ms. This signal is then used by the drive circuit to energize the coil of the brake 6, causing the brake arm to slowly open. The limit switch feedback angle is 8°, which is within the safe range of 5-10°. Because the weight of the car 5 (including two people, approximately 150kg) is greater than the weight of the counterweight, it begins to descend under the weight difference drive, with an initial speed of 0.1m / s.

[0106] Step S32: The dual redundant speed measuring unit 33 monitors in real time. The encoder measures the speed corresponding to the traction sheave rotation speed as 0.12m / s, and the laser speed meter measures the actual speed as 0.11m / s. The difference is 0.01m / s (≤0.1m / s), and the signal is valid.

[0107] As the car accelerates to 0.5 m / s, the emergency power control unit dynamically adjusts the PWM duty cycle to 35% to maintain smooth movement.

[0108] During operation, the encoder phase difference signal determines that the direction of movement is downward (consistent with expectations), and the status feedback unit 35 continuously monitors the brake status and confirms that it is normal.

[0109] Step S4: Level-floor door opening rescue, specifically including:

[0110] Step S41: When the car approaches the 12th floor leveling area, the permanent magnet sensor of the leveling position detection unit 41 triggers a "leveling signal". The leveling detection and door opening module 4 sends a "stop signal" to the car movement control module 3. The emergency power control unit 31 immediately stops the PWM output, and the brake 6 fully engages within 250ms under the action of the spring force. The limit switch feedback is 0V.

[0111] Step S43: The leveling detection and door opening module 4 sends an opening command to the door operator 42. The emergency power supply powers the door operator, and the car door and the landing door open synchronously, taking 1.2 seconds.

[0112] Voice prompt unit 43 plays a safety prompt: "The elevator has reached the floor level. Please leave in an orderly manner and watch your step." The Chinese prompt is repeated 3 times.

[0113] Step S44: After the infrared sensor of the door operator 42 detects that the person has left, it closes the door after a 5-second delay. The leveling detection and door opening module 4 outputs a "rescue completed signal". The system records the rescue data and stands by. During this rescue, the car moved for 28 seconds and the door opened for 15 seconds.

[0114] Example 3: An electronic device, comprising: one or more processors;

[0115] Memory, used to store one or more programs;

[0116] When one or more programs are executed by the processor, the processor performs the above-described method.

[0117] Example 4: A computer-readable storage medium having stored computer instructions thereon, which, when executed by a processor, implement the steps of the above method.

[0118] The specific embodiments described above are only used to illustrate the technical solutions and implementation details of the present invention in detail, and should not be construed as limiting the scope of the present invention. Without departing from the basic principles and core ideas of the present invention, those skilled in the art can make appropriate adjustments, modifications, or substitutions to the technical features in the above embodiments according to actual needs. All such changes or improvements based on the spirit of the present invention should be considered to fall within the protection scope of the present invention.

Claims

1. An intelligent elevator rescue and passenger release system, characterized in that, This includes a fault diagnosis module, a safe movement judgment and power control module, a car movement control module, and a leveling detection and door opening module, all connected in sequence by signals. The fault judgment module includes a high-definition camera and an intelligent analysis system installed inside the elevator car. The high-definition camera is used to collect panoramic images inside the car when the elevator stops abnormally and transmit them to the intelligent analysis system in real time. The intelligent analysis system extracts static and dynamic features of the human body, and outputs a trapped signal after dual-dimensional recognition and result verification. The safe movement judgment and power control module includes a safety parameter acquisition unit, a signal processing unit, a safety judgment unit, and a control output unit. The safety parameter acquisition unit acquires the wire rope tension signal and the guide rail status signal. After signal processing, the safety judgment unit completes multi-dimensional threshold judgment. After confirming that the elevator car can move safely, it disconnects the main power supply, starts the emergency power supply, and sends an alarm message. The car movement control module includes an emergency power control unit, a movement drive unit, a dual-redundant speed measurement unit, an overspeed protection unit, and a status feedback unit. The emergency power control unit outputs a PWM pulse signal to jog and release the elevator brake. The movement drive unit drives the car to move by relying on the weight difference between the car and the counterweight. The dual-redundant speed measurement unit monitors the elevator car's movement speed in real time. The overspeed protection unit cuts off the emergency power supply and causes the brake to engage when the speed exceeds a preset threshold. The leveling detection and door opening module is used to continuously detect the leveling position of the car. After reaching the leveling position, the car stops moving and the emergency power supply is used to power the door operator to open the car door and the landing door. During the door opening process, a safety prompt is sent through the voice prompt unit. The intelligent analysis system includes a preprocessing unit, a feature extraction unit, an AI dual-dimensional recognition unit, and a result verification unit. The preprocessing unit performs grayscale correction, Gaussian filtering, and low-light enhancement on the image, and outputs a valid frame image. The feature extraction unit is used to extract static and dynamic human body features from valid frame images; The AI ​​dual-dimensional recognition unit includes: The human target detection subunit uses a lightweight target detection algorithm to analyze static features and outputs the bounding box and confidence score of the human target. The abnormal action analysis subunit uses a temporal neural network to analyze dynamic features and identify preset abnormal actions such as patting the car wall, shouting, and repeatedly pressing buttons. The result verification unit is connected to the AI ​​dual-dimensional recognition unit. When the confidence level of human target detection is greater than or equal to the preset threshold and the abnormal action analysis meets the characteristics of entrapment, the entrapment signal is output to the safe movement judgment and power control module.

2. The elevator intelligent rescue and release system according to claim 1, characterized in that, The lightweight target detection algorithm is YOLOv5 or EfficientDet algorithm, the confidence threshold of the human target detection subunit is preset to be ≥0.7, and duplicate detection boxes are filtered by IOU threshold; The temporal neural network is an LSTM network. The abnormal action analysis subunit determines the following for patting the car wall: the hand is in contact with the car wall for more than 3 consecutive frames and the movement amplitude is ≥5cm. It determines the following for shouting: the head swing frequency is ≥2 times / second and the duration is ≥3 seconds.

3. The elevator intelligent rescue and release system according to claim 1, characterized in that, The safety parameter acquisition unit includes a wire rope tension sensor group installed in the elevator traction system and a guide rail status sensor group installed on the car. The wire rope tension sensor group is used to acquire the real-time tension signal of a single wire rope, and the guide rail status sensor group is used to acquire the guide rail flatness, verticality and lateral clearance signals. The signal processing unit is used to filter, temperature compensate, and perform analog-to-digital conversion on the acquired tension signal and guide rail status signal, and output a standardized digital signal. The safety judgment unit is connected to the signal processing unit and is used to: perform single-strand threshold judgment, tension uniformity judgment, and dynamic fluctuation judgment on the wire rope tension signal; perform flatness threshold judgment, verticality deviation judgment, and gap range judgment on the guide rail status signal; and output a "safe movement allowed" signal when all judgments meet the preset safety conditions. The control output unit is connected to the safety judgment unit signal. After receiving the "safe movement permitted" signal, the control output unit is used to disconnect the elevator main power switch, start the emergency power supply, and send alarm information containing the elevator location to the preset department.

4. The elevator intelligent rescue and release system according to claim 3, characterized in that, The wire rope tension sensor group adopts a pin-type tension sensor, with one sensor configured for each wire rope, and is installed on the wire rope idler roller shaft on both sides of the traction sheave. The guide rail status sensor group includes a laser displacement sensor and a tilt sensor. The laser displacement sensor is installed at the four corners of the top of the car and faces the working surface of the guide rail. The tilt sensor is installed on the upper and lower beams of the car and arranged along the extension direction of the guide rail.

5. The elevator intelligent rescue and release system according to claim 1, characterized in that, The emergency power control unit is used to receive the "safe movement permitted" signal from the safety judgment unit and output a pulse width modulation (PWM) signal to the elevator brake to release the brake in a jogging motion. The dual-redundant speed measurement unit includes at least two speed measurement systems: the first is an incremental encoder installed on the end of the traction sheave shaft, and the second is a laser Doppler speedometer installed in the car, which is used to collect the car's moving speed in real time and output the speed signal. The overspeed protection unit is signal-connected to the dual redundant speed measuring unit and is used to compare the collected speed signal with a preset safe speed threshold. When the speed exceeds the threshold and the duration is ≥100ms, the emergency power supply is cut off to cause the brake to engage. The status feedback unit is connected to the brake and the dual redundant speed measuring unit to monitor the brake's opening and closing status and the consistency of the speed measuring signal. In case of abnormality, the movement control function is locked. The status feedback unit monitors the opening and closing angle of the brake arm through the brake travel switch. When it detects that the brake has not been released as instructed, it immediately stops the pulse signal output and triggers an alarm.

6. The elevator intelligent rescue and release system according to claim 5, characterized in that, The dual-redundant speed measurement unit also includes a signal verification subunit, which is used to calculate the speed difference between the two speed measurement systems. If the difference exceeds 0.1 m / s, a backup judgment logic based on a preset acceleration threshold is activated. The emergency response time of the overspeed protection unit is ≤50ms, and the time from power cut-off to full braking of the brake is ≤300ms. The moving drive unit determines the direction of car movement by using the phase difference signal of the dual redundant speed measuring unit. If the direction does not match the expected direction of the weight difference drive, the overspeed protection unit is immediately triggered. After the brake is engaged, the status feedback unit outputs a "lock signal", which prevents the emergency power control unit from outputting a pulse signal again. The lock needs to be released by a manual reset command. The preset safe speed threshold of the overspeed protection unit is 110% of the elevator's rated speed, which is dynamically adjusted according to the car's load.

7. The elevator intelligent rescue and release system according to claim 1, characterized in that, The leveling position detection accuracy of the leveling detection and door opening module is ≤ ±5mm. After the door is opened, the door status is continuously monitored. If it is detected that the door is not fully open or that people are obstructing it, the power supply time is extended and the safety prompt is repeated. The voice prompt unit supports multiple languages, including Chinese and English, and the prompt content can be customized and edited through the property duty room control terminal; the safe movement judgment and power control module can store fault records and rescue data for the past year and supports wired or wireless network export.

8. A method for intelligent elevator rescue and release, applied to the intelligent elevator rescue and release system described in any one of claims 1-7, characterized in that, Includes the following steps: Step S1: Trapped Person Fault Recognition Trigger: When the elevator stops abnormally, the high-definition camera inside the car is activated to collect a panoramic image of the car. The intelligent analysis system preprocesses the image, extracts features, performs AI dual-dimensional recognition and verifies the results. After determining that there are people trapped inside the car, a trapped person signal is output. Step S2: Safe Movement Condition Determination and Power Control: After receiving the trapped person signal, the safety parameter acquisition unit collects the elevator wire rope tension signal and guide rail status signal. The signal processing unit performs filtering, temperature compensation, and analog-to-digital conversion on the signal. The safety judgment unit completes the safe movement condition determination based on multi-dimensional safety thresholds. If it is determined that the elevator can be moved safely, an alarm message containing the elevator's specific location and the time of the fault is sent to the preset rescue-related departments. At the same time, the elevator's main power switch is disconnected and the emergency power supply is activated. If it is determined that the elevator cannot be moved safely, specific fault type information is sent and the main power supply is kept on. Step S3: Car safety movement control: The elevator brake is released by jogging through the output of PWM pulse signal from the emergency power supply, and the car is driven to move by the weight difference between the car and the counterweight; a dual-redundant speed measurement system is used to collect the car's moving speed in real time, and the speed signal is compared with the preset safe speed threshold. If the speed exceeds the threshold and continues for a preset time, the emergency power supply is immediately cut off to make the brake engage. Step S4: Leveling and Door Opening Rescue: Continuously monitor the car's leveling position. When the car reaches the leveling position, stop the car's movement, supply power to the door operator via emergency power to open the car door and landing door, and simultaneously send safety prompts to the trapped personnel through the voice prompt unit to guide them to leave in an orderly manner, thus completing the rescue.

9. The elevator intelligent rescue and release method according to claim 8, characterized in that, In step S2, if the safety judgment unit determines that "it is not safe to move", it immediately sends the specific fault type to the preset department, keeps the main power supply on, and waits for manual rescue. In step S3, the opening and closing angle of the brake arm is monitored by the brake travel switch. If the brake arm is not released as instructed, the pulse output is stopped and three retries are triggered. If the retries fail, a "brake jamming" alarm message is sent.

Images