Detection of Elevator Cable Slippage Using Streaming Image Process Algorithm

. To


Introduction
In elevators, wire cables are largely used to support the cage and the compensation cable, which is connected to the compensation sheave.(3)(4)(5) Research on elevator cables has verified that the major risk during elevator operation may arise from cable slippage.(8)(9) To ensure the safety of elevator operation, International Standard ISO4101 was developed by Technical Committee ISO/TC 105, Steel Wire Ropes, and circulated to member bodies in October 1982. (10)Regular inspections are therefore mandatory to ensure a high level of safety for both personnel and equipment during elevator operation.According to this standard, the elevator cable is required to be replaced under any one of the following conditions: 1. more than 10% of the steel wire is fractured; 2. the rope has become thin with a diameter 7% less than the original size; 3. distortion exists; 4. rust erosion has occurred.13) The shortening of the life of multicable elevators as a result of problems such as vibration, noise, uneven wear of key components, and uneven distribution of rope tension is still unclarified in the elevator industry.(16) It was indicated that the elevator cable is a flexible wire with low damping, so it is prone to vibrations.In the case of a high-rise building, the large rope length may have a high possibility of resonance.In addition, an investigation on the vibration of elevator cables in a high-rise elevator system under earthquake excitation has been presented. (17)In this study, various cable lengths that may cause natural resonance and the effect of time-variant variables on the damping ratio were considered.The dynamics of the cables were also analyzed and solved numerically using the governing nonstationary, nonlinear equation.To reduce the vibrations caused by the resonance of the elevator cable owing to building sway, the vibration control approach of using a nonstationary robust control method was proposed. (18)Although several methods of estimating the vibration displacement using image processing or laser displacement meters with an observer or a Kalan filter were developed, observing the vibration of the elevator cables in an elevator shaft is a difficult task because of the movement of the elevator car and the thinness of the cable as an observation target. (19)ore importantly, the metal fatigue, along with abrasion, of the elevator cable may lead to the progressive loss of the metallic cross section. (20,21)Some features such as the roughened and pitted surfaces of the cables, reduced cable diameter, and broken wires can be seen.Therefore, the deterioration must be monitored before any unexpected damage or corrosion causes a fatal accident. (22)Ultrasound-guided wave-based inspection can be applied to monitor the cross section of wire cables.However, its efficacy in detecting defects in wire cables is low.For this reason, the finite element analysis of dispersion curves was conducted to study guided wave propagation in wire cables. (23)Recently, a new method of detecting broken steel wires on the cable surface using the conducting property of the cable has been reported. (24)It revealed that the spiral cable characteristic affects the eddy current response signal in the broken wires.On the basis of this principle, an eddy current differential probe was designed to improve the detection accuracy in a simulation.However, its applicability in practice is not yet confirmed.To balance the different tensions and the axial deformation of each cable, automatic balance equipment for cable-type elevators was proposed. (25)The results of using the Simscape numerical methodology showed that the imbalance rate of the system could be effectively limited to a low level.

System Structure
The proposed system structure is shown in Fig. 1.The main procedure is as follows.(1) A video of the elevator-cable main wheel operation is taken using an IP camera (IP Cam).( 2

Model principle
The practical elevator cable main wheel is a circle, as shown in Fig. 2. The circle is used to derive the mathematical equation for the proposed model. (26)n Fig. 3, an arbitrary point P(rcosθ, rsinθ) on the circle to the center of the circle, O(0, 0), is a constant radius, r, where the angle with the x-axis is θ (0 ≤ θ < 2π), and the perimeter l = rθ.Therefore, the circle equation is formed as x 2 + y 2 = r 2 , where x = rcosθ and y = rsinθ.
When a photograph of the main wheel is taken, the image will become an elliptical shape as shown in Fig. 4. Assume P(x, y) is an arbitrary ellipse point, where F 1 (c, 0) and F 2 (−c, 0) are two points of focus, O is a center point, a is the semi-major axis, b is the semi-minor axis, c is the semi-focus point, and p is the half-sinusoid.The ellipse trajectory equation is defined as where 2a is a constant length.
As shown in Fig. 5, the ellipse equation is

Proposed model
Initially, the circle is equally divided into 360°, as shown in Fig. 6.Its respective ellipse is shown in Fig. 7, where a red label (A) marks the reference point for the wheel rotation phase.Because of a slipping phenomenon, the A point is expected to exceed θ degree from the original location of 0°, as shown in Fig. 8.The number of rotations of the main wheel is N under a normal operation, where N is the standard number of rotations for the elevator car to travel from the highest level to the lowest level.The actual number of rotations is designated as M. In practice, M ≥ N, where M and N are integers.
The total cable slipping distance (L) can be calculated as (26) L where r is the main wheel radius and θ is the excess slipping angle.

Model Process
The proposed model process flowchart is shown in Fig. 9, and its major process is briefly described below.
(1) Take a video of the main wheel.

Dynamic Image Processing with Experimental Results
In this study, one surveillance IP Cam installed in front of the main wheel was used.The resolution of the surveillance IP Cam is 1920 ×1080 pixel, and the frame rate is 30 frames per second with the baud rate of 4096 kbps for image transmission.

Dynamic image processing
The main wheel image is shown in Fig. 10, where the black label indicates the reference point.
The main procedure to implement the proposed model is demonstrated as follows.(18) The location of the start angle is marked A in Fig. 28, where its value is 110.5°, as shown in the pink frame at the bottom of the figure.The elevator starts to operate from the start angle.(19) The location of the end angle is marked B in Fig. 29 when the elevator stops.Its value of 101.0°, is seen in the pink frame at the bottom of the figure.From Eq. ( 1), the slipping distance is obtained as 26.529 mm.Note that the number of main wheel clockwise (CS) rotations, i.e., 24, is the same as that of main wheel counterclockwise (CCS) rotations, as shown in the pink frame at the bottom left in the figure.

Experimental results
Twenty tests were carried out for the elevator operation of a round trip between the first floor and the fourth floor, which is at a height of about 12 m.Microsoft SQL Server was applied to save the measured results shown in Table 1, including start angle, end angle, and slipping distance.
From the results in Table 1, the crucial measurement factors, i.e., average value, standard deviation, and measurement uncertainty, were revealed.
The average value (x ) of the measurement is defined as where x i is the result of the ith measurement and n is the number of measurements.
The standard deviation (s) of the n measurements is defined as Accordingly, the measurement uncertainty (u) is defined as  The measurement data were substituted into Eqs.( 2)-( 4), and the results are shown in Table 2.As can be seen, the average slipping distance (x) is 26.599 mm.More importantly, the standard deviation (s) is as low as 1.153 mm, and the measurement uncertainty (u) is only 0.258 mm.These values confirm that the proposed model can achieve a high measurement accuracy.

Conclusions
Elevator cable slippage is regarded as one of the most influential factors that may severely affect the safety of elevator operation.Presently, slippage detection is still mostly performed by visual inspection in industry today.To solve this problem, the streaming image process model for elevator cable slippage detection has been successfully demonstrated to achieve the following: (1) The proposed model is insensitive to the position of wheel rotation.Here, only the wheel rotation start and end angles are needed to calculate the slipping distance by a simple arithmetic operation.(2) The standard deviation (s) is as low as 1.153 mm, and the measurement uncertainty (u) is only 0.258 mm.(3) The results obtained from elevator cable slippage measurement can be transmitted to the cloud instantly.(4) The cable slipping status can be monitored online via web-based GUI soon after the slipping distance is calculated at every elevator operation.Moreover, this proposed method can be applied to any type of elevator if the cable movement is pulled using a main wheel.The accuracy and precision when applied to other types of elevator will not be affected in practice owing to the model's insensitivity to the type of elevator.
) The collected image plus the elevator ID number and detection date with time are then sent to the dynamic image measuring model via Transmission Control Protocol/Internet Protocol (TCP/IP) using a router for further calculation.(3) The calculation results obtained using the proposed model are transmitted to a Microsoft SQL Server through the modem and router using TCP/IP.(4) The cable slipping distance data can be read by the graphical user interface (GUI) via the modem, ISP, and router.(1) IP Cam: It is used to transform the detected image into a video stream to be processed via TCP/IP by the computer.(2) Dynamic image measuring model: The slipping distance of the elevator cable is calculated on the basis of the video stream collected from the IP Cam.(3) Microsoft SQL Server: The cloud database is used to save the elevator ID number, detection date and time, cable slipping distance, and so forth.Simultaneously, it can provide an IP address for a remote authorized client (PC) to access the server.(4) GUI: It provides a friendly operation interface for reading the cable slipping information as well as defining the abnormal cable situation.
a > b > 0, if the focus is located on the x-axis.On the other hand, it is a > b > 0 and a 2 − b 2 = c 2 , if the focus is located on the y-axis.

( 3 )
Mark Up, Down, Left, Right, and Center reference points.(4) Equally divide the image into 360° to create the standard ellipse image.(5) Set the main wheel radius r and the standard number of rotations, N. (6) Perform image masking.(7) Set detection parameters.(8) Set the initial starting angle.(9) Calculate the number of rotations, M. (10) Set the angle for the end of rotation.(11) Compare the collected image with the standard image once the main wheel stops.(12) Calculate L = (M − N) × 2πr + rθ.

( 1 )
Enter the main operation panel, as shown in Fig. 11.(2) Path: Set the real-time streaming protocol (RTSP) address, as shown in Fig. 12. (3) Connect: Connect to the dedicated IP address by clicking the "Connect" button, as shown in Fig. 13.(4) Play: Play the video of the rotating main wheel, as shown in Fig. 14. (5) Flip: Flip the image, as shown in Fig. 15.(6) Point: Display the reference point of the image, as shown in Fig. 16.(7) Select the center reference point with pixel coordinates x: 501, y: 405, as shown in Fig. 17. (8) Auxiliary Line: Set the auxiliary line for Up, Down, Left, and Right (vertical or parallel mode), as shown in Fig. 18. (9) Select the Up reference point with pixel coordinates x: 501, y: 235, as shown in Fig. 19.(10) Select the Down reference point with pixel coordinates x: 501, y: 574, as shown in Fig. 20.(11) Select the Left reference point with pixel coordinates x: 375, y: 405, as shown in Fig. 21.(12) Select the Right reference point with pixel coordinates x: 629, y: 405, as shown in Fig. 22. (13) Close the auxiliary line, as shown in Fig. 23.(14) Set the start and end angles to calculate the number of main wheel rotations, as shown in Fig. 24.(15) Process Gaussian blur to reduce image noise, as shown in Fig. 25.(16) Implement the mask function, as shown in Fig. 26.(17) Start the detection process, as shown in Fig. 27.