Intelligent Elevator Safety Detection System Based on Cloud

Service Big Data

Su Zhengpeng

, Zhang Zehui

, Xie Jiaping

, Tao Fei

, Weng Yuejia

and Zhao Zhongwei

School of Information and Electronic Engineering, Zhejiang Gongshang University

Hangzhou,China

Keywords: Elevator detection, Big Data, C4.5 algorithm

Abstract: Elevator safety accidents often occur in people's daily lives. It is very important to perform 24-hour real-time

testing of elevators in daily operation. This paper proposes a smart elevator security detection system based on

cloud server distributed architecture and big data algorithm technology to solve the problem of elevator

security detection. The C4.5 algorithm can process incomplete data with high accuracy, and can quickly

analyze the operation of the elevator, which is convenient for maintenance personnel to quickly repair and

protect people's safety.

1 INTRODUCTION

As a special equipment for mechatronics, elevators

are inextricably linked with people's daily lives. For

high-rise buildings, the use of elevators as a means of

transportation greatly facilitates people's travel,

saving time and improving the efficiency of office

and life. While the elevator brings convenience to

people's work and life, like other special equipments,

its safe and reliable operation and hidden dangers

also attract people's attention. In the event of an

accident, it is easy to cause casualties and affect

public safety (Park and Yang,2010) (Hang and Guo-jun,

2012) (Xu and Zhao, 2014).

At present, the following treatment process is

often used for the troubleshooting of elevators: when

the elevator fails or an accident occurs, the elevator

owner or the property department makes a request to

the elevator maintenance department, and then the

maintenance personnel arrive at the scene to

troubleshoot, and the handling of the accident lags

behind.With the development of the Internet of

Things technology, the remote monitoring and

diagnosis system of the elevator will provide

guidance and support for the maintenance personnel

to eliminate the fault in the first time(Niu,Lee and

Yang,2018)(Ertuğrul Durak and Yurtseven,2016). Remote

Elevator Monitoring System (REMS) refers to the

remote monitoring, data management, maintenance,

statistics, analysis, fault alarm and rescue of multiple

elevators installed in a building in a certain

area(Lu ,Wang and Liu,2018)(Pisani and

Zucco,2018)(Katakura and Kuroda,2015)(Jin, Zhao and

Ji,2018). However, today's elevator remote

monitoring systems have more or less defects, such as:

the failure analysis of the elevator is not in place, the

maintenance is inconvenient, the test results are not

stored, and it is not convenient for personnel to check.

This paper proposes a new type of elevator

detection scheme, which is improved on the basis of

the elevator remote monitoring system.

Large-capacity data storage, cloud servers and smart

interconnects enable the system to analyze elevator

operation problems from large amounts of data while

processing big data, so that maintenance personnel

can prescribe the right medicine. The rapid analysis

and processing of the information enables rapid

response when the elevator fails, and timely feedback

to the computer, so that the maintenance personnel

can repair the elevator in the shortest time. The

detection system consists of front-end sensor

detection terminals, cloud servers, big data

recognition and analysis algorithms. Once the

elevator runs out of problems, the server or

maintenance enterprise manager will feedback the

results to the elevator maintenance engineer for the

first time, which will help the elevator repair to

quickly reflect and repair (Yi and Zhang,2018)(Crespo,

Kaczmarczyk and Picton,2018)(Sun,2017).

Zhengpeng, S., Zehui, Z., Jiaping, X., Fei, T., Yuejia, W. and Zhongwei, Z.

Intelligent Elevator Safety Detection System Based on Cloud Service Big Data.

DOI: 10.5220/0008097702510256

In Proceedings of the International Conference on Advances in Computer Technology, Information Science and Communications (CTISC 2019), pages 251-256

ISBN: 978-989-758-357-5

251

2 SYSTEM IMPLEMENTATION

PRINCIPLE

2.1 Program Function Block Diagram

Figure 1 is a schematic diagram of the structure of the

intelligent remote elevator monitoring system.

Figure 1: Schematic diagram of structure.

The system uses a speed sensor,a sound sensor,a

frequency sensor,a light level layer sensor, and a

current sensor as terminal data acquisition devices.

Use Arduino to process the collected data and send

the data to the cloud via arduino's wifi module. The

C4.5 algorithm is used to analyze the data transmitted

to the cloud, predict whether the elevator is faulty and

the cause of the failure, and transmit the elevator to

the maintenance department and the elevator

manufacturer.

The following figure is a big data algorithm

topology of the big data detection system:

Figure 2: Big data algorithm topology.

2.2 Algorithm Principle

The specific calculation process of the algorithm is as

follows:

Below we set:

(1) Acceleration(Ac)(m/s2): Ac > 1.5 is high,

0.9 < Ac < 1.5 is medium, Ac < 0.9 is low.

(2) LevelingAccuracy(LA)(mm): |LA| > 15 is

high, 10 < |LA| < 15 is medium, |LA| < 10 is low.

(3) UpstreamCurrent(UC)(A): UC > 11 is high,

9 < UC < 11 is medium, UC < 9 is low.

(4) VibrationFrequency(VF)(Hz) VF>10k is

high,1000<VF<10k is medium,VF<1000 is low.

(5) RunningNoise(RN)(dB) RN > 50 is high, 20

< RN < 50 is medium, RN < 20 is low.

(6) Classification(Cla): 1.circuit failure;

2.chain failure; 3.engine failure; 4.runner failure; 5.

no fault;

We did a series of experiments, the experimental

data shows the fault category statistics when different

variables are used. And the data table is placed in the

appendix. And we’ll use C4.5 algorithm to calculate

the data and draw the decision tree.

1. Computing the Information Entropy:

S1(circuit failure) = 446,

S2(chain failure) = 331,

S3(engine failure) = 2977,

S4(runner failure) = 1363,

S5(no fault) = 1087,

S = S1 + S2 + S3 + S4 + S5 = 6204

Info(D) = 





 







 = 1.9251

2. Computing the Information Entropy of Each

Attribute:

Info(UpstreamC urrent -h igh) = 1.6223

Info(UpstreamC urrent -medium) = 1.8478

Info(UpstreamC urrent -low) = 1.5399

Info(UpstreamC urrent )

= 2145/6204 * Info(UpstreamCurrent-high) +

1663/6204 * Info(UpstreamC urrent-medium) +

2396/6204 * Info(UpstreamCurrent-low)

= 1.6510

3. Computing the Information Gain:

Gain(UpstreamCurrent)

= Info(D) – Info(UpstreamCurrent) = 0.2755

4. Computing the Attribute Splitting Information

Measure:

H(UpstreamCurrent) = 1.5690

5. Computing the Information Gain Ratio:

IGR(Upstrea mCu rrent)

=Gain(UpstreamCurrent) /

H(UpstreamC urrent)

= 0.2755 / 1.5690

= 0.1756

Similarly, we can compute the IGR of other

attributes:

IGR(Acceleration) = 0.1635

IGR(VibrationFrequency) = 0.1534

IGR(RunningNoise) = 0.1265

IGR(Upstrea mCu rrent) = 0.1383

Based on the above calculation results, we can set

“UpstreamCurrent” as the root node of the decision

CTISC 2019 - International Conference on Advances in Computer Technology, Information Science and Communications

252

tree. Similarly, we can calculate and get all the leaf

nodes. And the final decision tree is as follows:

Figure 3: Decision tree.

1. By taking the measured data of the sensor into the

tree for analysis and comparison, it is possible to

predict the cause of the elevator failure.

3 ANALYSIS OF RESULTS

The system establishes the simulation environment

through MATLAB, and uses C4.5 algorithm and ID3

algorithm to test in the elevator load range. In the

simulation test, the actual load of the elevator is no

load, load 50%, load 70%, load 100%, and the

remaining variables are controlled to obtain the

following data:

Table 1: Data from simulation experiments.

It can be seen from the test data that within the

load range, the prediction of the C4.5 algorithm has

little error with the actual situation, and the error rate

is lower than that predicted by the ID3 algorithm,

which also indicates that the system is very good in

practical application. Performance. The following

figure shows the MATLAB simulation speed, uplink

current, and leveling accuracy.

Figure 4: Speed simulation diagram.

Figure 5: Uplink current simulation diagram.

Figure 6: Naughty precision simulation.

Intelligent Elevator Safety Detection System Based on Cloud Service Big Data

253

Figure 7: Vibration frequency simulation diagram.

Figure 8: Running noise simulation diagram.

It can be concluded from the simulation diagram

that the average acceleration of the elevator is

1.25m/s2, the upstream current is 8.3A, the leveling

accuracy is 16.7mm, the vibration frequency is

9900Hz, and the running noise is 51dB. It can be

obtained from data analysis, and the elevator has

runner failure.

4 CONCLUSIONS

With the wide application of elevators, elevator use

units should continuously strengthen the detection of

elevators in accordance with the actual situation,

eliminate various problems and hidden dangers in the

operation of elevators, avoid more safety accidents,

and ensure the personal safety of passengers. The

system detects whether there is a fault in the elevator

from the acceleration of the elevator running, the

displacement of the car and the current when the

elevator is running, and reflects the situation of the

elevator to the terminal of the maintenance

department in real time, reducing the detection of

manpower, and the safety of the elevator is improved.

ACKNOWLEDGEMENTS

National Undergraduate Training Program on

Innovation and Entrepreneurship（201810353035）

REFERENCES

Park S, Yang B. An implementation of risk-based

inspection for elevator maintenance[J]. Journal of

Mechanical Science and Technology, 2010,

24:2367-2376

Hang X, Guo-jun Z. Elevator remote monitoring system

based on Android[J]. Journal of Mechanical &

Electrical Engineering, 2012, 29(9).

Xu H, Zhao G. Wireless Monitoring System for Elevator on

Android Platform[M].Mechatronics and Automatic

Control Systems. Springer International Publishing,

2014: 373-381.

Niu G, Lee S S, Yang B S, et al. Decision fusion system for

fault diagnosis of elevator traction machine[J]. Journal

of Mechanical Science & Technology, 2018,

22(1):85-95.

Ertuğrul Durak, Yurtseven H A. Experimental study of the

tribological properties of an elevator’s brake linings[J].

Industrial Lubrication & Tribology, 2016,

68(6):683-688.

Lu Y S, Wang H W, Liu S H. An integrated accelerometer

for dynamic motion systems[J]. Measurement, 2018.

Pisani M, Zucco M. An accelerometer for spaceborne

application with interferometric readout[J].

Measurement, 2018.

Katakura D, Kuroda Y. 2A1-T07 Elevator detection using

the camera image and distance information[J]. 2015,

2015:_2A1-T07_1-_2A1-T07_2.

Jin X, Zhao K, Ji J, et al. Design and implementation of

Intelligent transplanting system based on photoelectric

sensor and PLC[J]. Future Generation Computer

Systems, 2018.

Yi Xin,Zhang Lei.Design and Research of Elevator

Environment Detection System Based on μC/OS[J].

Electronic Technology and Software Engineering,

2018, (4):94-95.

Crespo R S, Kaczmarczyk S, Picton P, et al. Modelling and

simulation of a stationary high-rise elevator system to

predict the dynamic interactions between its

components [J]. International Journal of Mechanical

Sciences, 2018, 137

Sun Sufen. Research on Elevator Operation Status

Monitoring and Fault Remote Alarm System[J]. China

High-tech Zone, 2017, (4): 94.

CTISC 2019 - International Conference on Advances in Computer Technology, Information Science and Communications

254

APPENDIX

Intelligent Elevator Safety Detection System Based on Cloud Service Big Data

255

CTISC 2019 - International Conference on Advances in Computer Technology, Information Science and Communications

256