WIRELESS COMMUNICATION AND MEMS SENSORS FOR CHEAPER CONDITION MONITORING AND PROGNOSTICS OF CHARGING CRANE J. Keski-Säntti1, T. Parkkila1, J. Leinonen2, and P. Leinonen3 1

VTT - Technical Research Centre of Finland P.O. Box 1100, FI-90571 Oulu, Finland [email protected] Tel: +358 20 722 2415 Fax: +358 20 722 2320 2

Department of Mechanical Engineering University of Oulu P.O. Box 4200, FI-90014 Oulu, Finland [email protected] 3

Ruukki Oyj P.O. Box 93. FI-92101 Raahe, Finland [email protected]

ABSTRACT The role of the charging cranes is important for the effective and reliable production at a steel mill. A broken crane affects to production and safety. Several checkups and terms to follow are done for safety and for usability of the crane. Continuous condition monitoring measurements for cranes has been quite problematic to carry out for travelling and for the dusty environment and large temperature variations. New technology at the wireless data transfer, sensors, and measurement systems can change the situation. This research focuses on testing MEMS sensors, antennas, and Nanonet TRX transmission technique for condition monitoring purposes. Also one of the targets is ability to use these methods at the prognostication. The testing of MEMS technology based acceleration sensors includes calibration and comparison with normally used sensors. Promising results have been achieved. Also usability of different kind of antennas at the charging crane was tested in a casting bay. Six antenna types were checked in different parts of the bay. The results of the studies gave us the ability to utilize measurement system with FPGA-SoC computing platform. Main advantages of the system are the price of the components which enables either cheaper condition monitoring systems or increasing the amount of measurements. As a part of PROGNOS project these measurements are also going to be utilized at the prognostics of charging crane.

KEYWORDS Condition monitoring, MEMS, Accelerometers, SoC, Prognostic, Charging crane 1 INTRODUCTION Different types of cranes are very common equipments in many types of seaports, construction sites, and industry. Basically they are used for performing quite simple tasks as carrying heavy loads from one point to another. The task might be simple, but the cranes can be only possibility to perform it. At the steel mill the charging cranes have important role for the effective and reliable production. A broken crane can cut off the production and at the worst case it can cause serious employment accidents. In order to ensure safe operations of charging cranes, there are several checkups and terms to follow. At most the purpose of these actions is for safety and doesn’t necessary guarantee the usability of the crane, which might be important part of the production chain. Thus the effective condition monitoring of cranes is necessary for both production and safety. At this case the target of the condition monitoring is travelling charging crane at the casting bay, where the environment is demanding for the measurements because of the dust and large temperature variations. Also the crane is moving at a quite large area, which means complicated structure if communication is implemented by wire. Therefore continuous measurements for cranes have been quite problematic and expensive in practice. Another reason is the price of suitable sensor technology and system implementing. The utilisation of wireless technology has suffered well known limiting factors compared with communication by wire, slower data transfer, lower connection quality and retention, problems in data security and power supply. These problems are now solved in many cases. Wireless data transfer is today enough fast for many solutions, which also makes possible to get connection quality to higher level. There are many ways to improve data security problems like encryption or spread spectrum techniques, but they can slower data transfer. Power supply is still quite problematic. Despite of the progression wireless area needs better batteries, techniques to minimize power usage, or ways to produce the power where it is needed, like it has been implemented in pressure sensors of tyres (Laatikainen, 2005). Measurement and Sensor Laboratory in University of Oulu and VTT Technical Research Centre of Finland have developed systems based on system-on-chip (SoC) technology for wireless comparative measuring. SoC-technology provides an easy and dynamically reconfigurable platform for developers, allowing multiple interfaces, peripherals and/or operators to share the same silicon at different times. Practically the amount or quality of measurements is not limiting the possibilities of designer so much than earlier. New devices can be connected to the system without changing external physical configuration. The needed logic for interface can be configured by programmable FGPA-chip. Also the signal processing is embedded to the system, which enables data compression, filtering, and FFTcalculation. (Karjalainen & Laukkanen, 2005)

2 WIRELESS MEASURING SYSTEM DESCRIPTION Wireless measuring system was selected for testing at the target environment because it was much faster, cheaper, easier to carry out, and suitable for temporary measurements, so the system is transferable from one test point to another. Also the system can be expanded quite freely if it is taken to permanent usage. At these tests the system was small and simple including main units as wireless sending node, wireless base station connected through Ethernet to laptop PC, which works as data collector and control unit. Wireless sensing node includes MEMS (Micro-Electro-Mechanical-Systems) accelerometers, power supply, voltage regulators, AD-converter, microcontroller, and nanoNET TRX radio module. Power supply is possible to perform either by battery or external DC terminal. Signal sampling and sampled data

transmission was performed by 2.4 GHz nanoNET radio technique. AVR microcontroller based transmitter buffers samples before transmitting. Sampling frequency and other parameters can be controlled through radio communications. Same nanoNET radio technique as in sensing node is used in wireless base station for communication. The radio data is received by Altera Cyclone II FPGA chip technique based receiver. After that UDP/IPprotocol is used for packing received data as socket packets and transmitting them to laptop PC through Ethernet by IP-address. The whole wireless measurement system is presented as layout in figure 1.

Figure 1. Layout of the developed wireless measurement system.

3 RADIO TECHNIQUES AND ANTENNA TESTING For to verify the ideas of the system and its suitability at the target environment radio technique tests was performed at the casting bay between charging crane and control room. Receiver was based at FPGA chip technique as presented in figure 1 and it was connected through Ethernet to laptop PC, where the data was stored. Transmitter was also like in figure 1, but without accelerometers in use so that the sent signals were given. The antenna at the transmitter was thin small antenna, but at the receiver tested five antenna types as presented in figure 2. At the testing wasn’t used retransmission feature of the radio protocol. If the data package wasn’t received it was marked as lost. That way the function of the radio could be tested explicit.

Figure 2. Antennas from left to right: thin small antenna, thick small antenna and three beam antennas sector, square panel and long panel. Collected results of the antenna testing are shown in table 1. The best results at the test runs were achieved using thick small antenna at the receiver. The antenna is omnidirectional with small amplification. Other antenna types had stronger signal strength especially at the longer distances, but error probability was higher. At the beam antennas the problem is their sensitive for incident angle. Table 1. Results of the antenna testing, including error probability and average signal strength. Antenna type Thin small antenna

Error [%] MSS

50 68 79,1

Distance [m] 35 16 82,9

Thick small antenna

Error [%] MSS

36 79,4

12 85

Beam antenna, sector (horizontal)

Error [%] MSS

47 88,6

72 91

97 82,9

Beam antenna, sector (vertical)

Error [%] MSS

69 81,3

53 92

21 90,1

Beam antenna, square panel

Error [%] MSS

51 95,9

44 88

Beam antenna, long panel

Error [%]

47

16

MSS

91,8

88

Error [%] = error probability MSS = mean signal strength

25 11 85,7

15 57 84,4

4 87,9

0 88,1

One example of test runs is presented at the figure 3, received data packets, errors, and signal strength at 35 meters distance with thick small antenna. This figure and other same type figures show how well those small omnidirectional antennas work. More data packets are received, because they are not angle sensitive and they can be achieved also by reflection. When the measuring environment is like in this case, the crane is all the time moving and transmitter and receiver are in different heights, the incident angel varies a lot. This should be noted when constructing wireless network and especially when the intention is to use beam antennas. Thick small antenna, distance 35m 10000

packet index / error number

9000 8000 7000 6000 data packets

5000

error

4000 3000 2000 1000 0 1

1001

2001

3001

4001

5001

6001

7001

8001

received data packets

Thick small antenna, distance 35m 90 88

signal strength

86 84 signal strength

82 80 78 76 74 1

1001

2001

3001

4001

5001

6001

7001

8001

received data packets

Figure 3. Received data packages, errors, and signal strength of thick small antenna testing.

4 SENSOR SELECTIONS AND CALIBRATION Currently the MEMS technology based sensors are used in many solutions, where is need for cheap, but working equipment. At the industry and maintenance the accuracy and quality of measurements is important and the conventional sensors are still more used than MEMS sensors, despite of the fact that prices are totally different. MEMS sensor prices are counted in euros when conventional sensors cost hundreds. One main criterion at the testing was the price of the sensors. Other criteria were suitability for

the environment and planned measurements. Selected MEMS acceleration sensor was KIONIX KXPA4 +/- 6g, which was aluminium encapsulated. Calibration was performed at standardized calibrator. The results of calibration showed that the sensor worked well. After calibration the MEMS acceleration sensors was tested at the real target for comparison with normally used piezoelectric sensor. The results presented at the figures 4 and 5 bring out that MEMS sensor works well.

Figure 4. MEMS accelerometer measurement of motor bearing.

Figure 5. Piezoelectric accelerometer measurement of motor bearing.

5 WIRELESS MEASUREMENT SYSTEM TESTING After all parts of the measurement system was tested and they seemed to work well, it was time to check if system is operating as a whole. Test was performed at the shaking box which had ability to feed up zaxis oriented acceleration and frequency at desired value. Aluminium protected MEMS acceleration sensor, KIONIX KXPA4 +/- 6g was glued into shaking box. Shaking frequencies were from 500 Hz to 3000 Hz at the 1G max acceleration. In chapter 2 presented wireless measurement system was used for measuring and data was saved to laptop PC. Digital oscilloscope was used for comparison of analogy signal. Saved data was analyzed by Matlab-program and as result frequency domain spectrum for stored signals at highest impulse are presented in figures 6 and 7.

Figure 6. Acceleration measurement by wireless system, impulse 3000 Hz.

Figure 7. Acceleration measurement by oscilloscope, impulse 3000 Hz

According to figures 6 and 7 and other analyzed measurements can be drawn a conclusion that frequency components over 2500 Hz are folding to lower frequencies at wireless transmitter and thus they are corrupting the frequency domain spectrum. Accelerometer sensors one pole low pass filter at 1700 Hz is not enough for sampling frequency 5208.33. Therefore usage of Atmega32 chip technique based wireless measurement transmitter with that maximum sampling frequency requires stronger filtering of measured analog signal.

6 DISCUSSIONS Wireless measurement systems and MEMS technology have developed very fast recently. Nowadays is possible to produce vast amounts of different kind of data and transfer it much cheaper than ever. Future visions show that progression is getting even faster. Utilisation of MEMS sensors, wireless technique, and data storing possibilities is all the time growing. That affects also to development of condition monitoring. (Holmberg et al, 2005) In future that can mean vast savings, but also new monitoring targets, or larger and more complicated systems. Vast savings can be obtained if the wireless measurement systems start to work as planned. In that case they are easy to installation, their connections to systems are smart, power supply problems have been solved, and they don’t need cabling work. However, it is also possible that new measurements are included to old systems without ability to handle the growing data flow. Anyway the systems are getting larger and more complicated, which means that the systems and equipments at the industry must have good interoperability for to effective and safe production. Requirements focus especially on data producing, transferring, and data storing capacity. Intelligent sensor systems can minimize the transferred data so that only necessary or demanded data is sent and stored. The developed wireless measurement system has shown that above mentioned issues can be solved. SoCtechnology provides an easy and dynamically reconfigurable platform. New devices can be connected to the system without changing external physical configuration. The needed logic for interface can be configured by programmable FGPA-chip. Also the signal processing is embedded to the system, which enables data compression, filtering, and FFT-calculation. So, the amount of transferred data can be controlled. Also the system can work years without battery changing if it is doing certain measurements at the required times. Main advantages of the system are the price of the components which enables either cheaper condition monitoring systems or increasing the amount of measurements. The results of the measurements can be used in condition monitoring. As a part of PROGNOS project these measurements are also going to be utilized at the prognostics of charging crane.

7 CONCLUSION This paper discusses wireless measurement system for condition monitoring purposes. Parts of that system: MEMS sensors, antennas, and transmission technique have been tested separately as well as the function of the whole system. The testing of MEMS technology based acceleration sensors includes calibration and comparison with normally used sensors. Antenna testing was performed using common types. Promising results have been achieved. According to our research this type of wireless measurement system is possible to be used in practice. As a prototype system most of the costs comes from work. The technique is expandable and reproducible. Main advantages of the system are the price of the components which enables either cheaper condition monitoring systems or increasing the amount of measurements. Also this kind of system is possible to use in places which hasn’t earlier achieved.

REFERENCES Holmberg, K., Helle, A. and Halme, J. (2005) “Prognostics for Industrial Machinery Availability”, Maintenance, Condition Monitoring and Diagnostics – International Seminar. POHTO, Oulu, 2829.9.2005, pp 17-30. Karjalainen, S. and Laukkanen, J. (2005) SoC-teknologian hyödyntäminen langattomassa mittaus järjestelmässä, Automation 05, Seminar Days 6. – 8.9.2005, Helsinki, pp. 305-310. Laatikainen, H. (2005), “Measuring tyre pressure for improved road safety”, VTI articles, Release Date 9 March 2005. Available http://www.vti.fi/newsen/newsen_2_4.html