Paweł Pawlik - AGH University of Science and Technology, Faculty of Mechanical Engineering, Department of Mechanics and Vibroacoustics
Dariusz Dabrowski - AGH University of Science and Technology, Faculty of Mechanical Engineering, Department of Mechanics and Vibroacoustics
Machine condition monitoring in mines is economically critical. The costs of frequent unplanned downtimes exceed the costs of repairs due to production reduction. Machine condition monitoring systems help predict failures for different machine parts.
In the Konin opencast mine, we developed a machine condition monitoring system for the bevel planetary gearbox, which is part of the power transmission system of the bucket-wheel excavator KWK-1500s (see Figure 1).
Because of the industrial conditions, the system had to be resistant to shock, vibration, and temperature. It also needed to ensure failure-free operation with the guarantee of diagnostics tasks. Due to these requirements, we chose the NI cRIO-9022 controller and NI cRIO-9114 chassis for our system.
We used NI CompactRIO and the NI 9234 C Series modules to acquire vibration signals. Accelerometers on the gearbox casing measured the signals. The CompactRIO controller facilitated communication with the programmable logic controller via Modbus TCP/IP protocol. As a result, the user could read excavator operation conditions such as oil temperature, oil pressure, power, and rotational speed from controller registers.
The CompactRIO device connected to the industrial panel computer IPPC-6192A by TCP/IP protocol. The system used a touch panel computer for data logging, results presentation, and communication with PC computers that had access privileges.
We configured the FPGA to perform parallel data acquisition. Afterward, the system sent data to the real-time controller via first-in-first-out (FIFO) memory buffer, which was configured in the DMA mode.1
The main application ran on the real-time controller as a multithreaded structure (see Figure 2). The structure featured four basic threads: acquisition, ACQ programmable logic controller (PLC), analysis, and communication with the panel PC. The system listed the threads from the highest priority to the lowest. With the multithreaded structure, we could add functionality to the system, such as acquisition analysis or communication, without interfering in the existing program code.
In the structure, the communication with the PC panel thread was the master because it sent commands to the other threads and was responsible for communication with the panel PC. The industrial panel computer used Secure Digital (SD) disks for data storage where the raw signals and all calculated estimates were archived. The system used Technical Data Management Streaming (TDMS) binary files for data logging and acquisition parameters storage.
The touch panel PC connected to the mine intranet for data transmission from the excavator to the PCs. Users with privileges could remotely control system parameters.
Diagnostic Signal Analysis
The planetary bevel KPB 190-214 transmission gear (nominal power 630 kW, transmission ratio 190 and input shaft rotation 990 rpm) is especially designed for drive systems of bucket-wheel excavators in open-pit mining.2 The gearbox is divided into four parts from a diagnostics point of view: input shaft, bevel gear, first planetary gear stage, and second planetary gear stage. Figure 3 shows the front panel of the PC application.
Next, the system calculates the characteristic frequencies of each part of the system and analyzes the estimates of RMS, crest factor, and kurtosis to determine the technical state of the gearbox. To calculate these estimates, we built the function “diagnostics parameters KPB214” (see Figure 4). In this function, raw signals from acceleration sensors mounted on the gearbox casing are filtered by passband cascade filters to remove all frequencies above the fourth harmonics.
Due to fluctuations of rotational speed caused by the operating conditions of the bucket-wheel excavator, the system calculates the filter coefficients in the function. Each element is considered separately, so at the end, the system obtains four filtered signals and sets of estimates. The estimate set consists of variables such as RMS, crest factor, and kurtosis. The function also gathers information about values of RMS in particular bands.
Rugged, Flexible System
We developed a machine condition monitoring system for the bevel planetary gearbox on the bucket-wheel excavator KWK-1500s in the Konin opencast mine. We built a rugged system for harsh industrial conditions with the CompactRIO controller. We can remotely control the system parameters and observation of registered data via the mine’s intranet by VPN connection. With continuous data logging, we can develop new diagnostics procedures and, after verification of new algorithms, incorporate them into the system.
1 National Instruments, NI LabVIEW for CompactRIO Developer’s Guide, 2012
2 http://www.flsmidth.com, 19.09.2012
AGH University of Science and Technology, Faculty of Mechanical Engineering, Department of Mechanics and Vibroacoustics