Railway Diagnostic System Based on Vibration and Acoustic Measurements

István Szatmári, PhD, evopro Innovation Kft.

"The advantage of the chosen modular instrumentation is that we can carry out multichannel, high-precision, and synchronized measurements. The reconfigurable chassis with a Virtex-5 FPGA core provides high computing power, and the real-time module running VxWorks ensures deterministic and reliable real-time applications."

- István Szatmári, PhD, evopro Innovation Kft.

The Challenge:

Developing a reliable measuring system to support a spatially distributed, multichannel, wide-frequency range sound and vibration analytics system for detecting wheel imperfections of railroad cars.

The Solution:

Using an NI CompactRIO embedded real-time controller with NI LabVIEW system design software to accomplish several complex measurements and execute real-time analysis with the built-in FPGA.

Evopro Innovation LLC is a member of evopro-group holding and a professional engineering services company in the field of industrial automation and information technology. Evopro group provides complete technological solutions, planning, testing, and commissioning services for automation around the world. Evopro Innovation focuses on challenging, research-oriented problems in which the latest industrial technologies do not provide sufficient solutions. Evopro Innovation offers its engineering services in embedded hardware design and software development, cloud-based systems, high-performance computing and data analytics, and mobile applications.

 

Railway Monitoring System and Wheel Diagnostic Research Project

Evopro Innovation developed and built the first dynamic in-motion railway measuring system (eRDM) operating on the Hungarian railway track network. The system determines wheel, axle, and carriage loads and identifies overloaded and unbalanced railcars. The basic accuracy of the wagon load measurement is within two percent in the speed range of 5 to 120 km/h. The system consists of two sets of eight digital sensor modules mounted on the rail that are capable of detecting wheel deformations. The rail-side sensor network connects to a wireless control and data gateway that sends data to the remote server. Figure 1 shows sensor modules attached along the railway tracks.

 

 

 

We decided to extend the functionalities of the system with more diagnostic capabilities, such as damaged wheel detection. Wheels and wheel sets of freight and passenger cars are exposed to high dynamic loads. They are safety-critical components that require regular monitoring. We wanted automatic inspection and early warning capabilities since the discovery of potential wheel damage is crucial for safety.

 

The objective of the research project was to examine the capabilities of detecting wheel and bogie imperfections by measuring and processing vibration and acoustic data. We needed to develop a prototype of a cost-effective, low-power diagnostic system that attaches to a rail track and can detect such errors. We needed either a stand-alone solution or an extension of the present eRDM system. Developing an efficient detection algorithm required a large number of field measurements that posed technical challenges, such as a wide speed range of the railcars (5 to 100 km/h), spatially distributed measurement positions (in different configuration of sensors placement alongside the railway tracks – such as 2x8 or 1x16 sensors), synchronized data acquisition, and a wide frequency range (up to 200 kHz), that is all ensured in a reliable measuring system.

 

Developing the Real-Time Data Acquisition and Examination System

We considered several possibilities before choosing NI modular instrumentation and the graphical programming approach. Proper sensor handling and data acquisition required additional workload that diverted our focus from the research problem. The regional NI office offered us a trial of the NI 9234 input module and the NI cDAQ-9171 chassis, which made it possible to connect and operate the IEPE-based vibration and acoustic sensors. We used NI SignalExpress software to rapidly prototype the measurement setup. Finally, we decided that the NI CompactRIO platform was an ideal solution for a reliable measurement environment. The CompactRIO platform included a Virtex-5 FPGA in the backplane of the chassis that could enhance the functionalities of the measurement system with real-time signal analysis and early evaluating capabilities of different detection algorithms.

 

Using the LabVIEW graphical interface, FPGA programming required considerably less development time compared to VHDL coding. Also, we did not need to construct custom hardware, which reduced development time and effort. Figure 2 shows the screenshot of the data acquisition panel controlling four channels in LabVIEW. The example illustrates one audio channel and channels of a 3-axis vibration sensor. Acquired data can be stored for later offline investigation or analyzed in real-time either in a temporal or frequency domain. This might include applying different digital filters or operators such as fast Fourier transforms.

 

 

 

The measurement system we chose consisted of an NI cRIO-9022 embedded real-time controller and an NI cRIO-9114 reconfigurable chassis. We used the NI 9234 input module for vibration and audio measurement. For the higher frequency acoustic emission sensor, we selected the NI 9223 analogue input module. Figure 3 shows the experimental setup built in the laboratory, which measures a single channel audio signal and three channels of vibration signals.

 

 

 

The advantage of the chosen modular instrumentation is that we can carry out multichannel, high-precision, and synchronized measurements. The reconfigurable chassis with a Virtex-5 FPGA core provides high computing power, and the real-time module running VxWorks ensures deterministic and reliable real-time applications. In addition to data acquisition and data logging, the system is suitable for on-site analysis and real-time testing of the developed algorithms. The various detection algorithms are implemented in a distributed environment. The preprocessing, such as the filters, takes place on the FPGA, and the more complex parts are executed on the embedded real-time controller.

 

Conclusion

The CompactRIO platform and LabVIEW environment helped evopro Innovation develop and apply a complete measurement system for railway sound and vibration analysis. This solution greatly reduced development time and the NI off-the-shelf, modular technology and graphical programming approach has hidden the low-level complexity of sensor handling and data acquisition. Therefore, our team could focus on system-level tasks, such as selecting proper sensors and experimenting and determining optimal locations and orientations. In addition to rapid development of the measurement system, the CompactRIO platform incorporated an FPGA that helped us implement and experiment railway wheel error detection algorithms at an early phase of research work.  

 

Author Information:

István Szatmári, PhD
evopro Innovation Kft.
Hauszmann Alajos u. 2.
Budapest 1116
Hungary
Tel: +36 1 464 7959
Fax: +36 1 279 3971
szatmari.istvan@evopro.hu

Figure 1. Sensor Modules of the eRDM Attached Along the Railway Tracks
Figure 2. A Data Acquisition Panel Showing a 4-Channel Setup in LabVIEW
Figure 3. Experimental Setup for Vibration and Acoustic Measurements Based on the CompactRIO Platform