The ability to interface real-time instrumentation data from remote hardware to local data analysis systems is important in the space industry. NASA, for example, is currently developing efficient systems to interface with real-time data for command, control, and monitoring of rocket and supporting hardware systems. Kennedy Space Center uses Allen-Bradley programmable logic controller (PLC)-based control systems to interact with remote hardware at launch, including cryogenic, pneumatic, sound suppression, and other hardware subsystems. Since PLCs can utilize OLE For Process Control (OPC) communication protocols to interact with remote systems, using signal processing applications that are OPC-capable can greatly enhance the monitoring, processing, and data analysis of remote aerospace hardware signals and industrial processes.
Researchers at the University of Central Florida (UCF) have found that by incorporating analysis and monitoring algorithms such as Inductive Monitoring System (IMS), neural networks, and recent advances in deep learning within the architecture’s signal processing system engineers have a flexible and powerful end-to-end data analysis and monitoring system for instrumented remote aerospace hardware.
Space Launch System (SLS) rocket engine RS-25 data, for example, can be received and analyzed by the pro-posed system/architecture during testing and launch countdown operations. Space station hardware may also be wirelessly monitored by distributed remote tablets interfacing with the proposed architecture. Anomaly detection and advanced monitoring systems are key capabilities for future human missions beyond Earth orbit.
A Hybrid Architecture for Monitoring and Anomaly Detection of Aerospace Systems (HAMADAS) encompasses hybrid systems within different layers of the design. In the signal/data analysis system layer (Simulink and MATLAB), the implementation of data processing and analysis logic, using different types of languages to process PLC data, provides a flexible and expandable mechanism for engineering analysis. Using C++ language legacy routines as well as MATLAB routines inside of a Simulink model creates a hybrid model that combines three programing languages: Simulink graphical, MATLAB scripting, and C++. This configuration takes advantage of a large array of analysis functions, visualization, anomaly detection, pattern recognition, and built-in tools available in different programming languages within Simulink models.
The four basic system components of HAMADAS are instrumented hardware, PLC, signal analysis, and remote tablets. This architecture, however, can be prototyped and tested even when real instrumented space hardware and PLCs are not available. UCF researchers tested the interactions between the analysis system, the cloud, tablets, and data within a simulated PLC environment producing simulated data streams.
Interfacing real-time data streams from PLC controller systems to an engineering data analysis and visualization system creates a flexible structure for interactivity. MATLAB and Simulink are very powerful and efficient tools that can be interfaced with real-time data. These data can then be processed and analyzed by the native processing system functions, tools, and capabilities.
Interfacing remote wireless tablets in the cloud with an architecture that connects a PLC controller system to a data analysis and monitoring system is a powerful human-machine interface. This interface configuration requires a combination of application software and technologies to interface with the Cloud.
Dropbox windows application software provided the interface from the Windows-based platform to the cloud by using a dedicated Dropbox windows folder configured for input/output transactions between PC and the cloud.
Remote tablets can receive raw data from the PLC by using the analysis system as a pass through. The diagram on this page shows a simple method to broadcast raw and processed data from the data analysis system to the cloud. In this configuration, the data analysis system receives PLC raw instrumentations data/tags via OPC. Then the data is passed through as vectors, which are stored in the cloud. Data vectors are then available to tablets using tablet apps which interface with the cloud.
The research conducted by the UCF team investigated the implementation of IMS’ anomaly detection legacy routines within the proposed architecture’s data analysis and visualization component. Due to interactions within the IMS logic routines, full implementation of anomaly detection IMS within the proposed architecture requires further understanding of the limitations within the legacy tool configuration of compiled Simulink S-functions. For example, some of the file I/O constructs in IMS’ routines may not be supported and would require modification of legacy logic before implementation in Simulink.
This article is based on SAE International technical paper 2013-01-2090 by Edwin A. Cortes and Luis Rabelo of the University of Central Florida.