The Defence and Science Technology Organisation (DSTO) and the Centre for Hypersonics at the University of Queensland conduct SCRAMJET (Supersonic Combustion Ramjet) engine experiments to gain a greater understanding of hypersonic flight. Future engine experiments will lead to an inevitable increase in the instrumentation requirements for the system. This thesis explores suitable data acquisition methods for future hypersonic test flights. There are five fundamental criteria that must be met in the development of data acquisition networks for hypersonic test vehicles. These criteria are telemetry bandwidth, fault tolerance, timeliness of data for flight control tasks, scalability, and power efficiency. The two most important criteria (telemetry bandwidth and fault tolerance) are addressed in this thesis by investigating data compression techniques and fault tolerant data acquisition networks.
Three broad data compression techniques (Delta, Run Length, and Huffman Compression) are investigated firstly by validating the algorithms using previous SCRAMJET flight data. Simulated data that is representative of future flights are also used to compare the compression techniques. Delta compression algorithms perform well for the majority of the datasets tested. Typical compression ratios of datasets tested range from approximately 2:1 to 8:1. Huffman encoding algorithms produce slightly higher compression ratios than the delta compression algorithms. Typical compression ratios for the datasets tested range from approximately 3:1 to 14:1. Run length encoding algorithms do not perform without data averaging. By removing data acquisition noise from the signal, compression ratios as high as 60:1 are achieved.
The efficacy of any data compression algorithm is dependent on the structure of the data that is to be compressed. Data compression will inevitably have a significant role to play in any data acquisition network. There is no single answer to the data compression problem, rather a selection of techniques and algorithms that can be optimised for the particular environment are required. Delta compression based on a three-sample moving window predictor is the currently preferred
algorithm since it yields high compression ratios and is relatively simple to implement.
High Speed Data Acquisition Networks are reasonably simple networks due to their primary requirement of transmitting data from many sensors in one direction towards a single data sink, typically the telemetry downlink or mass storage device. When the Data Acquisition Network requires any level of fault tolerance the paradigm becomes more complex. A network model is developed to create a simulation tool to evaluate alternative fault tolerant network structures. The simulation tool is used to compare 12 different networks resulting in 3960 simulations. Network performance parameters such as Data Loss Rate and Link Load Ratios are measured for the comparisons. Ring topologies with carefully selected redundant links produce the best levels of fault tolerance at a reasonable cost.
This thesis provides an analysis of data compression as it relates to the application of high speed data acquisition for SCRAMJET engines. A significant contribution of this thesis is the development of a simulation framework that can increase the efficacy of the instrumentation system design process for SCRAMJET engines.