| dc.description.abstract |
Gas sensing has a large number of applications ranging from greenhouse gases (GHGs) monitoring, industrial safety, industrial process control, breath analysis, study of biogenic gases, space exploration applications, food processing industry, air pollution monitoring, etc. Tunable Diode Laser Spectroscopy (TDLS) is a powerful technique for measuring various parameters of gases like concentration, pressure, temperature, flow rate, etc. This technique is highly sensitive and accurate. It is capable of making measurements in harsh environments and is not affected from electromagnetic interference (EMI).
Portable and compact gas sensors are important in field applications. For instance, to monitor the level of GHGs in the atmosphere, the sensor should be portable to collect data and map the level of GHGs in a particular region. Such sensors can also be mounted on an Unmanned Aerial Vehicle (UAV). For such applications, the size and weight of the system plays an important role as the weight that can be carried on such an UAV is limited. Another important factor is the speed of processing which decides the spatial resolution as well as the time resolution of the data collected across a particular region. Power consumption of such systems should also be considered.
The main components in a TDLS setup are the laser and detector which are very small and can be deployed on the field. But, the support electronics like the function generator, oscilloscope, lock-in amplifier (LIA) and computer makes the system bulky and prevents it to be used in field applications. To employ the TDLS system on the field, the support electronics need to be made compact so that the system can be transformed into a hand-held device. Various micro ontrollers available in the market can be used to develop an embedded system for the data acquisition and signal processing which can significantly reduce the overall size and cost of the TDLS system. A micro-controller Raspberry Pi is used in this study for data acquisition and signal processing to extract final concentration and pressure values of gas from the data. The ease of importing LAB programs into Octave programs reduces the development time. The Raspberry Pi can also be onnected to internet using either a Wi-Fi adapter or ethernet. The capability of connecting aspberry Pi to the internet can also be exploited to obtain data from remote location by uploading the data on the server. This data can be accessed wirelessly by an Android App which will access data from the server and provide statistical information on a cell-phone. The typical time resolution of data acquisition and processing to extract concentration and pressure values of gas by direct detection method of TDLS using Raspberry Pi is around 41 s (for 1000 data points). This time resolution is enough for monitoring slow varying processes. However, for real time-varying processes, the time resolution needs to be significantly improved. Higher time resolution is also required for a multi-point gas sensing system or simultaneous detection of multiple gas species. This is possible using a Digital Signal Processor (DSP) instead of a micro-controller such as Raspberry Pi. A DSP can be used as a stand-alone equipment which can eliminate the need for different equipment such as function generator, oscilloscope, LIA and a computer which will reduce the size and cost of a TDLS system significantly. A DSP TMS320F28377D from exas Instruments was selected for this purpose. There is no provision for importing MATLAB
programs in it. Therefore, all the algorithms need to be written either in assembly language or
C, C++ language with limited libraries which increases the development time of such an mbedded system. However, the speed of processing on a DSP is significantly higher than the
traditional micro-controllers and thus, providing much higher time resolution.
The processing of the TDLS signals to obtain gas parameters such as concentration, pressure, temperature, etc. involves a curve fitting program which fits the experimental data to the simulated lineshapes such as Lorentzian, Gaussian, Voigt, etc. generated from a molecular spectroscopic database such as HITRAN. The generation of Voigt lineshape is time consuming and acts as a bottleneck in achieving higher processing speed. The Voigt lineshape is a convolution of the Lorentzian and Gaussian lineshapes and there is no analytical formula for it. Therefore, it is generated by numerical integration. Empirical approximations have been proposed in the literature for rapid simulation of Voigt lineshape. One of those approximations
is the generation of the Voigt lineshape by a linear combination of the Lorentzian and the Gaussian lineshape which is implemented in this study and a comparison has been made between the Voigt lineshape generated by convolution and empirical approximation. The aim of this thesis work was to design an embedded system for portable TDLS systems. Micro-controller Raspberry Pi and DSP TMS320F28377D have been explored in this work. Also, empirical approximations for Voigt profile have been implemented to improve the speed of processing. |
en_US |