Using Least-Square Spectral Analysis Method for Time Series with Datum Shifts
Advancement in the field of technology such as processing power in computers and servers has provided us with an opportunity to improve the performance of legacy system programs. This will improve the performance of these systems along with increase in ma
- PDF / 1,047,259 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 86 Downloads / 249 Views
Abstract Advancement in the field of technology such as processing power in computers and servers has provided us with an opportunity to improve the performance of legacy system programs. This will improve the performance of these systems along with increase in maintainability and life expectancy. But it is not a simple task to simply pick a legacy system and convert it. In this project the algorithm to compute least-squares spectrum LSSA, which is used for analyzing huge data and plot the spectrum chart will be converted from its original FORTRAN program to C program and implement parallel processing in order to improve it using various available methodologies. Keywords LSSA
f2c Par4all FFT Spectrum
1 Introduction Researchers and engineers are often expected to work with experimental time series which have many missing data points. At times, there are time series that are unequally spaced, like that of weather observations, or sometimes to avoid aliasing variable sampling rate may be introduced intentionally. Further, advancements in development of instruments have led us to understand the physical phenomena better and hence resulting in more accurate data. So over long periods the time series collected will be unequally weighted and hence will be nonstationary. There may also be disturbances due to repair and replacement of instruments, which may cause datum shifts [1] in the series (Figs. 1, 2, 3 and 4). Ingudam Gomita (&) Sandeep Chauhan R.K. Sagar Amity University, Noida, India e-mail: [email protected] Sandeep Chauhan e-mail: [email protected] R.K. Sagar e-mail: [email protected] © Springer Science+Business Media Singapore 2016 M. Pant et al. (eds.), Proceedings of Fifth International Conference on Soft Computing for Problem Solving, Advances in Intelligent Systems and Computing 437, DOI 10.1007/978-981-10-0451-3_5
39
40
Ingudam Gomita et al.
Fig. 1 f2c conversion process
Fig. 2 Conversion using par4all
Entirety of the techniques of fashionable spectral techniques lies with the fast fourier transformation algorithms. These algorithms are used for determining the power spectrum. The FFT is mainly used for two reasons—(i) it is computationally efficient and (ii) for a huge class of signal processes it gives reasonable results (Figs. 5, 6 and 7).
Using Least-Square Spectral Analysis Method …
41
Fig. 3 LSSA.OUT output
However, it is not a perfect method for spectral analysis. It has certain limitations— the most salient being the fact that the data need to be equally weighted and equally spaced. So in these cases it becomes inevitable to process the data and as a result it performs poorly and unsatisfactorily. An alternative to the FFT was developed by Vanicek (1969, 1971) [2]. It is known as least-square spectral analysis (LSSA). It overcomes the methods of the fourier methods. The data need not be equally spaced or weighted. Gaps and datum shifts do not make a difference here. The LSSA has been revisited [3] highlighting its significant properties and concentrating on the co
Data Loading...
