Integrated Prediction Error Filter Analysis Application On Lake Towuti

Every naturally-occurred phenomenon on earth is related to cyclicity. On a larger scale, it can be defined as the occurrence of climate periodicity which is caused by the revolution of earth towards the sun. It can also be examined on a smaller process such as the days and nights cycles, as the effect of earths rotation. This research will specifically discuss about the cyclicity of grain size changes in sediments from lake Towuti, Indonesia. The cyclicity of the sediments is deduced using Integrated Prediction Error Filter Analysis (INPEFA) trends. The INPEFA trends are used to analyse the probability of sediments distribution by simply calculating the cumulative errors between predicted and actual data. Unlike any other implementation of INPEFA that mainly observing lithology controlled by sea level changes, this research is aimed at applying INPEFA to enhance well correlation process across an area that is strongly influenced by rainfall intensity and some climatically-driven processes. By correlating the sediment units, the lateral distribution of the climatically-driven diatom ooze will eventually help a better understanding of paleoclimate events on lake Towuti. This research is aimed at constructing and applying systematic algorithm on INPEFA logs calcultation. Two main cores that construct the INPEFA calculation are trend removal process and autoregressive coefficients calculation using Burgs method. When dealing with real datasets the trend removal process is an imperative process to prevent ambiguous INPEFA trend. Moreover, the use of trend removal process is also favourable in interpreting INPEFA trends for various cyclicity periods, this is achieved by varying the input parameters on the trend removal process. Autoregressive coefficients calculation on the other hand is the keystone that constructs the INPEFA logs calculation process. Well correlations process is successfully achieved through interpreting the INPEFA trends logs. Validation of the INPEFA logs shows good correlation between the result and core sample from lake Towuti with widely-distributed tephra being the main key validator. The changes in INPEFA trends is interpreted to be linked with the change in grain size and also in sediments impedance. Comparing and validating the INPEFA trends with two seismic traces from the lake reveals that the turning point of INPEFA trends are associated with strong reflection on the seismic traces. We approach the building of pseudo-INPEFA section through applying optimum Wiener filter (OWF) during the multi-attribures analysis. The lateral continuation of predicted pseudo-log was improved, overall correlation showed an increase by 15% and a decreased in error value by 25%.

A Simplified Method of Trend Removal to Determine Noise Observed During a Supercapacitor’s Discharging

In this paper, new method of trend removal is proposed. This is a simplified method based on Empirical Mode Decomposition (EMD). The method was applied for voltage time series observed during supercapacitor discharging process. It assured the determination of an additive noise component after subtracting the identified trend component. We analyzed voltage time series observed between the terminals of the supercapacitor when discharged by a loading resistance [Formula: see text]. The steps of the proposed method are presented in detail. The results are compared with the results obtained for polynomial approximation. Statistical parameters (kurtosis, skewness) of the histograms of the identified noise component were estimated to evaluate the quality of the proposed detrending method. The method was adjusted to the analyzed data by selecting a parameter of the applied envelope function of the EMD method. We conclude that the proposed method is faster and more efficient for detecting the additive noise component than the competitive polynomial approximation. The identified noise component may be used to evaluate the State of Health of tested supercapacitors and therefore requires fast algorithms with efficient detection.