Comparative Analysis of Classifiers for Prediction of Epileptic Seizures

Epilepsy is a neurological disease in which people suffer from seizure attack and lose the normal function of brain. Almost 50 million people have epilepsy in the world due to which it has become the most common neurological disease. Early prediction of epilepsy helps patients to avoid epilepsy and live normal life. Many studies have been conducted for the early prediction of epilepsy. However, selection of the most appropriate classifier has always been a question that needs to be resolved. In this study, we are using six classifiers of machine learning which are KNN, Naïve Bayes, Linear Classification Model, Discriminant Analysis Model, Support Vector Machine and Decision Tree, to find the best classifier for the prediction of epileptic seizures, in term of accuracy. Dataset from “Kaggle” was used. Preprocessing and cross-validation of the data was carried out for training and testing of classifiers. The results depict that Naive Bayes classifier has a better average accuracy of 95.739% as compared to other classifiers. The future work of this study is to implement the suggested model in real time, so that the workload of medical members could be reduced.

Investigation on Mechanisms of Dynamic Formation of Criteria Gaseous Pollutants in Cng Fired Automobile Engine

The need of alternate clean transport fuels is exponentially increasing due to stringent environmental regulations of vehicular emissions and alarmingly depleting the current resources of traditional transport fuels such as gasoline, diesel fuels. The world statistical data depicts that public transport vehicles are increasingly converted to Compressed Natural Gas (CNG) due to its environment friendly nature. The literature survey depicts that research mainly focused on fluid dynamics aspects (such as turbulence) and flame features (flame development, flame propagation, flame geometry etc) of combustion in various combustions systems. In this regard, a number of studies are reported in the literature in which the combustion of fuels (mostly single component such as methane, ethane, propane, octane, ethanol, pentane, hydrogen etc) was simulated using the kinetic reactions mechanisms. The limited work is reported in the literature related to the formation of pollutants due to the combustion of CNG (a multi component gas mixture) in automobile engines (powered by IC engines). In present research, the combustion of CNG is simulated using the kinetic reaction mechanisms in Internal Combustion (IC) engines. These mechanisms are primarily investigated to predict the formation of gaseous pollutant such as Carbon monoxide (CO), Oxides nitrogen (NO & NO 2 ) and ammonia (NH 3 ) due to combustion of CNG in IC engine. A number of reaction mechanisms were developed and analyzed under the selected simulation conditions describing the practical operating conditions of the automobile engine. The reaction mechanisms were developed by the coupling of two reaction mechanisms (i) a hydrocarbon reaction mechanism generated by EXGAS (an automatic mechanism generation tool) and (ii) Leeds NO x mechanisms. Each of the mechanisms was consisting of the hundreds of the elementary reactions of types including Unimolecular viiiinitiations, Bimolecular Decompositions to initiations, Additions, Isomerization, Beta-scissions, o-rings, Branching, Metatheses, Combination, Dismutation. These mechanism also contain a number of the species/radicals/intermediates such as Monohydroperoxides (OOH), Dihydroperoxydes (OOH) 2 , Allylic molecules YH, , Ethers (O), Ketones (CO), aldehydes (CHO). Each of the reaction was containing the kinetic data (Arrhenius rate parameters; A, b or β and E a ) required to determine the rate constant (k) using the Arrhenius Rate Law and the species thermo-chemical data (NASA Coefficients). Each of the proposed mechanisms was implemented in IC engine module of Chemkin 4.1.1 (a kinetic simulation package) for further analysis and the four detailed reaction mechanisms successfully predicted the combustion profiles of pressure, temperature and selected pollutant species. These are represented by Mechanism-I, Mechanism-II and Mechanism-III Mechanism-IV in this report. Mechanism-I is a comprehensive reaction mechanism containing reactions feasible at range of temperature conditions (below 800 K and above 1000 K). This mechanism is composed of 935 elementary reactions and 185 species. Mechanism-II is a high temperature (above 1000 K) reaction mechanism and consists of 124 species and 792 elementary reactions. This mechanism composed of that type of reaction feasibly at high temperature during the combustion of natural gas. Mechanism-III is a low temperature (below 800 K) reaction mechanism and consists of 152 species and 864 elementary reaction. Mechanism-IV is developed by the simplification of Mechanism-I by the chemical lumping technique and is consisting of only 72 species and 208 elementary reactions. In the simulation study, the common inputs were; (i) fuel composition (CH 4 , 89.03; C 2 H 6 , 1.5; C 3 H 8 , 0.27%; C 4 H 10 , 0.17 %, N 2 , 7.20% & CO 2 ; 2.60% by vol.); (ii) engine geometrical parameters (cylinder displacement volume, 63.0 cm 3 , connecting rod to crank radius ratio, 1.632 etc). Each of the proposed mechanisms of were investigated by (a) ixParametric Analysis (b) Rate of Production Analysis (ROP) (c) Sensitivity Analysis and (d) Uncertainty Analysis. In Parametric Analysis of proposed mechanism, the effect of engine operating parameters such as engine speed, fuel to air equivalence ratio, compression ratio, initial inlet temperature and pressure of feed mixture on the in-cylinder pressure, temperature and pollutant species profiles were analyzed. This analysis determined that output simulation profiles (of in-cylinder pressure, temperature, pollutant species) is greatly affected by the engine speed and fuel to equivalence ratio under the selected simulation condition. The rate of production analysis of each the mechanisms was carried out to identify the reactions involved in the formation of selected pollutant species in addition to the major combustion products (i.e. CO 2 & H 2 O). In this analysis, the total rate of production and normalized rate of production coefficient were calculated for each of the elementary reaction of each mechanism at two temperature condictions of 1500 K and 4000 K. The Sensitivity Analysis showed the dependency (sensitivity) of the output concentrations of pollutant species to the rate constants of the reactions involved. This effect was quantified by determined the “Logarithmic Normalized Sensitivity Coefficients” for each of the reaction involved and showed by the sensitivity bar plot. The Uncertainty Analysis was carried out to determine the uncertainties in the output concentrations of pollutant species due to (i) operating parameters (such as engine speed, fuel to air equivalence ratio and compression ratio) and (ii) due to kinetic parameters (Arrhenius parameters, A, β, E a ) for each reaction was studied. In simulation studies, the adiabatic flame temperature of natural gas combustion predicted are order of ~6300 K, 4400 K, 6200 K and 8200 K for Mechanism-I, Mechanism- II, Mechanism-III and Mechanism-IV respectively. It was also observed that adiabatic flame temperatures increase with increasing initial gas temperature. The in-cylinder xtemperature and pressure were predicted as 4554.738 K and 39.776 atm when compression ratio was 10.51 for Mechanism-I at equivalence ratio of 1.3 (under fuel rich operation), compression ratio of 10.5 (design value for the tested engine), about 3000 rpm engine speed. When combustion in IC engine was simulated with kinetic Mechanism-II (High temperature mechanism), the maximum peak temperature and pressure was achieved at equivalence ratio of 1.3, compression ratio of 10.51, and low engine speed of about 2000 rpm, and initial inlet temperature of 1500 K. The simulation with Mechanism-III illustrates that the maximum peak temperature (3526.161 K) and pressure (31.27 atm) in the combustion chamber were achieved at equivalence ratio of 1.4, compression ratio of 10.51, engine speed of 1500 rpm (low speed) and at initial inlet temperature of 2300 K. and pressure. The Mechanism-IV shows that the maximum peak temperature (4277.804 K) and pressure (41.84569 atm) was achieved when equivalence ratio (Fuel/air) was ≈1.3, compression ratio of ≈10.51, engine speed of ≈ 3000 rpm and initial inlet temperature of ≈1000 K. For experimental measurements, an experimental setup was developed to study the effect of various operating parameters on the CNG combustion in an automobile engine (a type an IC engine) and to validate the simulation result obtained by the proposed kinetic mechanisms. In this experimental study, the in-cylinder profiles of temperature, pressure and pollutant species (CO, NO, NO 2 & NH 3 ) were recorded under various operating conditions of an automobile engine. The simulation data for each of the proposed mechanism is compared with experimental data for and an appropriate mechanism of CNG combustion is selected which showed the closer agreement with the experimental results. The average measured cylinder pressure varied from 0.61 atm to 32.62 atm for six consecutive engine cycles. The highest concentrations of NO x were near the stoichiometric conditions and then become lower while CO level shows increasing trend. The modeled xidata was compared with the experimental data (measured when engine was operated at 3000 rpm, φ=1.0, P inlet =0.67 atm) for each proposed mechanisms. The simulated pressure & temperature profiles of Mechanism-I exhibited the closer agreement with those of the experimental measured profiles while the pollutant species profiles significantly deviated. The deviation in the species profile caused because of the reactions involved in the formation/destruction under given conditions. Similarly, the profiles of Mechanism-II (high temperature above 1000K) and Mechanism-III (low temperature below 800 K) exhibited the early start of the combustion which was not supported by the experimental measurements. On the basis of these discrepancies, it is conclude that Mechanism-I, Mechanism-II & Mechanism-III were failed in the viable prediction of the formation pollutants and the experimental measurements did not validated simulation result. In spite of the existence of some discrepancies among the simulation profiles, Mechanism-IV (consisting of 208 elementary reactions & 72 species) exhibits the closer agreement with the experimental data under the given engine operating conditions. This mechanism is containing the reactions feasible at range of temperature conditions of low (below 800 K) to high (1000 K). In this mechanism, major primary types of reactions include; Unimolecular initiations, Bimolecular initiations, Beta-scissions, Oxidation, Branching, Metatheses, Combination and Dismutation. On the basis of this, it is concluded that Mechanism-IV is consisting of those kinds elementary reactions (both primary & secondary type) involved in the combustion of CNG in the automobile engine and is capable of predicting the formation of the selected criteria gaseous pollutants." xml:lang="en_US