Bhaskar D. Kulkarni

Particle swarm and ant colony algorithms hybridized for improved continuous optimization

This paper proposes PSACO (particle swarm ant colony optimization) algorithm for highly non-convex optimization problems. Both particle swarm optimization (PSO) and ant colony optimization (ACO) are co-operative, population-based global search swarm intelligence metaheuristics. PSO is inspired by social behavior of bird flocking or fish schooling, while ACO imitates foraging behavior of real life ants. In this study, we explore a simple pheromone-guided mechanism to improve the performance of PSO method for optimization of multimodal continuous functions. The proposed PSACO algorithm is tested on several benchmark functions from the usual literature. Numerical results comparisons with different metaheuristics demonstrate the effectiveness and efficiency of the proposed PSACO method.

Ant colony framework for optimal design and scheduling of batch plants

This paper presents a new co-operative search approach, the ant colony optimisation paradigm, for the optimal design of batch chemical processes and illustrates it by solving (1) the combinatorial optimisation problem of multiproduct batch scheduling and (2) the continuous function optimisation problem for the design of multiproduct batch plant with single product campaigns and horizon constraints. The ant algorithm is simple to implement and results of the case studies show its ability to provide speedy and accurate solutions.

Using pseudo amino acid composition to predict protein subnuclear localization: Approached with PSSM

Identification of Nuclear protein localization assumes significance as it can provide in depth insight for genome regulation and function annotation of novel proteins. A multiclass SVM classifier with various input features was employed for nuclear protein compartment identification. The input features include factor solution scores and evolutionary information (position specific scoring matrix (PSSM) score) apart from conventional dipeptide composition and pseudo amino acid composition. All the SVM classifiers with different sets of input features performed better than the previously available prediction classifiers. The jack-knife success rate thus obtained on the benchmark dataset constructed by Shen and Chou [Shen, H.B., Chou, K.C., 2005, Predicting protein subnuclear location with optimized evidence-theoretic K-nearest classifier and pseudo amino acid composition. Biochem. Biophys. Res. Commun. 337, 752-756] is 71.23%, indicating that the novel pseudo amino acid composition approach with PSSM and SVM classifier is very promising and may at least play a complimentary role to the existing methods.

Knowledge incorporated support vector machines to detect faults in Tennessee Eastman Process

A support vector machine with knowledge incorporation is applied to detect the faults in Tennessee Eastman Process, a benchmark problem in chemical engineering. The knowledge incorporated algorithm takes advantage of the information on horizontal translation invariance in tangent direction of the instances in dataset. This essentially changes the representation of the input data while training the algorithm. These local translations do not alter the class membership of the instances in the dataset. The results on binary as well as multiple fault detection justify the use of knowledge incorporation.

A support vector machine-based method for predicting the propensity of a protein to be soluble or to form inclusion body on overexpression in Escherichia coli

MOTIVATION Inclusion body formation has been a major deterrent for overexpression studies since a large number of proteins form insoluble inclusion bodies when overexpressed in Escherichia coli. The formation of inclusion bodies is known to be an outcome of improper protein folding; thus the composition and arrangement of amino acids in the proteins would be a major influencing factor in deciding its aggregation propensity. There is a significant need for a prediction algorithm that would enable the rational identification of both mutants and also the ideal protein candidates for mutations that would confer higher solubility-on-overexpression instead of the presently used trial-and-error procedures. RESULTS Six physicochemical properties together with residue and dipeptide-compositions have been used to develop a support vector machine-based classifier to predict the overexpression status in E.coli. The prediction accuracy is approximately 72% suggesting that it performs reasonably well in predicting the propensity of a protein to be soluble or to form inclusion bodies. The algorithm could also correctly predict the change in solubility for most of the point mutations reported in literature. This algorithm can be a useful tool in screening protein libraries to identify soluble variants of proteins.

Estimation of Effective Transport Properties in Packed Bed Reactors

Packed bed reactors are commonly used for carrying out heterogeneous chemical reactions, and the attendant complexities arising out of the interactions of the momentum, heat, and mass transport processes make the design of such units a very involved and cumbersome task. The strategy for rational design depends strongly on the nature of the reaction scheme and its sensitivity to perturbations in operating conditions. Models ranging from simple one-dimensional pseudohomogeneous to two-dimensional heterogeneous are used to describe process behavior. Reviews detailing these aspects of the packed-bed have appeared from time to time.

Hybrid process modeling and optimization strategies integrating neural networks/support vector regression and genetic algorithms: study of benzene isopropylation on Hbeta catalyst

This paper presents a comparative study of two artificial intelligence based hybrid process modeling and optimization strategies, namely ANN-GA and SVR-GA, for modeling and optimization of benzene isopropylation on Hbeta catalytic process. In the ANN-GA approach [Ind. Eng. Chem. Res. 41 (2002) 2159], an artificial neural network model is constructed for correlating process data comprising values of operating and output variables. Next, model inputs describing process operating variables are optimized using genetic algorithms (GAs) with a view to maximize the process performance. The GA possesses certain unique advantages over the commonly used gradient-based deterministic optimization algorithms. In the second hybrid methodology, a novel machine learning formalism, namely support vector regression (SVR), has been utilized for developing process models and the input space of these models is optimized again using GAs. The SVR-GA is a new strategy for chemical process modeling and optimization. The major advantage of the two hybrid strategies is that modeling and optimization can be conducted exclusively from the historic process data wherein the detailed knowledge of process phenomenology (reaction mechanism, rate constants, etc.) is not required. Using ANN-GA and SVR-GA strategies, a number of sets of optimized operating conditions leading to maximized yield and selectivity of the benzene isopropylation reaction product, namely cumene, were obtained. The optimized solutions when verified experimentally resulted in a significant improvement in the cumene yield and selectivity.

Dynamic optimization of chemical processes using ant colony framework

Ant colony framework is illustrated by considering dynamic optimization of six important bench marking examples. This new computational tool is simple to implement and can tackle problems with state as well as terminal constraints in a straightforward fashion. It requires fewer grid points to reach the global optimum at relatively very low computational effort. The examples with varying degree of complexities, analyzed here, illustrate its potential for solving a large class of process optimization problems in chemical engineering.

Application of multiresolution analysis for simultaneous measurement of gas and liquid velocities and fractional gas hold-up in bubble column using LDA

A new strategy using wavelet transforms for multiresolution analysis of bubble column hydrodynamics from laser Doppler anemometer measurements of the axial velocity has been formulated. The advantage of the methodology is that we may infer intrinsic features of the bubble column hydrodynamics by accurately detecting the passage of bubbles at local positions in the column. The detection of bubbles was possible by denoising local time series data and analyzing for its intermittent behavior. The methodology allows for calculating the changes in the local average liquid velocities and local gas hold-up and was related to the intermittent nature of turbulence in the system. The variation in these properties and flow behavior as a function of radial distance have been studied.

Feature extraction and denoising using kernel PCA

Kernel PCA methodology, an elegant nonlinear generalization of the linear PCA, is illustrated by considering the examples of (i) denoising chaotic time series and, (ii) prediction of properties of polymer nanocomposites developed in our laboratory. Kernel PCA captures the dominant nonlinear features of the original data by transforming it to a high dimensional feature space. An appropriately defined kernel function allows the computations to be performed in the original input space and facilitates extraction of substantially higher number of principal components enabling excellent denoising and feature extraction capabilities. Use of simple matrix algebra in simulations makes the method an attractive alternative to some hard optimization based methodologies.