K. S. Tang

Genetic algorithms and their applications

This article introduces the genetic algorithm (GA) as an emerging optimization algorithm for signal processing. After a discussion of traditional optimization techniques, it reviews the fundamental operations of a simple GA and discusses procedures to improve its functionality. The properties of the GA that relate to signal processing are summarized, and a number of applications, such as IIR adaptive filtering, time delay estimation, active noise control, and speech processing, that are being successfully implemented are described.

Multi-product planning and scheduling using genetic algorithm approach

Abstract Earliness and tardiness production scheduling and planning (ETPSP) have been studied by a number of researchers in recent years. However, the existing researches have been limited to the study of machine scheduling, and the effects of multi-product production, with the considerations of machine scheduling and lot-size and capacity are not being investigated. One of the reasons for this is the complexity of solving large-scale discrete problems where restrictions of linearity, convexity and differentiability prevail. Classical optimization methods have proved inadequate and an alternative approach is investigated here. A new extensive model of ETPSP is developed in this paper to address the multi-product production environment. A genetic algorithm (GA) is applied in order to obtain an optimal solution for this large-scale problem. The investigation demonstrates the use of a comprehensive model to represent a real life manufacturing environment and illustrates the fact that a solution can be effectively and efficiently obtained using the GA approach.

Genetic algorithm to production planning and scheduling problems for manufacturing systems

Fundamental and extended multi-objective (MO) models are designed to address earliness/tardiness production scheduling planning (ETPSP) problems with multi-process capacity balance, multi-product production and lot-size consideration. A canonical genetic algorithm (GA) approach and a prospective multi-objective GA (MOGA) approach are proposed as solutions for different practical problems. Simulation results as well as comparisons with other techniques demonstrate the effectiveness of the MOGA approach, which is a noted improvement to any of the existing techniques, and also in practice provides a new trend of integrating manufacturing resource planning (MRPII) with just-in-time (JIT) in the production planning procedure.

Parallel Genetic-Based Hybrid Pattern Matching Algorithm for Isolated Word Recognition

Dynamic Time Warping (DTW) is a common technique widely used for nonlinear time normalization of different utterances in many speech recognition systems. Two major problems are usually encountered when the DTW is applied for recognizing speech utterances: (i) the normalization factors used in a warping path; and (ii) finding the K-best warping paths. Although DTW is modified to compute multiple warping paths by using the Tree-Trellis Search (TTS) algorithm, the use of actual normalization factor still remains a major problem for the DTW. In this paper, a Parallel Genetic Time Warping (PGTW) is proposed to solve the above said problems. A database extracted from the TIMIT speech database of 95 isolated words is set up for evaluating the performance of the PGTW. In the database, each of the first 15 words had 70 different utterances, and the remaining 80 words had only one utterance. For each of the 15 words, one utterance is arbitrarily selected as the test template for recognition. Distance measure for each test template to the utterances of the same word and to those of the 80 words is calculated with three different time warping algorithms: TTS, PGTW and Sequential Genetic Time Warping (SGTW). A Normal Distribution Model based on Rabiner23 is used to evaluate the performance of the three algorithms analytically. The analyzed results showed that the PGTW had performed better than the TTS. It also showed that the PGTW had very similar results as the SGTW, but about 30% CPU time is saved in the single processor system.

Multiobjective optimization of radio-to-fiber repeater placement a jumping gene algorithm

This paper considers the radio-to-fiber repeater placement problem in wireless local loop (WLL) Systems. The severe problem that the WLL systems encountered is that the large diffraction loss from rooftop to street occurs at its frequency band, 2.3 GHz. The radio-to-fiber repeaters can be used for the remedy of this situation. Unlike the conventional WLL systems, the total system cost of this option depends on the additional repeaters and optical fibers (links). Thus, our objective is to minimize the total repeater cost and total link cost simultaneously by selecting optimal locations for the repeaters. It is a multiobjective problem in which a tradeoff between the total repeater cost and total link cost can thus be made. A new jumping gene paradigm called jumping-gene genetic algorithm (JGGA) is proposed to solve this conflicting dilemma. The main feature of JGGA is that it only consists of a simple operation in which a transposition of the gene(s) is induced within the same or another chromosome within the framework of genetic algorithm. The algorithm has been tested by using two specific performance metrics in evaluating the quality of obtained sets of non-dominated solutions. Simulation results revealed from this study that JGGA is able to find non-dominated solutions with better convergence and diversity than other multiobjective evolutionary algorithms.

A realizable architecture for genetic algorithm parallelism

Abstract A novel hardware architecture is specifically designed here to fulfil the realization of parallelism of genetic algorithms. It is a modular structure which consists of three individual processing units, namely the Genetic Operator, the Fitness Evaluator and the Objective Function Sequencer. Each unit is implemented by the use of an FPGA chip. Due to its modular structure, this design possesses a unique flexible and scalable feature that is capable of handling various engineering applications. Such a scalable feature can markedly improve the computing speed of this hardware simply by the increase in the fitness evaluators. The results obtained are very encouraging for future development, particularly where genetic algorithms are used in real-time system applications.

A graphical teaching platform for genetic algorithms

Despite the increasing numbers of genetic algorithms (GA) applications, we have yet to witness a well-designed teaching kit developed for education. Such a missing link between research and education will hinder the further development of GA, especially for newcomers and other potential but unexplored research areas. A software teaching kit on Microsoft Windows 95 platform is hence developed based on the biomorph process. The main theme of this software is to create a creature, an insect, according to the creatures specifications and features. The user can play the game by selecting various parameters of an insect provided in the kit. Throughout the GA evolutionary processes, the preferred insect is to be created. In such, user can learn about the GA according to the pre-designed selection of the basic functions such as crossover, mutation, selection and so on.

Hierarchical Genetic Algorithms in Computational Intelligence

It is anticipated that future engineering design will relegate its own disciplinary concepts and will become heavily involved with computational intelligence (CI). This trend of development is understandable since computing power has become so much faster and cheaper nowadays, such that a required solution can be automatically obtained even when this is based upon a computationally intensive scheme.

Genetic Algorithms in Filtering

In filtering design, we often encounter the dilemma of choosing a minimum order of a model for the representation of a signal so that a minimum use of the computational power is ensured. In the case of real time processing, this requirement is necessary and obvious. Achieving this goal is by no means easy. This can be very involved in circumstances where the performance criteria are discrete, complex, multiobjective and very often, nonlinear.

Genetic Algorithms in Production Planning and Scheduling Problems

In any manufacturing system, an effective production planning and scheduling programme is always desirable. Such a scheme involves the power to solve several mathematical intangible equations. Ad hoc solutions may be obtained but often fail to address various involved issues.