Hybrid System for Ship-Aided Design Automation
Maria Meler-Kapcia · InTech eBooks · 2011
IntroductionIn carrying out the most important tasks of a shortage of ship design is the lack of formalized application methods, mathematical models and advanced computer support.Decisions and adopted solutions are often based on knowledge resulting from experience and intuition of designers.Use of information on previously executed projects of similar ships allow expert systems using the Case Based Reasoning method (CBR), which is a relatively new way of solving problems related to databases and knowledge bases.This facilitates the efficient design of the ship as soon as possible [1].A similar role has been neural networks, which can be taught on the basis of representative examples, and the results obtained from other sources (eg during the operation of the ship).To achieve this purpose, a hybrid support system for ship design based on the methodology of CBR with some artificial intelligence tools such as expert system Exsys Developer along with fuzzy logic, relational Access database, and artificial neural network with backward propagation of errors.Hybrid systems forming a new class of artificial intelligence tools have to combine the capabilities of each of the tools used to solve specific problems.In the simplest case of the hybrid system is a combination of classical techniques of expert systems with neural networks [2], which was applied in developed computer-aided design system.This system is intended to be ship-aided design automation, where most projects are using pre-built similar ships.The scope of the system, in addition to computer-aided design automation, were also aided design of ships at the initial stage, which determines the main parameters of the ship.In order to find solutions to similar, previously used on ships developed its own algorithm for multiobjective optimization of weighted gains to search a database of similar ships [13].The proposed algorithm was applied to computer-aided design ship's engine room automation, where the similarity may be of partial for example main propulsion (MP), power plants and individual installations, and the weighted sum of partial similarities is the similarity summary of the whole ship.Using this algorithm, the selection is made of methods for calculating the similarity presented in the literature, adapted to the design of ships and their own methods, has not been used, based on the use of functions: rectangular, trapezoidal, triangular and Gaussian [13].Using these methods, similarity analysis was conducted for the selection of ships, power and speed of main engine (ME).This analysis is intended comparison of selected methods and the selection of the best of them for computer-aided design automation engine room in the database application and expert system.Based on the results obtained are searched in a database similar ships, ie ships with the greatest similarity weighted summary.In the case of www.intechopen.comExpert Systems for Human, Materials and Automation 260 unsatisfactory results in the calculation of similarity, as a complementary, provided for a neural network learning algorithm.This algorithm is implemented in the system of Access and can be used for the selected database and its fields of any numeric type. How to referenceIn order to correctly reference this scholarly work, feel free to copy and paste the following: