It is desirable to know the characteristics inherent to the structure, so that the best train can be selected from the numerous distinct gear trains with the same number of links and degree of freedom. To achieve this, quantitative measures are developed in a simple way, using some of the principles of genetic process. These can be used for identifying a distinct train and to know the general characteristics of a gear train like speed ratio, transmission efficiency, etc.

Generation of epicyclic gear trains up to eight links has been reported by many investigations (Buchsbaum and Freudenstein, 1970; Ravisankar and Mruthyunjaya, 1985; Tsai, 1989; and Hsu and Hsu, 1997). In order to isolate the distinct trains, all the methods need a test for isomorphism. This has got more significance when the designer is able to compare the numerous distinct gear trains with the same number of links and degree of freedom (df). With the increasing application of neural networks in engineering and realizing its potential and lack of application in the area of kinematics, an attempt was made by Srinivasa Rao et al. (2010) to develop an algorithm for the structural comparison of in-parallel robotic manipulators. The present work extends the above algorithm to gear trains, taking care of the fact that the graphs representing gear trains contain edges of different types unlike in linkages.

Numerical strings are proposed, which will test both displacement and rotational isomorphism uniquely with least effort; also with the fitness, a concept that emerges from the genetic algorithm, distinct gear trains can be compared.

Mechanical Engineering Journal, Exit Blade Angles, Centrifugal Pump, Laser Doppler Velocimeter, ANSYS Turbo Grid, Power Consumption, Engine Oil, Software ANSYS, Kinematic Viscosities, Pump Vendor Catalogs.