ANN needs more data values as much as possible. Even the accuracy of data training depends on number of data values. In the perspective of simplicity and adaptability fuzzy can be selected. Nevertheless for the accuracy ANN may be the best with higher number of trials.
From the point of minimal number of experimental data's, fuzzy logic will be the ultimate choice. Fuzzy logic is used to reduce the fuzziness in the experimental data's. The output of the fuzzy logic mainly depends on the membership function chosen and rules framed to reduce the fuzziness. Framing the rules depends on the experience gained in that area, as you are a experienced researcher in composite machining the task of framing rules will be easier.
On the other hand, Artificial neural network requires large number of data's to train the model and also dependent on the parameter values (learning rate and momentum coefficient) used during training.
According to your question Fuzzy Logic vs. Artificial Neural Network
I also think in the way of the common point of view. Fuzzy Logic.
I hope you will extend your question in way of how much or how many.
I do not know. I would be happy to learn the researchers ideas, experiences and also publications in the way of how much and how many (number of experiments are small).
What is the definition of small in here? How small. Any range, any threshold and so forth.
Dear Sir,, In the of event of less no of exptl details, fuzzy logic method for modeling the experiment in 3D. i presume less than 10 results is small according to you?
Fuzzy is better for less no of experiments. The reason behind is that fuzzy makes definite decisions with ambiguous data, whether the data size is big or small. However, to minimize the error in fuzzy modeling, the no of experiments needed is to be optimised based on repeated trials only.
But, ANN needs more number of experiments comparitively as it relies completely on the training process. Less no of experiments mean less training in case of irregular data, which in turn leads to conflicting results.
If number of experimental data pieces is low you can use additionally expert knowledge in form of fuzzy rule base. Then probability of achieving better results inproblem solvuing is higher.
It dependes in what type of fuzzy logic or ANN uses. Also depend of several factors. One of the principal factors is the modelling phase.
the selction of the model depends in what do you want?
for instance you can read the following papers:
Montes Dorantes P.N. , Hernández García H.M., de la Rosa Elizondo J., Méndez G.M., Nieto González J.P. (2013). Sistemas difusos para monitoreo y control de metalurgia secundaria. Memorias del congreso internacional de metalurgia y materiales. Congreso 35. 1(1), pp. 354-363. ISSN: 2007-9540
Available at: http://its.mx/2014/pdf/35CIMM.pdf
Montes Dorantes P.N., Jiménez Gómez M.A., Cantú Rodríguez X., Méndez G.M. (2014). A comparative Study of Type 1 Singleton Fuzzy Logic Systems in Machining Application. Research in Computing Science 82, pp. 107-119.
Available at: http://www.rcs.cic.ipn.mx/rcs/2014_82/A%20Comparative%20Study%20of%20Type%201%20Singleton%20Fuzzy%20Logic%20Systems%20in%20Machining%20Application.pdf
Montes Dorantes P.N., Jiménez Gómez M.A, Méndez G.M. , Nieto González J.P., de la Rosa Elizondo J. (2015). One step models for soft computing techniques. Industrial application to image processing in quality assurance process. International Journal of advanced Manufacturing technology (IJAMT, Springer), 81(5), pp.771-778. DOI: 10.1007/s00170-015-7101-7
Available at: http://link.springer.com/article/10.1007/s00170-015-7101-7
P. N. M. Dorantes and G. M. Méndez. (2015). Non-iterative Radial Basis Function Neural Networks to Quality Control via Image Processing, IEEE LATIN AMERICA TRANSACTIONS, 13(10), pp. 3457-3451.
Available at: http://www.ewh.ieee.org/reg/9/etrans/ieee/issues/vol13/vol13issue10Oct.2015/13TLA10_37MontesDorantes.pdf
in personal experience the modelling phase is the most important factor to do a system with good results with limited number of experiments.