To achieve the objective of this study an experimental programme with small diameter cores to investigate the influence of various parameters was carried out to examine the results in relation to the recommended procedures was carried out. Three different concrete mixtures were produced. Ordinary Portland cement, natural aggregate, river sand were used in concrete mixtures. Aggregates of 20 mm maximum size were used in these concrete mixturesDifferent length to diameter (L/d) of 2, 1.
5, and 1 with and without reinforcements were used. 12 mm diameter bars are used in this present study to evaluate the compressive strength of concrete. The cores with having diameter of 300 mm, 225 mm, 150 mm were cast and moist cured under laboratory conditions until being tested. The compressive strength test results are the average of six specimens. Totally 192 core sample specimens were tested in this experimental investigation. Ultrasonic pulse velocity (UPV) tests were carried out on the specimens prepared by three different concrete mixtures.In neural network analyses, the problem can be defined as a non-linear input-output relation between the influencing factors (L/d) ratio, grade of concrete and orientation of the bar) and compressive strength values.
Numbers of trials were carried out in the MATLAB neural network toolbox environment for the determination of hidden neuron number of the hidden layers. The general architecture of MATAB programme is shown in Fig no: 1.
