Ghodratollah Talebnia1 - Department of Accounting, Olom v TahghighSat Branch, Islamic Azad University,Tehranan, Iran
Kamal Zareimoravej2 - Department of Accounting, Hamedan Branch, Islamic Azad University, Hamedan, Iran
Mohamadmehdi Shakori3 - Department of Accounting, Hamedan Branch, Islamic Azad University, Hamedan, Iran

Neurotic network is a machine for modeling human mind. Data - processing systemin human mind is non-linear, parallel and highly complex. Therefore, neuroticnetwork is a parallel-distributed processing system which has been made of simpleprocessing unites called neuron. This network accepts a series of inputs and then doesa learning process based on weights of neurotic network and finally chooses the bestoutputs among the inputs(1). While calculating the Economic Value Added (EVA),we need through information of basic financial statements along with their relatedexplanatory notes (2). In this research, two models of neurotic networks have beenused to approximate the relationship between indexes pertaining to structure of thecapital and Economic Value Added (EVA). After examining both models, it wasrevealed that the multi-layer Prespetron model, comparing with the neurotic networkof radiant-base functions, has better performance in calculating the desired functionand providing more efficient data. In the presented models in this research, thesignificant relationship between the structure of the capital and Economic ValueAdded (EVA) is proven by acquisition of a calculating function with %1 error rate.The acquired results depict that the relationship between the second variable and EVAis stronger than the relationship between the first variable and the third one. Bymaking use of both variables we can have a more accurate calculation of EVAallowing %1 error rate. Of course, it must be noted that the more input presentationwill lead to less erroneous system

Capital Structure; Economic Value Added; Multi-layer PrespetronNeurotic Network