Estimation of ground enhancing compound performance using Artificial Neural Network

Artificial neural network methodology for the estimation of ground enhancing compounds resistance

“…This work is a continuation and expansion of the works presented in [22, 23]. It is a first attempt to forecast the behaviour and the performance of such materials over time, as their importance is great for the improvement of soil electric characteristics and the decrease of ground resistance, all over the world.…”

“…ANNs are programmed computational models that aim to replicate the function of the human brain. They have gained wide acceptance due to their features that include: solving complex problems, identifying non‐linear relationships among data that are known to be difficult to model using classical methods, ability to generalise and learn (produce adequate responses to unknown situations), and capability of greater fault tolerance [23]. For this reason, ANNs evolved as a quite useful and easy to handle tool of artificial intelligence for the approximation of relationships among quantities, that otherwise would have been difficult to model.…”

“…Considering and filtering all the training techniques of BP algorithm that have been tested and studied in various ANN structures in previous research works [21–24], this paper focuses on the training algorithms (Table 1) which present more interest, convenience, speed and best results for the ANN training on this particular problem. For each training algorithm, the respective parameters of the network are optimised through successive trials, for example, the number of neurons in the hidden layer or the optimum combination of activation functions for the hidden and output layers.…”

“…So far, researchers such as Salam et al [18], Amaral et al [19] and Gouda et al [20] have successfully used ANNs attempting to model the correlation among ground resistance, soil resistivity and length of buried electrodes in soil, injection current frequency and peak current. This paper is an extension to the research presented in [21–24], in an effort to consider concisely all the previous work on ANN modelling and experiment on a modified architecture of a greater number of input and output variables, tested under faster training algorithms.…”

Research and applications of ground enhancing compounds in grounding systems

“…The influence of the additional compound's volume has been examined in [53,58] by using finiteelement method analysis in a simulation software, while Chen et al [54] and Al-Arainy et al [55][56][57] have employed analytical mathematical expressions. Considering the high non-linearities in the ground resistance-soil and/or compound resistivity relationship, some researchers turned to computational intelligence techniques, employing ANNs, wavelet neural networks, inductive machine learning and genetic programming [59][60][61][62][63][64]. These methods approach the underlying function, comprising possible nonlinearities, with great accuracy, through consecutive training and testing of the networks, with a way that is not, actually, clear and well known.…”