The use of neural networks methodology is not as common in the investigation and prediction noise as statistical analysis. The application of artificial neural networks for prediction of power tiller noise is set out in the present paper. The sound pressure signals for noise analysis were obtained in a field experiment using a 13-hp power tiller. During measurement and recording of the sound pressure signals of the power tiller, the engine speeds and gear ratios were varied to cover the most normal range of the power tiller operation in transportation conditions for the asphalt, dirt rural roads, and grassland. Signals recorded in the time domain were converted to the frequency domain with the help of a specially developed Fast Fourier Transform (FFT) program. The narrow band signals were further processed to obtain overall sound pressure levels in A-weighting. Altogether, 48 patterns were generated for training and evaluation of artificial neural networks. Artificial neural networks were designed based on three neurons in the input layer and one neuron in the output layer. The results showed that multi layer perceptron networks with a training ALGORITHM of BACK PROPAGATION were best for accurate prediction of power tiller overall noise. The minimum RMSE and R2 for the four-layer perceptron network with a sigmoid activation function, Extended Delta-Bar-Delta (Ext. DBD) learning rule with three neurons in the first hidden layer and two neurons in the second hidden layer, were 0.0198 and 0.992, respectively.

BACK PROPAGATION (BP) ALGORITHM is a systematic method for training multi-layerneural networks, which, despite many successful applications. also has many drawBACKs. For complex problems, BACK PROPAGATION may requie a long time to train the networks and it is possible that no training occurs at all. Long train in time can be the result of non-optimal parameters. It is not easy to choose an appropriate value for the parameters of a particular problem and the parameters are usually determined by rail and error. If the parameters are not chosen appropriately, slow convergence paralysis and continuous instability can result [1-4]. Moreover, the best values for the parameters at the beginning of training may not be good enough later. In this paper A technique has been incorp rated into BP ALGORITHM for adaptation of steepness parameter and momentum factor in order to achieve a higher rate of convergence. Through interconnection of Fixed Structure Learn in Automata (FSLA) to the feed forward neural networks. Learning automata scheme is applied in order to adjust these parameters based on the observation of random response of neural networks. The main motivation in using learning automata as an adaptation ALGORITHM is in is capability of global optimization when dealing with multi-modal surfaces. The feasibility of he proposed method is shown through simulations on three learning problems: exclusive-or encoding problem and digit recognition. These problems are chosen because they have different error surfaces and collectively present an environment that is suitable to determine the effect of the proposed method. The simulation results show that the adaptation of these parameters using his method increases not only the convergence rate of learning but also the likelihood of escaping the local minima. Computer simulations provided in this paper indicate that at least a magnitude of savings in running time can be achieved when FSLA is used for the adaptation of momentum factor and steepness parameters. Furthermore simulations demonstrate that the FSLA approach performs much better than the Variable Structure Learning Automata (VSLA) approach reported in [1,2].

The concept of BACK wave PROPAGATION is developed as an inversion method to estimate acoustic parameters of the tissue and material from measurements of the acoustic field for a known source-receiver. A phase-regulated technique is introduced to measure the sensitivity of the BACK-wave PROPAGATION inversion method for estimating weakly sensitive acoustic model parameters. The paper demonstrates theoretically that the sensitivity can be measured by a sensitivity factor using the phase-regulation procedure. The paper also demonstrates that the spatial resolution of BACK propagated signal energy that is focused at the known source location is increased when sensitivity factor increases. This result leads to the definition of a criterion based on the spatial distribution of the signal energy around the source location. The criterion is formulated based on the spatial variance of the BACK propagated pressure field in a windows around the known source location. Use of the BACK wave PROPAGATION inversion method is demonstrated for estimating the acoustic model parameters, including type of materials, tissues, water depth, compressional speed of the sediment layer, and mineral density as poro-elastic media.

A method is used to obtain the fundamental frequency of a retaining wall quite accurately and carry out a dynamic analysis of such wall based on modal response technique. The present procedure establishes both the general and particular cases of dynamic response of retaining wall based on improved Rayleigh-Ritz method. The wall will be assumed to be a flexural member. The fundamental frequency of the retaining wall with soil mass has been computed. The results based on proposed method are then used to BACK PROPAGATION neural network (BPN). In the present work, the fundamental frequency of a retaining wall is calculated by BPN. A significant benefit of BPN is its ability to learn relationships between variables with repeated exposure to those variables. Therefore, instead of deriving an analytical relationship from mathematical formulations, the BPN learns the relationship through an adaptive training process. Numerical example shows the merit of the BPN.