Search

This paper aims at introduction, design, and validation of a temperature estimation algorithm for an infrared dryer. The proposed algorithm estimates the temperature at one or some points in a thermal system (e.g., an infrared dryer) based on the measured temperature at a number of other points. In this research, the designed algorithm estimates the temperature of a single point; however, the methodology can be evidently extended to multiple points. Inspired by direct and inverse heat transfer models, a mathematical model is presented for this purpose. This model is developed and identified using artificial neural network (ANN) technique and laboratory experimental data. The proposed method exhibits excellent accuracy with no need to thermo-physical properties of the system. [ABSTRACT FROM AUTHOR]

This paper proposes a vibration-based fault-diagnosis method for mechanical parts. This method, after algorithm development, only requires a single inexpensive test to inspect the part which could take as short as half a second. The algorithm is developed in three major stages, (i) exciting specimens without or with known faults using a controlled force and recording acceleration of a single point for a short time (ii) finding a signature for each faulty specimen, using Fourier transform and statistical analysis. (iii) Developing a multi-layer perceptron, as a mathematical model, using the results of stage (ii). The elements of a part signature are the inputs to the model. The location (and possibly size and shape factor) of the fault is model output. Stage (i) can be performed experimentally or alternatively with a validated FEM, one experiment or simulation per specimen. The proposed technique was examined to locate (isolate) a fault on an automobile cylinder head. The presented accuracy is considerable, and the data collected at fairly low frequency range (below 1200 Hz) were found to be sufficient for this technique. In the case study of this paper, possible fault locations are on a line; as a result, fault location has one dimension. It is shown that the technique can be extended to higher dimensions. [ABSTRACT FROM AUTHOR]

Abstract: This paper deals with modeling a power plant component with mild nonlinear characteristics using a modified neural network structure. The hidden layer of the proposed neural network has a combination of neurons with linear and nonlinear activation functions. This approach is particularly suitable for nonlinear system with a low grade of nonlinearity, which can not be modeled satisfactorily by neural networks with purely nonlinear hidden layers or by the method of least square of errors (the ideal modeling method of linear systems). In this approach, two channels are installed in a hidden layer of the neural network to cover both linear and nonlinear behavior of systems. If the nonlinear characteristics of the system (i.e. de-superheater) are not negligible, then the nonlinear channel of the neural network is activated; that is, after training, the connections in nonlinear channel get considerable weights. The approach was applied to a de-superheater of a 325MW power generating plant. The actual plant response, obtained from field experiments, is compared with the response of the proposed model and the responses of linear and neuro-fuzzy models as well as a neural network with purely nonlinear hidden layer. A better accuracy is observed using the proposed approach. [Copyright &y& Elsevier]

Highlights: [•] A grounded-load digital charge-based displacement estimator (GDCDE) is presented. [•] An artificial neural network (ANN) was also designed to estimate the piezoelectric tube's displacement using the piezoelectric voltage at low frequencies of excitation. [•] A complementary filter combines the proposed GDCDE and the ANN to estimate displacement over a wide bandwidth and to overcome drift. [•] The discrepancy between the desired and estimated displacement is fed back to the piezoelectric actuator using proportional control. [Copyright &y& Elsevier]