In order to check the applicability of Artificial Intelligent (AI) techniques to act as reliable inverse models to solve the multi-input/multi-output heat flux estimation classes of inverse heat transfer problems (IHTPs), in a newly reconstructed experimental setup, a two-input/two-two output (TITO) heat flux estimation problem was defined in which the radiation acts as the main mode of thermal energy. A simple three-layer perceptron Artificial Neural Network (ANN) was designed, trained, and employed to estimate the input powers (represent emitted heats-heat fluxes from two halogen lamps) to irradiative batch drying process. To this end, different input power functions (signals) were input to the furnace/dryer's halogen lamps, and the resultant temperature histories were measured and recorded for two different points of the dryer/furnace. After determining the required parameters, the recorded data were prepared and arranged to be used for inverse modelling purposes. Next, an ANN was designed and trained to play the role of the inverse heat transfer model. The results showed that ANNs are applicable to solve heat flux estimation classes of IHTPs.

There are two major approaches in sequential (real-time) heat flux estimation problems using measured temperatures: (i) development of inverse heat transfer models that directly estimate heat flux and (ii) use of a combination of a direct heat transfer model (which estimates temperature using heat flux information) and an optimization algorithm. In physics-based solutions, using thermodynamics and heat transfer laws, the first approach is considered ill-posed and challenging, and the second approach is more popular. However, the use of artificial intelligence (AI) techniques has recently facilitated heat transfer inverse modelling, even for complex irradiative systems. Many of the claimed advantages of AI inverse models of irradiative systems result from the use of AI techniques rather than the inverse modelling approach. This research presents a rational comparison between the aforementioned approaches for an irradiative thermal system, both using AI techniques, for the first time. The results show that inverse models are superior because of their higher accuracy and shorter estimation delay time.

In this research, a hybrid control system is proposed to address the temperature control of an infrared dryer. The control system includes a feedback-predictive controller and a neural network steady state control law. The feedback-predictive controller outputs the amplified value of the predicted error as the transient control command. The predictive model was employed to suppress the undesirable effect of the dead-time of the system. A multilayer perceptron was designed and tested based on a control equilibrium point and steady state control to be used as a feedforward controller. The stability of the control system in a continuous domain was proved with no limit on the amplification gain of the predictive-feedback controller. In other words, there is no concern about losing stability with accelerating convergence towards the reference. The entire control system was constructed in Simulink and compiled to a C code and applied on the experimental setup. Experimental results are outstanding in comparison with the results of an interactively tuned IMC-based PID controller.

The present work has focused on a comparison between commonly employed artificial neural networks (ANNs) in engineering applications to identify the most efficient ANN for the inverse modelling of an irradiating furnace/ dryer in terms of accuracy and computing time. To this end, several ANNs were designed, trained and employed to estimate the heat emitted during the irradiative batch drying process with the aid of NeuroSolution (R).As part of the study, different ANNs were designed and trained to play the role of the inverse heat transfer model. The reasons for exploiting these ANNs were derived from various studies in the literature, in which ANNs were employed for engineering modelling purposes. The results showed that the multiple layer perceptron (MLP) with the Levenberg-Marquadt (LM) in the back propagation (BP) was the best ANN among the methods evaluated to solve the inverse heat estimation problems used in irradiative batch drying processes. An important advantage of the ANN method in comparison with the classical inverse heat transfer modelling approaches is that a detailed knowledge of geometrical and thermal properties of the system (such as wall conductivity, emissivity, etc.) is not required. Such properties are difficult to measure and may undergo significant changes during the temperature transient mode.In this study, genetic algorithms (GAs) have been employed to determine the key parameters of the employed ANNs. These parameters are normally found heuristically or by a trial and error brute force process. The results demonstrate that the aforementioned parameters may be estimated much more accurately and faster by the GA method. The performance of the networks has been improved as well and the number of required hidden layers has been discovered using a non trial-error method, which eliminates time-consuming repeating procedures and produces more accurate results. (C) 2014 Elsevier Ltd. All rights reserved.

In this work, a variety of new approaches are developed and results are compared for solving inverse heat transfer problems where radiation is the dominant mode of thermal energy transport. An artificial neural network (ANN), two hybrid methods of genetic algorithms and artificial neural networks (GA–ANNs), and an adaptive neuro-fuzzy inference system network (ANFIS) were designed. These were trained and then employed to estimate the required input power in an irradiative batch drying process. A comparison of the results shows that the most accurate method is ANFIS but the number of parameters in ANFIS is larger than ANNs. Consequently, the ANFIS solution is time consuming in this application; however other neuro-fuzzy techniques may require fewer parameters and these will be considered in future studies. For the studied ANNs, the hybrid method of GA–ANN is optimal using the Levenberg–Marquardt optimization algorithm during back propagation in terms of accuracy and network's performance.

In this paper, for the purpose of temperature control of a batch dryer, a PID controller is designed using genetic algorithm (GA) so that minimises performance(fitness) function which involves controller energy consumption and change in control input in three different operation areas. Having a varsity of operation areas in tunning and testing the controller in an area different not used during tunning procedure makes the proposed designed controller more reliable then the controller is implemented on the system through an I/O card and Real-Time windows target toolbox of MATLAB software. The merit of the designed GA-based PID controller is indicated compared to IMC-based PID.

In this paper, Artificial Neural Networks (ANN), as an intelligent technique, is employed for modelling of batch drying processes where the radiation is the dominant mode of heat transfer.The case study of this research is an infrared dryer. First, a data collection loop is designed by MATLAB/simulink software package, and then this loop is connected to the experimental setup (the dryer) through an I/O card via Real Time Windows Target (RTWT) toolbox of MATLAB. A variety of input signals (input voltages to the halogen lamp of the dryer) are applied and their corresponding temperature history of a point on the bottom surface of the dryer is recorded. After estimating the order, the appropriate sampling time and the dead-time, the recorded data are arranged for system identification purposes. Then, an artificial neural network (ANN) is designed for the furnace system and trained. In conclusion, the obtained results, after checking the trained neural network, show that the methodology of ANN is applicable to predict the behaviour of the batch drying process for several minutes very accurately. Moreover, in comparison with classical heat transfer modelling approaches, the proposed method does not require any knowledge about the mechanical properties of the system like conduction or emissivity coefficients which are usually inaccurate and subject to change. Furthermore, after off-line training process, the designed ANN model can predict the temperature much faster than classical heat transfer methods due to much less required computation.

In this research, a fuzzy knowledge-base controller is designed for yaw control of model helicopter. At the next stage, an adjusting algorithm is presented to reduce the influence of high inertia on fuzzy controlled systems. Inertia may cause significant overshoot, which is undesirable and difficult to eliminate. In order to solve this problem, a simple algorithm is presented to reduce the control input by adjusting the fuzzy controller parameters while the system is getting close to the desired condition. Implementing this approach (including a lateral algorithm to reset the parameters in special conditions) for yaw angle control of a model helicopter reduces the overshoot and energy consumption considerably without significant decrease of the settling time.

Heat flux function estimation problems are inverse heat conduction problems which heat flux functions (boundary or initial conditions) are the unknowns and temperature distribution (at present and earlier times) is available. There are several algorithms for solving this type of IHC problems. In all of these methods the computational cost is very heavy and all of common IHCP algorithm requires finding the solution of the direct heat conduction problem numerous times. In this paper the neural networks is utilized to estimate the "filter coefficients" needed to estimate heat flux in a particular system. In developing the training phase of the network inspiration is drawn from the Burgraff's exact solution of the IHCP as well as the filter method. Thus, the estimation phase neither requires any temperature field nor the sensitivity coefficients calculations which are common in classical methods. The neural network used in this work is a 2-layer perceptron. It is shown via classical triangular heat flux test cases that the method can yield very accurate, very efficient as well as stable estimations.