Abstract
Recent advances in the applications of ANN have demonstrated successful cases in time series analysis, data mining, civil engineering, financial analysis, music creation, fishing prediction, production scheduling, intruder detection, etc., making them an important tool for research and development[1]. ANN and evolutionary computation(EC) techniques have been employed successfully in solving real-world problems including those with a temporal component[2]. In another work[3], a hybrid method based on a combination of evolutionary computation and neural network(NN) has been used to predict time series. In the world of databases, various ANN-based strategies have been used for knowledge search and extraction[4]. Intelligent neural systems have been constructed with the aid of genetic algorithm-based EC techniques and these systems have been applied in breast cancer diagnosis[5]. Genetic algorithms(GA) have been applied to develop a general method of selecting the most relevant subset of variables in the field of analytical chemistry to classify apple beverages[6]. New ANN methods enable civil engineers to use computing in different ways. Besides as a tool in urban storm drainage[7], ANN and Genetic Programming(GP) have been implemented in the prediction and modelling of the flow of a typical urban basin [8]. In the latter case, it was shown that these two techniques could be combined in order to design a real-time alarm system for floods or subsidence warning in various types of urban basins. ANN models for consistency, measured by slump, in the case of conventional concrete have also been developed[9]. In a time series prediction of the quarterly values of the medical component of the Consumer Price Index(CPI), the results obtained with both neural and functional networks have been shown to be quite similar[10]. Dimensionality reduction, variable reduction, hybrid networks, normal fuzzy and ANN have been applied to predict bond rating[11]. A recent online survey through the ISI Web of Knowledge using keywords such as “ANN” and “thermal design” would reveal only ten relevant SCI publications[12]. In the area of food processing, ANN was used to predict the maximum or minimum temperature reached in the sample after pressurization and the time needed for thermal re-equilibration[13]. The accurate determination of thermophysical properties of milk is very important for design, simulation, optimization, and control of food processing such as evaporation, heat exchanging, spray drying, and so forth. Generally, polynomial methods are used for prediction of these properties based on empirical correlation to experimental data. However, it was found that ANN presented a better prediction capability of specific heat, thermal conductivity, and density of milk than polynomial modeling and it was suggested as a reasonable alternative to empirical modeling for thermophysical properties of foods[14]. Numerical simulation of natural circulation boiling water reactor is important in order to study its performance for different designs and under various off-design conditions. It was found that very fast numerical simulations, useful for extensive parametric studies and for solving design optimization problems, can be achieved by using an ANN model of the system[15]. ANN models and GA were applied for developing prediction models and for optimization of constant temperature retort thermal processing of conduction heating foods[16]. ANN technique has been used as a new approach to determine the exergy losses of an ejector-absorption heat transformer (EAHT)[17]. The results show that the ANN approach has the advantages of computational speed, low cost for feasibility, rapid turnaround, which is especially important during iterative design phases, and easy of design by operators with little technical experience. Computational fluid dynamics approach is often employed for heat transfer analysis of a ball grid array(BGA) package that is widely used in the modern electronics industry. Owing to the complicated geometric configuration of the BGA package, an ANN was trained to establish the relationship between the geometry input and the thermal resistance output[18]. The results of this study provide the electronic packaging industry with a reliable and rapid method for heat dissipation design of BGA packages. Thermal spraying is a versatile technique of coating manufacturing implementing large variety of materials and processes. An ANN was developed to relate processing parameters to properties of alumina-titania ceramic coatings[19]. Predicted results show globally a well agreement with the experimental values. It can be seen that applications of ANN in thermal design is scarce and this article aims to explore the application of an ANN in gas-fired cooktop burner design.
TopIntroduction
Recent advances in the applications of ANN have demonstrated successful cases in time series analysis, data mining, civil engineering, financial analysis, music creation, fishing prediction, production scheduling, intruder detection, etc., making them an important tool for research and development[1]. ANN and evolutionary computation(EC) techniques have been employed successfully in solving real-world problems including those with a temporal component[2]. In another work[3], a hybrid method based on a combination of evolutionary computation and neural network(NN) has been used to predict time series.
In the world of databases, various ANN-based strategies have been used for knowledge search and extraction[4]. Intelligent neural systems have been constructed with the aid of genetic algorithm-based EC techniques and these systems have been applied in breast cancer diagnosis[5]. Genetic algorithms(GA) have been applied to develop a general method of selecting the most relevant subset of variables in the field of analytical chemistry to classify apple beverages[6]. New ANN methods enable civil engineers to use computing in different ways. Besides as a tool in urban storm drainage[7], ANN and Genetic Programming(GP) have been implemented in the prediction and modelling of the flow of a typical urban basin [8]. In the latter case, it was shown that these two techniques could be combined in order to design a real-time alarm system for floods or subsidence warning in various types of urban basins. ANN models for consistency, measured by slump, in the case of conventional concrete have also been developed[9]. In a time series prediction of the quarterly values of the medical component of the Consumer Price Index(CPI), the results obtained with both neural and functional networks have been shown to be quite similar[10]. Dimensionality reduction, variable reduction, hybrid networks, normal fuzzy and ANN have been applied to predict bond rating[11].
A recent online survey through the ISI Web of Knowledge using keywords such as “ANN” and “thermal design” would reveal only ten relevant SCI publications[12]. In the area of food processing, ANN was used to predict the maximum or minimum temperature reached in the sample after pressurization and the time needed for thermal re-equilibration[13]. The accurate determination of thermophysical properties of milk is very important for design, simulation, optimization, and control of food processing such as evaporation, heat exchanging, spray drying, and so forth. Generally, polynomial methods are used for prediction of these properties based on empirical correlation to experimental data. However, it was found that ANN presented a better prediction capability of specific heat, thermal conductivity, and density of milk than polynomial modeling and it was suggested as a reasonable alternative to empirical modeling for thermophysical properties of foods[14].
Numerical simulation of natural circulation boiling water reactor is important in order to study its performance for different designs and under various off-design conditions. It was found that very fast numerical simulations, useful for extensive parametric studies and for solving design optimization problems, can be achieved by using an ANN model of the system[15]. ANN models and GA were applied for developing prediction models and for optimization of constant temperature retort thermal processing of conduction heating foods[16]. ANN technique has been used as a new approach to determine the exergy losses of an ejector-absorption heat transformer (EAHT)[17]. The results show that the ANN approach has the advantages of computational speed, low cost for feasibility, rapid turnaround, which is especially important during iterative design phases, and easy of design by operators with little technical experience.
Computational fluid dynamics approach is often employed for heat transfer analysis of a ball grid array(BGA) package that is widely used in the modern electronics industry. Owing to the complicated geometric configuration of the BGA package, an ANN was trained to establish the relationship between the geometry input and the thermal resistance output[18]. The results of this study provide the electronic packaging industry with a reliable and rapid method for heat dissipation design of BGA packages. Thermal spraying is a versatile technique of coating manufacturing implementing large variety of materials and processes. An ANN was developed to relate processing parameters to properties of alumina-titania ceramic coatings[19]. Predicted results show globally a well agreement with the experimental values.
It can be seen that applications of ANN in thermal design is scarce and this article aims to explore the application of an ANN in gas-fired cooktop burner design.
Key Terms in this Chapter
Designed of Experiments: A set of tests conducted under controlled conditions in which multiple levels of a set of factors are manipulated and the resulting response(s) of a system or process is measured or observed.
Multilayer Perceptrons (MLPs): They are feedforward neural networks trained with the standard backpropagation algorithm. They are supervised networks so they require a desired response to be trained. They learn how to transform input data into a desired response, so they are widely used for pattern classification. With one or two hidden layers, they can approximate virtually any input-output map.
Backpropagation: A supervised learning technique used for training artificial neural networks. It is most useful for feed-forward networks (networks that have no feedback, or simply, that have no connections that loop). The term is an abbreviation for “backwards propagation of errors”. Backpropagation requires that the transfer function used by the artificial neurons (or “nodes”) be differentiable.
Response: The (dependent) variable(s) measured or observed that is the result of a test conducted at a specific set of factor levels.
Multiple Correlation, R2: the percent of variance in the dependent variable explained collectively by all of the independent variables.
Levels: The sets of values for each factor.
Reynolds Number(Re): the ratio of inertial forces to viscous forces and consequently it quantifies the relative importance of these two types of forces for given flow conditions. Thus, it is used to identify different flow regimes, such as laminar or turbulent flow.
Full Factorial Design: This design allows a designer to adequately quantify a response with a reasonable number of tests. In general, full factorial designs require three levels for each factor thus allowing one to evaluate second order models.
Factors: The set of (independent) variables that are believed to affect the response of a system or process.