In spite of continuing advances in optimal solution techniques for optimization and control problems, many practical combinatorial problems remain too large or too complex to be solved by these known techniques. Thus, a heuristic approach (Neural Network Model) is often the only viable alternative. Neural Network Models offer the most unified approach to building truly intelligent systems which can provide good optimal solution for many applications. In this work we propose a hybrid (Kalkoh) neural network algorithm which is being used to model and solve the continuous stirred tank mixer (CSTM) problem. The hybrid algorithm is robust and converges fast without being trapped into a local minimal as is the case with the popular back-propagation neural network. The characteristic equations governing the dynamics of the Continuous Stirred Tank Mixer/Reactor and the controller were formulated and tested and found to be consistently stable.
Genetic Algorithms are one of the most powerful techniques in optimization and search problems. The Objective was, to understand this powerful technique and to explain it in better way so as to expand its field of application.The algorithm is so versatile that it can be used in any field.The objective of the GA is to find an optimal solution to a problem. Since Gas is heuristic procedures, they are not guaranteed to find the optimal solution, but they are able to find very good solutions for a wide range of problems.The use of both, genetic algorithms and artificial neural networks, was originally motivated by the astonishing success of these concepts in there biological counterparts.Despite their totally deferent approaches, both can merely be seen as optimization methods, which are used in a wide range of applications.They are capable to finding solution to hard NP-based Problems. Neural Networks utilizing back propagation based learning have promisingly showed results to a vast variety of function and problems. TSP is one such classical problem for theoretical computation.
Many problems arising in science and engineering aim to find a function which is the optimal value of a specified functional. Some examples include optimal control, inverse analysis and optimal shape design. Only some of these, regarded as variational problems, can be solved analytically, and the only general technique is to approximate the solution using direct methods. Unfortunately, variational problems are very difficult to solve, and it becomes necessary to innovate in the field of numerical methods in order to overcome the difficulties. The objective of this PhD Thesis is to develop a conceptual theory of neural networks from the perspective of functional analysis and variational calculus. Within this formulation, learning means to solve a variational problem by minimizing an objective functional associated to the neural network. The choice of the objective functional depends on the particular application. On the other side, its evaluation might need the integration of functions, ordinary differential equations or partial differential equations. As it will be shown, neural networks are able to deal with a wide range of applications in mathematics and physics.
Advanced Manufacturing Technology, deals with high flexibility, high speed, least waste and computer based soft-computing techniques and application of hybridized artificial intelligence. Organizations continually strive to achieve their goal, but the contention in any manageable system is limited in achieving more of its goal by very small number of constraints. Theory of Constraints represented as linear programming mathematical model provide an explicit solution. The constraints in the Product-Mix problem are further explored and exploited, which contain fuzzy, imprecise, vague and uncertain parameters of the system, are optimized using fuzzy numbers, rough sets and evolutionary algorithms. Data model using the artificial neural networks, the fusion of neural networks and fuzzy adapt to the imprecise and fuzzy parameters is demonstrated. Flexible Manufacturing System has been tested and verified for the product mix problem, and the ease of implementation show the usefulness to obtain an optimal solution and decision making in real world problems. The paradigm shift in the aspect of manufacturing, soft computing is a road to model the Factory of the Future.
The learning capabilities of artificial neural networks (ANNs) to identify and emulate the behavior of complicated nonlinear systems have made them effective tools that can be utilized in intelligent adaptive control strategies. The use of ANNs in the design of trajectory tracking controllers for robotic manipulators is dated back to the 1980s. Due to the flexibility of their structure as well as the continuous development and enhancement of their self-training algorithms, the use of ANNs in the field of robotic manipulator trajectory tracking control is being considered an important research area. This textbook explains in great detail the process of designing an effective controller to enhance the trajectory tracking performance of a two degree of freedom (2-DOF) robotic arm using neural networks. Feed-forward ANNs were used in both model-based and non-model-based control strategies. Since it also includes a deep explanation of the modeling of the 2-DOF robotic arm system including its actuating DC-motors and their control using a PD controller, this textbook can also serve as an effective educational tool for both undergraduate and graduate electrical engineering students.
The present book deals with how to use the precious wireless resources that are usually wasted by under-utilization of networks. We have been particularly interested to all resources that can be used in an opportunistic fashion using different technologies. We have designed new schemes for better and more efficient use of wireless systems by providing mathematical frameworks. In this sense we focused on cognitive communication networks, where a cellular service provider can reserve a portion of its resources to secondary users or virtual suppliers. Delay tolerant networks are used as an alternative to the significant increase of the traffic load in cellular networks. In areas where the network infrastructure implementation is geographically difficult, the use of ad-hoc networks seems an appropriate solution. Indeed, we have developed a new analytical model of IEEE 802.11e used in this type of network protocol.
In this book, we derive a sufficient condition for asymptotic stability of the zero solution of delay-difference system of Hopfield neural networks and cellular neural networks in the terms of certain matrix inequalities by using the second Lyapunov method. The result has been applied to obtain new stability conditions for some classes of delay-difference equation such as delay-difference system of Hopfield neural networks with multiple delays, delay-difference control system of Hopfield neural networks, delay-difference control system of Hopfield neural networks with multiple delays, delay-difference system of Hopfield neural networks with time-varying delay, delay-difference system of Hopfield neural networks with multiple time-varying delays, delay-difference system of cellular neural networks with multiple delays, delay-difference control system of cellular neural networks, delay-difference control system of cellular neural networks with multiple delays, delay-difference system of cellular neural networks with time-varying delay and delay-difference system of cellular neural networks with multiple time-varying delays in the terms of certain matrix inequalities.
Electric load forecasting is an important research field in electric power industry. It plays a crucial role in solving a wide range of tasks of short-term planning and operating control of electric power system operating modes. Load forecasting is carried out in different time spans. Load forecasting within a current day – operating forecasting; one-day-week-month-ahead load forecasting – short-term load forecasting; one-month-quarter-year-ahead load forecasting – long-term load forecasting. So far a great number of both conventional and non-conventional electric load forecasting methods and models have been developed. The work presents research results of electric load forecasting for electrical power systems using artificial neural networks and fuzzy logic as one of the most advanced and perspective directions of solving this task. A theoretical approach to the issues discussed is combined with the data of experimental studies implemented with application of load curves of regional electrical power systems. The book is addressed to specialists and researchers concerned with operational control modes of electric power systems.
The aim of this B is to design fast thesisd forward neural networks to present a method to solve initial value problem for ordinary differential equations. That is to develop an algorithm which can speedup the solution times, reduce solver failures, and increase possibility of obtaining the globally optimal solution. The applicability of this approach ranges from single ordinary differential equations, to systems of ordinary differential equations with initial condition . Also, a variant types of compute the search direction ?k of conjugate gradient training algorithm are introduced and we describing several different training algorithms, many modified and new algorithms have been proposed for training Feed Forward Neural Network(FFNN), many of them having a very fast convergence rate for reasonable size networks. In all of these algorithms we use the gradient of the performance function( energy function) to determine how to adjust the weights such that the performance function is minimized, where the back propagation algorithm has been used to increase the speed of training. Finally, we illustrate the method by solving a variety of model problems.
The aim of this book is to design fast feed forward neural networks to present a method to solve two point boundary value problems for ordinary differential equations, that is, design a fully connected networks contains links between all nodes in adjacent layers which can speedup the solution times, reduce solver failures, and increase possibility of obtaining the globally optimal solution. We training suggested network by Levenberg – Marquardt, BFGS Quasi-Newton, Bayesian regularization, CG training algorithm with Polak-Ribiere update procedure then speeding suggested networks by modification these training algorithm, many of them having a very fast convergence rate for reasonable size networks. The above modify algorithms have a variety of different computation and storage requirements, however non of the above algorithms has a global properties, such as stability and convergence, which suited to all problems, and all the above algorithms applied in solving two point boundary value problem . Finally, we illustrate the suggested network by solving a variety of model problems and present comparisons with solutions obtained using other different method .
The framework of this study is to convert observed measurements of reservoir data into characteristic information of the reservoir. Artificial neural network (ANN) technology is utilized in mapping/interpolating the non-linear complex relationship between observed measurements and reservoir characteristics. The proposed ANN methodology is applied towards analysing the pressure transient measurements collected from isotropic and anisotropic faulted dual-porosity gas reservoirs, as an inverse solution to formation characteristics, such as the permeability and porosity of the fracture and matrix systems, distance to the fault, orientation of the fault with respect to the principal flow directions, and sealing capacity of the fault are predicted using the reservoir fluid, rock, and bottom-hole pressure as the principal inputs. The main focus of this study is to develop a suitable network to obtain accurate prediction about desired reservoir characteristics of dual-porosity tight gas systems with a fault, and demonstrate the efficient processing power of ANN on this class of reservoir problems.
The aim of hydro system scheduling problem is to find out the periodic water releases from each reservoir and through each power house so as to optimize the total benefit of hydro generated energy. The major focus of the work contained here is to develop and explain methods for solving different types of problems concerning optimal operation of interconnected hydro power plants. A decomposition method is explained which optimizes the mid-term operation of a practical reservoir system. Also a hybrid method based on decomposition and two-phase neural network is discussed to solve this problem. The objective here is to maximize annual hydro-generated energy. Effective solution technique based on neural nets is elaborated to solve the multiobjective hydro scheduling problem. Here the objectives are to maximize the annual hydropower generation and to satisfy the irrigation requirements as far as possible. A fuzzy-neural model is formulated for operating a reservoir type hydro plant with random inflows. Inflow sequences with specified mean and standard deviation are randomly generated using normal distribution. Two short-term scheduling examples are solved using two-phase neural network
Artificial neural networks are suitable for many tasks in pattern recognition and machine learning. Unlike conventional techniques for time series analysis, an artificial neural network needs little information about the time series data and can be applied to a broad range of problems. The usage of artificial neural networks for time series analysis relies purely on the data that were observed. As Radial Basis networks with one hidden layer is capable of approximating any measurable function. An artificial neural network is powerful enough to represent any form of time series. The capability to generalize allows artificial neural networks to learn even in the case of noisy and/or missing data. Another advantage over linear models is the network's ability to represent nonlinear time series. Prediction of tides is very much essential for human activities and to reduce the construction cost in marine environment. This book presents an application of the artificial neural network with Radial basis function for accurate prediction of tides. This neural network model predicts the time series data of hourly tides directly while using an an efficient learning process.
A midcourse guidance law based on minimum flight time optimal control formulation is developed based on singular perturbation technique (SPT). Singular perturbation techniques are used to simplify the optimal control problem. The resulting near-optimal control problem does not require the solution of any TPBVP. The guidance law, which is obtained through application of SPT on the optimal control problem, is tested using five degree-of-freedom (5 DOF) mathematical model of an air-to-air missile. The concept of minimum flight time and conserved energy during the terminal part of flight is studied for Medium Range Air-to-Air Missile (MRAAM) through the interception problem against non-maneuvering target by controlling the thrust magnitude and the burning time for the booster and sustainer missile motors. It can be noted that the missile performance using the technique of pulse ignition control motor (PICM) through the guidance gives much improvement over the boost-sustain missile motor.
Congestion affects the performance of a wireless sensor network in two aspects: increased data loss and reduced lifetime. All data generated in wireless sensor networks may not be alike; some data may be more important than others and hence may have different delivery requirements. As deployment sizes and data rates grow, congestion arises as a major problem in these networks. This congestion leads to indiscriminate dropping of data, i.e. data of high importance might be dropped while others of less importance are delivered. In this Book, I take a look at data delivery issues in the presence of Congestion Aware Routing Protocol designed earlier for wireless sensor networks. I propose a new data prioritization and a priority aware routing protocol - Advanced Congestion Aware Routing (ACAR) which slacking the packet dropping rate of data packets.