Yasar Becerikli

Intelligent optimal control with dynamic neural networks

The application of neural networks technology to dynamic system control has been constrained by the non-dynamic nature of popular network architectures. Many of difficulties are-large network sizes (i.e. curse of dimensionality), long training times, etc. These problems can be overcome with dynamic neural networks (DNN). In this study, intelligent optimal control problem is considered as a nonlinear optimization with dynamic equality constraints, and DNN as a control trajectory priming system. The resulting algorithm operates as an auto-trainer for DNN (a self-learning structure) and generates optimal feed-forward control trajectories in a significantly smaller number of iterations. In this way, optimal control trajectories are encapsulated and generalized by DNN. The time varying optimal feedback gains are also generated along the trajectory as byproducts. Speeding up trajectory calculations opens up avenues for real-time intelligent optimal control with virtual global feedback. We used direct-descent-curvature algorithm with some modifications (we called modified-descend-controller-MDC algorithm) for the optimal control computations. The algorithm has generated numerically very robust solutions with respect to conjugate points. The adjoint theory has been used in the training of DNN which is considered as a quasi-linear dynamic system. The updating of weights (identification of parameters) are based on Broyden-Fletcher-Goldfarb-Shanno BFGS method. Simulation results are given for an intelligent optimal control system controlling a difficult nonlinear second-order system using fully connected three-neuron DNN.

Neural network implementation in hardware using FPGAs

The usage of the FPGA (Field Programmable Gate Array) for neural network implementation provides flexibility in programmable systems. For the neural network based instrument prototype in real time application, conventional specific VLSI neural chip design suffers the limitation in time and cost. With low precision artificial neural network design, FPGAs have higher speed and smaller size for real time application than the VLSI design. In addition, artificial neural network based on FPGAs has fairly achieved with classification application. The programmability of reconfigurable FPGAs yields the availability of fast special purpose hardware for wide applications. Its programmability could set the conditions to explore new neural network algorithms and problems of a scale that would not be feasible with conventional processor. The goal of this work is to realize the hardware implementation of neural network using FPGAs. Digital system architecture is presented using Very High Speed Integrated Circuits Hardware Description Language (VHDL) and is implemented in FPGA chip. The design was tested on a FPGA demo board.

Wavelet networks for nonlinear system modeling

This study presents a nonlinear systems and function learning by using wavelet network. Wavelet networks are as neural network for training and structural approach. But, training algorithms of wavelet networks is required a smaller number of iterations when the compared with neural networks. Gaussian-based mother wavelet function is used as an activation function. Wavelet networks have three main parameters; dilation, translation, and connection parameters (weights). Initial values of these parameters are randomly selected. They are optimized during training (learning) phase. Because of random selection of all initial values, it may not be suitable for process modeling. Because wavelet functions are rapidly vanishing functions. For this reason heuristic procedure has been used. In this study serial-parallel identification model has been applied to system modeling. This structure does not utilize feedback. Real system outputs have been exercised for prediction of the future system outputs. So that stability and approximation of the network is guaranteed. Gradient methods have been applied for parameters updating with momentum term. Quadratic cost function is used for error minimization. Three example problems have been examined in the simulation. They are static nonlinear functions and discrete dynamic nonlinear system.

Nonlinear predictive control of a drying process using genetic algorithms.

A nonlinear predictive control technique is developed to determine the optimal drying profile for a drying process. A complete nonlinear model of the bakers yeast drying process is used for predicting the future control actions. To minimize the difference between the model predictions and the desired trajectory throughout finite horizon, an objective function is described. The optimization problem is solved using a genetic algorithm due to the successful overconventional optimization techniques in the applications of the complex optimization problems. The control scheme comprises a drying process, a nonlinear prediction model, an optimizer, and a genetic search block. The nonlinear predictive control method proposed in this paper is applied to the bakers yeast drying process. The results show significant enhancement of the manufacturing quality, considerable decrease of the energy consumption and drying time, obtained by the proposed nonlinear predictive control.

Intelligent adaptive nonlinear flight control for a high performance aircraft with neural networks

This paper describes the development of a neural network (NN) based adaptive flight control system for a high performance aircraft. The main contribution of this work is that the proposed control system is able to compensate the system uncertainties, adapt to the changes in flight conditions, and accommodate the system failures. The underlying study can be considered in two phases. The objective of the first phase is to model the dynamic behavior of a nonlinear F-16 model using NNs. Therefore a NN-based adaptive identification model is developed for three angular rates of the aircraft. An on-line training procedure is developed to adapt the changes in the system dynamics and improve the identification accuracy. In this procedure, a first-in first-out stack is used to store a certain history of the input-output data. The training is performed over the whole data in the stack at every stage. To speed up the convergence rate and enhance the accuracy for achieving the on-line learning, the Levenberg-Marquardt optimization method with a trust region approach is adapted to train the NNs. The objective of the second phase is to develop intelligent flight controllers. A NN-based adaptive PID control scheme that is composed of an emulator NN, an estimator NN, and a discrete time PID controller is developed. The emulator NN is used to calculate the system Jacobian required to train the estimator NN. The estimator NN, which is trained on-line by propagating the output error through the emulator, is used to adjust the PID gains. The NN-based adaptive PID control system is applied to control three angular rates of the nonlinear F-16 model. The body-axis pitch, roll, and yaw rates are fed back via the PID controllers to the elevator, aileron, and rudder actuators, respectively. The resulting control system has learning, adaptation, and fault-tolerant abilities. It avoids the storage and interpolation requirements for the too many controller parameters of a typical flight control system. Performance of the control system is successfully tested by performing several six-degrees-of-freedom nonlinear simulations.

A new fuzzy approach for edge detection

An edge detection is one of the most important tasks in image processing. Image segmentation, registration and identification are based on edge detection. In the literature, there is some techniques developed to achive this task such as Sobel, Prewitt, Laplacian and Laplacian of Gaussian. In this paper, a novel knowledge-based approach which have been used to realize control techniques for past years is proposed for edge detection. Some of the classical techniques are used with certain parameters such as threshold and σ to implement edge detection process. The another restricts about classial approach, results generally have fixed edge thickness. The rule-based approach offers most advantages such as giving permission to adapt some parameters easily. The edges thickness can be changed easily by adding new rules or changing output parameters. That is to say rule-based approach has flexible structure which can be adapted any time or any where easily and new fuzzy approach produces nice result as well as classical techniques at least.

Modeling and prediction with a class of time delay dynamic neural networks

In this paper, we propose a time delay dynamic neural network (TDDNN) to track and predict a chaotic time series systems. The application of artificial neural networks to dynamical systems has been constrained by the non-dynamical nature of popular network architectures. Many of the drawbacks caused by the algebraic structures can be overcome with TDDNNs. TDDNNs have time delay elements in their states. This approach provides the natural properties of physical systems. The minimization of a quadratic performance index is considered for trajectory tracking applications. Gradient computations are presented based on adjoint sensitivity analysis. The computational complexity is significantly less than direct method, but it requires a backward integration capability. We used Levenberg-Marquardt parameter updating method.

On three intelligent systems: dynamic neural, fuzzy, and wavelet networks for training trajectory

Intelligent systems cover a wide range of technologies related to hard sciences, such as modeling and control theory, and soft sciences, such as the artificial intelligence (AI). Intelligent systems, including neural networks (NNs), fuzzy logic (FL), and wavelet techniques, utilize the concepts of biological systems and human cognitive capabilities. These three systems have been recognized as a robust and attractive alternative to the some of the classical modeling and control methods. The application of classical NNs, FL, and wavelet technology to dynamic system modeling and control has been constrained by the non-dynamic nature of their popular architectures. The major drawbacks of these architectures are the curse of dimensionality, such as the requirement of too many parameters in NNs, the use of large rule bases in FL, the large number of wavelets, and the long training times, etc. These problems can be overcome with dynamic network structures, referred to as dynamic neural networks (DNNs), dynamic fuzzy networks (DFNs), and dynamic wavelet networks (DWNs), which have unconstrained connectivity and dynamic neural, fuzzy, and wavelet processing units, called “neurons”, “feurons”, and “wavelons”, respectively. The structure of dynamic networks are based on Hopfield networks. Here, we present a comparative study of DNNs, DFNs, and DWNs for non-linear dynamical system modeling. All three dynamic networks have a lag dynamic, an activation function, and interconnection weights. The network weights are adjusted using fast training (optimization) algorithms (quasi-Newton methods). Also, it has been shown that all dynamic networks can be effectively used in non-linear system modeling, and that DWNs result in the best capacity. But all networks have non-linearity properties in non-linear systems. In this study, all dynamic networks are considered as a non-linear optimization with dynamic equality constraints for non-linear system modeling. They encapsulate and generalize the target trajectories. The adjoint theory, whose computational complexity is significantly less than the direct method, has been used in the training of the networks. The updating of weights (identification of network parameters) is based on Broyden–Fletcher–Goldfarb–Shanno method. First, phase portrait examples are given. From this, it has been shown that they have oscillatory and chaotic properties. A dynamical system with discrete events is modeled using the above network structure. There is a localization property at discrete event instants for time and frequency in this example.

Trajectory tracking with dynamic neural networks

The application of artificial neural networks to dynamical systems has been constrained by the non-dynamical nature of popular network architectures. Many of the difficulties that ensue-large network sizes, long training times, the need to predetermine buffer lengths-can be overcome with dynamic neural networks. The minimization of a quadratic performance index is considered for trajectory tracking or process simulation applications. Two approaches for gradient computation are discussed: forward and adjoint sensitivity analysis. The computational complexity of the latter is significantly less, but it requires a backward integration capability. We also discuss two parameter updating methods: the gradient descent method and Levenberg-Marquardt approach.

Generalized modeling principles of a nonlinear system with a dynamic fuzzy network

Generalized modeling principles of a nonlinear system with a dynamic fuzzy network (DFN)- a network with unconstrained connectivity and with dynamic fuzzy processing units called ‘feurons’, have been given. DFN model has been trained both in open loop and closed loop forms to satisfy these principles. Several system trajectories with a PRBS input have been used for open loop training. DFN model obtained from open loop training was used in a closed loop training with an extended Kalman filter (EKF) in an observer design. For gradient computations adjoint sensitivity method has been used.