Daniel Patiño

A New Approach for Time Series Forecasting: Bayesian Enhanced by Fractional Brownian Motion with Application to Rainfall Series

A new predictor algorithm based on Bayesian enhanced approach (BEA) for long-term chaotic time series using artificial neural networks (ANN) is presented. The technique based on stochastic models uses Bayesian inference by means of Fractional Brownian Motion as model data and Beta model as prior information. However, the need of experimental data for specifying and estimating causal models has not changed. Indeed, Bayes method provides another way to incorporate prior knowledge in forecasting models; the simplest representations of prior knowledge in forecasting models are hard to beat in many forecasting situations, either because prior knowledge is insufficient to improve on models or because prior knowledge leads to the conclusion that the situation is stable. This work contributes with long-term time series prediction, to give forecast horizons up to 18 steps ahead. Thus, the forecasted values and validation data are presented by solutions of benchmark chaotic series such as Mackey-Glass, Lorenz, Henon, Logistic, Rossler, Ikeda, Quadratic one-dimensional map series and monthly cumulative rainfall collected from Despenaderos, Cordoba, Argentina. The computational results are evaluated against several non-linear ANN predictors proposed before on high roughness series that shows a better performance of Bayesian Enhanced approach in long-term forecasting.

A New Approach for Nonlinear Multivariable Fed-Batch Bioprocess Trajectory Tracking Control

This paper proposes a new control law based on linear algebra. This technique allows nonlinear path tracking in multivariable and complex systems. This new methodology consists in finding the control action to make the system follow predefined concentration profiles solving a system of linear equations. The controller parameters are selected with a Monte Carlo algorithm so as to minimize a previously defined cost index. The control scheme is applied to a fed-batch penicillin production process. Different tests are shown to prove the controller effectiveness, such as adding parametric uncertainty, perturbations in the control action and in the initial conditions. Moreover, a comparison with other controllers from the literature is made, showing the better performance of the present approach.

On the approximate suboptimal control by neural network - rainfall observer

This paper presents an approach for approximate suboptimal control of nonlinear systems with constraints by neural network based rainfall observer for guiding crop growth in extensive agriculture. We propose a neural-network rainfall observer approximation by means of historical rainfall information. The goal is to obtain a close-loop operation with rainfall information, whose design is based on optimal control theory. Thus, the neurocontroller design proposed helps to drive the growth development of the cultivation as cost function and final state errors are minimized by physical constraints on the process variables. Therefore, it is possible to establish the control scheme and policy according to the criterion that generates the highest profit margin in the process. The contribution shows an optimal policy to guide the crop from an initial to a desired state. The estimates are consistent in a weak sense, and the question whether they are pointwise consistent is still open. Nevertheless, in order to assess the performance and practical tractability of the neurocontroller, real data and computational results are shown for soybean crop at Santa Francisca, Cordoba, Argentina.

Controller design for tracking paths in nonlinear biochemical processes

The control of fed-batch bioprocess is a current challenge. Mathematical models are highly rigid systems of nonlinear differential equations with strict physical limitations. In this paper a simple and efficient technique for tracking optimal profiles with minimal error is developed. It is based on linear algebra for the calculation of control actions, by solving a system of linear equations. The performance of the designed controller is tested through simulations (adding parametric uncertainty and perturbations in the initial conditions), which show very satisfactory results.

A combined approach for long-term series prediction: Renyi permutation entropy with BEA predictor filter

In order to predict long-term series, a Bayesian enhanced approach (BEA) combining permutation entropy (BEMA) is presented. The motivation of the proposed filter is to predict long-term time series by changing the structure of the predictor filter according to data model selected, then computational results are evaluated on high roughness time series selected from benchmark, in which they are compared with recent artificial neural networks (ANN) nonlinear filters such as Bayesian Enhanced approach (BEA) and Bayesian Approach (BA). These results support the applicability of permutation entropy in analyzing the dynamic behavior of chaotic time series for long-term series predictions.

Controller design by monitoring desired concentration profiles for the penicillin production in a feed batch reactor

The objective of this work is to design a controller for process variables for the penicillin production, carried out in a fed-batch reactor, following predefined profiles. The technique is based on linear algebra, which allows the design of multivariable controllers and highly nonlinear systems. To achieve this, it is necessary to possess a mathematical model that adequately represents the process and the concentration profiles that the system should follow. Simulation results are shown for different initial conditions.

Numerical solution to uncertainties in the finite time optimal control systems design

Application of Optimal Control Laws in autonomous systems with finite energy sources, such as freestanding and aerial robots, is a real need for today. This paper presents a finite horizon optimal controlled design with fixed initial and final states, for such complex and autonomous self-propelled systems. While its design methodology is well known, inconsistencies and singularities may occur in the control law computation, mainly for higher order systems. It is a critical issue in real-time design, which is not mentioned and treated in classical optimal control theory. In this paper, we propose and show, how the problem of the numerical spread error in the path computation a singularity can be reached. This type of problem is critical and unwanted for real-time design, in which is necessary to recalculate the path to get the desired final state online. A solution for this problem is presented, and computational simulation is also shown.