Cyril Oswald

Learning entropy for novelty detection a cognitive approach for adaptive filters

This paper recalls the practical calculation of Learning Entropy (LE) for novelty detection, extends it for various gradient techniques and discusses its use for multivariate dynamical systems with ability of distinguishing between data perturbations or system-function perturbations. LG has been recently introduced for novelty detection in time series via supervised incremental learning of polynomial filters, i.e. higher-order neural units (HONU). This paper demonstrates LG also on enhanced gradient descent adaptation techniques that are adopted and summarized for HONU. As an aside, LG is proposed as a new performance index of adaptive filters. Then, we discuss Principal Component Analysis and Kernel PCA for HONU as a potential method to suppress detection of data-measurement perturbations and to enforce LG for system-perturbation novelties.

Transfer Issues of Control Optimizing Combustion from Small-scale to Medium-scale Biomass-fired Boilers

Abstract The quick development of biomass fired boilers increased the attention being paid to the ecological and economical aspects of their operation. A very important role in the process of adapting to harder regulations has been the control technology replacing older simple solutions. Lately, we have been offered a very interesting opportunity to make use of our experience with small-scale boiler control which has been gradually equipped with instruments for more sophisticated experiments and to transfer verified control algorithms to a district heating boiler of medium-scale size. The following text describes what has been achieved in experiments with the small-scale boiler and how this can be transferred to bigger scale one. An algorithm to optimize the combustion process from the efficiency point of view, as well as harmful emission reduction is described in detail.

Evaluation of open-loop frequency indicators in a closed loop for PID controller autotuning

In industrial practice, PID controllers are extensively used because they can be successfully implemented without knowing a model of the object or process which is controlled. Despite of the fact that many PID controller tuning algorithms have been developed since the Ziegler and Nichols tuning procedures were published in 1942, their experimental method continues to be used up to now. There are two main reasons why newer methods are not widely-used: the methods are based on a model of the controlled plant, and/or it is necessary to stop the control process during controller retuning. Our method presented in this paper is based on an experimentally performed evaluation of excited frequency responses with the aim of achieving recommended values of one or more control quality indicators known from the course of the Nyquist plot, e.g. Frequency response evaluation can be performed in a closed loop with no controller function degradation, and no model is necessary for frequency response evaluation. In this way, the main problems of other tuning methods are eliminated. Since there is no need for a model, the tuning method can be used in a control loop involving nonlinearities even in the controller.

Neural Network Evaluation of Combustion Process for Continuous Control of Small Scale Biomass Fired Boilers

Abstract Boilers based on combustion of biomass become widely used as a heating source nowadays. The modern ones are typically controlled automatically. The control algorithm of those boilers is crucial in reaching optimal operational conditions by means of maximal efficiency and minimal environmental impact. On the other hand, the acquisition costs of these advanced devices should be maintained at reasonable level. This paper deals with implementation of modern proper control algorithm and obtaining the necessary input values that cannot be easily measured operationally by a direct measurement.

Optimal Combustion Conditions for a Small-scale Biomass Boiler

This paper reports on an attempt to achieve maximum efficiency and lowest possible emissions for a small-scale biomass boiler. This aim can be attained only by changing the control algorithm of the boiler, and in this way not raising the acquisition costs for the boiler. This paper describes the experimental facility, the problems that arose while establishing the facility, and how we have dealt with them. The focus is on discontinuities arising after periodic grate sweeping, and on finding the parameters of the PID control loops. Alongside these methods, which need a lambda probe signal for proper functionality, we inroduce another method, which is able to intercept the optimal combustion working point without the need to use a lambda sensor.

Case Study of Learning Entropy for Adaptive Novelty Detection in Solid-Fuel Combustion Control

This paper deals with the case study of usability of the Learning Entropy approach for the adaptive novelty detection in MIMO dynamical systems. The novelty detection is studied for typical parameters of linear systems including time delay. The solid-fuel combustion process is selected as a representative of typical non-linear dynamic MIMO system. The complex mathematical model of a biomass-fired 100kW boiler is used for verification of the potentials of the proposed method, and the motivation for novelty detection in solid-fuel combustion processes is discussed in this paper.

SMALL AND MEDIUM SCALE BIOMASS COMBUSTION CONTROL POSSIBILITIES

Biomass combustion is one of the most versatile and spread heat source for residential usage in the world. Nowadays a lot of new combustion devices in developed countries are controlled automatically. The control algorithm used in these devices is crucial in attaining optimal service economy with minimal environmental impact. This contribution deals with two ways how to implement proper control algorithms to small and medium scale boilers. The first method is by applying modern methods of controller development and the second by adapting tested algorithms from full-scale boilers. The used algorithms are implemented and tested on experimental devices based on common market products.