Danilo Pelusi

Optimization of a fuzzy logic controller using genetic algorithms

The design of a fuzzy controller suffers from choice problems of fuzzy input and output membership functions and rules inference system definition. Generally, such procedures are implemented by trial and error iterations which do not assure an optimal fuzzy controller design. Moreover the fuzzy features of control system depend by the specific application of fuzzy controller. There are several techniques reported in recent literature that use Genetic Algorithms to optimize a fuzzy logic controller. This paper proposes a methodology to optimize fuzzy logic parameters based on Genetic Algorithms. The scheme is applied to the problem of electrical signal frequency driving for signals acquisition experiments. The fuzzy logic controller is tuned by Genetic Algorithms until to achieve the optimal parameters. The tuning design approach offers a complete and fast way to design an optimal fuzzy system. Moreover, the results show that the optimized fuzzy controller gives better performance than a conventional fuzzy controller also in terms of rise and settling time.

Optimal Control Algorithms for Second Order Systems

Proportional Integral Derivative (PID) controllers are widely used in industrial processes for their simplicity and robustness. The main application problems are the tuning of PID parameters to obtain good settling time, rise time and overshoot. The challenge is to improve the timing parameters to achieve optimal control performances. Remarkable findings are obtained through the use of Artificial Intelligence techniques as Fuzzy Logic, Genetic Algorithms and Neural Networks. The combination of these theories can give good results in terms of settling time, rise time and overshoot. In this study, suitable controllers able of improving timing performance of second order plants are proposed. The results show that the PID controller has good overshoot values and shows optimal robustness. The genetic-fuzzy controller gives a good value of settling time and a very good overshoot value. The neural-fuzzy controller gives the best timing parameters improving the control performances of the others two approaches. Further improvements are achieved designing a real-time optimization algorithm which works on a genetic-neuro-fuzzy controller.

Improving Settling and Rise Times of Controllers via Intelligent Algorithms

Generally, conventional controllers are characterized by too longs settling and rise times. In order to solve this problem, suitable fuzzy logic controllers have been designed. However, some intelligent techniques can be added during the controllers designing phase. In the literature, the employed methods are Genetic Algorithms and Neural Networks. The first ones are good search methods whereas the others ones have the capability to learn from data. In this paper, an optimized genetic-neuro-fuzzy controller is proposed. This controller works in according with a real-time optimization algorithm which optimally combines the features of Fuzzy Logic, Genetic Algorithms and Neural Networks. The genetic procedures search the optimal membership functions whereas the neural methods optimize the fuzzy rules. The target is to reduce the settling time and rise time with overshoot equal to zero. The novelty of this approach is that the optimization procedures occur at the same time and not separately. The results show that the settling time and the rise time are reduced by comparing them with the same quantities of optimized PD and PID controllers. Moreover, the designed controller improves the timing performance of conventional and intelligent controllers.

The International Robotic Antarctic Infrared Telescope (IRAIT)

Thanks to exceptional coldness, low sky brightness and low content of water vapour of the above atmosphere Dome C, one of the three highest peaks of the large Antarctic plateau, is likely to be the best site on Earth for thermal infrared observations (2.3-300 μm) as well as for the far infrared range (30 μm-1mm). IRAIT (International Robotic Antarctic Infrared Telescope) will be the first European Infrared telescope operating at Dome C. It will be delivered to Antarctica at the end of 2006, will reach Dome C at the end of 2007 and the first winter-over operation will start in spring 2008. IRAIT will offer a unique opportunity for astronomers to test and verify the astronomical quality of the site and it will be a useful test-instrument for a new generation of Antarctic telescopes and focal plane instrumentations. We give here a general overview of the project and of the logistics and transportation options adopted to facilitate the installation of IRAIT at Dome C. We summarize the results of the electrical, electronics and networking tests and of the sky polarization measurements carried out at Dome C during the 2005-2006 summer-campaign. We also present the 25 cm optical telescope (small-IRAIT project) that will installed at Dome C during the Antarctic summer 2006-2007 and that will start observations during the 2007 Antarctic winter when a member of the IRAIT collaboration will join the Italian-French Dome C winter-over team.

On designing optimal control systems through genetic and neuro-fuzzy techniques

Many industrial processes are affected by flow disturbances and sensor noise. To maintain optimal timing performances, the control system needs to adapt continuously to these changes. The goodness of a control system depends on timing parameters such as settling time, rise time and overshoot. Avoiding undesirable overshoot, longer settling times and vibration from a state to another one, the designed control system gives optimal control performances. Control problems can be overcome using computational intelligence procedures. The target of this work is to find optimal combinations of intelligent techniques such as fuzzy logic, Genetic Algorithms and neural networks to obtain good control performances. The membership functions of the designed fuzzy controllers are optimized through Genetic Algorithms. Moreover, the fuzzy rules weights are tuned both Genetic Algorithms and neural networks. In this way, the control system has the capability to learn from data. The results show that our controllers improve the timing performances of conventional controllers. Moreover, the fuzzy rules weights optimization with Genetic Algorithms is improved using neural networks techniques which suitably tune the weights.

Genetic-Neuro-Fuzzy Controllers for Second Order Control Systems

Overshoot, settling and rise time define the timing parameters of a control system. The main challenge is to attempt to reduce these parameters to achieve good control performances. The target is to obtain the optimal timing values. In this paper, three different approaches are presented to improve the control performances of second order control systems. The first approach is related to the design of a PID controller based on Ziegler-Nichols tuning formula. An optimal fuzzy controller optimized through Genetic Algorithms represents the second approach. Following this way, the best membership functions are chosen with the help of the darwinian theory of natural selection. The third approach uses the neural networks to achieve adaptive neuro-fuzzy controllers. In this way, the fuzzy controller assumes self-tuning capability. The results show that the designed PID controller has a very slow rise time. The genetic-fuzzy controller gives good values of overshoot and settling time. The best global results are achieved by neuro-fuzzy controller which presents good values of overshoot, settling and rise time. Moreover, our neuro-fuzzy controller improves the results of some conventional PID and fuzzy controllers.

Revised Gravitational Search Algorithms Based on Evolutionary-Fuzzy Systems

The choice of the best optimization algorithm is a hard issue, and it sometime depends on specific problem. The Gravitational Search Algorithm (GSA) is a search algorithm based on the law of gravity, which states that each particle attracts every other particle with a force called gravitational force. Some revised versions of GSA have been proposed by using intelligent techniques. This work proposes some GSA versions based on fuzzy techniques powered by evolutionary methods, such as Genetic Algorithms (GA), Particle Swarm Optimization (PSO) and Differential Evolution (DE), to improve GSA. The designed algorithms tune a suitable parameter of GSA through a fuzzy controller whose membership functions are optimized by GA, PSO and DE. The results show that Fuzzy Gravitational Search Algorithm (FGSA) optimized by DE is optimal for unimodal functions, whereas FGSA optimized through GA is good for multimodal functions.

Fuzzy algorithm control effectiveness on drum boiler simulated dynamics

The optimal control of parameters in a system assumes an important role in industrial processes. Models based on boiler-turbine plant are proposed in various applications. The target of this paper is to apply intelligent techniques on a boiler simulator to improve the speed and precision of control. In order to release such task, genetic-fuzzy controllers able of tuning the time duration of specific boiler drum signals, are designed. The results show good precision and relevant speed of control system improving the control performances of classical control structures.

Designing Neural Networks to improve timing performances of intelligent controllers

Abstract Intelligent techniques are applied to improve the control methods of physical quantities. Many researchers employ the fuzzy logic combined with genetic procedures to achieve good control results. Such intelligent techniques can be enhanced using Neural Networks. In this paper, an optimization algorithm to define the best training set for suitable Neural Networks is designed. The network trained with the optimal sample is introduced into genetic-fuzzy controllers to improve the timing performances. The results show that the genetic-neuro-fuzzy controllers are better than genetic-fuzzy controllers in terms of settling time and rise time.

Optimal trading rules at hourly frequency in the foreign exchange markets

An accurate measure of profitability of Technical Analysis, free of “data snooping”, requires the separation of the Training Set (where the parameters of the technical filter are obtained) from the Trading Set (where the profit results of this technical filter are studied, using parameters obtained in the former). The next task is how to obtain the “best” parameters for high profits. Following the suggestions of the literature, we used a Genetic Algorithm (GA) to spot the “best” parameters in the Training Set to be used, separately and independently, in the Trading Set. This paper presents quantitative results in the use of one GA applied to the Dual Moving Average Crossover rule (DMAC) applied to hourly data of the Euro-Dollar exchange rate between 1999 and 2006. One important feature of the paper is the use of a GA in an unconstrained and constrained optimization set-up. The first optimization aims at obtaining the highest profit rates. The second one looks for smoother profit rates. We study the impact of these two techniques on a kind of mean-variance relationship of profit rates. Unconstrained optimization yields an yearly average profits of 16.8%; the constrained one gets 13.4% (but with much lower volatility of cumulative profits overtime).