Search

Because there is a couple between the heating layer and the cooling one of the boiler in a PCT-II process control system, tuning the PID parameters is quite difficult and takes a long time to control the temperatures of the boiler besides some steady-state error. In this paper a decoupling control method based on the neural network is presented. The algorithm adopts tandem structure with the PID controller and an artificial neural network. And some appropriate compensation is used to eliminate the influence among the control variables, which can control the temperature effectively in the system. The experiment shows that the decoupling controller could almost eliminate the coupling between the variables and the steady-state output error, and achieve a better decoupling control by using the artificial neural networks to adapt to the nonlinear, time-varying characteristics.

This paper introduces the principle of the time-recording on automobile driver training, discusses an error tolerance plan of data storage countering some possible inferences of the system and analyses of the double-layered (16, 11, 4) minimal optimal odd-weight error-correcting code. From the analysis, one can conclude that this code has the merit of low redundancy and strong ability of anti-inference and detecting and correcting error.

During the inversion process of an SPWM inverter, it is very important to obtain the moments of a power switch. The magnitude of the harmonic waves can be diminished by the good moments. In this paper, the new method of reducing the influence of all harmonic waves, which can obtain the optimal switching time by using genetic algorithms, is presented. Simulation results show that this method can suppress effectively low-order harmonic wave with the limited items.

This paper analyses mind-evolution-based machine learning (MEBML) that has been proposed recently. The paper first discusses the practical problems in the implement of MEBML in numerical optimization. Then the paper gives the criterion used to judge whether a group is mature, also the paper proposes a adaptive method to adjust the parameters in similar taxis. The results of the experiment of numerical optimization are given. The experiment shows that the global convergence rate and computation efficiency are both improved above 20% compared with standard genetic algorithm. The improvement in convergence rate and efficiency is due to the distinctive structure of MEBML and the introduction of similar taxis and dissimilation.

This paper describes the foundation of proposing MEBML (mind-evolution-based machine learning) that was recently presented. Then the paper analyses the performance mechanism of MEBML which is different from GA (genetic algorithm), and its characteristics. With its own distinctive mechanism MEBML improved the efficiency and convergence rate greatly compared with GA. The paper also summarizes the recent development of researches on MEBML, which includes several different strategies of similarity and dissimilarity, the proof of convergence of MEBML and its applications. In addition the authors discuss the future work of MEBML.