In this paper, the problem of adaptively identifying time-varying nonlinear systems is considered. For that purpose, the discrete time-varying Volterra series is employed as a system model, and the filtered weighted least squares (FWLS) algorithm, developed for adaptive identification of linear time-varying systems, is utilized for the adaptive identification of time-varying quadratic Volterra systems. To demonstrate the performance of the proposed approach, some simulation results are provided. Note that the FWLS algorithm, decomposing the conventional weighted basis function (WBF) algorithm into a cascade of two (i.e., estimation and filtering) procedures, leads to fast parameter tracking with low computational burden, and the proposed approach can be easily extended to the adaptive identification of time-varying higher-order Volterra systems.

In rotating systems, backlash impose limitations on the quality of control. System with gear is an example where this is a well-known limitation. In order to increase the controller performance, we design a fuzzy system to compensate the backlash effect. We prove that under certain conditions the fuzzy compensator guarantees that the backlash output converges to the desired trajectory. Simulation results show that the fuzzy compensator is robust to the backlash parameter.

There are two important things for the general purpose neural network processor. The first is a capability to build various structures of neural network, and the second is to be able to support suitable learning method for that neural network. Some way to process various learning algorithms is required for on-chip learning, because the more neural network types are to be handled, the more learning methods need to be built into. In this paper, an improved hardware structure is proposed to compute various kinds of learning algorithms flexibly. The hardware structure is based on the existing modular neural network structure. It doesn't need to add a new circuit or a new program for the learning process. It is shown that rearrangements of the existing processing elements can produce several neural network learning modules. The performance and utilization of this module are analyzed by comparing with other neural network chips.

This paper infers two patterns of timeout from the characteristics of TCP Reno and confirms the existence of the patterns by conducting actual experiments. All timeouts can be classified into either of these patterns by using the history of RTT value. Based on the observed patterns, the paper proposes two algorithms to improve the performance of TCP Reno. Experimental results show that, when compared with TCP Reno, the proposed algorithms improve the bandwidth utilization by 3 to 12 percent. The paper provides good examples of how timeout-based and delay-based congestion control can efficiently work together.

In this paper, studied error of various signal interpolation methods in vibration signal analysis with digital order tracking. Because, interpolation errors are related with sampling rate and amount of calculation. Appled Signal interpolation methods are Lagrange, Newton and Cubic-spline. This paper proposed more proper interpolation method. Also, we suggest guideline for adaptive application of signal interpolation methods with Calculated results.

In active noise control (ANC) systems, the error-reduction performance of the conventional Filtered-X Least Mean Square (FXLMS) algorithm may be affected by nonlinear distortions in the secondary path such as in the power amplifiers, loudspeakers and transducers. In this paper, a nonlinear FXLMS algorithm with high error-reduction performance is proposed to compensate for undesirable nonlinearities in the secondary-path of ANC systems by employing the inverse Volterra filtering approach. In particular, the proposed approach is based on the utilization of the conventional P-th order inverse approach to nonlinearity compensation in the secondary path of ANC systems. Finally, the simulation results showed that the proposed approach yields a better nonlinearity compensation performance for the ANC systems with a nonlinear secondary path than the conventional FXLMS.

This paper proposes an algorithm for EEG pattern classification using the Multi-template method, which is a kind of speaker adaptation method for speech signal processing. 10-channel EEG signals are collected in various environments. The linear prediction coefficients of the EEGs are extracted as the feature parameter of human sensibility. The human sensibility classification algorithm is developed using neural networks. Using EEGs of comfortable or uncomfortable seats, the proposed algorithm showed about 75% of classification performance in subject-independent test. In the tests using EEG signals according to room temperature and humidity variations, the proposed algorithm showed good performance in tracking of pleasantness changes and the subject-independent tests produced similar performances with subject-dependent ones.