Multilayer discrete-time neural-net controller with guaranteedperformance

A family of novel multilayer discrete-time neural-net (NN) controllers is presented for the control of a class of multi-input multi-output (MIMO) dynamical systems. The neural net controller includes modified delta rule weight tuning and exhibits a learning while-functioning-features. The structure of the NN controller is derived using a filtered error/passivity approach. Linearity in the parameters is not required and certainty equivalence is not used. This overcomes several limitations of standard adaptive control. The notion of persistency of excitation (PE) for multilayer NN is defined and explored. New online improved tuning algorithms for discrete-time systems are derived, which are similar to sigma or epsilon-modification for the case of continuous-time systems, that include a modification to the learning rate parameter plus a correction term. These algorithms guarantee tracking as well as bounded NN weights in nonideal situations so that PE is not needed. An extension of these novel weight tuning updates to NN with an arbitrary number of hidden layers is discussed. The notions of discrete-time passive NN, dissipative NN, and robust NN are introduced. The NN makes the closed-loop system passive.     

Self-aligned SiGe NPN transistors with 285 GHz f/sub MAX/ and 207 GHz f/sub T/ in a manufacturable technology

This paper reports on SiGe NPN HBTs with unity gain cutoff frequency (f_T) of 207 GHz and an f_MAX extrapolated from Mason's unilateral gain of 285 GHz. f_MAX extrapolated from maximum available gain is 194 GHz. Transistors sized 0.12×2.5 μm² have these characteristics at a linear current of 1.0 mA/μm (8.3 mA/μm²). Smaller transistors (0.12×0.5 μm²) have an f_T of 180 GHz at 800 μA current. The devices have a pinched base sheet resistance of 2.5 kΩ/sq. and an open-base breakdown voltage BV_CEO of 1.7 V. The improved performance is a result of a new self-aligned device structure that minimizes parasitic resistance and capacitance without affecting f_T at small lateral dimensions     

Microprocessor-Based Universal Regulator Using Dual-Rate Sampling

Standard single-rate sampling self-tuning regulators may not be able to effectively control fast-acting systems if the minimum computation time required is larger than the largest sampling period permissible. A dual-rate sampling self-tuning regulator that overcomes this problem by allowing a data sampling rate to be chosen consistent with the system time constants and a control output interval based on the computation time is described in this paper. This algorithm is also very suitable for implementation using fixed word-length processors as the quantization problems are minimized. The algorithm has been implemented on a microprocessor. Implementation details with emphasis on data acquisition, interrupt controlled sampling, time sufficiency test and scaling, and experimental results are also given.     

Spin-wave excitations in amorphous ferromagnets

Mössbauer measurements have been performed on a number of metallic glasses. The temperature dependence of average hyperfine or internal magnetic fieldH _eff(T) arises from long wavelength spin wave excitations in these glasses. Values ofB _3/2 andC _5/2 are in general much higher than those observed for crystalline ferromagnets indicating higher density of states for spin waves in amorphous ferromagnets.     

Application of the finite-element-synthesis model to multidimensional reactor transient problems without feedback

The finite-element-synthesis model has been employed to solve time-dependent multigroup diffusion equations with multiple delayed-neutron precursor groups. Feedback effects are not considered. Precursor equations are analytically integrated within a time step, assuming linear variation of fission rate during the time interval. A fully implicit scheme is used for the time integration of the mixing functions. A coarse-mesh rebalancing technique is found to considerably accelerate the convergence of the inhomogeneous source problem of each time step. Many 2-D and 3-D problems were analysed with the present calculational model. The results are in good agreement with those of other less approximate methods, except for the problems in which the reflector zone is perturbed. There is a considerable saving in computer time due to the synthesis approximation.     

Carbon supported nickel phosphide as efficient electrocatalyst for hydrogen and oxygen evolution reactions

Hydrogen production through water splitting is an efficient and green technology for fulfilling future energy demands. Carbon nanotubes (CNT) supported Ni₂P has been synthesized through a simpler hydrothermal method. Ni₂P/CNT has been employed as efficient electrocatalysts for hydrogen and oxygen evolution reactions in acidic and alkaline media respectively. The electrocatalyst has exhibited low overpotential of 137 and 360 mV for hydrogen and oxygen evolution reactions respectively at 10 mA cm^?2. Lower Tafel slopes, improved electrochemical active surface area, enhanced stability have also been observed. Advantages of carbon support in terms of activity and stability have been described by comparing with unsupported electrocatalyst.