STATISTICAL ASSOCIATION LEARNING OF THE MARKOV DECISION PROCESS

A neural computing approach to the Markov decision problem is presented. The method makes use of historical records of rewards as inputs and average long-run incomes per epoch as targets for training a backpropagation network to associate the two quantities, without a priori knowledge of the state transition probabilities. Estimation of the long-run income, given a new reward matrix as input, is interpreted as a statistical-association learning problem. Following the training, the relevant functional relationship between the income (output) and the reward (input), that has been learned by the network, can be used to compute an unknown expected income. The present study examines network generalization to new inputs, considering the effects of different topological designs and the characteristics of training samples on neural computing accuracy. Successful performance of the best design, which is determined directly from the underlying mathematical model of the Markov decision problem, is demonstrated by a computer simulation experiment.     

Crystallite growth characteristics in superfine zirconia powders and compacts

Crystallite growth characteristics of coprecipitated superfine zirconia powders have been investigated. It was found that the crystallite growth in powders follows a cubic law at 800 and 1000°C; however, the crystallite size data for compacts of both Y-TZP and YSZ cannot be fitted with a traditional parabolic or cubic law, but with a linear relation between crystallite size and the logarithm of time. In addition, it was also found that the degree of agglomeration of the powders can affect the crystallite growth.     

Prolonged release Matrix Pellets Prepared by Melt Pelletization I. Process Variables

A melt pelletization process was investigated in an 8 litre laboratory scale high shear mixer using a formulation with paracetamol, glyceryl monostearate 40-50, and microcrystalline wax. The effects of jacket temperature, product temperature during massing, product load, massing time and impeller speed were investigated by means of factorially designed experiments. The maximum yield of pellets in the range of 500-1400μm was found to approx. 90%. For process conditions preventing deposition of moist mass, the process was found to be reproducible. Impeller speed and massing time were found to be important process variables. Remarkably low in vitro drug release rates were observed in USP-dissolution tests.     

Effects of load modeling in security constrained OPF studies

Results of computational experiments carried out on an adaptation of the AEP/IEEE 30 bus test system are reported. Some selected buses were modeled using an exponential type of load model. Three cases are considered. In the first case it is assumed that the specified load at modeled buses is obtained with unity voltage. In the second case it is assumed that the transformer taps have been adjusted to give all industrial-type consumers one per unit at the low-voltage panel when the high-side voltage corresponds to the standard optimal power flow (OPF) solution. The third case differs from the second in that the specified power demand is assumed to take place when the high-side voltages correspond to the intact case of the standard security constrained OPF solution. It is concluded that a decrease in fuel cost can be obtained in some instances when load models are incorporated in security constrained OPF studies during contingencies only. In situations where a decrease in fuel cost is obtained in this manner, the magnitude of decrease depends on the total percentage of load fed by fixed tap transformers and the sensitivity of these loads to modeling     

An integrated air-gap-capacitor pressure sensor and digital readoutwith sub-100 attofarad resolution

The fabrication and characterization of an integrated air-gap-capacitor pressure sensor are presented. The capacitor fabrication process uses standard IC processing to create NMOS circuits, and an added polysilicon layer to create poly-to-n⁺ capacitors with a 0.6-μm-thick dielectric using deposited oxide. Subsequent processing is used to produce deformable, parallel-plate, air-gap capacitors on the front side alongside MOS circuits. Sensor chips are fabricated using 100-μm×100-μm, 100-fF air-gap capacitors with on-chip circuitry. The sensor chip is a part of a capacitive measurement system that uses a charge-redistribution sense technique to achieve very high capacitance resolution. The behavior of the pressure sensor chips was studied as a function of applied pressure in the 0-240-kPa (0-35-psi) range. Measurements indicate a sensitivity of 0.93 mV/kPa (6.40 mV/psi) with a deflection of 10 nm/kPa (70 nm/psi) at 0-69 kPa (0-10 psi). Standard deviations indicate a static pressure resolution of 0.54 kPa (0.078 psi), which translates to 30 attofarads at a sampling frequency of 11 kHz     

Landsat-7 ETM+ on-orbit reflective-band radiometric stability and absolute calibration

Launched in April 1999, the Landsat-7 Enhanced Thematic Mapper Plus (ETM+) instrument is in its sixth year of operation. The ETM+ instrument has been the most stable of any of the Landsat instruments. To date, the best onboard calibration source for the reflective bands has been the Full Aperture Solar Calibrator, a solar-diffuser-based system, which has indicated changes of between 1% to 2% per year in the ETM+ gain for bands 1-4 and 8 and less than 0.5%/year for bands 5 and 7. However, most of this change is believed to be caused by changes in the solar diffuser panel, as opposed to a change in the instrument's gain. This belief is based partially on vicarious calibrations and observations of "invariant sites", hyperarid sites of the Sahara and Arabia. Weighted average slopes determined from these datasets suggest changes of 0.0% to 0.4% per year for bands 1-4 and 8 and 0.4% to 0.5% per year for bands 5 and 7. Absolute calibration of the reflective bands of the ETM+ is consistent with vicarious observations and other sensors generally at the 5% level, though there appear to be some systematic differences.     

Three-dimensional inverse scattering of a dielectric target embedded in a lossy half-space

A modified iterative Born method is applied for three-dimensional inversion of a lossless dielectric target embedded in a lossy half-space. The forward solver employs a modified form of the extended Born method, and the half-space Green's function is computed efficiently via the complex-image technique. Example results are shown, with all scattering data based on a computational model, utilizing a rigorous forward solver distinct from that employed in the inversion. In addition, distinct gridding schemes are used in the forward and inverse solvers. Simple Tikhonov regularization is found to yield adequate results for inversion of noisy data.     

A reliable and manufacturable method to induce a stress of >1 GPa on a P-channel MOSFET in high volume manufacturing

This letter discusses a reliable and manufacturable integration technique to induce greater than 1 GPa of stress into a p-channel MOSFET, which will be required to increase the drive current beyond 1 mA//spl mu/m at the sub-90-nm process generation. Uniaxial compressive stress is introduced into the p-channel by both a selective deposition of SiGe in the source/drain and an engineered 2.5-GPa compressively stressed nitride. The highest to date compressively stressed SiN film is obtained by heavy ion bombardment during the deposition of the film.     

Transmission improvement in ultralong dispersion-managed soliton WDM systems by using pulses with different widths

It was shown that one can improve transmission performance in ultra-long-haul wavelength-division-multiplexed (WDM) systems by matching the input pulsewidth to the path-averaged dispersion (PAD) in the line. Wider pulses should be used for higher values of PAD, and vice versa. It was also noted that this pulsewidth selection is only effective in the dispersion-managed soliton (DMS) propagation regime but not in the chirped return-to-zero (CRZ) regime.