Modeling and exploiting behavior patterns in dynamic environments

This paper presents a new approach to improving the effectiveness of autonomous systems that deal with dynamic environments. The basis of the approach is to find repeating patterns of behavior in the dynamic elements of the system, and then to use predictions of the repeating elements to better plan goal directed behavior. It is a layered approach involving classifying, modeling, predicting and exploiting. Classifying involves using observations to place the moving elements into previously defined classes. Modeling involves recording features of the behavior on a coarse grained grid. Exploitation is achieved by integrating predictions from the model into the behavior selection module to improve the utility of the robot's actions. This is in contrast to typical approaches that use the model to select between different strategies or plays. Three methods of adaptation to the dynamic features of the environment are explored. The effectiveness of each method is determined using statistical tests over a number of repeated experiments. The work is presented in the context of predicting opponent behavior in the highly dynamic and multi-agent robot soccer domain (RoboCup)

Dynamical local models for segmentation and prediction of financial time series

In the analysis and prediction of many real-world time series, the assumption of stationarity is not valid. A special form of non-stationarity, where the underlying generator switches between (approximately) stationary regimes, seems particularly appropriate for financial markets. We introduce a new model which combines a dynamic switching (controlled by a hidden Markov model) and a non-linear dynamical system. We show how to train this hybrid model in a maximum likelihood approach and evaluate its performance on both synthetic and financial data.

Majority versus minority influence and prediction of behavioral intentions and behavior

Two experiments examined the effects of majority and minority influence on attitude-consistent behavioral intentions. In the first experiment, when attitudes were changed via minority influence there was a greater likelihood to engage in an attitude-consistent behavioral intention than when attitudes were changed via majority influence. This suggests that minority influence leads to stronger attitudes (based on systematic processing) that are more predictive of behavioral intentions, while attitude change via majority influence is due to compliance through non-systematic processing. Further support for this interpretation comes from the finding that the amount of message-congruent elaboration mediated behavioral intention. When there was no attitude change, there was no impact on behavioral intention to engage in an attitude-consistent behavior. Experiment 2 explored the role of personal relevance of the topic and also included a real behavioral measure. When the topic was of low personal relevance, the same pattern was found as Experiment 1. When the topic was of high personal relevance, thus increasing the motivation to engage in systematic processing, attitudes changed by both a majority and minority source increased behavioral intention and actual behavior. The results are consistent with the view that both majorities and minorities can lead to different processes and consequences under different situations. © 2006 Elsevier Inc. All rights reserved.

CFD-modeling and experiments of insulation debris transport phenomena in water flow

The investigation of insulation debris generation, transport, and sedimentation becomes more important with regard to reactor safety research for pressurized water reactors and boiling water reactors when considering the long-term behavior of emergency core coolant systems during all types of loss-of-coolant accidents (LOCAs). The insulation debris released near the break during a LOCA incident consists of a mixture of disparate particle populations that varies with size, shape, consistency, and other properties. Some fractions of the released insulation debris can be transported into the reactor sump, where it may perturb/impinge on the emergency core cooling systems. Open questions of generic interest are, for example, the particle load on strainers and corresponding pressure drop, the sedimentation of the insulation debris in a water pool, and its possible resuspension and transport in the sump water flow. A joint research project on such questions is being performed in cooperation with the University of Applied Sciences Zittau/Görlitz. The project deals with the experimental investigation and the development of computational fluid dynamics (CFD) models for the description of particle transport phenomena in coolant flow. While the experiments are performed at the University of Applied Sciences Zittau/Görlitz, the theoretical work is concentrated at Forschungszentrum Dresden-Rossendorf. In the current paper the basic concepts for CFD modeling are described and feasibility studies including the conceptual design of the experiments are presented.

Design, modeling and construction of a novel ablative fast pyrolysis reactor and product collection system

The objective of this work was to design, construct and commission a new ablative pyrolysis reactor and a high efficiency product collection system. The reactor was to have a nominal throughput of 10 kg/11r of dry biomass and be inherently scalable up to an industrial scale application of 10 tones/hr. The whole process consists of a bladed ablative pyrolysis reactor, two high efficiency cyclones for char removal and a disk and doughnut quench column combined with a wet walled electrostatic precipitator, which is directly mounted on top, for liquids collection. In order to aid design and scale-up calculations, detailed mathematical modelling was undertaken of the reaction system enabling sizes, efficiencies and operating conditions to be determined. Specifically, a modular approach was taken due to the iterative nature of some of the design methodologies, with the output from one module being the input to the next. Separate modules were developed for the determination of the biomass ablation rate, specification of the reactor capacity, cyclone design, quench column design and electrostatic precipitator design. These models enabled a rigorous design protocol to be developed capable of specifying the required reactor and product collection system size for specified biomass throughputs, operating conditions and collection efficiencies. The reactor proved capable of generating an ablation rate of 0.63 mm/s for pine wood at a temperature of 525 'DC with a relative velocity between the heated surface and reacting biomass particle of 12.1 m/s. The reactor achieved a maximum throughput of 2.3 kg/hr, which was the maximum the biomass feeder could supply. The reactor is capable of being operated at a far higher throughput but this would require a new feeder and drive motor to be purchased. Modelling showed that the reactor is capable of achieving a reactor throughput of approximately 30 kg/hr. This is an area that should be considered for the future as the reactor is currently operating well below its theoretical maximum. Calculations show that the current product collection system could operate efficiently up to a maximum feed rate of 10 kg/Fir, provided the inert gas supply was adjusted accordingly to keep the vapour residence time in the electrostatic precipitator above one second. Operation above 10 kg/hr would require some modifications to the product collection system. Eight experimental runs were documented and considered successful, more were attempted but due to equipment failure had to be abandoned. This does not detract from the fact that the reactor and product collection system design was extremely efficient. The maximum total liquid yield was 64.9 % liquid yields on a dry wood fed basis. It is considered that the liquid yield would have been higher had there been sufficient development time to overcome certain operational difficulties and if longer operating runs had been attempted to offset product losses occurring due to the difficulties in collecting all available product from a large scale collection unit. The liquids collection system was highly efficient and modeling determined a liquid collection efficiency of above 99% on a mass basis. This was validated due to the fact that a dry ice/acetone condenser and a cotton wool filter downstream of the collection unit enabled mass measurements of the amount of condensable product exiting the product collection unit. This showed that the collection efficiency was in excess of 99% on a mass basis.