Please choose your delivery country and your customer group
Thrust has been to develop those new concepts of probability that show promise of being able to model some of the significant sources of uncertainty encountered in a power system or electric utility operation that are not well-modeled by the usual concept of numerical probability as formalized by Kolmogorov. Examples of relevant sources of uncertainty that are not amenable to conventional probabilistic models include most variables encountered in long term forecasts and in modeling such natural phenomena as weather and the availability and distribution of fuel resources. Our research has primarily focussed on the relatively new and little explored concept of interval-valued or upper and lower probability (U/LP). The axiomatic structure of U/LP may be thought of as a pair of nonnegative numbers in the unit interval that simultaneously represent the tendency for an event in question (interval midpoint, say) to occur and our lack of knowledge concerning this tendency (interval width). We have contributed to the development of the theory associated with U/LP in such areas as the generation of useful and tractable specific models (e.g., analogues to specific probability models like the Gaussian), the calculus of the random quantities described by U/LP (e.g., sums of random variables), and attempted to develop better definitions of independence and conditioning (learning from experience). We have applied the concept of U/LP to the problem of aggregating group/expert opinion so as to better represent the state of knowledge of an electric utility planning group doing long range planning and forecasting. We believe that our research has been successful in this area in that we can formally represent both consensus and disagreement in the opinions of a group of experts whereas the approaches based upon the conventional notion of numerical probability cannot display disagreement. (ERA citation 10:031161)