The blackouts that plagued Chicago and other cities this summer could soon be a thing of the past. An engineering consortium led by Purdue University professor Lefteri H. Tsoukalas is developing a computerized system that imbues the power grid with intelligence. The system accounts for environmental conditions and predicts the power needs of its customers based on their usage history. It automatically meets those demands by managing the distribution of electricity in the energy grid. If demand outweighs supply, say on a hot summer day, the system knows to supplement the grid with power from small backup generators.
The smart system, dubbed Telos, uses "neuro-fuzzy" technology, which combines neural networks and fuzzy-set logic to make its predictions. Neural networks mimic the human brain's ability to spot emerging trends based on past experience, while fuzzy logic emulates the human tendency to think in general terms, not precise mathematical formulas. This enables Telos to make quick decisions without solving equations that specify local conditions in minute detail.
Today, utility company workers have used temperature readings and weather forecasts to estimate
customer' energy demands. That can be dangerous: Distant parts of the same service area can experience distinctly different weather conditions.
Commonwealth Edison Co. and the Tennessee Valley Authority, both consortium members, are planning to test Telos by 2001. If it works, the consortium will make the technology available to utilities nationwide.