OCR Text |
Show R a V S O i l (Pavilion~) Pavilion Technologies, Inc. Process Insights™ Adaptive Software for Optimization and Control If you're in manufacturing, you well understand that real-time optimization and control are critical for reducing costs, improving quality, and minimizing environmental impact. Toward these ends, plants have been extensively equipped with low-cost, reliable sensors and with sophisticated distributed-control systems. In short, the average plant has been transformed into a data-rich environment, in which many megabytes of production data are stored daily. The technology and tools to effectively use this ocean of data, however, have lagged far behind. What if you could use this data to accurately model a production unit-even one with many variables and nonlinearities-in a short time without disrupting production? What if you could optimize this unit in real time and accurately predict process events long before they occur? What if you could do all this without knowing the interrelationships between process variables-in effect, without specifying any rules, first principles, or statistical correlations? What if all this were possible with just a few simple commands? Process Insights from Pavilion Technologies has already enabled several leading petrochemical companies to do just this. The results have been compelling. For example, in one 30-year old chemical process line that had already been exhaustively modeled, Process Insights improved product quality by 3 0 % . In another large chemical unit operation, Process Insights has saved hundreds of thousands of dollars annually. In fact, the competitive advantages created by Process Insights are such that they are never published because of proprietary reasons. How is Process Insights able to optimize complex processes in just a few days, when older modeling and tuning technologies that require many months or even years of costly development have often failed at such tasks? First, Process Insights is adaptive. Rather than being programmed with rules or algorithms, Process Insights modifies its internal structure in response to historical production data. From this data, Process Insights "learns" the complex mathematical relationships between system inputs and outputs. Using Process Insights, you can model and optimize systems for which the rules or principles of operation are poorly understood. Second, Process Insights automates the powerful new technologies of neural networks, fuzzy logic, and chaotic systems theory and makes them available to the non-expert through a set of simple commands. N o knowledge of the underlying technologies is required. Process Insights places the burden of modeling and optimization on the computer rather than on the production engineer. |