Real-time architecture for knowledge processing systems

The paper describes an architecture for knowledge processing systems which is designed to satisfy hard real-time requirements. To achieve this, the problem of knowledge processing is separated into interacting subsystems, each processing more specific knowledge. These subsystems efficiently evaluate their knowledge according to the real-time ARON-technique (alternatives regularly organized and numbered) which is described in more detail. Proposed systems enable users to specify operational knowledge without the aid of a knowledge engineer. Implementation is easy on various computing systems, even directly in hardware. The applicability and efficiency of the suggested architecture is demonstrated by two examples of knowledge based control, namely control of a simulated biological tank reactor and of a real three phase AC motor.<
>