A KNOWLEDGE-BASED CONTROL METHOD - APPLICATION TO ACCELERATOR EQUIPMENT SETUP

An executable description language for defining the structure of large installations and their corresponding control algorithms is presented . This method of knowledge representation is based on software entitie s called ''goal procedures'' for real-time control in an Object-Orient ed manner. The knowledge interpreter is driven by events coming from t he control system and is capable of forward graph traversal and backtr acking. The accelerator equipment setup problem concerns a large colle ction of hardware and corresponding parameter descriptions, equipment initialization and testing, data loading and diagnosis. The proposed m ethod and tools are applied to the CERN/PS accelerator setup for cold start and recovery. This system includes descriptions of the CAMAC har dware and accelerator equipment and the possibility of controlling and diagnosing all stages of the setup process.