| 摘要 |
A system and method for the automated maintenance of a computer system. A scheduler periodically activates sensors. When activated, the sensors gather information about various aspects of the computer system. The sensors store this information in a knowledge database. The knowledge database also contains cases, questions, and actions. The cases describe potential computer problems and solutions. The questions are used to diagnose the problems while the actions describe steps that can be taken to solve the diagnosed problems. If the information gathered by the sensors indicates a problem with the computer system, then the sensors activate an artificial intelligence engine. The engine uses the information in the knowledge database to evaluate certain cases. If information necessary to evaluate a case is not in the knowledge database, then the engine activates a sensor to gather the information. As the cases are evaluated, the confidence levels of certain other cases, questions, and actions increase. If the confidence level of a case or question rises above a predetermined threshold, then the engine evaluates that case or question. If the confidence level of an action rises above the threshold, then that action represents the likely solution to the problem. Accordingly, the engine activates a sensor to perform the action. If no case, question, or action rises above the threshold, then the knowledge database does not contain enough information to solve the problem. In such a case, the engine saves the state of the computer system and knowledge database. Then, a human expert can update the database with the knowledge necessary to solve the problem.
|
