A knowledge-based or expert system has been developed for intelligent monitoring and control of industrial pipeline network operations. The expert system would perform the supervisory and decision-support tasks based on the expertise and operating procedures that are documented in the maintainable knowledge base. Since it is the first expert system in the pipeline network applications, the pipeline network of a municipal water supply system was chosen as a testing domain due mainly to the experts' availability and safety reason. The stages of system development are described from the knowledge acquisition to the implementation stage. The paper presents an engineering concept of energy management that was applied to build part of the knowledge base in the system. The potential advantages of the expert systems are also listed at the end of this paper.
The pipeline network is one of the most essential components in oil and gas transportation industries. For municipal utilities, the pipeline network system is also an important component that distributes water and natural gas to industrial and business consumers as well as residents throughout the city. The pipeline network system normally consists of a series of pipes (with components such as elbows, tees, valves, etc.), pressurized equipment (pumps or compressors), storage rooms (reservoirs or tanks), and field instruments (flow meters, pressure gauges, level sensors, etc.). Generally, all components are installed in the network at various locations so that a remote monitoring and control system is absolutely necessary. The status of equipment and measurement signals from field instruments are typically sent through modems and telephone lines to a main control station. Operators at the station will monitor the incoming data: when emergencies or changing of process variables occur, the operators will take a series of actions to control the process equipment in order to ensure smooth operations.
For the large-scale pipeline network, there would be hundreds of data signals reported to the main control station at every time interval. Ideally, operators should observe all input data reported on the screen(s), analyze them correctly and use them to make proper decision within a few seconds. Of course, the operators who are experts would know how to select the necessary data for solving each type of problems Although expertise is transferable to new operators, some is tacit and difficult to understand within the training courses. The trainees also need a certain period of time to absorb the material and to get used to the system.
An expert system for monitoring and control of the pipeline network operations is useful when the most experienced operators will soon resign or retire, and their knowledge and expertise has not been documented and transferred to aspiring operators. This scenario provides the motivation for capturing expertise and heuristic reasoning of the experts on pipeline network operations using artificial intelligent technologies. The paper describes our efforts at this task.
The domain addressed is a pipeline network of a municipal water supply system of typical moderate-sized prairie cities in North America.