Analysis by an expert system of initial irrigation canal renovation project planning

Initial irrigation canal renovation project planning stages were specifically analyzed since these stages are not conventionally systematically performed. Following extensive interviews with project planning experts, two main observations surfaced, i.e., experts make plants to cope with estimated future degraded canal conditions rather than present conditions, and they heavily rely on a set of easily perceived logics between task completion effects.A model system is developed which is set into operation when a present state list (vector), which includes seven values representing the results of a present condition evaluation, is input to the model. The larger the degree of observed problems, the larger the assigned values. The model system subsequently estimates a future state by adding estimated future degradation factors to values in the list.The actual project plan is considered as an aggregate of countermeasures (task elements) which reduce future troubles. A task element's effect is expressed by a task effect list which is subtracted from the future state list, with the smaller the remaining values, the better effect the task element has on the project. This process is repeated using a generic decision tree formed by various task elements to select the best, most effective set of task elements. Each element in the tree is also related to previously assigned monetary and environmental costs. The model then recommends several best combinations using the selected best task elements, thereby forming an entire set of project plans by calculating the combined effects of the selected result lists.This model system was applied to several irrigation canals in the Tone River Basin, Japan and produced practical results which showed good agreement with expected expert decisions.