Remote downstream feedback control of branching canal networks.

摘要

The main purpose of an irrigation water delivery system is to deliver water to users at the desired time, rate, frequency, and duration. Most operators of irrigation water delivery systems operate the canal using manual techniques, which is a difficult and time-consuming process. Thus, some canal operators have turned to automatic control techniques in an attempt to more efficiently control irrigation water delivery systems. Although many attempts have been made to automatically control in-line irrigation water delivery systems, nothing has been done to control an entire branching network of canals. In practice it is desirable to automate an entire canal network instead of just one branch of the network.; Recently, researchers at the U.S. Water Conservation Laboratory (USWCL) in Phoenix, AZ have developed an automatic control system for in-line canals that appears promising because of the ease with which the feedback controller can be tuned. Model Predictive Control (MPC) is a control algorithm that has been applied in many industrial settings. Recently, MPC has also been applied to the automatic control of in-line irrigation water delivery systems. Both the USWCL automatic control system and MPC consist of three parts: (1) a feedback controller to account for unknown disturbances and flow measurement errors, (2) a feedforward controller based on the volume compensation method to route known flow changes through the system, and (3) a local flow control at the individual check structures.; The USWCL automatic control system and MPC were examined and modified so that they could be applied to branching canal networks. Simulations were performed on a large portion of the irrigation water delivery system operated by the Salt River Project to verify that the modified controllers could effectively control branching canal networks. Both the USWCL automatic control system and MPC were able to control the water levels in the branching canal network. Typically, the USWCL automatic control system performed better than MPC because of inadequacies in the process model that was used to predict the system output in the future. When the feedforward portions of the control algorithms were not used, then MPC performed better than the USWCL automatic control system.