The Study Of Learning Control Algorithm For Automatic Irrigation Ditch Systems

China is abundant in water resources,but its water resources are unevenly distributed in space.Therefore,it is often required to transfer water from a long or short distance,and employ irrigation ditch to bring water from the source region to farmland.The automation of irrigation ditch helps to control the operation of ditch in a scientific way,prevent insufficient water supply or the waste of water resources during irrigation,liberate labor force and reduce operation costs caused by overdue dispatch.Farm irrigation is featured by periodicity in time and repeatability in process.The irrigation ditch system is also non-linear,indeterminate and strong coupled.Aimed at the above-mentioned characteristics,this essay has adopted the iterative learning control method to control the operation of irrigation ditch,and it mainly concludes the following contents:Firstly,it has studied the iterative learning control related to the dam-river ditch irrigation system.After simplifying the Saint-Venant equations,it has linearized the equation of the dam-river ditch irrigation system and obtained the Hayami partial differential equation.In space,the Hayami equation is discretized through employing backward difference scheme.By superposing vector quantities,the state-space mathematical model between the gate opening gauge and flow in the upriver is established and the D-type algorithm of iterative learning control is designed.In addition,the convergence condition of the algorithm is also given and proven.According to the results,through the iterative learning control of the gate opening gauge in the upriver,the output flow in ditch can completely track the expected flow.Secondly,it has studied the iterative learning control related to the prismatic single-ditch irrigation system.The Saint-Venant equations that describes the prismatic single-ditch irrigation system has been discretized in time and space.The discretization scheme is the Crank-Nicolson scheme with unconditional stability.The state-space mathematical model between the flow regulated by gate and water level relation in the upper and lower reaches is established and the P-type algorithm of iterative learning control is designed.In addition,the convergence condition of the algorithm is also given and proven.According to the results,through the iterative learning control of the gate both in the upriver and in the downriver,the flow in ditch can completely track the expected flow.Thirdly,it has studied the iterative learning control related to the large prismatic multiple-ditch irrigation system that is more common in reality.The Saint-Venant equations that describes the prismatic multiple-ditch irrigation system has been discretized in time and space based on the Crank-Nicolson scheme.The state-space mathematical model between the flow regulated by all gates and water level relation is established and the P-type algorithm of iterative learning control is designed.In addition,the convergence condition of the algorithm is also proven.According to the results,through the iterative learning control of all gates,the flow in ditch can completely track the expected flow.Furthermore,with regard to the iterative learning control of dam-river ditch irrigation system and prismatic single-ditch irrigation system,it has also provided the algorithm flow chart and simulation examples of iterative learning.The simulation result has further shown the effectiveness of the proposed iterative learning control method.