Modeling And Control Of Water Separating Slurry Pump

Recently, there comes up a new method of transportation, the slurry pipeline transportation. The advantage of this method is small investment, low operating cost and convenient management. And its key equipment is pumping equipment. It directly affects the slurry pipeline performance. The water separating slurry pump is kind of that equipment, which has the advantage of high lift, large flow, high concentration and long life. Its control requirement is complicated, e.g. control action complex, large change of operation state, poor regularity. It is the system of nonlinear and time varying, so it is difficult to use conventional controller. The performance of water separating slurry pump is slurry mixing and water loss, avoiding slurry mixing and reducing water loss.At first, we analyze the movement of separating ball in the separating tank. Secondly, using hydro mechanical principle, we can get the model of water separating slurry pump in charging slurry process and discharging slurry process. The model is nonlinear and dynamic. By modeling, we can get the cause of mixing slurry and the factors which affect water loss.In the premise of period control, the plan of water separating slurry pump is designed, making the slurry level unchanged, making separating ball move in the center of separating tank and making the speed of separating ball change slowly. So the slurry can be delivered to the tailing reservoir. At the same time, the slurry mixing is avoided and the water loss is decreased.First of all, the water separating slurry pump is controlled by PID controller, but the result is not very good, because the model has the characteristics of nonlinear and time varying. It mainly reflects that the control effect is not very good when the external slurry flow changed. And we attempt to use T-S fuzzy RBF neural network controller instead of PID controller, because T-S fuzzy model can modify the parameters on line and RBF neural network is equal to Fuzzy inference system when RBF neural network satisfy certain conditions. The advantage of T-S fuzzy RBF neural network is robustness, insensitive to disturbance signal, and self-adaptive to the external slurry flow change, which can adjust the parameters automatically. By simulating, we can find that the charging slurry flow can better follow the external slurry, the center position performance is small and water loss is low, we can get a better result using T-S fuzzy RBF neural network.