| In coal preparation plant, slime water treatment system plays a very important role for the reason that the process water must be recycled. Generally, the turbidity of overflow water in thickening tank should below the required level. Unqualified recycled water will have impact on the production quotas such as separating effect, heavy medium consumption and product moisture. Especially, the severe disorder of slime water treatment system will result in anomaly, even shut down, of the overall coal separating system. Due to the great environmental concerns and potential economic reward, in the past decades many efforts have been made by both here and abroad researcher to study the methods of slime water treatment. However, most plants still rely on experienced workers'manual adjustment to control the addition of flocculant nowadays, which may lead to unstable water turbidity. In some factories, the automatic dosing has adopted, but the efficient control schemes to the great inertia and long delay of flocculating sedimentation process is lacking. Therefore, it is required to further study the advanced control methods for the specific problem.In this paper, we make the following contributions:1. We do some experiments on the samples of the coal slime water. From the parameters measured in the experiments, we analyze the properties of coal clime water and the multiple factors related to its sedimentation process. Based on such observations, we conclude that the particular challenges in effective control:large inertia and long latency. It is found that the existing schemes underestimate these problems.2. To solve the system's large latency problem, we propose a novel turbidity detection method, which senses the turbidity at three locations in the vertical direction along the thickener. Compared with the traditional method only observing the turbidity at the overflow pipe, the new scheme is able to discover the sedimentation's situation with reduced delay. We develop grey prediction algorithm to foretell the turbidity in advance so that the delay problem is much alleviated.3. Based on theories about grey prediction and fuzzy PID, we present a new flocculating sedimentation strategy with feed-forward and feedback. The scheme is different from the existing schemes in two ways. Firstly, the traditional fuzzy control method is substituted by the fuzzy PID algorithm, which can adjust the PID parameters in an online manner according to the fuzzy rules. The advantages are twofold:it could avoid the steady state error problem for fuzzy control methods; and it makes the PID parameter setting convenient. Secondly, the turbidity variation vector in the sinking region replaces the turbidity error differentials at clarity region as the second input for the fuzzy controller. Since the trend of turbidity change could be inferred, the large-delay system can be controlled in advance. Besides, the control strategy is practical to implement in harsh industrial environment.4. Through extensive simulations, we study the impacts of a number of settings on the accuracy of prediction, like the grey prediction model dimensions, the fluctuation of data sequence, and prediction step size. Based on PID fuzzy deduction rules, we make domain analysis and design member functions. A virtual model of coal slime water sedimentation has been derived by experiments and system identification method, which discloses an input-output relation of sedimentation process. The results show that the proposed control method is effective for large-delay system, even with interference signals.5. Finally, we design and implement the automatic control system of flocculating sedimentation process which is composed of the display and adjusting unit on-site, lower machine PLC and upper machine PC. With operation ability of PC, grey forecasting process is realized by Mingled-Programming between MATLAB and configuration software. System Sensor data acquisition and the dosage of the fuzzy control were done by PLC. Upper machine and lower machine can exchange data and control information through the industrial Ethernet network. |
