Research On Coal Slime Floatation Equipment With Dosing Expert System And Parameters Measuring

Floatation is one of efficient ways of coal slime select, automatic dosing in floatation process is an important means to ensure floatation clean coal ash and to raise clean coal yield. At present, conventional feed-forward control methods cannot meet the influence from coal quality fluctuation in coal preparation plant, more dosage control operation rely on the experiences of workers, large subjective factors, severe fluctuations in product indicators, pharmaceutical consumption keeps high.Expert system is an important branch of artificial intelligent control field, suitable for complex industrial process which is difficult to establish an accurate mathematical model for control. Expert system has been widely used, and is a proven and effective method. Many factors affect the floatation process, according to the fact that the worker who dosing manually rely on experience get better product than the conventional feed forward control, use for reference expert system theory and technology, make use of expert systems to improve the automatic dosing system of coal slime floatation process and enhance the system adaptability on the impact from coal quality fluctuation.With the standard methods and procedures of ES, to design and develop an expert system of coal slime floatation dosing. First of all, based on the laboratory experiments of coal samples, optimize reagent addition institution and obtain a primary expert rules, and then expand and alter the rules according to the knowledge of industry site floatation drivers and experienced technical staff (experts) in the field, establish the Changchun coal preparation plant ES knowledge base, at last, take the system to industrial actuality, when running in the scene, this is still need further refinement and amendment of expert database, inference engine and the conflict resolution mechanism.Floatation indicators as well as floatation clean coal ash and yield cannot be measured by currently on-line sensors. The rise and development of Soft sensor theory and technology supply new ways to solve problems. Soft sensor is built on the primary automation can only achieve the implementation, ES establishment of floatation dosing system, to provide essentially the necessary basic conditions for soft sensor implementation.Based on the review and summary of Soft sensor technology, with the aim of industrial systems development, it focus on the study of floatation indicators simulation with the three algorithms:Partial Least Square(PLS), Back Propagation neural network (BP) and General regression neural network(GRNN) which are maturity and reliable. Computer simulation with concentration of floatation feed, and the two dosages as auxiliary variables, floatation clean coal ash and yield as the dominant variable, simulation results show that:targets for soft sensor modeling of coal floatation indicators, PLS algorithm is not applicable, BP algorithm requires a large amount of data is also unrealistic, and GRNN soft sensor modeling method is feasible and appropriate. Simulation results also show that:the model has a great influence on floatation clean coal ash forecast accuracy when the coal changed, so in the further research and development of this system should add the mechanism which can choice feed coal and switch the appropriate model, to satisfy the scene require of the floatation ash prediction accuracy.Based on the simulation, the work of develop soft sensor platform has been accomplished, include design and selection of the control system and actuator. Actuator is the key related to the success or failure of the system, through laboratory tests on extreme agent conditions and tests on measurement accuracy, experiments prove that the design and selection is reasonable and efficient. For the lack of PLC and Kingview in conducting a complex algorithm, the thesis proposes a technology way of software integrate development, through the DDE technology to achieve an organic combination of MATLAB and Kingview environments. It focuses on the data sharing and exchange methods between Upper and Lower, MATLAB and Kingview, to improve efficiency, and ensure the quality of the algorithm.The coal floatation dosing expert system and floatation indicators soft sensor model, discussed in this thesis, are about to use in industrial actuality, as the expert rules of ES have not yet received the test of industrial actuality, and the soft sensor model library of floatation indicators is relative simple, in the face of complex coal variations and working condition changes, how to improve the accuracy and stability of soft sensor model, ensure stable and reliable operation of the system still need to do further research and exploration.