Based On Rough Set Lead And Zinc Smelting Process Incomplete Information Intelligent Processing Methods And Applications

The Lead-Zinc Smelting Process of Imperial Smelting Furnace is a complicated process with complex mechanism, bad environment, multi-parameters, strong nonlinear, time varying and time-delay, strong coupling, and there exists incomplete and uncertain information in the process. On the basis of analysis of the information characteristics, the intelligent methods based on rough set is studied to deal with the incomplete information in the smelting process.Considering the data-missing and the various types of attributes of process information, the similar relationship between different objects is discussed under qualitative attribute or quantitative attribute. An expanded multi-attribute similar relationship is defined under these two attributes, and the similitude degree is imported to describe the relationship between different objects. Based on the fused similar relationship and the similitude degree, a new data complement method is proposed, and the ensample validating process and results show its validity and feasibility.Considering the uncertain data in the process information system, the heuristic reduction algorithm based on rough set theory is presented to deal with the uncertain information. After analyzing the shortage of algorithm in real application, an advanced attribute-importance computing method is proposed based on expert experiences and the dependence of attributes, in which the expert experiences is an important guide of attribute importance. Based on the new computing method, an improved heuristic reduction algorithm is proposed, and the ensample validating process and results show its validity and feasibility.The intelligent methods are applied to the fault diagnosis of the smelting process, and a fault diagnosis model based on rough set and neural networks is proposed by analyzing the fault in the process. The multi-attribute based data complement method above is firstly used to complete the missing data of the incomplete information system, and the advanced heuristic reduction algorithm is then used to reduce the redundant information. Then, the gained relative complete information system serves as the training data of fault diagnosis neural network. The simulation results show that, by using the intelligent method when modeling, the structure of neural network model is optimized and simplified, and the diagnosis efficiency is improved effectively.