Fault Diagnosis Technology And Application Of Pb-Zn Manufacture Process In The Imperial Smelting Furnace

The smelting manufacture with Imperial Smelting Furnace(ISF) is the main part of Lead-Zinc pyrometallurgical smelter processes, which is a complicated physicochemical reaction process accompanied with hyperthermia, multiphase, including many characteristics such as multivariate, nonlinearity, strong coupling , large delay , uncertainty and so on, the key process parameters in which have no way to check directly or to be described with accurate mathematic model as well. Most of parameters are decided empirically by operators currently, so the production process is not able to run steadily, and the Furnace conditions fluctuate greatly, the faults arise frequently, the period of damping down is short and the technical index in the production process fluctuates, all of those contribute a lot to the poor output and quality of Lead-Zinc. Therefore, to ensure smooth and stable operation in ISF and decrease the frequency of blast to avoid faults arising make sense in many aspects, such as Energy saving and consumption reducing, promoting the efficiency of the resource and smelting units as well as exploiting the potential crop in smelting process sufficiently, upgrading the technical and economic indexes in smelting process, promoting sustainable development of the enterprises and so on.The corresponding fault diagnosis methods and technology in the melting process in order to improve the productivity of the imperial smelting furnace are researched and presented in this thesis, which aims to get energy saving and consumption reducing with high-grade and high-yield products. Firstly, the research actuality in fault diagnosis and the characteristic of melting process of imperial smelting furnace are summarized. Secondly, several corresponding fault diagnosis technology are presented after various faults are carried out analysis and classification in detail making use of the technology of statistics, data mining and case reasoning respectively. These technology introduced are respectively based on Principal Component Analysis, on incomplete data sets, on the integration of the procedure parameters forecasting and case reasoning. All of those are applied to the fault diagnosis and decision-making system in the melting process of the imperial smelting furnace, and take good effects. The main research contributions of this thesis are as follows:(1) It Analyses the melting process mechanism of the imperial smelting furnace in detail, makes a futher study of the fault characteristic, causes and treatment in the melting process, and then divides the faults into two major categories and 16 sub-categories based on the foregoing analysis.(2) It makes use of PCA(Principal Component Analysis) analytical method firstly to extract the major monitoring variables which are closely correlate to the faults since there are lot of monitoring variables which are complicate to the corresponding faults. The traditional PCA method can only monitor if the procedures have change, without providing the cause to the change, which is difficult to be used in progress to faults identification. Hence, to replace the traditional SPE(Squared Prediction Error) statistic, a new statistic variable is constructed by using the predicted residual of the process variables which significantly correlate to the pivot elements, calculating the multiple correlation coefficient between the process variables and the pivot elements, and choosing these variables correlate to pivot element, which constructs an improved PCA. This method is able to discover ten sub-categories faults in time such as the cone fault, blast tuyere faults, air leakage faults, water loss faults and so on. This fault diagnosis method turns out to be well done in application to ISF, which gives a successful example on using of improved PCA in engineering.(3) Since the hang-ups problem in ISF appears frequently, a method of fault diagnosis owing to rules reasoning according to the characteristic of these faults is brought forward. Firstly, an improved method is presented to discrete the property variables based on subtraction clustering for the problem of continuous variables discovered according to the knowledge of the hang-ups faults diagnosis, and provides the algorithm to make the continuous property variables crisp discrimation and fuzzy discrimation; Secondly, a knowledge acquisition technology based on incomplete data set via the concepts of the equivalent class in rough set theory is proposed, aiming to solving the problem of incomplete data set. This technology solves the problem of being unable to evaluate the sealed eigenvalue of the training instance property while extracting the rules at the same time. The hang-ups rules library of imperial smelting furnace constructed based on this method is able to estimate the trends of hang-ups and the arising position, which leads to taking correspond measures to prevent the situation from being severe, supplying a new approach to the research and application in knowledge acquisition in incompleted data set.(4) The accretion in the furnace is one of the grave faults of imperial smelting furnace, this thesis analyses the cause and the characteristic of the furnace throat accretion, flue accretion, furnace accretion, condense accretion and so on, brings forward the method of using the gas permeability and the variety Zinc-output per unit time to describe the accretion states, as well as the intelligent diagnosis method integrating the key parameters forecasting with case reasoning. This method turn out to be able to built the gas permeability forecasting model to realized the ISF permeability forecasting base on neural networks and gray theory, built Zinc-output forecasting model online making use of vector machine of knowledge, taking the gas permeability and Zinc-output per unit time as eigenvalues and using the history faults data to simulate the field experts how to treat these faults to build an intelligent fault diagnosis model based on the parameter forecasting integrate with case reasoning. The practice shows that this method is well fit for accretion diagnosis and decision-making system.(5) It constructs the intelligent integration system structure for the fault diagnosis system aiming to the fault diagnosis characteristic of the imperial smelting furnace, and designs the intelligent integration system for fault diagnosis and decision-making. This intelligent system is able to predict the furnace running status and trends, is also able to alarm and diagnose abnormal parameters and abnormal furnace status, and provide the operational guidance to avoid the furnace operating status of blindness. Consequently, it effectively improves the output and quality of the Lead-Zinc to gain bigger economic effect. The application effect in practice indicates that the technology routes and the intelligent fault diagnosis methods proposed in this thesis are available and robust, they are able to be propagated in the quasi complex colored metal melting process conveniently.