Font Size: a A A

Based On The Improved Price Sensitive Support Vector Machine (svm) Evaluation Model Of Copper Flash Smelting Condition Of Research And Application

Posted on:2013-10-17Degree:MasterType:Thesis
Country:ChinaCandidate:S L WuFull Text:PDF
GTID:2241330374488109Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
In the process of copper flash smelting, the matte temperature, matte grade and ratio of Fe to SiO2are three synthesis criterions that can reflect the comprehensive quality of production process. Owing to the lag of the three manual test parameters, it’s difficult to indicate the level of working-status timely, making the operation parameters difficult to adjust, and wasting a lot of resource. But in practical production, lots of industrial operating data are accumulated, which contains abundant potential information that can reflect the current operation level. Digging out the evaluation criterion of the working-status from these data, has important sense to guide the practical production operation parameter adjustment and improve production efficiency. Therefore, based on historical production data and the mechanism analysis of production process, a method of modeling working-status assessment based on the improved cost-sensitive support vector machine is proposed in this paper.First of all, based on the mechanism analysis of copper flash smelting process, the main factors which impact copper flash smelting process condition are analyzed, and on this basis, comprehensive status evaluation standard of the copper flash smelting process is provided, and technical difficulties of working-status assessment are discussed.To estimate the working-status of copper flash smelting process timely and effectively, a method of working-status assessment based on improved cost-sensitive support vector machine is proposed in this paper. In the method, a local kernel function and a global kernel function are fused to form a new kernel function, in order to improve learning and generalization ability of support vector machine, then the particle swarm optimization is used to get the best kernel function weights, parameters, as well as the cost of misclassification, so as to avoid the human factor in setting the parameters of support vector machine. Based on the proposed method, and combined with the practical copper flash smelting process, the working-status assessment model of copper flash smelting process based on improved cost-sensitive support vector machine is established. The effectiveness and practicality of the method are fully proved by historical production data simulation.Finally, by the operation data of copper flash smelting process and the working-status assessment model that has been established, the copper flash smelting process working-status assessment system is developed which includes architecture design, implementation steps and so on.
Keywords/Search Tags:copper flash smelting, working-status identification, support vector machine, cost-sensitive learning
PDF Full Text Request
Related items