(中南工業(yè)大學(xué) 熱工設(shè)備仿真與優(yōu)化研究所， 長(zhǎng)沙 410083)
摘 要: 根據(jù)灰色理論，采用新陳代謝灰色建模法對(duì)銅锍品位的歷史數(shù)據(jù)建立了GM(1， 1)模型群，并對(duì)各維模型進(jìn)行了精度檢驗(yàn)，計(jì)算表明維數(shù)為4～6時(shí)模型精度達(dá)A級(jí)， 維數(shù)繼續(xù)增大則模型的精度變差。 選出精度高的模型對(duì)當(dāng)前加入連續(xù)吹煉爐的銅锍的品位進(jìn)行預(yù)測(cè)并做均值化處理， 采用此法對(duì)現(xiàn)場(chǎng)90余班次的數(shù)據(jù)進(jìn)行了預(yù)測(cè)計(jì)算， 與化驗(yàn)值相比， 預(yù)測(cè)值的平均絕對(duì)誤差在±0.5%以內(nèi)。
關(guān)鍵字: 灰色預(yù)測(cè)； 灰色模型； 銅锍品位；吹煉
（Institute on Simulation & Optimization of PyroInstallations,Central South University of Technology， Changsha 410083, P.R.China）
Abstract:Based on the grey theory, by adopting metabolism grey modeling method, a series of GM(1, 1) models for the historical matte grade data were developed, and the accuracy of each model was also tested.The result shows that the grade of the models' accuracy is A as the dimension of the sequence is from 4 to 6. The grade of matte currently added in the continuum converting furnace can be gained by averaging the predicted value of the highaccuracy models. The comparison betwe, en the predicted value and the assay result shows an average absolute error only ±0.5%.
Key words: grey predict; gray model; matte grade; converting
