采用支持向量機建立了丙酮精制過程的產(chǎn)品質(zhì)量與生產(chǎn)工藝參數(shù)之間的預(yù)測模型，并
將其與反向傳播神經(jīng)網(wǎng)絡(luò)和徑向基神經(jīng)網(wǎng)絡(luò)模型相比較。在實際工業(yè)數(shù)據(jù)上進行的實驗結(jié)果表明，支持向量機模型對丙酮純度具有良好的預(yù)測效果，性能優(yōu)于反向傳播神經(jīng)網(wǎng)絡(luò)和徑向基網(wǎng)絡(luò)模型。
關(guān)鍵詞：支持向量機; 神經(jīng)網(wǎng)絡(luò); 丙酮精制

Abstract: Support vector machine (SVM) is applied to build a prediction model of the acetone
quality in an industrial acetone refining process. The model is also compared with the models
based on back-propagation neural network (BPNN) and radial basis function neural network
(RBFNN). The experimental results on the real industrial data demonstrate that the model based on SVM achieves good performance and has less prediction errors than those of BPNN and RBFNN models.
Key words: support vector machine; neural network; acetone refining process