论文部分内容阅读
针对顶吹熔炼系统喷枪寿命预测原始生产数据含噪量大以及单一预测模型容易失效的问题,提出一种基于核主元分析法(KPCA)的PSO-WLSSVM组合预测方法。首先利用KPCA对原始生产数据进行去噪处理,通过贡献率对样本降维,提取样本中的非线性主元信息,然后用粒子群算法(PSO)优化WLSSVM的两大主要参数,从而避免了经验法的弊端。与PSO-WLSSVM和WLSSVM模型进行对比,实验结果表明,KPCA-PSO-WLSSVM模型对喷枪寿命预测的可信度和准确度较高,为下一步提出维修策略奠定了基础。
Aiming at the problem of high noise in the raw production data of the spray gun life prediction system of the top-blown smelting system and the failure of the single prediction model, a combined PSO-WLSSVM forecasting method based on kernel principal component analysis (KPCA) is proposed. Firstly, the KPCA is used to denoise the original production data, reduce the dimension of the sample by the contribution rate, extract the nonlinear principal component information from the sample, and then optimize the two main parameters of WLSSVM by using Particle Swarm Optimization (PSO), thus avoiding the experience Law malpractice. Compared with the PSO-WLSSVM and WLSSVM models, the experimental results show that the KPCA-PSO-WLSSVM model has a higher reliability and accuracy in predicting the life expectancy of the spray gun, which lays the foundation for the next step of the maintenance strategy.