论文部分内容阅读
时间序列预测技术可实现过程参数未来变化趋势的早期预报,从而为分析判断工况是否正常、确定转入下一工序的时机提供依据。针对间歇过程数据长度短、非线性、动态、不同批次数据不等长等特点,提出了一种基于相空间重构-最小二乘支持向量机的非线性时间序列预测方法。首先将多批次数据随机的拼接组成长数据向量,差分处理后采用相空间重构关联积分C-C方法计算该序列的延迟时间τ和嵌入维数m,从而构建训练集和检验集,然后采用最小二乘支持向量机算法建立预测模型。对某间歇蒸馏过程上升气温度建立的5步预测模型可用于生产现场的在线预报。
Time series forecasting technology can realize the early forecasting of the trend of the process parameters in the future, so as to provide basis for analyzing and judging whether the working conditions are normal or not and determining the timing of the next process. Aiming at the short data length of batch process, nonlinearity, dynamics and different batches of data, a nonlinear time series prediction method based on phase space reconstruction - least square support vector machine is proposed. Firstly, the long data vector is composed by splicing multiple batches of data randomly, and the delay time τ and embedding dimension m of the sequence are calculated by using the phase-space reconstruction correlation score CC after differential processing, so that the training set and the test set are constructed, and then the minimum Two-Support Vector Machine Algorithm to Establish Predictive Model. The 5-step prediction model established for the rising temperature of a batch distillation process can be used for online forecasting at the production site.