铜精炼过程中液化气流量实测数据有效性检测方法

来源期刊：中南大学学报(自然科学版)2010年第4期

论文作者：鄂加强 李娟 邓元望 彭雨 李光明 贾国海

文章页码：1410 - 1417

关键词：神经网络；混沌优化算法；模糊；有效性

Key words：neural network; chaotic optimization algorithm; fuzzy; validity

摘    要：针对铜精炼过程中液化气流量测量系统各传感器本身不精确、易受环境噪声以及人为干扰等因素的影响等问题，利用模糊综合评判理论和自动获取权重的混沌优化神经网络方法，通过对隶属函数的选择和综合评判权重的自适应调整，识别突变引起的误差以及连续长时间出现的误差并剔除变化较小的误差，从而提出一种铜精炼过程中液化气流量实测数据有效性检测方法。研究结果表明：在铜精炼过程中，经有效性检验的液化气标准板孔流量测量模型的质量流量最大相对误差小于2.75%，与不经有效性检验的质量流量最小相对误差相比至少低2.0%，有效地解决了实际复杂环境中多传感器实时数据采集的野值所导致的实测结果精度降低问题。

Abstract: Based on the effects such as the imprecision of sensors, environmental noise and human-caused disturbance in measurement system about liquefied petroleum gas flux in copper refining process, a validity test method of measurement data about liquefied petroleum gas flux in copper refining process was presented by using fuzzy comprehensive evaluation and chaotic optimization neural network algorithm with the ability of getting weight automatically, the errors from signal jump and the longtime errors were recognized and the errors of small rate were eliminated using subject function selection and adaptive adjustment of comprehensive evaluation. The results show that maximum relative error can be less than 2.75% and the relative error is less than 2.0% compared with that not being tested, which can solve the problems of low precision caused by the outliers from multi-sensor information fusion data in complicated environment.

基金信息：湖南省自然科学基金资助项目