摘要
本文针对具有小样本数据的N型热电偶在应用中存在的问题,提出了一种基于最小二乘支持向量机对热电偶进行非线性校正的方法,并与以往采用的BP网络、RBF网络和ANFIS校正方法进行了比较。结果表明,采用最小二乘支持向量机的校正精度高于以上3种校正方法;同时以阳极焙烧过程中料箱温度为对象进行了仿真和实际应用研究,取得了满意的结果。
Aiming at the problem existed in the application of N type thermocouple with small data samples, the least squares support vector machine （LSSVM） method for correcting nonlinear error of thermocouple sensor is introduced. This method is compared with some commonly used calibration methods, such as BP neural network, RBF neural network and ANFIS method. The result of experiment shows that the nonlinear calibration method based on LSSVM has higher precision than the methods based on BP, RBF or ANFIS. LSSVM method was used to test fire-path temperature in anode baking process, and satisfactory result was achieved. Test result proves that the method is effective.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2007年第4期640-644,共5页
Chinese Journal of Scientific Instrument
基金
国家科技攻关计划基金(2002BA901A28)
甘肃省省长基金(GS015-A52-012)资助项目
