During the course of calculating the rice evapotranspiration using weather factors,we often find that some independent variables have multiple correlation.The phenomena can lead to the traditional multivariate regress...During the course of calculating the rice evapotranspiration using weather factors,we often find that some independent variables have multiple correlation.The phenomena can lead to the traditional multivariate regression model which based on least square method distortion.And the stability of the model will be lost.The model will be built based on partial least square regression in the paper,through applying the idea of main component analyze and typical correlation analyze,the writer picks up some component from original material.Thus,the writer builds up the model of rice evapotranspiration to solve the multiple correlation among the independent variables (some weather factors).At last,the writer analyses the model in some parts,and gains the satisfied result.展开更多
煤炭灰分值是衡量煤炭质量的关键指标之一,灰分含量和性质对燃烧设备、环境、后续的加工利用都有着极大影响。针对目前煤炭灰分检测方法的滞后性、劳动密集型问题,提出了一种基于XRF光谱的预处理(Preprocessing,PRE)与偏最小二乘法(Part...煤炭灰分值是衡量煤炭质量的关键指标之一,灰分含量和性质对燃烧设备、环境、后续的加工利用都有着极大影响。针对目前煤炭灰分检测方法的滞后性、劳动密集型问题,提出了一种基于XRF光谱的预处理(Preprocessing,PRE)与偏最小二乘法(Partial Least Squares,PLS)相结合的XRF煤炭灰分智能预测算法。通过将XRF技术获取煤炭样品的光谱数据输入PLS主模型初步预测灰分,再将相关校正参数输入补偿优化模型中,最终将两者相加得到预测灰分值。试验结果表明:相对于偏最小二乘法回归、神经网络回归模型,PRE-PLS模型决定系数为0.9951,均方根误差为0.9411,平均绝对误差为0.7332%,表明该模型具备较高的精度,能够胜任现场检测工作,为生产提供可靠指导。展开更多
文摘During the course of calculating the rice evapotranspiration using weather factors,we often find that some independent variables have multiple correlation.The phenomena can lead to the traditional multivariate regression model which based on least square method distortion.And the stability of the model will be lost.The model will be built based on partial least square regression in the paper,through applying the idea of main component analyze and typical correlation analyze,the writer picks up some component from original material.Thus,the writer builds up the model of rice evapotranspiration to solve the multiple correlation among the independent variables (some weather factors).At last,the writer analyses the model in some parts,and gains the satisfied result.
文摘煤炭灰分值是衡量煤炭质量的关键指标之一,灰分含量和性质对燃烧设备、环境、后续的加工利用都有着极大影响。针对目前煤炭灰分检测方法的滞后性、劳动密集型问题,提出了一种基于XRF光谱的预处理(Preprocessing,PRE)与偏最小二乘法(Partial Least Squares,PLS)相结合的XRF煤炭灰分智能预测算法。通过将XRF技术获取煤炭样品的光谱数据输入PLS主模型初步预测灰分,再将相关校正参数输入补偿优化模型中,最终将两者相加得到预测灰分值。试验结果表明:相对于偏最小二乘法回归、神经网络回归模型,PRE-PLS模型决定系数为0.9951,均方根误差为0.9411,平均绝对误差为0.7332%,表明该模型具备较高的精度,能够胜任现场检测工作,为生产提供可靠指导。
