【摘要】 获取高精度表层土壤水分数据能揭示不同土壤结构的水分动态变化状况,对于研究土壤水分植被承载力具有重要意义。传统的测量数据校准方法采用最小二乘估计,估计结果对于土壤水分数据中的误差较为敏感,当水分测量数据中存在显著误差时会导致错误的校准结果,降低了校准精度,且校准效率低。针对这些问题,提出一种基于粒子群优化的表层土壤承载力水分测量数据快速校准方法。采用三截尾估计取代最小二乘估计作为测量数据校准的目标函数,以水分测量数据的随机误差遵从正态分布为约束条件,构建三截尾鲁棒土壤水分数据校准模型。采用粒子群算法对校准模型进行求解,引入不确定知识概念,在种群迭代搜索过程中限定个体搜索中心及搜索范围,使得粒子以更快、更有效的方式向最优解逼近,获得更为准确的校准值。实验结果表明,所提方法相比当前测量数据校准方法的校准效率最高提高了28.5%左右,校准精度较高,可用于实际测量。更多还原

【Abstract】 Estimation result of traditional calibration method of moisture measurement data in surface soil is sensitive to error of the moisture data, which leads to wrong estimation result and poor calibration precision if the moisture measurement data has obvious error. Furthermore, the calibration efficiency is low. Therefore, this paper proposes a rapid calibration method based on particle swarm optimization(PSO). Our research used three-truncation estimation instead of least square estimation as objective function of measurement data calibration and built robust calibration model of oil moisture data with three truncations taking random error of the measurement data obeying normal distribution as constraint condition. The research used the PSO to solve the calibration model and introduced uncertainty knowledge conception. The research restricted individual search center and search range during species iterative search to make particle approach optimal solution faster and more effectively. Thus, we obtained more accurate calibration result. Simulation results show that calibration efficiency of the method increases by 28.5 percent compared with current method. The calibration precision is much higher. The method is available for actual measurement.更多还原