针对传统GPS高程异常反演方法中存在的不足,首先采用遗传算法优化最小二乘支持向量机的参数,然后用于反演GPS高程异常。实验表明,在有限样本情况下,GA—LSSVM模型不仅发挥了LSSVM处理小样本数据的能力,而且通过GA优化后的LSSVM能够选择出合适的参数,反演得到高精度的GPS高程异常。通过对测试样本的反演精度进行评定,得到的GA—LSSVM反演GPS高程异常可以推广到实际应用中。
Against the shortcoming of the traditional GPS elevation anomaly calculation methods, used the GA-LSSVM to calculate the GPS elevation anomaly. The experiment shows that GA can select the suitable parameters of the LSSVM. In the case of limited sample,the GA-LSSVM model can overcome the disadvantage of the small sample, and obtaion the accurate results. By evaluating the accuracy of the results, the GA-LSSVM can be extended to practical applications.