Abstract

In the modeling approach of Jupiter’s radiation belt, the accurate calculation of magnetic coordinates from geographic coordinates is the basis. In the previous studies, the L-shell parameters are always calculated based on the assumption of a dipole field though the accuracy of this method is low. We present a new L-shell calculation method based on the magnetic field lines tracing method and ANN (Artificial Neural Network). In this method, a compromise between calculation accuracy and speed is achieved. This method consists of a classifier and a predictor. The Classifier is a BP (Back Propagation) ANN based on AdaBoost algorithm and the Predictor is a BP ANN optimized by GA (Genetic Algorithm). The Classifier is used to identify whether the coordinates are within Jupiter’s inner magnetosphere. If so, the Predictor is used to calculate the L-shell parameters. The error rates of the Classifier and the Predictor are 3 and 7%, relatively. In an example of the Juno’s orbit, the calculation speed of this ANN-based method is about 3 orders higher than that based on the magnetic field lines tracing method.

中文翻译：

---

利用人工神经网络快速计算木星磁层中的磁坐标

摘要

在木星辐射带的建模方法中，从地理坐标精确计算磁坐标是基础。在以往的研究中，L-壳层参数的计算总是基于偶极场的假设，尽管这种方法的准确性较低。我们提出了一个新的L-基于磁场线追踪方法和ANN（人工神经网络）的壳计算方法。在这种方法中，实现了计算精度和速度之间的折衷。该方法由分类器和预测器组成。Classifier是基于AdaBoost算法的BP（Back Propagation）ANN，Predictor是GA（Genetic Algorithm）优化的BP ANN。分类器用于识别坐标是否在木星的内部磁层内。如果是，则使用预测器计算L壳参数。分类器和预测器的错误率分别为 3% 和 7%。在朱诺轨道的一个例子中，这种基于人工神经网络的方法的计算速度比基于磁力线追踪方法的计算速度快3个数量级。