基于选权迭代法的既有铁路平面线形拟合方法
The linear fitting method of existing railway horizontal alignment based on iteration method with variable weights

当采用最小二乘法进行既有铁路平面线形拟合时,需首先判断测点所在位置的平面线形(即线形识别)并将实测数据分组,然后依据相应拟合模型对各组实测数据进行线形拟合。然而,既有铁路平面线形的实测坐标数据不可避免地存在偶然误差或粗差,这将对线形识别产生不利影响,从而导致实测数据分组不准确,进而造成线形拟合的效果较差。鉴于以上原因,提出基于选权迭代法的既有铁路平面线形拟合方法。该方法将各分组内不属于待拟合线形内的测点视作含粗差的测点,并利用选权迭代法具备的较强抗差能力对其进行检测和剔除,最终实现无需精确分组的既有铁路平面线形拟合,获得可靠的线形拟合参数估值。
When existing railway plane is fitted least square multiplication, the first step is to determine horizontal alignment of measuring points’ locations( i.e., linear identification), and cluster the measured data. Then according to the corresponding fitting model, the measured data groups are fitted linearly. However, there are unavoidable errors and gross errors in the measured coordinate data of existing railway horizontal alignment, which have adverse effects on linear identification. It will lead to inaccurate measured data grouping and cause poor quality of linear fitting. For these reasons, we propose the linear fitting method of existing railway horizontal alignment in this paper based on iteration method with variable weights. In this method, it regards the measured points beyond being fitted linearly in different groups as measured points with gross errors. It detects and eliminates points by using the robust correction ability of iteration method with variable weights. Ultimately, it can realize existing railway horizontal alignment without accurate grouping, and obtain the reliable parameters valuation of linear fitting