Representativeness-based optimal sampling method

Abstract

The invention discloses a representativeness-based optimal sampling method. A point with high local representativeness is first selected by virtue of a stagnation point concept, and a first sampling point is selected out by utilizing similarity of vectors formed by target variables in a time sequence; and when subsequent sampling points are selected, information quantities contained in the selected points are removed in sequence. The method comprises the steps of calculating orthogonal components, relative to the vectors of the selected sampling points, of original vectors in sequence; then selecting a new sampling point by utilizing similarity between the orthogonal components; calculating accumulative representativeness, first calculating the accumulative representativeness of the sampling point to each sub-pixel, and then taking a mathematic mean value of the accumulative representativeness of all sub-pixels in a region as the accumulative representativeness of the region; and ending the sampling until the accumulative representativeness of the region meets a certain requirement. According to the method, the ground observation on the sampling points can accurately reflect a spatial and temporal distribution characteristic of the target variables in the sampling region, the number of the sampling points is reduced as far as possible, and the ground observation cost is reduced.

Description

technical field [0001] The invention relates to the technical field of space sampling, in particular to an optimal sampling method based on representativeness. Background technique [0002] In spatial sampling, commonly used point selection methods include random selection method, enumeration method, sequential method, computer evolution algorithm, etc. The random selection method, as the name suggests, randomly selects sampling points from the sampling area. When adding new sampling points cannot improve the sample accuracy and cannot reflect the goal of optimal sampling, it is often used. When selecting a new point, the sequential method takes into account the influence of the selected point, uses the Kriging variance as the weight of the unsampled point, and selects the point with the largest weight as the new sampling point until the total accuracy of the selected sample point is Reaching a given level or adding new samples cannot significantly improve the overall accur...

AI Technical Summary

Problems solved by technology

The simulated annealing method (Ge Yong et al., 2012) in the computer evolutionary algorithm is a commonly used optimal sampling method at present. Its advantage is that it can search for the global optimal solution with high efficiency. However, this algorithm has several limitations: ①The input parameters can only refer to the spatial distribution information of a single time phase; ②The optimal solution is affected by the initial parameter field, and has a certain degree of randomness, and the results of the two samplings before and after may be completely different; ③The number of selected points must be controlled before selecting points , unable to update the information increment brought by the new sensor

However, there is no sampling method in the prior art that can achieve the above effects

Images