Track prediction method and system

Abstract

The invention discloses a track prediction method and system, relates to the technical field of track prediction, and aims to solve the problem that variable-duration track prediction cannot be directly carried out based on sparse non-uniform time sequence historical data. The track prediction method comprises the steps of firstly obtaining a plurality of pieces of historical track point data, expanding original features of each piece of historical track point data to obtain expanded features, then constructing a multi-dimensional track feature matrix according to all the expanded features and prediction moments, and finally taking the multi-dimensional track feature matrix as an input; using a prediction model to acquire the track position at the prediction time. Sparse non-uniform time sequence data can be directly processed, the requirement for data sampling is lower, and position prediction of a certain non-equal interval moment in the future can be achieved. The flight path prediction method and system provided by the invention are used for performing variable-duration flight path prediction.

Description

technical field [0001] The invention relates to the technical field of track prediction, in particular to a time-varying track prediction method and system for sparse and uneven time series data. Background technique [0002] At present, many scholars have carried out a lot of research on track prediction. However, the commonly used forecasting methods need to homogenize the historical data, that is, use the interpolation algorithm to fit the historical data, sample uniformly, and obtain historical data with equal time intervals, so as to predict the trajectory based on the historical data with equal time intervals. Such a prediction method reduces the accuracy of the data by operating in stages, and the accuracy of the interpolation data will limit the accuracy of the prediction. Moreover, the commonly used forecasting methods are mainly for forecasting at a certain point in the future. In the algorithm model, the time interval between the forecasting time and the last tim...

AI Technical Summary

Problems solved by technology

However, the commonly used forecasting methods need to homogenize the historical data, that is, use the interpolation algorithm to fit the historical data, sample uniformly, and obtain historical data with equal time intervals, so as to predict the trajectory based on the historical data with equal time intervals. Such a prediction method reduces the data accuracy by operating in stages, and the interpolation data accuracy will limit the prediction accuracy

Moreover, the commonly used forecasting methods are mainly for forecasting at a certain point in the future. In the algorithm model, the time interval between the forecasting time and the last time of the historical data needs to be set in advance, and it is impossible to realize the demand for variable-time forecasting.

Images