用滑动窗多项式拟合法实时预测运动目标轨迹

提出了滑动窗多项式拟合数学模型。该模型用多项式拟合法处理一段目标位置历史数据序列,计算目标未来时刻位置,并随时刻推移不断更新目标数据,实现了实时预测。该模型误差小且稳定,克服了一般拟合预测中误差随时间推移增大的缺点。仿真结果说明,该方法参数固定、计算量小、速度快、精度高,利于实时预测跟踪。

Prediction of moving target trajectory with sliding window polynomial fitting

A mathematical model for sliding window polynomial fitting is proposed. With the use of the model, a real-time prediction is implemented through processing the historic data sequence of the target position with polynomial fitting, calculating its future position at one time and renewing target data continuously with the lapse of time. The error of the model is small and its performance is stable. The problem that error increases with the lapse of time existed in common fitting prediction has been overcome. The simulation results demonstrate that the method has advantages of constant parameters, small calculation quantity, fast-speed and high-accuracy, it is good at real-time predicting tracking.