中文摘要：

采用基于时间序列的乘积季节ARIMA（求和自回归移动平均）模型以及无人艇模型规则波试验数据，研究了水面无人艇运动极短期预报技术。采用经趋势差分和季节差分后的ARMA（自回归移动平均）模型，运用最小信息准则和白噪声检验方法，验证所选择的最佳模型，并对无人艇进行了20步极短期运动预报。计算结果表明：无人艇船模加速度、升沉、纵摇的前lO步短期预报最大误差均不超过6％，随着预测步数的增加，误差有扩大的趋势，加速度的后10步短期预报最大误差达到16．68％。可见，极短期预报技术有效。

英文摘要：

Multiple seasonal ARIMA（auto regressive integrated moving average） model based on time series was used, the extreme short-time prediction technology of unmanned surface vehicle motion was studied based on the seakeeping test data of boat model in regular wave. ARMA（auto regressive moving average） models with trend difference and seasonal difference were adopted, Akaike information criterion and white noise inspection method were carried out, and the chosen best model was validated. Extreme short-time prediction in twenty steps was made for unmanned surface vehicle. Calculation result indicates that the maximum errors are not more than 6 % in the former ten-step predictions of acceleration, heave and pitch. With the increase of prediction step, the error has enlarging trend, the maximum error reaches to 16.68~ in the following ten-step prediction of acceleration. So extreme short-time prediction technology is effective. 3 tabs, 13 figs, 15 refs.