【摘要】 本文采用时间序列分析方法和回归分析方法,对泌乳曲线的数学模型进行了研究,同时采用Kalnan滤波技术提早估计了模型参数;探讨了提早选择种公牛等问题。结果表明,不完全Г函数存在一定误差,平稳自回归模型具有显明的改进效果。在泌乳曲线的模型模拟研究中,应按产犊季节分别研究;采用最小二乘方法求解模型参数,提早预测产量误差较大,运用现代控制理论建立滤波控制系统,用Kalman滤波估计模型参数,比较最小二乘法具有独特优越性。当由Kalman滤波提早估计模型参数后,用平稳自回归模型预测全期产量误差较小。根据预测的总产乳量,运用BLUP原理估计公牛育种值,可提早选择种公牛。更多还原

【Abstract】 Time series analysis, regression analysis and Kalman filter analysis were used in this study. The stationary autoregressive model and the linear regression on the logarithmic transformation of the incomplete gamma function were fitted. The former was better than the latter. The bias for predicting 308-day yield was low when model parameters, estimated by Kalman filter, were high by the least squares. The effect of calving season on predicting 308-day yield appeared to be obvious. When heritability was unknown, the maximum likelihood solution gave the estimators of BLUP breeding values. By the model and method given in this paper, selecting bulls with the data of 90-day lactation was feasible.更多还原