提出了建立门限自回归模型（ＴＡＲ）的一套简便通用的方法。用作者提出的改进遗传算法，可同时优化门限值和自回归系数，成功地解决了ＴＡＲ建模过程所涉及的大量复杂寻优工作这一难题，为ＴＡＲ模型的广泛应用提供了强有力的工具。实例计算的结果说明了这套方法的可行性和有效性，同时也说明了，通过门限值的控制作用，ＴＡＲ模型可以有效地利用如海洋资源所隐含的时序分段相依性这一重要信息，限制了模型误差，从而保证了ＴＡＲ模型预测性能的稳健性，提高了预测精度。该方法具有通用性，在各种非线性时序预测中具有重要的理论意义和应用价值 A simple and general method for establishing threshold auto-regressive model (TAR) is proposed. With the improved genetic algorithm proposed by the author, the threshold value and autoregressive coefficient can be optimized simultaneously, successfully solving the large number of complex optimization tasks involved in the TAR modeling process and providing a powerful tool. The result of example calculation shows the feasibility and validity of this method. At the same time, it also shows that TAR model can effectively utilize the importance of timing dependencies as implied by marine resources through the control of thresholds Information, which limits the model error, thus ensuring the robustness of the TAR model in predicting the performance and improving the prediction accuracy. The method is universal and has important theoretical significance and application value in various nonlinear time series prediction