Identification of time delay is a prerequisite for the realisation of control compensation. The existing researches generally assume that the system input should be monotonic, which is unsuitable for online identification due to system saturation. To solve this problem, relaxed input conditions are developed in this Letter. Then a modified least-square algorithm with forgetting factors, which has a strong ability in tracking time-varying delays, is presented. Meanwhile, two extra terms whose design parameters are adjusted online according to the input condition are added into the covariance update law to guarantee the boundedness and robustness of the algorithm. Simulation results demonstrate the superiority of the proposed method and the reasonableness of the developed input conditions.

中文翻译：

---

在宽松输入条件下在线识别时变时滞系统

识别时间延迟是实现控制补偿的前提。现有研究一般认为系统输入应该是单调的，由于系统饱和，不适合在线识别。为了解决这个问题，在本函中开发了宽松的输入条件。然后提出了一种具有遗忘因子的改进最小二乘算法，该算法具有很强的跟踪时变时延的能力。同时，将两个额外的项（其设计参数根据输入条件进行在线调整）添加到协方差更新定律中，以确保算法的有界性和鲁棒性。仿真结果证明了该方法的优越性和所开发输入条件的合理性。