中文摘要：

以Hindmarsh-Rose混沌神经元模型为例,讨论了基于自适应同步识别混沌系统多个参数方法的优化问题.在构造的李亚普诺夫函数中引入可调节的增益系数来控制系统同步和参数观测器的暂态过程长短.在应用单个控制器和5个参数观测器来同步和识别Hindmarsh-Rose混沌神经元中5个未知参数时发现最小参数的识别结果出现了振荡而其他参数都能准确识别现象,分析其原因可能在于要识别的目标参数的巨大差异性.通过增加控制器的个数（选择两个控制器）可以克服这个困难.研究发现增益系数太小不能实现完全同步和参数的准确识别,当增益系数太大则延长了识别参数的暂态过程.在恰当的增益系数下可以在比较短的暂态过程下准确识别系统参数.进一步讨论了系统参数发生阶跃变化时系统参数的识别问题,数值计算结果验证了该方法的可靠性和有效性.

英文摘要：

Optimization of self-adaptive syn chronization is investigated to estimate a group of five unknown parameters in one certain chaotic neuron model, which is described by the Hindmarsh-Rose. Two controllable gain coefficients are introduced into the Lyapunov function, which is necessary to get the form of parameter observers and controllers for parameter estimation and synchronization, to adjust the transient period for complete syn chronization and parameter identification. It is found that the identified results for the minimal parameter （three orders of magnitude less than the maximal parameter） oscillate with time （the estimated results for this parameter is not exact） while the four remaining parameters are estimated very well when one controller and five parameter observers are used to work on the driven system （response syste m）. To the best of our knowledge, it could result from the great difference of five target parameters （values）. As a result, this problem could be solved when two controllers and five parameter observers are used to change the driven system and all the unknown parameters are identified with high precision. Furthermore, longer transient period for parameter estimation and complete synchronizatio n is required when too strong gain coefficients are used, whils parameters can not be estimated exactly if too weak gain coefficients are used. Therefore, appropriate gain coefficients are critical to achieve the shortest transient period for parameter identification and complete synchronization of chaotic systems, and th e optimization of gain coefficients depends on the model being studied . Furthermore, it is confirmed by our numerical results that this scheme is effective and reliable to estimate the parameters even if some parameters jump suddenly.