英语翻译The main approach is based on the use of linear matrix inequality (LMI) techniques [6],where conditions valid for any hi verifying the convex sum property are used.The fact that the actual scalar functions hi (·) are not introduced in th

英语翻译The main approach is based on the use of linear matrix inequality (LMI) techniques [6],where conditions valid for any hi verifying the convex sum property are used.The fact that the actual scalar functions hi (·) are not introduced in th
英语翻译
The main approach is based on the use of linear matrix inequality (LMI) techniques [6],where conditions valid for any hi verifying the convex sum property are used.The fact that the actual scalar functions hi (·) are not introduced in the LMI conditions will give only sufficient results.
In the pioneering works (see [52] and the references therein),the basic problem is solved without taking into account additional aspects like performances and robustness specifications and the use of state observers.The extension to output feedback with a state observer is quite straightforward when the premise variables z in (3) are measurable,in which case the separation principle holds [27].In [52] and [2] dynamic output-feedback control is considered.
It should be stressed that contrary to claims often made in the literature,fuzzy controllers are not ‘inherently robust’,unless an uncertainty description is explicitly associated with the fuzzy model and taken into account in the design procedure.Norm-bounded uncertainties were introduced in [66,57,50,13].Pole
placement constraints for the individual linear models can be used to impose performance requirements [29,26].
Following the first works on nonlinear models in the form of (3),several extensions were explored.Some ofthem are direct extension ofpre vious results,for example TS time-delay models with or without uncertainties [52,63].Others try to extend to this representation standard results from the linear framework,
for example control laws based on a descriptor form of CTS models or optimal fuzzy control design [52].
The methods remain conservative and the challenge is to find some new ways to reduce this conservatism.One possibility is to modify the Lyapunov function.Alternatives for the classical quadratic Lyapunov function (7) were proposed; a piecewise quadratic Lyapunov function [7,18,31],and a fuzzy Lyapunov function [4,53] for CTS models and a non-quadratic function for DTS models [21].Another approach is to use some matrix properties:specific transformations,elimination lemma [58]and/or relaxation of the basic inequality (6),see for example [51,34].

英语翻译The main approach is based on the use of linear matrix inequality (LMI) techniques [6],where conditions valid for any hi verifying the convex sum property are used.The fact that the actual scalar functions hi (·) are not introduced in th
The main approach is based on the use of linear matrix inequality (LMI) techniques [6],where conditions valid for any hi verifying the convex sum property are used.The fact that the actual scalar functions hi (·) are not introduced in the LMI conditions will give only sufficient results.主要方法是基于使用线性矩阵不等式（ LMI ）技术[ 6 ] ,那里的条件有效期为任何高科技核实凸一笔财产的使用.事实上,实际标职能高科技（ ）不介绍了基于LMI条件将只有足够的结果.
In the pioneering works (see [52] and the references therein),the basic problem is solved without taking into account additional aspects like performances and robustness specifications and the use of state observers在开拓工程（见[ 52 ]和提述） ,基本问题得到解决,而不考虑其他方面一样表现和稳健性的规格和使用国家观察员The extension to output feedback with a state observer is quite straightforward when the premise variables z in (3) are measurable,in which case the separation principle holds [27].In [52] and [2] dynamic output-feedback control is considered.已延长至输出反馈与状态观测器是非常简单的前提时,变量z系列（ 3 ）可衡量的,在这种情况下,拥有分离的原则[ 27 ] .在[ 52 ]和[ 2 ]动态输出反馈控制的考虑.It should be stressed that contrary to claims often made in the literature,fuzzy controllers are not ‘inherently robust’,unless an uncertainty description is explicitly associated with the fuzzy model and taken into account in the design procedure.应当强调的是相反的索赔往往在文献中,模糊控制器的不是'天生强大的' ,否则不确定性的描述是明确与模糊模型,并考虑到设计的程序.Norm-bounded uncertainties were introduced in [66,57,50,13].Pole
placement constraints for the individual linear models can be used to impose performance requirements [29,26].
Following the first works on nonlinear models in the form of (3),several extensions were explored.Some ofthem are direct extension ofpre vious results,范数有界不确定性介绍[ 66,57,50,13 ] .北极
安置限制个人线性模型可以用来强加的性能要求[ 29,26 ] .
继第一个工程,非线性模型的形式（ 3 ） ,延长了一些探索.一些ofthem是直接延长ofpre vious结果 for example TS time-delay models with or without uncertainties [52,63].Others try to extend to this representation standard results from the linear framework,
for example control laws based on a descriptor form of CTS models or optimal fuzzy control design [52].
The methods remain conservative and the challenge is to find some new ways to reduce this conservatism.One possibility is to modify the Lyapunov function.Alternatives for the classical quadratic Lyapunov function (7) were proposed; a piecewise quadratic Lyapunov function [7,18,31],and a fuzzy Lyapunov function [4,53] for CTS models and a non-quadratic function for DTS models [21].Another approach is to use some matrix properties:specific transformations,elimination lemma [58]and/or relaxation of the basic inequality (6),see for example [51,34].例如的TS时间延迟模型或不确定性[ 52,63 ] .其他尝试向这一结果的代表性标准的线性框架,
例如管制法律的基础上描述形式的壳聚糖模式或最优模糊控制设计[ 52 ] .
这些方法仍然是保守和面临的挑战是找到了一些新的方法来减少这种保守主义.一种可能性是修改Lyapunov函数.替代传统的二次Lyapunov函数（ 7 ）提出了一个分段二次Lyapunov函数[ 7,18,31 ] ,和一个模糊Lyapunov函数[ 4,53 ]为旅模型和一个非二次函数的DTS模型[ 21 ] .另一种办法是使用某种矩阵属性：具体的变革,消除引理[ 58 ]和/或松弛的基本不等式（ 6 ） ,例如,见[ 51,34 ] .