Question

请各位大侠们帮我解释一下下面一段英文：

Fig. 3 shows responses of the system’s temperature T to step changes on the setpoint TSP at t = 10 hr, for both linear and nonlinear controller designs. It is evident that there are set-point changes (e.g. –
8.7 %) for which the controller of Fig. 2 performs clearly better than the corresponding linear controller. The following remarks are in order:
• Fig. 3 shows that for the nonlinear controller design the mapping TSP/Tapproaches the linear transfer function s + 1 . A small discrepancy is due to the fact that the RNN only approximates the first-principles CSTR model. Good closed-loop performance in the presence of this plant/model mismatch is evidence, albeit not proof, of robustness of the proposed nonlinear controller.Fig. 4 shows responses of the RNN states for the setpoint change TSP = 500 •K (Fig. 3).
• The zeros of the transfer function (U – Us)(s) , listed below, show that P is locally stable,implying that the mapping between v and u is locally stable.Fig. 5 shows u to be bounded and within feasible bounds for the setpoint change TSP = 500 •K(Fig. 3).

• The responses of the transfer functio (o + 1s + 2s2) and the exact-linearized CSTR to v, with v taking random values in [– 2, 2], are compared in Fig. 6. The discrepancy is due to the approximation of CSTR dynamics by the RNN.

• Small perturbations to the values of the CSTR parameters resulted in no appreciable deterioration of performance. For example, 40 hours after a setpoint change to 500 •K, TI and FI were changed from 8008 mol/m3 and 1.133 m3/hr to 7800 mol/m3 and 1.0 m3/hr, respectively. The behavior of the closed-loop CSTR is shown in Fig. 7. The superiority of the RNN-based controller is clear.

Answer 1

图3演示了系统的反应温度跳变T在设定的1 / 4茶匙、在T = 10人力资源,因为这两者都线性和非线性控制器的设计。很明显,有设定值变化(例句。-
87 %),在图2的负责人进行清楚比其线性化控制器。以下的评论被依次是:
•图3显示了非线性控制器设计的映射TSP / Tapproaches线性传递函数是+ 1。一个小的差异是由于这样的事实,RNN CSTR采用的仅仅是近似模型。良好的闭环表现在存在这种植物/模型误差的证据,虽然尚未证明,所提出的鲁棒性的非线性controller.Fig。4显示的响应的美国RNN改变设定值TSP = 500•凯西(图3)。
•满足传递函数(U -美国)(s),下面列出,表明,P当地稳定,这意味着之间的映射stable.Fig v和你的局部你。5给出了有界,在可行的范围内的改变设定值TSP = 500•凯西(图3)。

•的反应转移函数(o + 1 +2s2 CSTR)和exact-linearized v级,v随机抽价值在[- 2、2]的基础上,进行了比较在图6。这种偏差是由于近似的RNN CSTR动力学。

•小的动摇的价值观CSTR参数没有明显的下降导致的工作业绩。例如,40个小时后改变设定值500•凯西、钛发生了改变,并探讨8008组分,从每立方米1.133立方米/人力资源每立方米个组分和1.0立方米/人力资源等。闭环CSTR的行为被显示在图7。RNN-based控制器的优越性是清楚的。

追问

RNN 是递归神经网络 CSTR是个特定的控制系统 你是用机器翻译的吧！存在好多问题

追答

那就这样吧