Imminized Artificial Neural Systems - Extensions, Analysis, and Applications

7#L L$2 D N X$|x D 6*`$6*6 6`，6 6 6 Krishnakumar 9415351免疫学与人工适应系统所要求的强健适应本质有许多相似之处。其中一些相似之处是在NSF研究启动拨款下进行的，并在两个问题上得到了证明：(A)不确定系统的在线辨识和自适应控制；以及9b)使用人在回路中的模拟来综合复杂控制器。使用无线电控制的飞机进行实验验证目前正在进行中。该项目将扩展最初的概念，以包括(A)分析iANS的能力(完备性、健壮性和适应性)和限制(计算速度、存储等)；(B)高阶构件识别；(C)包括免疫系统的几个微观特征；以及(D)探索其在适应性批评功能中的有用性。作为该项目的一部分，将开发和测试一种完全自主的遥控飞机闭环控制系统。为了实现这一目标，阿拉巴马大学已经在努力建立一个专门的实验室设施。其他应用将包括航天器控制和操作的不确定性管理。免疫的人工神经系统在完全开发和测试后，可以用于解决无数问题。其中一些是：复杂过程建模、高性能飞机/导弹控制、自主无人机任务控制、自主侦察机/旋翼机操作、复杂设计和制造过程，以及自主空间探索和机器人规划。人工自适应系统所要求的自适应特性。其中一些相似之处是在美国国家科学基金会的研究启动拨款下进行的，并在两个探测器L O上进行了演示@b L！L O L 9：~H H(F G(H H(d‘@=/R B H--激光作家时报+J e J S，艾丽西亚·哈里斯艾丽西亚·哈里斯

7 # L L $ $ $ $ $ 2 D D D D N X X $ | x D 6 * ` $ 6 * 6 6 ` , 6 6 6 6 6 6 Krishnakumar 9415351 The science of immunology has many parallels to the robust adaptive nature desired in aretificial adaptive systems. Some of these parallels were investigated under an NSF researcch initiation grant and were demonstrated on two problems: (a) on-line identification and adaptive control of an uncertain system; and 9b) synthesis of complex controllers using man-in-the loop simulations. Experimental vertification using radio controlled aircraft is currentyly in progress. This project will extend the initial concept to include (a) analysis if IANS' capabilities (completeness, robustness, and adaptiveness) and limitations ( Computational speed, storage, etc.); (b) higher order building blocks identification; (c) inclusion of several micro-features of the immune system; and (d) explore its usefulness in adaptive critic functions. As part of this project, a 'Completely autonomous remotely piloted aircraft closed-loop control system will be developed and tested. Towards this goal, efforts are already underway with the development of a dedicated laboratory facility at the University of Alabama. Other applications will include uncertainty management for spacecraft control and operations. Immunized artificial neural systems, when fully developed and tested, could be used in a myriad of problems. Some of these are: complex process modeling, high performance aircraft/missile control, autonomous UAV mission control, autonomous scout aircraft/rotorcraft operations, complex design ad manufacturing processes, and autonomous space explorations and robot planning. u st adaptive nature desired in aretificial adaptive systems. Some of these parallels were investigated under an NSF researcch initiation grant and were demons trated on two prob L O @ b L ! ! ! ! L L L O L ! 9 :~ H H ( F G ( H H ( d ' @ = / R B H -:LaserWriter Times + + + J e J S , krish.abs Alicia Harris Alicia Harris