|
Fornew readers and those who request to be “好友 good friends” please read my 公告栏 first.
A recent article about the early days and the father of control theory
http://blog.sciencenet.cn/blog-673362-674628.html
and the related comment #8 by “beyondcontrol” led me to write this article to clarify
what I know about this topic.
At the time of late forties, there was a confluence of events:
1. the US was flush from its WWII success in the development of servomechanism
2. the digital computer pioneered by Aiken and Eckert-Mauchly just debuted.
3. The now famous paper by Vannevar Bush “Science, the endless frontier”
The prevailing sentiment at that time was that wonderful machines and computers
will soon take over and artificial intelligence is just around the corner (note 1). In this
atmosphere, the book by Wiener, given his fame as a mathematician and contributions
in stochastic processes as used in designing servomechanisms, naturally hit a sweet
spot inthe public’s imagination. But on this topic, Wiener is primarily a visionary and philosopher. He did not have anything concrete to offer. On the other hand, Shannon’s invention of Information Theory offered quantitative result on the
measure of information and provided means to design and evaluate codes for communication. Thus, he is rightly called the father of Information Theory.Wiener can
only be credited with the invention of the word “Cybernetics”.
Also at that time and into the middle and late fifties,the mainstream control theory is an outgrowth of the theory of servomechanism.The dominant tools are Laplace transform,
complex variable, and frequency domain analysis. Thus when Tsien, who gained fame
in fluid and aero dynamics, published his book “Engineering Cybernetics” in 1955 he
was an outsider tocontrol. And the mathematics he used were unfamiliar to the
control field; norwas aerospace a US national goal (note 2). He was thus unappreciated. Furthermore, he returned to China shortly afterward and no one was there to publicize his work. In 1959, with catching up to USSR as the national goal and the publications of Kalman and his filter, the modern era of control theory started. He can rightly be credited as the father of modern control theory. My mentor, A.E. Bryson in his famous 1962 paper on optimal control finally realized in practice what Tsien forsaw in 1955. Wikipedia thus also named himas the father of optimal control http://blog.sciencenet.cn/blog-1565-577590.html (50+ years of control research atHarvard in 5+ minutes).
In short, visionary may forecast things to come, but it is the practically useful accomplishments that society
pays for and rewards. I would say history has beenfair to all concerns. My personal opinion and view.
Note 1. Of course more than half a century later AI is still out of reach except for special
cases.
http://blog.sciencenet.cn/blog-1565-414924.html
Note 2. Sputnik was launched almost three yearslater in 1957
Chinese translation below ably provided by "beyondcontrol"
你可以赞赏远见卓识,但社会关注实用性
博主丁国如的一篇最近关于控制理论早期和控制理论之父的博文:对“维纳”的新认识
http://blog.sciencenet.cn/blog-673362-674628.html
以及 “beyondcontrol”的相关评论8,使得我写这篇文章,去澄清在这个论题上我所知道的事情。
上个世纪四十年代晚期,有一系列历史事件同时出现:
1. 由于二战中伺服机构的成功发展,美国如日中天。(译者注:伺服机构即伺服随动系统,可用于火炮控制等。以下圆括号里内容是译者加上的)
2. (1947年)由 Aiken and Eckert-Mauchly 首创的数字计算机被推出。
3. (1945年)Vannevar Bush 向美国总统提交了传世报告“科学,无尽的前沿”。
当时流行的观点是那美妙的机器和电脑将很快接管世界,人工智能指日可待[注1]。
在这样的氛围下,维纳的书,不仅给他带来作为一个数学家的声誉(note by ho: Should read 因為他在數學界的名氮以及展示了其使他在随机过程的贡献用于伺服机构设计的贡献,自然也击中了公众想象力的要害。但在这个论题上,维纳主要是一个空想家(预言家Note by ho: may be 遠見者 is more appropriate)和哲学家。他没有提供任何具体的方法。另一方面,香农信息论的发明对信息的度量提供了定量结果,同时为通信中设计和评估编码提供了方法。因此,他公正地被称为信息论之父,而仅仅术语“控制论”的发明归功于维纳。
同时,在四十年代晚期和进入到五十年代后期,主流控制理论成长于伺服随动理论。主要工具是拉普拉斯变换,复变量函数,和频域分析。因此,1955年,在流体和空气动力学声名鹊起的钱学森发表《工程控制论》时,钱对于搞控制的人来说是一个局外人。他所用的数学对控制领域工作者来说是不熟悉的;(钱所做的)也不是美国国家航天的目标[注2]。
因此他(以及他的工作)没有引起关注和赏识。此外,他不久回到中国,没有人宣传他的工作。1959,与赶上苏联作为国家目标相伴,随着(R.E. Kalman)卡尔曼工作和他的滤波器(算法)的发表,现代控制理论的时代开始了。卡尔曼被公正地誉为现代控制理论之父。我(指何)的导师, Arthur E. Bryson http://en.wikipedia.org/wiki/Arthur_E._Bryson在他著名的1962年关于最优控制的文章里,最终在实践中(或实际上)实现了钱学森1955年(书中)的预见。在维基百科里也因此称我的导师是最优控制之父;见http://blog.sciencenet.cn/blog-1565-577590.html [50多年的在哈佛的控制研究,一个5分钟多的报告]。
简而言之,空想家可以预见到未来的事情,但是专业团体只奖赏实际有用的成就并为其买单。我想说,对于所关心的事情历史已经是公平的。这是我个人的意见和观点。
Note 1. 当然,超过半个多世纪后,人工智能还未实现,除了特殊情况,参见http://blog.sciencenet.cn/blog-1565-414924.html
注 2. 几乎是3年后的1957年,Sputnik (“旅行伴侣”苏联第一颗人造卫星)发射成功。
beyondcontrol 的注释:1)何教授的导师 Bryson 1951年毕业于加州理工学院 Caltech,他的导师是 Hans W. Liepmann. 钱学森于1949年的夏天回到 Caltech 就任冠名讲座教授。Liepmann是钱的同事,钱的1955年(1954年出版不知道为什么何先生用1955年Note by Ho: Faulty memoryon my part)书里引用过 Liepmann 机翼在湍流中扰动速度的功率谱的工作。Bryson的博士论文似乎不是控制方面。2) Bryson 和 W. F. Denham 1962年的那篇文章 “A Steepest-Ascent Method for Solving OptimumProgramming Problems” (ASME Journal of Applied Mechanics) 是自动控制25篇经典著作之一,该文章现在找不到,不知道是否引用了钱的书。Bryson 与何的1969年的书:Applied Optimal Control: Optimization, Estimation andControl 未见引用钱的书 Note by blogger:It is onlywith 20/20 hindsight after many decades and during by first visit to China in1979 that I was reminded of Tsien’s contribution and his book by reading hispaper about system engineering published, if I remember correctly, in 大公報。3)钱书的内容有多少是原创的,对beyondcontrol 来说目前未解之谜,这个并不难,有时间再看。钱书的内容如下:
1 引言
2 拉氏变换法
3 输入、输出和传递函数
4 反馈伺服系统
5 不互相影响的控制
6 交流伺服系统与震荡控制伺服系统
7 采样伺服系统
8 有时滞的线性系统
9 平稳随机输入下的线性系统
10 继电器伺服系统
11 非线性系统
12 变系数线性系统
13 利用摄动理论的控制设计
14 满足指定积分条件的控制设计
15 自动寻求最优运转点的控制系统
16 噪声过滤的设计原理
17 自行镇定和适应环境的系统
18 误差的控制
以上来自与戴汝为等人的译本。原英文版和戴的译本及何与其导师的书可在网上下载。
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-12-23 06:32
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社