An Overview of Evolutionary Algorithms for Control Systems Engineering

An Overview of Evolutionary Algorithms for Control Systems Engineering

控制系统工程进化算法综述
Andrew Chipperfield and Peter Fleming 安德鲁·奇伯菲尔德和彼得·弗莱明
Department of Automatic Control and Systems Engineering University of Sheffield, Mappin Street
谢菲尔德大学自动控制和系统工程系，Mappin街

---

Abstract: Developments in computational models of evolutionary processes have led to the realisation of powerful, robust and general optimization and adaptive systems collectively called evolutionary algorithms. In this paper we provide an overview of evolutionary algorithms and consider the features and characteristics that are particularly appropriate for control engineering applications. The versatile and robust qualities of these algorithms are considered and a number of application areas described.
Keywords: Evolutionary algorithms; genetic algorithms; optimization;
摘要：进化过程计算模型的发展使得强大、健壮和通用的优化和自适应系统得以实现，统称为进化算法。在本文中，我们提供了进化算法的概述，并考虑了特别适合于控制工程应用的特性和特性。考虑了这些算法的通用性和鲁棒性，并描述了一些应用领域。
关键词：进化算法；遗传算法；优化；

---

1. INTRODUCTION

1简介
Stochastic search and optimization methods, based on the principles of natural biological evolution, have received considerable and increasing interest over the past decade.
基于自然生物进化原理的随机搜索和优化方法，在过去的十年里受到了广泛的关注。

Stochastic随机指标;随机;随机性;随机指数;随机的 optimization最佳化；优选法；最恰当; 最佳条件选择；求最佳参数
optimization最佳化；优选法；最恰当; 最佳条件选择；求最佳参数
based on以……为基础；在……基础上，以……为依据
principles道德原则; 行为准则; 规范; 法则; 原则; 原理; 观念; 理由，信条; principle的复数
received被承认的; 被一致认可的; 拿到; 接到; 收到; 体验; 受到; 对…作出反应; receive的过去分词和过去式
considerable相当多的
increasing interest利息增加
past过去的; 昔日的; 刚过去的; 刚结束的; 从前的; 以往的; 过去; 昔日; 过去的事情; 过去的经历; 晚于; 在…之后; 在另一边; 到另一侧; 多于; 超过; 从一侧到另一侧; 经过; 过去，逝去
decade十年，十年期

Broadly, evolutionary algorithms (EAs) are a class of search techniques that includes genetic algorithms (GAs) (Holland, 1975), evolutionary programming (EP) (Fogel et al., 1966), evolution strategies (ES) (Rechenberg, 1973) and genetic programming (GP) (Koza, 1991).
广义地说，进化算法是一类搜索技术，包括遗传算法（GAs）（Holland，1975）、进化规划（EP）（Fogel等人，1966）、进化策略（ES）（Rechenberg，1973）和遗传规划（GP）（Koza，1991）。
EAs operate on a population of potential solutions applying the principle of survival of the fittest to produce successively better approximations to a solution. At each generation of the EA, a new set of approximations is created by the process of selecting individuals according to their level of fitness in the problem domain and reproducing them using operators borrowed from natural genetics.
EAs运用适者生存原理对一组潜在解进行运算，从而对一个解进行连续的更好的近似。在每一代的进化算法中，根据个体在问题域中的适应度水平选择个体，并使用从自然遗传学中借用的算子进行复制，从而产生一组新的近似值。

operate on给……做手术; 产生作用
survival of the fittest适者生存
successively连续; 先后; 渐渐
approximations近似值; 粗略估算; 类似事物; approximation的复数
set of的; 集合; 释……的集合
according to据; 按; 依照; 按照; 根据
reproducing复制; 再生产; 再制造; 使再次发生; 再现; 繁殖; 生育; reproduce的现在分词

This process leads to the evolution of populations of individuals that are better suited to their environment than the individuals from which they were created, just as in natural adaptation.
这一过程导致了个体种群的进化，这些个体比它们产生的个体更适合它们的环境，就像在自然适应中一样。
Compared to traditional optimization methods, such as calculus-based and enumerative strategies, EAs are robust, global and may be generally applied without recourse to domain-specific heuristics.
与传统的优化方法（如基于微积分和枚举策略）相比，进化算法具有鲁棒性、全局性和不依赖于特定领域的启发式方法而得到广泛应用。

optimization最佳化；优选法；最恰当; 最佳条件选择；求最佳参数
such as例如; 像; 象…这样; 诸如…之类
enumerative列举的，计数的，点数的
strategies策略; 计策; 行动计划; 策划; 规划; 部署; 统筹安排; 战略; 战略部署; strategy的复数
robust强健的; 强壮的; 结实的; 耐用的; 坚固的; 强劲的; 富有活力的
may be多半
heuristics探索法; 启发式

For these and other reasons the control community has been quick to realise the potential of EAs.
由于这些和其他原因，控制界很快意识到了EAs的潜力。
In this paper we consider the important characteristics of EAs and their relevance to problems in control engineering.
本文讨论了EAs的重要特性及其与控制工程问题的相关性。

characteristics特征; 特点; 品质; characteristic的复数
relevance重要性; 意义; 相关性

Recognising both the power and shortcomings of conventional search and optimization methods, this paper will consider some control problems that are not susceptible to efficient solution by conventional approaches.
在认识到传统搜索和优化方法的优点和缺点的基础上，本文将考虑一些常规方法无法有效解决的控制问题。

shortcomings缺点; 短处; shortcoming的复数
conventional依照惯例的; 遵循习俗的; 墨守成规的; 普通平凡的; 传统的; 习惯的; 非核的; 常规的
optimization最佳化；优选法；最恰当; 最佳条件选择；求最佳参数
methods方法; 办法; 措施; 条理; 有条不紊; method的复数
paper纸; 纸张; 报纸; 文件; 文献; 贴壁纸
consider仔细考虑，细想; 认为; 以为; 觉得; 体谅; 考虑到; 顾及
problems棘手的问题; 难题; 困难; 逻辑题; 数学题; problem的复数
susceptible易受影响; 敏感; 过敏; 好动感情的; 感情丰富的; 善感的; 容许…的; 可能…的; 可以…的
approaches靠近，接近; 接洽; 建议; 要求; 接近; 方式，方法，态度; approach的第三人称单数和复数

2. EVOLUTIONARY ALGORITHMS

Evolutionary algorithms are based on computational models of fundamental evolutionary processes such as selection, recombination and mutation. Fig. 1 gives an overview of a general evolutionary algorithm. Individuals, or current approximations, are encoded as strings composed over some alphabet(s), e.g. binary, integer, real- valued etc., and an initial population is produced by randomly sampling these strings. Once a population has been produced it may be evaluated using an objective function which characterises an individuals performance in the problem domain. The objective function is also
used as the basis for selection and determines how well an individual performs in its environment. A fitness value is then derived from the raw performance measure given by the objective function and is used to bias the selection process. Highly fit individuals will be assigned a higher probability of being selected for reproduction than individuals with a lower fitness value. Therefore, the average performance of individuals can be expected to increase as the fitter individuals are more likely to be selected for reproduction and the lower fitness individuals get discarded. Note that individuals may be selected more than once at any generation (iteration) of the EA.
Selected individuals are then reproduced, usually in pairs, through the application of genetic operators. These operators are applied to pairs of individuals with a given probability and result in new offspring that contain material exchanged from their parents. The offspring from reproduction are then further perturbed by mutation. These new individuals then make up the next generation. These processes of selection, reproduction and evaluation are then repeated until some termination criteria are satisfied, e.g. a certain number of generations completed, a mean deviation in the performance of individuals in the population or when a particular point in the search space is reached.
procedure EA {

procedure EA { 
	t = 0; 
	initialize P(t); 
	evaluate P(t); 
 	while not finished do { 
	t = t + 1; 
		select P(t) from P(t-1); 
		reproduce pairs in P(t); 
		mutate P(t); 
		evaluate P(t); 
	} 
}

Figure 1: A Simple Evolutionary Algorithm

Although similar at the highest level, many variations exist in EAs.
Evolutionary Programming (EP), arising from the desire to generate machine intelligence, typically uses a representation tailored to the problem domain. For example, in numerical optimization vectors of real-valued numbers would be used and combinatorial problems would employ ordered lists. Given a population size of N, all N individuals are selected as parents and a representation specific mutation operator used to generate N offspring. The N offspring would then be evaluated and the next generation selected using a fitness-based probabilistic function on these 2N individuals. The mutation operator in EP is often adaptive and different adaptation rates may be used for each decision variable within an individual.

3. EVOLUTIONARY ALGORITHM ISSUES FOR CONTROL

4. CONTROL APPLICATIONS OF EVOLUTIONARY ALGORITHMS

4.1 Control System Design

4.2 Multiobjective Optimization

4.3 Robust Control

4.4 System Identification

4.5 System Integration

4.6 Real-Time and Adaptive Control

5. CONCLUDING REMARKS

6. ACKNOWLEDGEMENT

7. REFERENCES