What is Intelligent Design?

The intelligent design ^[1] hypothesis (also known as intelligent design), a political doctrine that states that "some characteristics of the universe and living things can be explained more intelligently by intelligent reasons than by undirected natural selection." The main proponents of the hypothesis are mainly Christian think tanks. They believe that the intelligent design hypothesis is based on political doctrines of religious belief and that the evolutionary origin of living things is regarded as God or God's will and arrangement.

Intelligent design

Intelligent design refers to the application of modern information technology, the use of computers to simulate human thinking activities, to improve the computer's intelligence level, so that the computer can more and better undertake various complex tasks in the design process, becoming an important auxiliary tool for designers.

1) Guided by design methodology. The development of intelligent design depends fundamentally on the understanding of the nature of design. Design methodology The in-depth study of design essence, process design thinking characteristics and methodology is the basic basis for intelligent design to simulate artificial design.

2) Use artificial intelligence technology as a means of implementation. With the powerful function of expert system technology in knowledge processing, combined with artificial neural network and machine learning technology, it can better support the design process automation.

3) With tradition

Based on the research status and development trends of intelligent design at home and abroad, intelligent design can be divided into three levels according to design capabilities: conventional design, associative design, and evolutionary design.

Conventional design

That is, the design attributes, design process, and design strategy have been planned. The intelligent system uses the inference engine to call symbolic models (such as rules, semantic networks, and frameworks) for design. At present, most of the intelligent design systems put into use at home and abroad belong to this category, such as the Japanese NEC Corporation s Wirex system for VLSI product layout design, the standard V-belt drive design expert system (JDDES) developed by Huazhong University of Science and Technology, and the intelligent CAD system for pressure vessels. Wait. This type of intelligent system can often only solve well-defined and well-structured conventional problems, so it is called conventional design.

Lenovo Design

The current research can be divided into two categories: one is to use existing design cases in the project to compare and obtain guidance information for existing designs. This requires collecting a large number of good and comparable design cases, which is difficult for most problems The other is to use artificial neural network numerical processing capabilities to obtain implicit knowledge about design from experimental data and calculation data to guide the design. With the help of other examples and design data, this type of design achieves a certain breakthrough in conventional design, which is called Lenovo design.

Evolutionary design

Genetic algorithm (GA) is a highly parallel, random, adaptive search algorithm that draws on the natural selection and natural evolutionary mechanisms of the biological world. In the early 1980s, genetic algorithms have been widely used in areas such as manual search and function optimization, and have been extended to many fields such as computer science and mechanical engineering. In the 1990s, the research of genetic algorithms was based on the principle of population evolution, and expanded into the fields of evolutionary programming (EP, Evolutionary programming), evolutionary strategies (ES, Evolutionary strategies), and they are also called evolutionary computation (EC, (Evolutionary computation).

Evolutionary computing has extended intelligent design to evolutionary design. Its characteristics are:

* The design plan or design strategy is coded as a gene string to form a genetic population of design samples.

* The design evaluation function determines the pros and cons of the samples in the population and the direction of evolution.

* The evolutionary process is the process of sample reproduction, crossover and mutation.

Evolutionary design has little dependence on environmental knowledge, and the crossover and variation of good samples are often the source of design innovation. Therefore, "Artificial Interlligenceindesign'96" was held in 1996.

Principle design intelligent design

The result of the scheme design will affect the whole process of the design, and it will play a vital role in reducing costs, improving quality, and shortening the design cycle. Principle scheme design is the process of seeking original understanding and is the key to product innovation. The process of principle scheme design is total function analysis-function decomposition-function element (sub-function) solution-local solution combination-evaluation decision-best principle scheme. According to this design method, the core of the principle scheme design comes down to the principle-oriented principle-based solution. The design catalog for generalized sub-functions can comprehensively describe the requirements of sub-functions and the original understanding, and implicitly maps from physical effects to original understanding. It is the initial knowledge base document of the intelligent principle scheme design system. The intelligent system based on the design of the design of the scheme of the catalog can better realize the intelligent design of the concept.

Collaborative solution

ICAD should have multiple knowledge representation modes, multiple inference decision-making mechanisms, and multiple expert systems for collaborative solution functions. At the same time, a model based on knowledge programs and methods related to the theory should be combined into a collaborative solution system. Work together under the control of a scheduler to solve complex design problems together.

The ability of a single expert system to solve problems at one link is often limited to the coordination and adaptability with other links. In order to broaden the problem-solving field of expert systems, or to enable some interconnected fields to be solved by the same system, the concept of a so-called collaborative expert system has emerged. In this system, there are multiple expert systems working together, and this is the collaborative multi-expert system. The key to collaborative solution of multi-expert systems is to connect and cooperate with experts in the field of engineering design to solve problems. The consistency principle and evaluation strategy of information transmission in the collaborative solution process is to judge whether the work currently being performed is in the direction beneficial to the overall goal. Collaborative solution of multi-expert systems, in addition to implementing parallel features in this process, it is necessary to develop a software environment for solving multi-expert system collaborative problems with practical significance.

Knowledge acquisition, expression and expert system technology

Knowledge acquisition, expression and utilization of technical expert system technology is the foundation of ICAD. Its main development direction for CAD applications can be summarized as:

* The study of machine learning models aims to solve the problems of knowledge acquisition, refinement and structure.

* To deepen the reasoning technology, we must have monotonic reasoning for the forward, reverse, and bidirectional reasoning process control modes, and focus on non-inductive, non-monotonic, and neural network-based reasoning.

* Comprehensive knowledge expression mode, that is, how to construct a unified multi-knowledge table structure with deep and shallow knowledge.

* Research on structural systems based on distribution and parallel thinking.

* Blackboard structure model

The key technologies of the intelligent design system include: re-understanding of the design process, design knowledge representation, multi-expert system collaboration technology, redesign and self-learning mechanism, comprehensive application of multiple inference mechanisms, intelligent human-machine interface, etc.

1) Rethinking the design process

The development of intelligent design systems depends on an understanding of the design process itself. Although people have made a lot of explorations in design methods, design procedures, and design laws, from a computerization perspective, the current design methodology is far from being able to meet the needs of design technology development, and there is still a need to explore designs suitable for computer processing. Theory and design patterns.

2) Design knowledge representation

The design process is a very complicated process that involves the application of many different types of knowledge, so a single knowledge representation is not enough to effectively express a variety of design knowledge. How to build an effective knowledge representation model and an effective knowledge representation is always The key to the success of a design expert system.

3) Multi-expert system collaboration technology

The more complex design process can generally be decomposed into several links, each link corresponds to an expert system, multiple expert systems collaborate and share information, and use fuzzy evaluation and artificial neural networks to effectively solve the multi-disciplinary and multi-disciplinary design process. Goal decision and optimization problems.

4) Redesign and self-learning mechanism

When the design result fails to meet the requirements, the system should be able to return to the corresponding level for redesign to complete local and global redesign tasks. At the same time, inductive reasoning and analogical reasoning can be used to obtain new knowledge, summarize experiences, continuously expand the knowledge base, and achieve self-improvement through re-learning.

5) Comprehensive application of multiple inference mechanisms

In intelligent design systems,

The emergence of intelligent design can be traced back to the period when the expert system technology was first applied. Its initial form has adopted a symbolic reasoning technology in a single field of knowledge-the design-type expert system, which is important for design automation technology from information processing automation to knowledge processing automation. Meaning, but the design-type expert system is just to solve the local needs of some difficult problems in the design, it is only the initial stage of intelligent design.

In the past 10 years, the rapid development of CIMS has brought new challenges to intelligent design. In an environment like CIMS, product design, as a key link in enterprise production, is even more important. In order to fundamentally strengthen the company's ability to respond quickly to market demands and competitiveness, people have put forward higher requirements for design automation. On the basis of the computer providing knowledge processing automation (which can be completed by a design-type expert system), decision-making automation is achieved, that is, helping human design experts make decisions in design activities. It should be pointed out that the decision automation mentioned here is by no means the exclusion of human experts. On the contrary, in a large-scale integrated environment, the role of people in the system will become more important. Human experts will always be the most creative source of knowledge and key decision makers in the system. Therefore, a complex giant system such as CIMS must be an integrated intelligent system combining human and machine. In line with this, CIMS-oriented intelligent design has entered the advanced stage of intelligent design-human-computer intelligent design system. Although it also requires the use of expert system technology, it is only one of its own technical foundations, and there is a fundamental difference between a design-based expert system.

The core problem solved by design-type expert system is pattern design, and scheme design can be its typical representative. Different from the design-type expert system, the core problem to be solved by the human-machine intelligent design system is innovative design. This is because in a large-scale knowledge integration environment such as CIMS, design activities involve multi-domain and multi-disciplinary knowledge, and its influencing factors complex. The CIMS environment places higher requirements on the flexibility of design activities, and it is difficult to abstract a limited steady state model. In other words, even with the pride of design patterns, design patterns are ever-changing and almost impossible to exhaust. Such design activities must be more innovative, so innovative design is the core of the human-machine intelligent design system.

The design expert system and the human-computer intelligent design system have different cores, which can be derived from the differences in other aspects. For example, design expert systems generally only solve specific problems in a certain field, and are relatively isolated and It is closed and difficult to integrate with other knowledge systems, and the human-computer intelligent design system faces the entire design process and is an open architecture.

The development of intelligent design is linked to the development of CAD. At different stages of CAD development, the bearers of the intelligent part of the design activity are different. Traditional CAD systems can only handle computational tasks, and design intelligent activities are performed by human experts. In the ICAD stage, intelligent activities are completed by design-type expert systems. However, due to the limitation of the problem-solving ability of expert systems using single-domain symbolic reasoning technology, the scale and complexity of design objects (products) are limited. The design is mainly conventional design, but with the support of computer, the design efficiency is greatly improved. In the CIMS-oriented ICAD, that is, the I3CAD stage, due to the requirements of integration and openness, intelligent activities are jointly undertaken by humans and computers. This is the human-computer intelligent design system, which can not only be competent for conventional design, but also support innovative design. . Therefore, the human-computer intelligent design system is a software system designed for large-scale complex products. It is an integration-oriented decision-making automation and an advanced design automation.

(Copy paste think tank encyclopedia;)