The automation discipline is standing on a new “tuyere”, which is expected to promote a profound change of The Times.
“Since the creation of cybernetics by American scientists in the last century, the influence of the discipline of control and automation has crossed disciplinary boundaries and become an important condition and a significant symbol of human civilization progress and modernization.” Zheng Nanning, chairman of the Chinese Society of Automation and academician of the Chinese Academy of Engineering, said that in recent years, with the continuous progress of modern science and technology, automation, artificial intelligence and the Internet have given birth to the fourth industrial revolution, especially the deep integration of a new generation of information technology and manufacturing industry, accelerating industrial change. The development of science and technology cannot be separated from innovation, and automation science and engineering is not only the strongest driving force for technological change, but also a strong support for building a scientific and technological power.
9565-810F When attending the forum, Zheng Nanning condensed two basic questions that must be answered to develop a new generation of artificial intelligence: One is to answer what are the unique characteristics of human intelligence; Second, what does a computer need to be able to see and understand the world like a human?
“Humans rely on common sense a lot of the time to judge the properties or changes of things, and can act intelligently on a wide range of issues with very little data.” The list goes on and on. In daily life, hats with different shapes and functions, such as sun hats, hard hats, straw hats, etc., people can rely on common sense to identify immediately, but for computers, this is a topic that needs to rely on a lot of data to “learn”. From an academic point of view, this problem expresses the realization of the recognition of object attribute-class combinations. About 30,000 kinds of objects are designed in People’s Daily life, and each object has a variety of attributes. When attributes and objects are combined, there is an exponential explosion problem. For example, it is easy for people to judge a broken mobile phone and a mobile phone with a scratched screen surface, but for the machine to accurately identify, the modeling also involves the expression and transfer of specific concepts. Let the computer also have the same common sense reasoning and object attribute recognition ability as people, and teach the computer to learn like people, Zheng Nanning’s team is currently working on related issues.
9565-810F In addition to dealing with complex environments, game theory is also getting hot in the field of control systems. A foreign scholar has written a book “Reward Punishment”, analyzing that some reward behavior after excessive implementation (such as spawning the laziness of the rewarded person), not only did not achieve the preset goal, promote behavior change and reminder, but sometimes went to the opposite of the original intention of the reward. A similar “game” also occurs between the human immune system and the virus. The virus may mutate, or antibodies may be formed, which can cause the immune system to fight the virus at first, but then lose its effectiveness. These factors need to be taken into account if the relationship between human immunity and viruses needs to be mathematically constructed.
Taking a series of vivid cases as the introduction, Guo Lei, academician of the Chinese Academy of Sciences and researcher of the Academy of Mathematics and Systems Science of the Chinese Academy of Sciences, introduced when he attended the forum to make a report that the objects of traditional control theory research are mostly engineering systems, and the autonomous behavior or autonomous pursuit of objects is generally not included in mathematical modeling, which also limits the scope of control theory. It is difficult to apply directly to the terms of the economy, society and more “smart” engineering systems of the future. This also led his research team to propose a dynamic system framework for studying the game behavior of controlled objects.
9565-810F “This kind of system must be hierarchical, and the goal pursuit between different subjects at the bottom may not be the same, and between the bottom and the top may not be the same, so a macro-control is needed to ensure the overall ideal system behavior.” In Guo Lei’s view, in this basic research field, it is relatively easy to describe the target task, but there are still several problems to be solved in order to make substantial progress through research.
Adhere to the demand orientation, from the country’s urgent needs and long-term needs, and truly solve practical problems. According to the latest “Industrial Internet + Safety Production” action plan issued by the Ministry of Industry and Information Technology and the Ministry of Emergency Management, by the end of 2023, the coordinated development pattern of industrial Internet and safety production will basically take shape, the level of digital management, network coordination, and intelligent management and control will be significantly improved, and the level of intrinsic safety will be significantly enhanced. At the same time, the Ministry of Industry and Information Technology also issued the “Building materials industry intelligent manufacturing digital transformation and upgrading” action plan, the first is the “cement industry integrated system overall solution”, will be implemented within three years.