Traditional automation technology has achieved great success, but it also has its limitations. For example, classical control theory is often only for those physical objects which can be described by linear ordinary differential equations, but many physical systems can not be described by such mathematical models. Another limitation is that automatic control can only be used for the control objects whose structure and operation logic do not change, such as single equipment and valve, whose structure and operation logic are stable is a normal; however, for a large system such as a factory, some equipment failure and abnormal production are inevitable. For the above problems, digital method has great power.
digitization:
The key to advance lies in the in-depth application of computer and Internet
To promote digitization, in fact, is to promote the extensive and in-depth application of computers and the Internet.
First of all, digital modeling is powerful. With the help of computer, people have the ability to describe any mathematical formula, physical object, production and business process. All elements such as people, machines and materials in the factory can be described in the digital space. In the past, due to the limitation of storage and processing capacity, this ability could not be fully released. With the continuation of Moore's law, the emergence of Internet, cloud computing, big data and other technologies, people have the ability to fully describe the various business activities of manufacturing enterprises in the digital virtual world, and manage and control the physical world through this virtual world.
Second, computers and network systems can help people make decisions. When automatic control cannot be realized, human experts (such as dispatchers and managers) are generally required to deal with it. At this time, these systems can help managers obtain information, analyze problems and transmit instructions, so that human beings can make better decisions. From the perspective of cybernetics, human experts essentially act as a "controller".
Computer systems in manufacturing enterprises are often hierarchical. The most common practice of iron and steel enterprises is to divide them into four levels - L1 ~ L4. Among them, L1 and L2 systems for valve and equipment level control are generally called automation systems; L3 and L4 systems for workshops and factories are generally referred to as "information systems" or "management computers". It is easy to find that the low-level system often has a fast response speed but a small control range, while the high-level system has a slow response speed but a large control range. At the same time, the proportion of low-level computer automatic control is high, and the proportion of high-level computer automatic control is low. This phenomenon shows that for the computer system, fast response, high degree of automation and wide management range are contradictory.
This kind of contradiction is disadvantageous to the control of steel mills. In the steel plant, energy, environmental protection, quality, production organization often involve multiple departments and processes, which belongs to the problem of large system. Problems in a link often need to be solved through coordination. Academician Yin Ruiyu often said: "iron and steel enterprises are typical dissipative systems." If the problem is not handled in time, it may lead to the increase of energy consumption, the decline of quality and production efficiency. However, the emergence of key information on the production site is in the order of seconds and milliseconds, and the cross process and cross department coordination may delay several hours or even days. For example, problems occurring instantaneously in the continuous casting process may cause quality defects in cold rolled products after a few days.
Under the traditional computer system architecture, this kind of contradiction is difficult to solve comprehensively, because human experts can't deal with the second level and millisecond level information sent by thousands of devices and sensors, while machines are not good at flexibly handling cross process and cross department large system collaboration problems. Such problems need to be solved by digital methods in the intelligent era.
Intelligent:
Deep integration of management and control
Ge industrial Internet white paper points out that the industrial Internet has three elements: intelligent machines, advanced algorithms and people at work. Among them, intelligent machine is the device or sensor that can receive digital instructions and send relevant status, which is the source of data. It can be imagined that in a large enterprise such as a steel plant, there are thousands of sensor data, which can not be handled by human beings. For this reason, the advanced algorithm can find out the problems that need human attention and processing from the complex real-time data, and then hand them to the "people at work" to deal with. In this way, the problems that need to be handled by human beings are selected, and the real-time performance is not particularly strong, so human beings have the ability to deal with these problems.
In fact, if the problems in the factory are common, we can also turn the logic and methods of human experts to deal with problems into computer code, so that the machine can make decisions according to the ideas of human beings, which is the so-called "human intelligence into wit". In this way, we can further reduce the burden of human processing problems and improve the level of automation of decision-making - this is actually intelligent.
In this regard, people often have the question: what is the value of this practice? The significance of this approach lies in improving the management level of enterprises. We know that as long as human beings participate in the work, there are always management problems. In China, the so-called "low level of technology" is actually a problem of management or personnel operation. In some iron and steel enterprises, the cost loss caused by management problems far exceeds the profits of enterprises. When it comes to the interests of individuals and departments, management will become very difficult because management problems are often deliberately hidden; when it comes to the cooperation and sharing of multiple departments, the traditional management methods are often difficult to solve problems quickly.
However, with the above-mentioned digital and networked means, many human decision-making work can be handed over to machines or done with the help and "supervision" of machines. In this sense, the information of sensors can not only be used to describe equipment status and process conditions, but also to reflect the working state of human beings. They are like the "eyes" of managers in the production process, which can empower managers; for operators on the spot, they can directly realize the impact of operation on operation and management through digital means. For example, when the gas supply is insufficient, the production of secondary processes may be detrimental to the overall interests of the enterprise. These methods are helpful to maximize the overall interests of enterprises.
Mr. He Linsheng, the former deputy technical director of Baosteel, said that the constitution of Anshan Iron and steel company requires "managers to participate in labor and workers to participate in management.". Under the past technical conditions, this is very difficult to achieve. However, with the help of digital means, managers can see the scene and workers can understand the manager, so the above requirements can be easily realized.
In a sense, intelligence is the integration of management and control, automation system and information system. From the perspective of cybernetics, the traditional automation, informatization and intelligence are the unity of "perception, decision-making and implementation". However, automation is often aimed at small systems with strong real-time performance, and decision-making is automatic; informatization is often aimed at large-scale systems with weak real-time performance, and decision-makers are human beings; and intelligence is often for large-scale systems with strong real-time performance, and decision-making is the integration of human and computer. In other words, many management and decision-making problems which were originally in charge of by people have been realized automation and semi-automation.
In fact, similar ideas have been around for a long time. The idea of "data not landing" promoted by Baosteel at the beginning of its establishment contains a similar idea, and the idea of computer integrated manufacturing system (CIMS) is also similar to it. However, the technical conditions at that time made it difficult for these ideas to be implemented. The continuation of Moore's law has created today's opportunity.
In addition, in the process of promoting intelligence, it often involves the reconstruction of organizational process, the innovation of business model, and even the systematic optimization of physical equipment and production process. In this sense, automation is often a relatively simple technical problem, while the key to intelligence may be management or even strategic decision-making.
