How do we understand the major industrial data, enabling industrialized how intelligent manufacturing data, and how the Internet industry into the future on the basis of intelligent manufacturing? Path to produce industrial transformation and upgrading of the role of big data system software National Engineering Laboratory, will produce industrial transformation and upgrading of the role of the path, summarized as “Math” four quadrants. The so-called plus and minus is intelligent manufacturing. Intelligent Manufacturing more focused on internal things, intelligent manufacturing narrow manufacturing concern, namely production processes, intelligent manufacturing contains a broad corporate lifecycle, from research and design to manufacturing and then to operation and maintenance services. Intelligent Manufacturing nothing more than adding something in the existing process, Save something, it can be substantially summarized in two phrases: mention quality, efficiency, cost, risk control. Today, intelligent manufacturing do is add and subtract. But in this day and age is not enough light to do addition and subtraction, such as a private equity investment company, doing business every year a little additions CONTROL ENGINEERING China Copyright , investors may not be satisfied, but to achieve enterprise increase exponentially. How to achieve? Internet industry is likely to achieve the path multiplication and division. Multiplication is a plateau effect. Such as Taobao, for the innumerable shops in the shop to make money on its platform, it is a case. But in the industry, whether to build a platform for the Internet industry? Clothing industry as a case study. Traditional first generation of garment enterprises, have their own design, factories, stores, namely a complete industrial chain. The second generation of garment enterprises, give up the whole select OEM factory, turned to do marketing to stores as an asset. The age of the Internet clothing business, neither the plant nor the store, the cost is almost zero, all the stores rely on Taobao, is only responsible for the rapid design, to control the supply chain, the last of the “total plate” although not necessarily traditional companies so big, but the high profit margins. So the division is the business focus on their core competitiveness. Asset-light high-profit operations, this is the future of Chinese SME innovation and entrepreneurship Road. Internet platform to build industrial ecology, not to say that this is the only platform in order to make money, everyone on the platform are likely to make money.
three levels: the major industrial sector classification data big data system software contact with the National Engineering Laboratory, has done a lot of big industrial application data, and divided into three levels. The first level is the unit level, that is, forIndustrial equipment, equipment ROMS are not limited to, further comprising, failure analysis early warning of equipment failure, and optimize the operation of equipment, asset management, and so on. First, we need equipment running precise digital measurement, this measurement actually means is large continuous space industry data discrete. The continuous space is complex, and the physical quantity, accuracy, the number of sensors that can be measured is limited, so the whole spatial sampling can not be achieved. But with the improvement of the level of digitization, the process of information propulsion, intelligent application iteration , the next measurement will be upgraded. The second level is the plant hierarchy. This level is not concerned about individual equipment, and focus on operating efficiency of the entire plant, product quality and safety, environmental issues. Industry is supposed to include people, materials, technology, equipment, environmental factors, including, in complex dynamic systems capable of synergy. Assuming that the whole of China as a big factory, how to enhance their efficiency in the industrial chain? We do big data industry today, do “smart +”, that is, for this purpose. To answer the first data where, in fact, the data in any one place. Data on industrial pipe before the pipe is relatively rough, the traditional field of information technology is doing relatively better management information, and now a lot of industry data is only used for monitoring data and do the event of failure to do playback. The data used to how to do integration of the two (information technology and automated data fusion) has yet to be verified. The third level is how to get other related data? For example, construction excavator to automation, need to understand GIS data, environmental data, but these are not traditional manufacturing enterprises have data. This shows that the industry today connotation of big data than traditional data connotation is much greater. Automation and cross-border overall data, constitute the industrial system big data. Classification and industrial challenges of big data in fact, industrial data has three characteristics: first feature is multi-modal. Very simple and crude to the past data into structured data, semi-structured data, unstructured data, but industrial enterprises is not the case. Today saw a lot of different formats like, unstructured project data and really turn it on when is not the same. Efficiency depends unstructured data structured, can be structured only efficient use; second feature is a high throughput. Many devices are non-stop, all the data is 7 * 24Hours of continuous generation of very large; third characteristic is a strong correlation. In various sectors of industry , the data associated with different rules, rather than follow the Simple Syndication. It features large industrial data itself brings a lot of challenges. In addition to the challenge of data acquisition, and was followed by data analysis, challenging applications. Here edge biggest limitation is the causal relationship Control Engineering Copyright , that is data-driven approach can only tell us the association, but not causality can not tell us. Other products such as Taobao, only know the recommended related products, but do not care about this matter of cause and effect – why users is such a person. But this does not work in the industry, especially in terms of control, so that the model takes a long time analysis and verification. The presence of white-box models and industrial gray box model, i.e., white-box model of the mechanism of industrial, industrial enterprises based mechanism design step, the product structure and process, which is the first step. When they were finished design, operation will be a lot of uncertainties, these uncertainties eliminate rely on expert craftsmen experience, the whole process becomes more stable and efficient production, which is gray-box state. Not the mechanism itself and the knowledge to analyze and understand the data model is a black box model. The nature of big data and industry intelligence industry is that these experiences and knowledge to quantify it, tap the tacit knowledge without heart, the mouth, or try to data by the statistical method to find the relationship, and then returned to the artisans analysis. Industry is the industry that exists longer than the time information, more than the accumulation of information, while big data and artificial intelligence technology to bring just a small change to the industry, to help it try to eliminate uncertainty. Big data, application of artificial intelligence in the industry’s first intelligent manufacturing. For example, the yield fell to a machine, then the machine can guess the tool may be worn out, and offered to ATC, overheating or furnace temperature, the temperature will go down on the independent twice. If the device can inform autonomous, independent changes, rather than to a preset logical operation, this is intelligent. True digital workshop should be what? Divided into three levels: the first layer is large data integration. Big data to build data integration system, so that policy makers see each workshop what happened, what control parameter is, what test parameters Yes. In such a material as the center axis of the data integration process flowSystem that is able to provide more and better decision-making information to adjust work; second layer is the big data analysis. You can not be a good batch data and batch data for the difference between the addition and, to see the difference in the control parameter? Big Data can guess cause of the problem, at least you can sort, make adjustments sorted workers to do the inspection and adjustment; third layer is the mechanistic models. Can be constructed through a lot of data and feedback, a relatively accurate industrial enterprises, the forward simulation model and digital twin body, digital space for debugging, the final test at the factory, which is the digital twin bring intelligent system. So the logic of the Internet industry for intelligent manufacturing is what has changed? From a business point of view, the Internet industry more concerned about the boundary of the Internet industry, rather than focus on production processes within the enterprise, it can be summarized into three cross-border integration: First, cross-border business integration, through upstream and downstream industry chain expand their business boundaries, companies can try to upstream integration of upstream and downstream can also serve downstream, we are a large factory in the perspective of a collaborative industry chain; the second is the integration of cross-border data chain, business development has brought data boundary development, today’s data is not limited to existing enterprise data. For example, to serve the builders, requires environmental data, operational data, meteorological data; three is the most fundamental technical change , compared to the development of IT technology, industrial software and IT industry is not a development curve, but now can be lightweight cloud computing technology allows users to do such a development, which can produce the technology spillover opportunities in many areas. Industrial appear Internet platform, so that industrial enterprises can spend a lot of time developed a simulation model may settle into a small but new forms of industrial software.