Li Jie, a special lecture professor at the University of Cincinnati and consultant to the White House Information Physics Systems and the US Challenge Project, shared his views on industrial data and how artificial intelligence can improve industrial data analysis at the China Conference on the Application of Large Data in 2017.
Beijing, July 20, 2007Data analysis insights.
Industrial big data
I have worked in the U.S. industry and government for 37 years. IMS Center was established in 2000. At present, there are more than 90 enterprise partners in the world, such as Mitsubishi Electric, GE, Huawei, Zhonghua, Zhonghua, Sany Heavy Industries and so on. It took us 15 years to write this industrial big data.I didn’t even think about writing this book at all. Many companies say you’ve sorted it out. So this book was published in 2015 at the Shanghai Research and Development Center of General Electric Company of China. It basically left behind the legacy of the methods and logic that have been done in the past 15 years. There are some cases in it. “From the big””Data to Intelligent Manufacturing” tells many cases of intelligent manufacturing, such as semiconductors, automobiles, engines, robots, etc. are using the latest methods to solve problems that people could not solve in the past.
《 CPSNew Generation Industrial Intelligence is a new book published in 2017, and CPS is the future of intelligent systems that don’t stop at sensors or software. It’s a management system. The car runs into a pit. The next time you drive a mile down the road, you’ll be told there’s a pit in the road and share it with others.The other car knew there was a hole in the road. For example, you can save a lot of gas money, drive at an unstable pressure, you unconsciously waste five dollars a day, 150 dollars a month, and more than a thousand dollars a year, so you have no idea, then the CPS will tell you.
The big industrial data we are doing now basically revolves around the visible problems of pollution, efficiency, quality and production. These problems give us a lot of knowledge and experience, so the data approach is to find the reasons for these problems and reference, and ultimately produce value. Industrial bigData is often said in a word, is the “experience” to “fact” transformation, that is, I can change people’s experience into a mathematical or data results can prove the fact. Manufacturers prefer to concentrate on products, that is, egg yolks. After the engine takes off, a lot of “protein” data come out.”Protein” data can be used to analyze, you can find the reasons for aircraft take-off, landing, it is related to air humidity, and wind direction, so from take-off, flight, landing three stages to do classification, maintenance or security checks, so it is separated, so data classification, segmentation, sharing and so on are allIt can be done well.
2005Komatsu intelligent maintenance of large data, but the amount of data is too large, large is not a problem, the focus is not the data classification, segmentation, decomposition, you have to classify the data first, segmentation, decomposition and then pass out to do analysis, do not all pass on to do carding, this is very tired. So we use egg yolk and protein concept to dig.Digging data, so we accumulated experience. At present, Komatsu uses UAVs to model the site. After scanning the site, the automatic excavator dug the site overnight. This is intelligent.
Tires are the same, tire pressure is not the same, running time is not the same, how to adjust do not know, you spend more than 100 dollars a month, more than 10,000 dollars a year, so you can know which tire is not good, but this is something that people simply can not do, why do we say this thing? We talk about industry.Data, is to use a lot of data resources, we can use historical data, sensing data, and then do analysis methods, and then find out the reasons for his prevention, these are very important. It used to be all data uploaded, and that’s right, and it’s not true, depending on what industry you’re in, like Apple’s mobile data going to be in the cloudIt’s not a big deal, but the cost of high-speed rail data is very high. So there are three foundations for big data, DT, PT, AT, DT called data technology, PT called platform technology, AT called analysis technology. There are two analytical techniques, one is analytical tools, the other is analytical tools.One is the analytical process. Tools are like kitchen cooking tools, craft is what the chef should have, I give you a fish three people three different ways, three different flavors, so tools can be open source, technology can not be open source, because he is a craftsman spirit. Speaking of the three characteristics of DT, the next three “B, the first is data splitting. The second “B” is the advantages and disadvantages of the data, the third “B” is the background of the data, many industries have large amounts of data but no background, so can not be used, so industrial big data DT is not ordinary people can do.
The first layer is to see how to collect data, the second layer is the transformation of data to information content, the third layer is the virtual network content management, the fourth layer is the identification and decision-making of the problem, the fifth layer is the grouping of equipment, it is one layer at a time.For example, a lot of data does not need to be collected, it is directly completed on the processing side, there is no data essence after the cloud, but it has the value of analysis, it is different in nature, the cloud has been processed. For example, if the fish is ready, I can eat three fish. So you can do all this technology, but it has a system philosophy.Learning, we have many opportunities in our country, there are many small success, small success has also created a lot of wealth. Twenty-six years ago I said that data could one day be classified, split, decomposed, analyzed, and shared through a learning software. Then there will be a chip and software in different industrial applications including aircraft.Engine testing, wind power, high-speed rail, and large power stations, and the entire transmission of China Telecom, the purpose of our talk today is how artificial intelligence improves industrial data analysis, which is our topic today.
How AI improves industrial big data analysis
Artificial intelligence can’t intelligentize data, but it can intelligentize meaning. The first artificial intelligence I did in the United States was to analyze it with postal parcels, that is, automatically subpackage, automatically identify addresses, without the need for people to operate. Later, we used the software in industry, and we supervised the learning software.Unsupervised, there are a lot of software can be used, the purpose is to make the software tools can be understood.
For example, we use SOM, a classification learning method, which can reduce a lot of data on the one hand, on the other hand, classification, segmentation, such as bearing and distinguish whether the inner ring or the outer ring is a problem, just as the doctor hears the heart sound, he himself can not hear, he wants magnetic resonance to distinguish. So we canFound through this classification.
CPSIt can be more intelligent and manageable. CPS is a symmetrical system of entity system and information system. For example, when a car runs, there is an information management system that lets us know where the risk is highest, where the turning is often in traffic accidents. Before I arrived, I slowed down to 30 kilometers per hour a kilometer ago.It can ensure safety.
There are many kinds of data sources, the relationship between data, the meaning of data, so offshore wind power, when I see the wind speed is changing the angle is different, so according to this cut wind speed, power generation and wind speed, angle relationship, this is impossible for people to know things, when I know it can know later.The wind power is declining for three hours in a row, and I can do something tomorrow.
EMU is also the same. When high-speed rail runs, it uses modeling analysis directly. Every high-speed rail data is analyzed, and I know which road is not stable. In the future, the central train will do the “one belt and one way” plan of the whole world. Data can not be done without supporting platform. China can do the following two hundred km / h speed of high-speed rail bearing.China can’t do more than two hundred kilometers per hour. Now it has to run 460 kilometers per hour, which requires technology to support it.
2005We worked with a Toyota compressor in 1997. The compressor bearings are broken several times a year. It must be compressed when it runs. To be efficient, it will produce resonance. A resonance bearing will be blank. It only takes three to five seconds. So you need to monitor the energy release immediately to solve the problem.Yes, so we monitored the pressure characteristics, and the valve could be opened immediately if it was detected to be subtle. We use only four most important parameters from the 12 parameters, and we use SVM to do the model segmentation. Segmentation is not allowed to cause errors, if there are errors in the original situation, butThe fruit line is a little more careful, giving you a little more warning. There has been no failure in the past eleven years.
Nissan Robot Health Surveillance, Robot do a daily comparison, we can not monitor him, I do the comparison according to the point of movement, made a comparison after the establishment of a file, and other robots to compare. So he doesn’t need to be who he is. He doesn’t need to monitor. He doesn’t need data, but needs differentiation.And characteristics. We did actual production at the Canadian plant, and on October 20 last year we quickly caught two robots changing, and three days ago we knew they were going to break down.
China does not send surplus things to the “along the way” countries. In the early days, because of the fact that things were done a lot, there was excess, and the water tanks were full. Digging a ditch to let the water out was called “one belt and one road”. In fact, this is not the case, so we must take the system engineering out. Wisdom sea. We cooperate with Chinese ships to sea.Weather, weather, and intelligent modeling enable the ship to save fuel. This is the “egg yolk + protein” service, which saves the world’s ships oil, and customers will be happy to hand it over to you.
Now we’ve developed a lot of telecommunications sensors, so we’ve developed a skin sensor that sticks directly to the skin. As soon as your skin moves, I know the strength is not great, such as leg pain, so we can calculate the problem of muscle injury and whether the athletes are excessive.
Conclusion, if big data is fact, it is benefit.
Fenjiu, how to brew a good digital wine
This article is reproduced from WeChat public No. IBM enterprise consulting service.
“Xinghua village wine such as spring, “Guo Moruo praised the” Gan Ruquan “Shanxi Fen wine in 2017 ushered in a good day.
Since this year, the pace of reform of state-owned enterprises in Shanxi has accelerated. Fenjiu Group has attracted much attention as one of the star enterprises. At the same time, the liquor consumption and capital market warming also let Fenjiu Group harvest the net profit and share price doubled gratifying report card. Behind the bright results, Fenjiu Group is making use of the upgrading of information technology to complete quietly.Comprehensive changes in management, system and so on.
The flavor of Fen Liquor in the millennium is a little more technological and internet, and the strong and stable digital core system also makes Fen Liquor fragrant.
Transformation, opening up departmental barriers
In a recent IBM new generation of digital engine seminar, Fenjiu Group Information Center Director Liang Zhonghong described how Fenjiu boarded the IBM information express. From last September to March this year, Fenjiu Group joined hands with IBM to successfully deploy the SAP S/4HANA platform.The liquor group is a rush to “fashionable”, but also indicates that the development of enterprises has come to a new era.
Fenjiu Group is an old state-owned enterprise. It is also the first “liquor first stock” listed on A shares. Liang Zhonghong said that as a former wine model, “business leaders and employees want to win back the original market position, to achieve this goal, we found that in modern enterprise management, information technology is the most important thing.”.”
2017Fenjiu Group is one of the two key reform enterprises in Shanxi Province. The SASAC promulgated seven policies, which gave the enterprises great authorization. The bonus of the policies greatly stimulated Fenjiu Group from top to bottom. Obviously, for the transformation of information technology, Fenjiu Group is coming true!
“To realize informatization, the first thing to break is the barriers between the original departments and departments, between people, to eliminate the information island, to break through the barriers between departments, to enhance overall efficiency. Liang Zhonghong introduced that in the past, the monthly work of reconciliation statements between Fenjiu Group’s production and sales companies would have cost a lot.With lots of manpower and time, a financial team of more than 20 people needs a week to dock data because of the fragmentation of the system, and now with financial automation, only two people are needed. In this regard, Fenjiu Group’s financial staff lamented, “In the past we were only accounting, now is the real financial.”
Data enable business managers to see “frontline”.
It is reported that the whole system of Fenjiu Group has been in good operation from the on-line system up to now. For the head of Fenjiu Group and managers at all levels, as long as you sit in the office you can see the real-time operation of the company.
Fenjiu Group has an interesting name for this visualized data presentation system – “Fenjiu Management Cockpit”. Liang Zhonghong introduced that the data is playing a bigger role in the whole management and operation chain of Fenjiu Group. From purchasing to operation, the whole industry chain has been opened. At present, the IT part of the Group is cooperating with the business department to do so.Good data extraction, analysis and distribution, and in the continuous cycle to enhance the value of data, “perhaps a report, a data, the front door to taste the sweet, will be unable to stop.”
Behind the efficient data link is a unified data standard. IBM GBS Global Business Consulting Services Division Ma Ming disclosed that in the process of data standards unification, the IBM and Fenjiu Group team spent a lot of energy, “just like people have ID cards, products have their own ID cards, and in the past this was completely different from the unified.”The statistical caliber is also inconsistent. There are many “one thing, many yards” and “one yard, many things”. Through the MDG master data management platform, we uniquely encode the production data and carry out huge data cleaning work.
“Although it takes a lot of effort, it is a long and hard work. ” “Sometimes big data doesn’t necessarily have much data, but it’s more latitudes and riches that make it more valuable,” Ma said.
Liang Zhonghong indicated that at present, the solid foundation data has brought the rich reward in the daily work production.
IBM，“Cooperation between a century old shop
“We are standing on the shoulders of giants and catching up with the trend of the Internet. There are two giants in the red mouth of Liang, one is SAP, the other is IBM.
In recent years, Fenjiu Group’s strategic investment in IT has increased significantly. Liang Zhonghong said frankly, before choosing IBM and SAP, Fenjiu Group IT team has inspected Mengniu, Luzhou Laojiao and other similar enterprise cases, “IBM threshold is indeed higher than the general enterprise, but IBM in the control of the pass is alsoIt is more stringent than other enterprises. “
“IBMIt is a hundred-year-old shop, Fenjiu Group is also a hundred-year-old shop, are in their own industry reputation companies. Liang Zhonghong disclosed that in the previous research and communication, the IBM team from management thinking to team building and Fenjiu Group’s strategic direction is consistent with a higher degree, with a consensus, “ultimately through the tender link.”The cooperation with IBM has been achieved.
In addition, the strategic planning of Fenjiu Group in information technology has been looked at 3-5 years, even in the next 10 years, Liang Zhonghong believes that IBM can work together to complete the transformation and transformation of Fenjiu Group, because IBM is a company that can achieve “consultation + landing” and “eventually to fall into business,Hope brings value.
Flexible manufacturing reveals the embryonic form of Internet manufacturing
“Fen liquor must be drunk and Fen liquor must be drunk”is a popular advertising slogan of Shanxi Fen liquor. But in the eyes of consumers nowadays, what role does Fen liquor play? It is understood that Fen Liquor has begun to make a breakthrough in the personalized experience of its products. It is used for weddings and personalized wines. It engraves its own name on the wine altar and hasThe value of the “pot pulp”… Fenjiu is facing a broader consumer group to demonstrate its Internet manufacturing capabilities.
“We have studied Internet + for a long time. ” Liang Zhonghong said that after the establishment of SAP S/4HANA platform, with a real-time database, Fenjiu Group is planning to build a unified coordination system at the business level of opening up franchised stores, channel suppliers, etc., through a large platform to serve marketing teams, dealers, etc.Consumers bring benefits to companies and consumers.
Liang Zhonghong said that Fenjiu Group is to give consumers the most concerned questions. Through traceability systems, consumers can ensure that the source of authentic goods in the distribution channels is reliable; transparent pricing systems can make it clear that their consumption is not at a loss; and for distribution channels, a more reasonable profit distribution model.This way enables the channel to enjoy the fruits of enterprise development together.
Not only that, the demand for liquor channels is becoming increasingly short and flat, “Channel vendors are likely to suddenly place orders for a certain type of product, such as 1000 cases immediately, but only one week delivery, not the next week.” Ma Ming said that in order to adapt to changes in consumption, liquor producers are passing through.”Flexible manufacturing” to adapt to this shorter and shorter production window period.
“IBMCognitive computing and artificial intelligence will be able to predict the behavior of alcohol consumers and distributors in the future and achieve more accurate business intelligence applications. Ma Ming said.
In this regard, Liang Zhonghong said, “Artificial intelligence and other new technologies, we will try and apply point by point. A lot of new concepts, we will try, for example, new retail, we can not do a small platform, and the introduction of Taobao e-commerce model, drop-by-drop snatch mode, the United States take-out mode, we willThe way to go is definitely a mixed mode rather than a single one.
AI cyberspace physical operating system
AI-CPS OS
“Today, when the branch of AI Cyber Physical Operating System (AI-CPS OS), a new generation of technology + business operating system, is used, business leaders must understand how to infiltrate “technology” into “business” fields such as companies, products and so on.In the scenario, the use of AI-CPS OS to form a digital + Intelligent force, the realization of industry re-layout, enterprise restructuring and self-renewal.
AI-CPS OSThe real value does not come from the technology or function, but from the integration of automation + informationization, intelligence + products + services and data + analysis in a way that delivers unique competitive advantages, which can unleash new business and operational models. If we can not achieve a larger scale cross functional convergence, noThese will be impossible to subvert the status quo.
Leaders can not rely on a single strategic approach to deal with multi-dimensional digital change. Faced with the subversive digitization and intellectualization of the new generation of technology + business operating system AI-CPS OS, leaders must stay ahead of the rest of the industry, businesses and individuals at all levels:
-
Re industry layout:How do you change your outlook on the world? What kind of reflection do you have to the industry paradigm?
-
Restructuring enterprises:What changes do your business need to make? How are you going to redefine your company?
-
Rebuild yourself:What kind of person do you need to be? How do you do it if you want to reinvent yourself and take the lead in the era of digitalization and intelligence?
AI-CPS OSIt is a digital intelligent innovation platform. The design idea is seamless integration of large data, Internet of Things, block chains and artificial intelligence in the cloud. It can help enterprises integrate innovation results into their own business system and realize the advantages and synergies of various frontier technologies in the cloud. Digitalization + intelligence of AI-CPS OSThe intersection of quantity and industry, enterprise and individual forms a leadership model, which integrates digitalization into the core position of the leader’s enterprise and leadership style.
-
Fine:This power enables people to observe and perceive what is happening in the real world and digital world at a more realistic and detailed level, and then to understand and control product personalization, micro-business scene events and results more carefully.
-
Intelligence:The model changes with time (data), and the whole system has the ability of intelligence (self-learning).
-
High efficiency:Enterprises need to establish real-time or quasi-real-time data acquisition and transmission, model prediction and response decision-making capabilities, so that intelligence from a batch, phased behavior into a reachable real-time behavior.
-
Uncertainty:Digital change subverts and alters the way of thinking, structure, and practical experience that leaders once relied on, resulting in a subversive force of compound uncertainty. The main uncertainties lie in three areas: technology, culture and institution.
-
Boundary blurred:The continuous integration of the digital world and the real world into CPS has not only changed the core products, economic theorems and possibilities of known industries, but also blurred the boundaries between different industries. This effect is spreading rapidly to ecosystems, enterprises, customers and products.
AI-CPS OSThe formation of digitalization + intelligence power stimulates economic growth through three ways:
-
Create virtual labor force to undertake complex tasks requiring adaptability and agility, i. e. intelligent automation, to distinguish it from traditional automation solutions.
-
Replenish and upgrade existing labor and physical assets to improve capital efficiency;
-
The popularization of AI will promote the innovation of many industries and open up a new space for economic growth.
Recommendations for decision makers and business leaders:
-
Beyond automation, to open up new innovative models: the use of self-learning and self-control capabilities of dynamic machine intelligence, to create new business opportunities for enterprises;
-
Welcome the new generation of information technology to meet artificial intelligence: seamless integration of human intelligence and machine intelligence.
Assess future knowledge and skill types;
-
Develop ethical norms: develop ethical codes for AI ecosystems and open them to intelligent machines.
Identify clearer standards and best practices in the process of development.
-
Pay attention to redistribution effect: prepare for the possible impact of artificial intelligence, formulate strategic help to face
A higher risk of unemployment;
-
Developing new capabilities for digital + Intelligent enterprises: Employee teams need to actively grasp important human-specific capabilities such as judgment, communication, imagination and creativity. For Chinese enterprises, creating a culture of inclusiveness and diversity is also very important.
Confucius said, “gentlemen are different but different.” “The Analects of Confucius: Zilu” cloud computing, big data, the Internet of Things, block chains and artificial intelligence, like a gentleman’s general integration, together reflects that technology is productivity.
If the last great discovery of Columbo’s geography is to expand the physical space of mankind. So this time, the discovery of geography is expanding the digital space of people. In the mathematical space, the establishment of a new business civilization, so as to discover a new model of wealth for human society to bring new wealth space. Cloud computing, big data and thingsNetworking and block chains are the ships that enter this digital space, and artificial intelligence is the sail on that ship, the sail of Columbus!
As the core driving force of a new round of industrial transformation, AI-CPS OS, a new generation of technology and commercial AI-CPS OS, will further release the enormous energy accumulated in the past technological revolution and industrial transformation, and create a new powerful engine. Reconstructing economic activities such as production, distribution, exchange and consumptionLinks, from the macro to micro fields of intelligent new demand, the birth of new technologies, new products, new industries, new formats, new models. It will lead to major changes in the economic structure, profoundly change the way of life and thinking of human beings, and realize the overall leap of social productive forces.
Industrial intelligence officer AI-CPS
“Use”AI cyberspace physical operating system”（New technology + business operating system “AI-CPS OS”: cloud computing + big data + Internet of Things + block chains + artificial intelligence,Construct the cognitive computing and machine intelligence of state perception-real-time analysis-independent decision-making-precise execution-learning promotion in the scene; realize the industrial interconnected ecological chain of industrial transformation and upgrading, DT-driven business, value innovation and creation.。