Boston Consulting Group once pointed out in the report “Industry 4.0 — Future Productivity and Manufacturing Development Prospects” that based on the powerful computing power brought by cloud computing to society, ARTIFICIAL intelligence will bring 25% increase in production efficiency in China’s manufacturing industry, creating additional value of 6 trillion yuan.

In the past year, Ali cloud ET industrial brain has served Xiexin Photovoltaic, Zhongce rubber, Zhengtai New Energy, Panzhihua Group and dozens of leading enterprises in the field of industrial segmentation, in the rate of good, cost, efficiency and other aspects of the breakthrough, to help enterprises create profits of billions of yuan.

In 2018 Guangdong Cloud Conference, Ali Cloud machine intelligence chief scientist Min Wanli shared how ET industrial brain to bring value to the industrial Internet and the technical strength behind, is willing to work with the majority of developers to achieve a win-win situation, will bring intelligence to more industrial production lines.

Here are excerpts from the speech:

Data upgrade and intelligent dimensionality reduction to build an intelligent industrial Internet platform



From October 2016 officially ET city brain, we will put the cloud computing, artificial intelligence, Internet technology of grafting to the real scene, then we go to the fields, came to farmland, all this is not an accident, have internal logic, is along the direction of data flow, do rise and intelligent data dimension reduction.

The explanation is simple: With all kinds of Internet, Internet of things and platforms, why are we still producing second-rate products and third-rate prices? Since the beginning of the Westernization movement, we have been introducing production lines, using the strength of the barbarians to cure the barbarians, but have we learned that skill? We may introduce production manuals, operation manuals, but we never know where the trick is.

Because the tricks are gone from the data, there’s no cloud computing, data analytics or anything to wake them up.

Our journey is to accurately reduce dimension, accurately tell you which machine button to adjust the parameters to which value, the final result is good. No longer look at the manual of the old master’s way to inherit from generation to generation, but with the electronic way to continue forever.

Industrial intelligence has three main elements: data, technology and industry. If any black technology cannot be written into this formula, its value appeal is empty. Connecting for the sake of connecting can never cause people of the same mind, but can only be different bedfellows. So many platforms, why is the value not explicit? Because the formula wasn’t written well.



In August 2016, one of our engineers wrote the earliest version of ET Industrial Brain, but the compilation error did not run successfully, exploring the path of innovation will never be accomplished overnight.

Today, we have accelerated from 1% starting point, 1% good rate to 5% good rate, and then to 100% open, ET industrial brain platform has been opened, and successfully integrated into the just released Feilong industrial Internet platform.

We put the industry into another intelligent industry to the Internet, the Internet so you see on this title more than an “I”, the global Internet industry is very much, there are more than 150, but for the first time, we added a “I”, because from the alien ET finally arrived, actually this ET should be a developer, In that cold control room two years ago, and also from aliyun’s flying engineers wrote the code nine years ago. With a determination to challenge the impossible, with a determination to make technical judgments, we have finally achieved an intelligent industrial Internet.

Self-developed technology support: MaxCompute, PAI



Today, if you open the APP of Taobao and Tmall, when you see the customized home page, MaxCompute is behind you to do the calculation. Every small function is supported by strong technology.

Two years ago, MaxCompute helped our engineers find the most accurate 60 key parameters and two key links out of thousands of parameters in the pv industry, resulting in a 1% yield improvement.

Computing is not enough, we also need intelligence. The brawn and the brain need to go hand in hand. When you have computing power, where does the brain come from?

It comes from PAI, our ai autonomous machine learning platform, which is used by thousands of alibaba Group engineers every day, supporting recommendation algorithms, search sorting and other functions.

We open up these two “weapons” and graft them into the industry, and the industry data is the value of percentage. It can be seen that if we identify the various big data products and artificial intelligence products that have been developed internally, we can achieve perfection from the very beginning with a little spark.

We can make more platforms become intelligent industrial Internet platforms, we add Internet GIS to it, covering not only photovoltaic, steel, photoelectric, but also petrochemical, textile, cnooc and so on. More importantly, more developers can follow this path and explore the unknown territory with our MaxCompute and PAI, so that the relay of innovation and exploration can be passed down from generation to generation without suffering the pain of the engineer two years ago, which is the original purpose of today’s launch. Let more developers and industry experts to identify the pain points of the industry, make the value.

Work with talent and developers



On The Tianchi Platform, many developers from around the world have given insights into practical problems in dozens of niche industries over the past four years.

We have just completed an intelligent analysis of a new industry segment. The Tianchi Competition took the real business pain points of Jianmei Aluminum as the scene, realized accurate annotation of sample data for the first time, and invited more than 4000 developers from all over the world to jointly challenge the automatic detection of aluminum production defects with artificial intelligence algorithm.

It used to depend on people’s eyes to judge whether there is a flaw, what kind of flaw. But with aliyun’s ARTIFICIAL intelligence, the “eyes” are never tired, resting or distracted. They are always there to read data, with 100 percent coverage.

Today, I am very happy to share with you a data that in this competition, among more than 4000 contestants, their champion has been able to achieve defect detection accuracy of over 80% through ET Industrial Brain in just two months.

This is an industry that our engineers have never seen before, but these four thousand developers are delivering results like this today.

Imagine, if the next is steel, coal and other segmented industries, we can completely use similar ideas to graft industrial Internet platform, and the industrial chain of products for transformation and upgrading, intelligent transformation.



More than a century ago, we carried out the Westernization Movement and introduced industrial equipment, and the operators still had long pigtails. It was the first time that we came into contact with the achievements of the industrial revolution, and we went to learn how to operate the production line.

Later we found that maintaining a production line, production line keep health work, need some techniques to make the product qualified, operation to obtain higher roa, so after one hundred, when we could introduce the production line would require manual, requires very good developers, or use the tips, instructions, fault maintenance manuals, etc.

But you may still remember a story. In the early days of China’s reform and opening up, when our production line broke down, we had to have experts fly in and tell them where to fix a part. The manual didn’t teach you that. Why is that? Trick son, this is the skill of walking rivers and lakes, impossible to fly everywhere.

But today, with the ET industrial brain, the industrial Internet platform, and the intelligent industrial Internet, it is possible for us to let the machine tell us what kind of maintenance it needs and what kind of care it needs.

Why is that? Because the results of the data analysis are already there, the secrets, the facts are all there. Alphago has become a super generational chess player overnight, because all the boards have shown these truths, just didn’t have the computing power to challenge it before. If you graft the core capabilities of industrial Internet and Ali Cloud ARTIFICIAL intelligence to these steel and coal industries, and to the boiler industry of Disson Thermal energy, we get the picture on the far right.

This is a real case, their operators no longer need to look at the manual, the tablet will give accurate recommendation numbers, all calculated in real time by ET Industrial Brain. No more calling the old master to ask what to do, no more turning over hundreds of pages of manuals, all of today’s wisdom is condensed on this small screen. The smaller the screen, the higher the density of intelligence, the higher the grafting force.

We have this philosophy that when we go out to developers, we end up being the fulcrum of the 1 percent, because we’re just a drop in the ocean of the big picture, and the greater wisdom comes from all of you. The Great Wall, marathon, intelligent manufacturing, digital transformation and intelligent upgrading cannot be completed overnight by a single product, but need several generations of scientific and technological personnel to continuously relay.

Today we hand over our baton, with ET industrial brain as the core, we deliver the industrial Internet platform, and hope that more people can carry on. Of course, when we reach the end of this long run, we will give the world the answer. We have redefined what is called industry 4.0 and what is called intelligent manufacturing.