Intelligent manufacturing after years of discussion and evolution, the goal has been quite clear and is also an achievable vision, the truth is easy to understand, the introduction of AI, and then make more than human results, but when used in intelligent manufacturing, it seems to lack anything. For AI to succeed, there are three main components: algorithms, computing power, and data. The first two are easy to get if you have enough capital, but data is particularly troublesome.

The trouble is that the collection itself is a problem, and the data is not well understood. With current technology, language and images can be converted into data, but there are many types of machines, and the characteristics and requirements generated are not yet understood, and how to analyze them?

1794-IE12 Transformation starts with data

The factory must first digitize (digital transformation), collect good data, understand the data, and really have the opportunity to go to intelligent manufacturing.

It is not enough to collect only operational data, equipment and technology data should be collected completely. There are a lot of vendors out there offering solutions, but a lot of them are doing “visible digital transformation.” For example, which machine is not in processing, there is no material, there is no light and so on. But what is really needed should be the so-called “invisible digital transformation”, that is, the processing quality, efficiency, stability, etc., inside the machine, rather than simply examining the processing and not processing.

Of course, it is good to control the operation rate, but if the operation rate is 1794-IE12  low on any day, the operator does not know what is happening. So the future is the key, collect information that can grasp the future, through analysis to know why, the future these things can be under control.

Don’t treat digital transformation like a project

There are more than one or two links in the factory process, but in any case, we must first analyze the data to understand before we can optimize. At present, the common practice in the industry is to throw out the most complex and valuable problems, set up a project team, and deal with them in a project manner.

But such an approach can create a shift of responsibility within the organization, if not the people on the project team, then the digital transformation has nothing to do with it. In fact, in order for a factory to truly evolve and move toward Industry 4.0, there will be hundreds or even thousands of problems that need to be solved, that is, the so-called “long tail” problems, and the total value of these problems is likely to exceed the most expensive ones.

1794-IE12

Enterprises need a core data analysis process architecture to solve the long-tail problem. In the past, in the field of software development, enterprises have prevailed in a set of DevOps development and maintenance architecture, in order to systematize and scale to cope with frequent deployment needs.

1794-IE12 In the face of the improvement of data analysis efficiency requirements in the digital process, the concept of AnalyticOps is now followed by the DevOps architecture in the world, and a set of systematic platforms are used to build a scalable data analysis process to improve the efficiency of digitalization in various aspects of the enterprise.

For example, the theoretical structure of the product launched by Information Center is similar to that of AnalyticOps, and the intelligent decision-making platform JarviX is launched to cut into the needs of OT(operational technology) side, and lowers the use threshold of tools through AI enhanced analysis technology, so that OT side can complete the whole process of data analysis independently. To reduce the huge cross-departmental communication costs paid by enterprises for data analysis, 1794-IE12 compared with the past project team consisting of IT and DT personnel to do analysis, now OT personnel can directly do data analysis, and improve the efficiency of all aspects of the analysis process on a large scale.

The mechanics of learning will be the biggest challenge

To make better products, of course, you can improve the accuracy or stability of the machine, but how to use it is the meaning of data analysis, through analysis, you can add value, beyond the limit of everyone’s use, and see the future from the data, really solve the problem or continue to optimize the process.

Traditional masters store these machine usage methods in their heads and turn them into intuitive responses, of course, it is not excluded that the master also does some statistical analysis, but mainly relies on experience, and these things are more biased toward “personal repair.” In fact, the whole world now, through the innovation of hardware and software technology, can be more accurate.

There is another problem. The world is changing so fast that what old Master said is not wrong, but it is not necessarily right. One of the benefits of data is that the correct ingredients are definitely larger, because they reflect what’s actually inside the process. Like looking at whether a person is healthy or not, if energy measures all the data of the body, not just from the surface, it must be more accurate.