Alex Álvarez Salaverria
General Director
ECON Tech
/
View from the Top

Detection, Analysis, Prediction, Correction: the Future of AI

By Andrea Villar | Wed, 10/21/2020 - 05:00

Q: How is ECON Tech working to boost technology implementations among automotive suppliers?

A: We have broad experience in the field of technology. Our company has a group of engineers with more than 20 years of experience in industrial automation. ECON Tech also has an R&D center where we introduce ideas and research topics that may be attractive to customers. Technology is constantly evolving and these actions and experience are the main strengths of the company.

For many years now, some industries in Mexico have implemented digital transformation systems where everything is interconnected. The volume of data that is being stored is too big to be analyzed correctly. No matter how hard a company tries to analyze it to obtain insights, the immense world of data will only allow us to analyze a tiny part of it. This brought us to consider the next step needed to solve the problem. We started research at our R&D center with the purpose of creating a system for data analysis to provide customers and users with insights.

One of the solutions we have used to solve this problem besides big data is the analysis of data using artificial intelligence (AI). In the coming months and years, there will be a boom in this technology, not because of its novelty but because of the current computing capacity and volume of data driving its adoption. Rather than one person trying to analyze a certain volume of data, we are looking for AI to be able to analyze very large volumes of data and achieve insights that we cannot yet imagine. This solution will eventually give us answers about how things should be optimized. In this research, there is an immense range of solutions and algorithms that serve to create machine learning-based solutions for predicting the quality of a product. We have been working hard to define the digital transformation path from Industry 3.0 to Industry 4.0.

Another area we have been working on at the research center is Message Queuing Telemetry Transport (MQTT), a messaging protocol that was developed several years ago to deal with a hostile system, namely one with little communication capacity or bandwidth. This protocol has been very helpful to manage data because it allows handling large volumes of data by optimizing the use of network resources. Experimenting with this protocol and testing it on different hardware platforms led us to integrate many devices without using the normal Industry 3.0 scheme of Programmable Logic Controllers (PLC) as a communication Hub. However, the volume of information and the capacity of devices that are going to be interconnected forces us to migrate to new architectures such as Industry 4.0. The latter does not use PLCs as a hub for the exchange of information but different devices interconnected with innovative protocols.

Q: In which cases has ECON successfully implemented this solution?

A: We implemented this solution in the steel sector. As specialists in the industry, we understand the need to achieve better quality in steel production processes. The current market demands much more specialization and there are limitations in terms of the laboratory analyses that can be done during a process. Hence the idea of making quality predictions through the use of AI. Once we detect the problem, the system makes a prescriptive analysis to take action and help the process to get to the proposed objectives.

In the automotive industry, one of the most important factors is the measurement of performance through KPIs. Among the factors that can affect efficiency in a production chain are interruptions, failures, rejected parts, or machine efficiency. It is very difficult for someone in the company to analyze all these factors since the variables over time are many. Instead, AI can optimize performance measurements creating connections that would not be visible to a person. For example, AI can detect a rejected product, and when investigating further find that the root of the problem is the logistics of inventory management. AI has a great capacity to learn from historical data. At ECON, we work with Overall Equipment Effectiveness (OEE), which is the gold standard for measuring manufacturing productivity.

Q: What are the automotive industry's weaknesses in data analysis for process improvement?

A: What the automotive industry needs are a solid digital transformation strategy. Many companies want to have data analysis based on AI but they still do not have the structure for that. The roadmap to digital transformation consists of four steps. The first is to have the capacity to know what the failure was by integrating all systems into a unified data platform that manages all the information from the manufacturing process but also from the warehouses, logistics, and the quality department. Once everything is integrated, the information is available to the entire production chain. The second step is to understand why there was a malfunction. This is where descriptive analysis comes into play, helping to statistically analyze data to find out why a failure occurred.

The third step ensures that all systems are interconnected and that there is a quality data history. At this point, it is possible to start implementing solutions based on predictive analytics, which will foresee quality and performance problems and even production line failures. However, at ECON we did not stop there. By knowing what will happen, we developed a solution that allows the system to take action based on predictions to correct failures that will happen in the future.

 

ECON Tech offers innovative solutions in project development and engineering based on technology and research. The company specializes in automation, electrical systems and information technology, using IT and the Internet of Things to build smarter businesses

Andrea Villar Andrea Villar Journalist and Industry Analyst