The project provides a solution to the problem of analyzing big data collected from machinery and equipment of manufacturing plants through sensors. This is a real problem that all businesses are currently aiming for and is the foundation of “industrial data mining”. This research is completely consistent with the general trend of the “industrial revolution 4.0” taking place in Vietnam.
The Industrial Revolution 4.0 has created favorable conditions for smart factories to be born. A smart factory is where a Cyber Physical System – CPS communicates based on an Internet of Things (IoT) connection to support people and machines in performing work. CPS is a highly interconnected and integrated system of machine tools, personnel, management systems, and even customer service. This integration will create a big data environment. How to effectively monitor large amounts of high-frequency streaming data and perform mass data analysis will be a major challenge when developing smart factories. Therefore, there is an urgent need to develop a set of big data analytics tools that play an important role in the CPS system for the smart factory of Industry 4.0.
Reasons why data analytics is becoming increasingly important, especially in the field of Industry 4.0: 1) large amounts of data containing a lot of valuable information have not been fully used and exploited; 2) with the development of sensor technology, the cost and technology of data collection become easier; 3) software and hardware technologies such as cloud computing have increased the computing power of devices and can quickly process large amounts of data; 4) production processes are increasingly complex, and human experience cannot be relied on to manage the entire system.
This project researches and proposes a data analysis model for high-frequency large data series in the smart manufacturing industry. The proposed model consists of two stages. In phase 1, the large data series is detected and filtered to remove unusual features that could cause errors in production. In phase 2, these abnormal data are analyzed to find key factors from machine tools that affect product quality in the smart factory.
The big data analytics model proposed in this project is expected to become a core component of industrial big data analytics platforms, with tools that can quickly find the main causes of anomalies in large data series related to the production process at machine tools. Besides, the proposed data analysis model also meets the following goals: 1) providing a toolkit to meet the rapid processing requirements of industrial big data; 2) accelerate feature selection in batch and multidimensional data; and 3) infuse data analytics processes to analyze the causes and probabilistic needs of production state transitions.