From IIoT, AIoT to a Smart Factory

2023 年 9 月 27 日

Whether it’s been nearly a decade of Industry 4.0, or the recent years of big data, machine learning, and AIoT smart factories, the essential prerequisite behind their operations is data collection through the Internet of Things (IoT). Since IoT is so crucial, how should companies progressively move forward to ultimately achieve an AIoT smart factory? The following diagram illustrates the five steps of factory digital transformation

1st step : IIoT (Machine Networking)

Organizing the existing machine networking capabilities is the first and most time-consuming step, often the one most easily abandoned. As the saying goes, “The hardest part is to get started,” which is quite fitting. We recommend categorizing machines into the following five types as a starting point:

  1. Machines that are impossible to retrofit and have limited functionality, such as machines with only start and stop buttons and no display of production results.
  2. Machines that, while unable to directly connect to the network, consistently store production-related data in fixed directories.
  3. Machines that don’t store data in fixed directories but can capture production data through modifications.
  4. Machines with connectivity, but communication is unidirectional. In other words, the machine can send data to the system, and the system can place production parameters in a fixed directory on the machine for the machine to read and execute.
  5. Machines that, in addition to the capabilities of the fourth category, can also provide real-time production status in response to system queries, making them the ideal candidates for IoT connectivity.

2nd step: Data Storage

Before the advent of the big data trend, even though machines had data that could be stored, many factories, considering the cost of storage space, did not store machine data in databases. Instead, they typically retained files for several months to several years before deletion. In recent years, the emergence of big data databases has made it economically viable to store these vast volumes of machine production data. Storing this data is essential to enable subsequent data mining and AI applications.

3rd step: Real-time Dashboard

A real-time dashboard can be created even before storing big data. The data source relies on data input from production or equipment personnel, and the inability to provide real-time data is the biggest drawback. This also affects the accuracy of subsequent OEE (Overall Equipment Effectiveness) reports, resulting in discrepancies between actual output and machine specification output. With the implementation of IIoT, not only can machine conditions be reflected in real-time on the real-time dashboard, but it also ensures that subsequent reports are based on accurate data for calculation.

4th step: Data Mining

The other purpose of IIoT is to indentify the relationships among the data in the big data. For example, if the machine continuously sends out the specific alert messages within 5 minutes, there is a 90% probability that the breakdown will occur in 10 minutes. In such cases, we can provide early warnings to equipment engineers for preemptive action, reducing the frequency and downtime of machine failures, and improving production efficiency.

Learn about data platform. By integrating data across the systems, various enterprise applications such as AI, BI, apps, or internal systems can be satisfied. This resolves the issue of data redundancy and reduces data processing costs when planning enterprise services.

5th step: AIoT to Smart Factory

With the logical insights derived from big data analysis and real-time machine connectivity, the objectives of a smart factory encompass various capabilities. Whether it’s AI automating dispatching assignments for maximizing production capacity, adjusting production parameters in real-time based on current conditions, or performing rapid quality inspections using real-time image, they all fall within the purview of a smart factory. The 4th step of data mining and the 5th step of the smart factory are quite similar, both aimed at improving factory production efficiency based on big data. The difference lies in data mining, which uncovers data logic using existing data, whereas AI utilizes real-time data for future predictions.

Learn more about Equipment automation.

Using the example of a specific warning message continuously appearing on a machine within 5 minutes, with a 90% probability of a shutdown within 10 minutes, this logic was discovered during data mining. We can incorporate this logic into the program to monitor and send notifications in real-time, which is the result of data mining. However, if we do not continue data mining to discover new logics, the monitoring logic remains static, making monitoring a passive execution method.

With the implementation of AI, in addition to the logics discovered through existing data mining, AI continues to monitor real-time data and identify data correlation logic. This transition from passive monitoring to proactive monitoring allows for the early detection of issues, providing more time for resolution. You may be also interested in the benefits of AI platform.

NTT DATA Taiwan not only assists many enterprises in implementing systems such as ERP, MES, WMS, APS, but also provide the services of IIoT and setting up real-time dashboards, leadning the clients to build an AIoT smart factory in the end. We have comprehensive practical experience to speed up the digital transformation process for enterprises and reduce the period of trial and error.