Packaging company uses AI to reduce waste and reduce production process costs

SINGAPORE — For local packaging company Federal Packaging Industries, one minute of downtime in its paper box manufacturing process can have significant consequences.

This could lead to a waste of 10-20% of raw materials, energy and man-hours.

The company supplies packaging boxes for many companies, including multinational corporations and small and medium-sized businesses like itself.

To help reduce downtime costs, Federal Packaging is working with local technology company Evercomm to develop an artificial intelligence (AI) algorithm that can help detect bottlenecks and issues in the process. production of boxes at an early stage in order to avoid production interruptions.

This will be done with co-funding from the Advanced Digital Solutions program of the Infocomm Media Development Authority (IMDA), which helps SMEs to digitize themselves by adopting advanced technologies, such as robotics, AI and Internet-connected devices.

Under this initiative, SMEs can obtain up to 80% financial support for eligible digital solutions.

Last Friday (March 4), Minister of State for Communications and Information Tan Kiat How announced during his ministry’s budget debate that the government would expand the initiative’s current list of 30 solutions, with a focus on AI-based and cloud-based solutions. .

Nearly 500 companies have adopted or are in the process of implementing solutions under the initiative.

The program was announced in March 2020 to help businesses deepen their digital capabilities and navigate challenging situations so they can continue to operate with little disruption.

For Federal Packaging, the company hopes that by using the AI ​​solution co-funded by the program, it can detect machine problems earlier and fix them faster.

To help it develop the AI, sensors have been installed in the company’s machines. The sensors collect data such as machine temperature, energy consumption and water flow. Data collection began around the middle of last year.

Around six to 12 months of collected data are needed to train the AI ​​algorithm.

Once the AI ​​model is ready, Federal Packaging can use it to monitor problems with their machines so they can be addressed quickly to minimize production downtime and waste.

“When we have data on our machines, we will know very quickly if there are anomalies detected by the AI. This gives us more time to act sooner on any problem that arises, in addition to the actions we have already taken,” said Federal Lau Chee Herng, vision manager at Packaging.