Manufacturers often view AI as highly complex and expensive software that requires end-to-end systems across the company to function properly. The reality is that AI is more focused and achievable. It can work on the factory floor with the smallest structure and connect to machines through the Industrial Internet of Things (IIoT). In this article, the author makes key recommendations and three scenarios on how to use manufacturing intelligence in a real-world environment.

In a recent manufacturing insights survey on artificial intelligence (AI), 44% of respondents from automotive and manufacturing ranked artificial intelligence as a category that is “very important” to manufacturing functions over the next five years, and Nearly half to 49% of respondents consider artificial intelligence “absolutely vital to success”.

However, in many cases, it is difficult for manufacturers to understand AI, because the technology industry has used such a wide range of tools, and few people really understand how it is instantiated, except for some of the all-purpose resources that provide better business results. outer.

Manufacturers may actually think that artificial intelligence is very complex and expensive and requires the entire company’s end-to-end systems to work properly, which means costly updates to entire IT / OT operations. The truth is that artificial intelligence is more focused and easier to implement. Artificial intelligence can work in factories with minimal structure and connect with machines through the Industrial Internet of Things (IIoT).

When it comes to the implementation of artificial intelligence, the first thing an OEM needs to know is the type of use case to focus on. As part of the Internet of Things, most edge machines on the production floor are being reorganized to send data via wireless sensors. This data is then entered into a software suite for processing. The data entry process will become an ongoing process to create an ever-expanding data network. All of this data can be stored in the cloud for insight, making artificial intelligence-driven models possible.

The following three use cases can help dispel manufacturers’ concerns about AI capabilities:

1. Machine uptime

The consumer product packaging production line is 24 × 7, producing millions of cartons of different sizes for packaging different consumer products. It is important to keep production free of any failures or any quality issues. Speed ​​and quality matter. Manual monitoring is error prone, costly and inefficient.

The data collected through the IIoT system can provide 24/7 real-time insights on production line throughput and equipment failures through tailored visualizations and alerts. AI can finally help you understand how much data you want to collect. This data is processed on the edge gateway to quickly identify anomalies and send instant alerts. Larger data is aggregated in a cloud-based IoT platform for further predictive analysis and defined behavior and rule-based models. The system will provide a custom dashboard and report machine idle time, failure reason codes and overall OEE data. This allows managers to better plan their operations, avoiding machine idle time and performing predictive maintenance.

2. Cost optimization

American sensor manufacturer SpectraSymbol has been producing one of the best linear sensors and potentiometers in the industry to address the energy market. As a process, in remote oil wells, when oil and water are pumped into the tank, the oil and water levels need to be measured. Regarding this oil drilling operation, the company urgently needs to use IIoT data to extend the service life of marginal wells more economically, so as to continue to optimize costs. The biggest problem is that these wells have insufficient oil production and are not worthy of data sensors. Uniform investments, so their cost models must be reduced. These wells are also located in remote areas, adding cost and time challenges. The sensor installation costs for these wells are also very high, increasing the cost by 60%. For smaller operations and distant scrap wells, rapid ROI is the key to IoT implementation.

An IIoT software platform was built for SpectaSymbol’s multiple wells to store and process all machine data. It creates a “data lake” with related data stored in the cloud. Data analyzed through AI-driven machine learning has become a driving force for custom applications for business that are specifically designed to evaluate well performance and condition monitoring through AI analysis. As a result, specific reports are available to all stakeholders, and the operating time and performance of marginal wells are optimized.

3. Improve forecast quality

A chemical company, SRF, wants to increase its productivity and manufacturing operations through IoT-based digital transformation. To achieve this, SRF must connect key processes in the production of its packaging films and technical textiles. The goal is to improve quality by analyzing parameters that are critical to the manufacturing process, improve its fuel consumption and reduce power consumption, and reduce any line interruptions. SRF’s plant productivity can be improved by using condition monitoring to predict downtime. The result “data lake” created by the inputs of the manufacturing process has been integrated with SRF’s ERP to close the cycle throughout the manufacturing value chain.

Artificial intelligence is at the core of the project because it leverages machine learning technology to support a flexible set of multivariate statistical analysis. Specifically, real-time machine data is used as a feedback loop to more accurately define the best settings for the machine to ensure product quality and machine reliability. The result is that SRF monitors and analyzes parameters that are critical to machine health, and optimizes machine downtime by making predictions before failures occur.

The following process outlines a plan for improving forecast accuracy using artificial intelligence support systems:

1. Evaluate and characterize the current forecasting system.
2. Measure the current level of error.
3. Compare error levels with industry norms.
4. Specify new requirements.
5. Characterize the economic impact of improved forecasts.
6. Identify alternative AI forecasting options.
7. Select best approach(s).
8. Develop implementation schedule.
9. Identify potential bottlenecks and problem areas.
10. Implement new system and monitor performance.

Source :-

Leave a Comment

Scroll to Top