Recent incidents like the major fire at the Nan-Ya Plastics in Linkou, which is still under investigation, highlight the constant emergence of similar events. These incidents may stem from overuse of operating equipment or the absence of real-time monitoring systems guarding against these potential problems, gradually manifesting unseen threats.

 

To address such issues, Yu-Chen offers a professional system to help customers promptly grasp the status of on-site equipment. It can identify problematic components in the early stages, using vibration principles to unravel issues without relying on the long-term cost of manual inspections to reduce the probability of accidents. This not only enhances equipment efficiency but also enables timely scheduling of maintenance based on data analysis.

 

In the past, maintenance personnel might only discover equipment issues during major accidents or shutdowns. However, now there are convenient systems available for them to inspect relevant machinery, eliminating the need to ponder which machines to inspect and maintain daily. Instead, they can clearly identify which machines may develop problems and when, maximizing usage rates and extending equipment lifespan to the fullest.

How does Yu-Chen achieve this? In the next 5 minutes, we'll quickly delve into the secrets of predictive maintenance.
 

＊What if the Motor Behaves Badly or Has a Lot of Problems?

In the semiconductor technology industry, motors are essential to production operations, serving critical roles such as fans, pumps, conveyors, and air conditioning units. However, they are prone to accumulating hidden dangers.

 

Due to the high-intensity operating frequencies of rotating equipment like motors, minor issues may go unnoticed over prolonged use. Common problems include loose bearings, rotor failures, stator failures, and misaligned shafts. To address these potential issues, the traditional method involves periodic manual inspections by carrying testing equipment. However, as mentioned earlier, this method is no longer the optimal solution.

 

With the rise of smart factories and the application of AIoT, the concept of predictive maintenance is flourishing. According to a Deloitte report, predictive maintenance can increase capacity by 25%, reduce failures by 70%, and lower maintenance costs by 25%. Vibration monitoring has numerous successful cases in the market, covering industries such as energy, manufacturing, daily life, power, industry, and automotive. This is why vibration is crucial for monitoring motors.

 

＊Understanding Vibration Alone Isn't Enough, Can It Also Calculate Lifespan and Schedule Shifts for Equipment?

Why use vibration as a detection basis?

Because there is an ISO 10816 standard in the vibration field for assessing and monitoring the vibration of rotating machinery, providing evaluations of mechanical condition to reduce failure rates and increase equipment lifespan. From the diagram below, you can see different evaluation indicators based on different operating equipment, conditions, and frequencies. These are fundamental thresholds quantified by big data that can quickly determine whether the currently operating equipment has exceeded the normal warning range.

  
 

Generally, experts use spectrum analysis to extract underlying parameters from the vibration data generated by equipment and compare them with ISO standards. However, determining which frequency corresponds to which component is the essence of the technology and a secret that experts are unwilling to reveal. However, the vibration monitoring system developed by Yu-Chen allows you to become an expert in a second!

 

Yu-Chen spent several years collecting vibration big data and introduced AI artificial intelligence technology to improve accuracy. Based on ISO 10816 and ISO 20816 standards, a software system has been established to grasp the past, present, and future. It compares current operating conditions with past equipment vibration data and predicts future service life based on the two. After installation, you can remotely check whether the equipment is operating normally at home, without worrying about negligence by on-site personnel or a machine suddenly breaking down. It can also help you calculate the appropriate maintenance cycle for equipment, maximizing the service life of equipment.

 

It even uses special frequency division technology, meaning it no longer just analyzes equipment based on total vibration quantity, but can capture the frequency range corresponding to faults through spectral diagrams, identifying abnormal vibrations of specific components. This eliminates the need for users to worry about where the problem lies, directly addressing issues to save costs and time.

  
 

＊Four Major AI Technologies Save More Than Just Billions

The above mentioned convenient features hide many difficulties and tears, which are the four major AI technologies developed by Yu-Chen with great care. Each project contains some clever ideas to help future managers operate the software easily and intuitively, and to plan the most perfect maintenance schedule as early as possible.


1.Automatic Setting of Warning Values/Danger Values:
In addition to the total vibration quantity, there are no standardized specifications for the vibration quantities of other components except for ISO 10816. Therefore, maintenance personnel often rely on their own past experience to set upper limits. However, in practice, improper upper limit settings often lead to overly frequent false alarms, losing the purpose of early warning, or requiring a large amount of manpower to set reasonable upper limits when there are many sensors. Fortunately, Yu-Chen has combined a large amount of data sorting and SPC control boundary to automatically determine reasonable warning values and danger values for equipment operation through AI.

2.Remaining Useful Life Estimation (RUL):
Using collected historical vibration data, the system predicts the failure warning line of the equipment in a linear and nonlinear manner to reverse-calculate the remaining reasonable operating time. Of course, there are many models and formulas involved here, so I won't go into detail here. Simply put, this feature helps maintenance personnel quickly determine the time limit for the normal rotation of the equipment, whether it's in years, months, days, or even hours, reaching the effect of arranging maintenance and repairs in advance.
 

3.Equipment Health:
By using three indicators: the definition of the dangerous value in the ISO standard, the remaining life ratio, and the stability ratio, as well as dividing them into four colors, the system allows users to intuitively browse the equipment that needs attention in the entire factory. The digitized health value can also help maintenance personnel prioritize maintenance based on the order of maintenance, obtaining maximum benefits at minimal costs. The calculation of the three indicators needs to adjust different parameter settings according to the operating status of the equipment to achieve the best predictive value.
 

4.Failure Mode Analysis (FMA):
Based on past historical failure mode data and comparing it with current equipment operation data, the system establishes the probability of all potential failure factors. This is a growing feature that can incorporate newly obtained failure data into prediction models by adding new data to the database model and autonomously training through neural network models. It can be said that it becomes stronger as it fights, becoming the most powerful assistant in the factory equipment.
 

【Summary】

There are many rotating equipment in the factory area. If you rely solely on manual inspection and maintenance, it's time-consuming and labor-intensive. However, with Yu-Chen vibration monitoring system, you have data collection every 10 seconds, 24-hour online monitoring, and a simple and intuitive user interface, which is a good helper that meets the expectations of factory managers.

 

Overall, if manufacturers adopt predictive maintenance intelligent monitoring, it is estimated that maintenance costs can be reduced by 15-25%. Additionally, it can reduce machine failure repair time by 50% and machine failure probability by over 85%.

back