You can use current sensor as an alternative method or you can use both vibration and current together to evaluate which one has more significant impacts to the monitoring purpose. You can see some references https://www.mmsonline.com/blog/post/using-ai-to-predict-cnc-machine-spindle-issues-before-they-are-issues & Article Industrial Artificial Intelligence for Industry 4.0-based Ma...

Article Industrial AI

Dear Yazan,

The criteria never change - we know when we MUST change a part due to failure or wear. The PrM challenge is to optimize the intervention to best utilize the machine, parts, and personnel. WHEN you make a change will depend on many business factors, such as the cost of downtime, cost of bad product/waste, cost of maintenance, and parts.

I would think that part of the value of such a dissertation would be in developing such an algorithm for various values. The sensors just provide data points. The heart of PrM vs. reactive maintenance is how we use that data. As a start, you can look at things like the interface or tool temperature, speed of milling to a particular position (size), power cost per operation and things like that to determine when the machine part is beginning to wear, up to a failure point if desired.

The overall analysis of maintenance must compare the cost of the predictive intervention to the cost of the reactive intervention. In many cases, it really does make more sense to let a part fail before replacing it! But I imagine that is one of the reasons for the potential value of such a dissertation.

Best regards,

Steven

You can use current sensor as an alternative method or you can use both vibration and current together to evaluate which one has more significant impacts to the monitoring purpose. You can see some references https://www.mmsonline.com/blog/post/using-ai-to-predict-cnc-machine-spindle-issues-before-they-are-issues & Article Industrial Artificial Intelligence for Industry 4.0-based Ma...

Article Industrial AI

From Our Research Team. See Below.

Hi Yazan,

My name is Cyrus, I am a Ph.D. student in Prof. Jay Lee’s lab.  Please see my comments below in regards to the milling machine health monitoring project you have. I hope these comments would help you to resolve the issue and move forward with your project, however, after reading the comments if still there are additional questions, I would be happy to provide more information.

As you mentioned there are many contributing factors that may change the threshold used in triggering the maintenance alarm. In order to reduce the source of variations, I would suggest using a no-load (free-run) condition for data collection and health assessment. For instance, in a similar project, we ran a spindle at a constant speed under zero-load and collected data for machine learning tasks. After training different models, the same working condition (constant speed and zero-load) was used for testing the health condition of the system. This process removes lots of variations initiated from the changes in cutting parts, and complex operations.

Moreover, you can follow the general guidelines developed by the IMS center [1-3] to reduce uncertainties in the development of your machine learning models. These steps include 1- Identifying the critical assets 2- Setting up the DAQ system and monitoring agents for high-quality data collection 3- Preprocessing the data for removing noise and outliers and separating data into proper regimes. 4- Feature extraction and developing machine learning models suitable for your specific problem 5- Model evaluation and update if required.

[1] Lee, Jay, et al. "Prognostics and health management design for rotary machinery systems—Reviews, methodology, and applications." Mechanical systems and signal processing 42.1-2 (2014): 314-334.

[2] Jia, Xiaodong, et al. "A deviation based assessment methodology for multiple machine health patterns classification and fault detection." Mechanical Systems and Signal Processing 99 (2018): 244-261.

[3] Singh, Jaskaran, et al. "A systematic review of machine learning algorithms for PHM of rolling element bearings: fundamentals, concepts, and applications." Measurement Science and Technology (2020).

Best, Cyrus