2018年11月17日星期六

HKSQ/Minitab/TQM Seminar on Asset Management and Reliability: Machine Learning + AI

The seminar named “Asset Management and Reliability: Machine Learning + AI” coorganized by HKSQ, Minitab and TQM held on 17th Nov 2018. Before the seminar, I took a photo with Dr. Aaron Tong (Former chairman, HKSQ). 


HKSQ exco team took a group photo. Dr. Jane Wong (Exco member, HKSQ) joined the second part of the seminar.  Since the seminar was too attractive for our members, it separated into two session with three speakers (09:00 a.m. – 11:30 a.m. / 11:30 a.m. – 02:00 p.m.).


I also took a photo with speakers and coorganizers.

(Left: I, Prof. Peter W. Tse (CityU), Ms. Iza Ng, Mr. Ming Dong and Mr. Alex Kwan (Minitab))

I met Mr. Henry Li (Chairman, IISE)).


Before the seminar start, three guest speakers took a group photo.


Firstly, Dr. Aaron Tong gave a welcome speech and introduced the seminar to all participants.


Mr. Ming Dong (Senior Technical Training Manager, Minitab Pty Ltd) was the first speaker and his presentation entitled “Using Minitab and Machine Learning Tools to Increase Maintenance Reliability and Reduce Defective Products”. The main message of his talk included the current state of asset, asset management system and reliability applied into business.


Mr. Dong explained Reliability that was “The probability (0~1) that a system or product would perform its intended function under normal operation conditions for a specified period of time”.  There had four reasons to analyze reliability included “Predicting product reliability”, “Predicting or assessing warranty costs”, “Assessing the effect of proposed design changes” and “Assessing whether customers’ needs are met”.  Then he introduced the Bathtub Mode to under the shelf life.


After that Mr. Dong employed “Particular truck multiple failure modes” included Engine, Tire and Brake failure hour as example to using Minitab reliability software.  He also explained series system and parallel system on reliability.  He demonstrated Minitab using sample data and summarized its characteristic and capability. 


And then Mr. Ming Dong introduced “Salford Predictive Modeler (SPM)” that Salford Systems was a machine learning Software Company founded in 1983 by Dr. Dan Steinbery after he completed his PhD in Econometrics at Harvard.  Minitab acquired Salford System in March 2017.  SPM could be used for Machine Learning especially in manufacturing included operation, asset management & maintenance, and field service.


Salford Predictive Modeler aimed to optimize a large plant / company / service with many complex process steps, many predictors, huge data sets, many complex nonlinear effects or interactions, many missing values and outliers. The terminology of machine learning were described as following diagram.


At the end, Mr. Ming Dong using SPM for live demonstration.


Prof. Peter W. Tse (Director, Smart Engineering Asset Management (SEAM) Laboratory, CityU) was the second speaker and his topic named “AI in Reliability and Asset Management”.  Firstly, Prof. Tse quoted McKinsey’s AI Report to show us the AI technology trends included deep learning neural networks, transfer learning and reinforcement learning.  


Then Prof. Peter Tse briefed Industrial AI and its relationship in the following diagram.  That related to Sensor, IoT & Data, AI and Industrial Big Data Analytics. Prof. Tse also mentioned Reliability in Data 6Vs (i.e. Volume, Velocity, Variety, Veracity, Visibility and Value), 3Bs (Bad, Broken and Background) and 3Rs (Resources, Relationship and Reference). 


After that Prof. Tse briefed the Data-driven AI Algorithms separated into Supervised Learning and Unsupervised Learning.


He also mentioned the importance in knowing the collected data about their reliability and accuracy.  There are three data validation for Machine Learning including Detectability, Diagnosability and Prognosability.  


Finally, Prof. Tse shared a real case study about Slurry Pumps to predict its breakdown.  He used sample of frequency analysis of vibration data files collected from a health pump and a deteriorating pump.  Prof. Tse said most experts only considered low frequency that was not able to identify the problem but he employed high frequency that could identify the failure.  Then he proposed a real-time condition-based &proactive maintenance for the customer.  Lastly, he concluded that a fully automatic and intelligent system had been designed, tailor-made and implemented for predicting the remained useful life (RUL) of slurry pumps.  Hence, the occureence of sudden breakdown of pumps was avoided.  


Dr. Aaron Tong (Former Chairman, HKSQ and China Outstanding Quality Professional, CAQ (中國傑出質量人)) was the third speaker and his presentation entitled-QA vs +QA (減法質量保證 vs 傳統增量質量保證). 


Dr. Tong reviewed the traditional cost of quality and the idea approach expected to be reduced total cost of quality (COQ) through the failure cost.


In fact, increased appraisal and preventive cost could not be reduced the total COQ if you were manufacturer.  Because you increased cost and the benefit for your buyer (not end customer).  Your buyer only said “Thank You” and never increase the price to you.  Therefore, Dr. Tong propose minus QA (-QA).


His approach was to reduce failure but did not increase cost on parts or inspection.  Actually, he proposed to reduce the inspection and prevention cost.  The core theory was focused on the root causes of production and its propagation. He requested supplier to do it right and no need to do better or best. 


At the end, Dr. Aaron Tong shared some successful cases such as “SUPOR” that employed –QA concept to reduce the after-sale return rate from 3% to 0.9%


Reference:
HKSQ - www.hksq.org
Minitab - www.minitab.com
20180712: Minitab Seminar on “Machine Learning enhance Product Reliability from Prediction to Execution” - https://qualityalchemist.blogspot.com/2018/07/minitab-seminar-on-machine-learning.html
20180414: HKSQ Congratulate to Dr. Aaron Tong awarded China Outstanding Quality Professional (中國傑出質量人) 2017 and visit Beijing Industries - https://qualityalchemist.blogspot.com/2018/04/hksq-congratulate-to-dr-aaron-tong.html
20181108: The 3rd World Summit for Chinese Quality (WSCQ) – Day 1 (Part A) - https://qualityalchemist.blogspot.com/2018/11/the-3rd-world-summit-for-chinese.html

沒有留言:

LinkWithin

Related Posts with Thumbnails