M4E 2025 Quiz
Manufacturing 4 Everyone or M4E started three years ago as a way to transfer knowledge to my coworkers. It has been specially planned for those less erudite in manufacturing, to learn about these topics (but it doesn’t hurt to refresh some concepts!). Gradiant is an ICT R&D centre with some projects in the manufacturing and Industry 4.0 world. Hence the need to create a way for my colleagues to foster creativity, learning the language of manufacturing and many of the classical basic concepts of this world, but also of future manufacturing.
As a way to promote reading this newsletter on fridays, once in the year I organize a contest with 5 questions and 5 prizes, giving them to the coworker that provided the most valid answer to each question. Next you can find the questions from this year edition, that took place yesterday. Hope you can answer them! All can be answered by reviewing past M4E posts. Feel free to leave your answers in the comments or in the chat and discuss! I’ll provide the answers on friday.
In an electronics components factory, a shelf full of obsolete components that are no longer used is reviewed. The supervisor orders the disposal of everything that has not been utilized in the past year. What is being done? On the other hand, they also discover that the components are stored without any criteria, so the supervisor decides that the team should label and assign a specific place for each type of component. What is being done? Answer both questions in Japanese.
Suppose a part has technical specifications (in cm) of 6.00 ± 0.02. Considering that the cost implications of a deviation are 6800, what would be the economic loss incurred per part if the dimensions of a manufactured part are 6.05 cm? What method did you use to calculate it?
Gradiant is developing a system based on artificial intelligence to predict failures in production lines of a company manufacturing electronic boards. The goal is to anticipate soldering defects or component feeding issues by using sensor data, PLC logs, and historical records from the MES.
Before defining the final architecture of the system, the team carries out several actions to understand the client’s needs.
Interviews are conducted with maintenance managers to identify which failures cause the most downtime and their economic impact.
The team shadows plant personnel during real shifts to observe how they respond to line alarms and interpret messages from the current system.
An automatic analysis of logs and sensors is performed using a process mining tool to identify failure patterns.
A survey is conducted with end customers (those who purchase the assembled products) regarding their perception of the finished product's quality.
Internal audits and documented non-conformities from the quality department are reviewed.
Which of the above actions correspond to a VoC methodology? Reason your answer.
In the following VSM, there is a value of 0.48 days between the drilling and packaging process. Where does this value come from? And the value of 1200?
Gradiant is developing a video analysis system with drones to detect intrusions in critical infrastructures. During field tests, the team detects that the system generates many false alarms when there are small animals, wind, or shadow movements.
To solve the problem, they have applied a methodology whose steps are listed below in a non-sequential order. Provide the name of the methodology and arrange these steps. A keen eye will notice an error.
A. Real videos are collected from 5 facilities monitored by drones over 15 days, and more than 400 alarm events are manually classified.
B. The project problem is defined: “Reduce the rate of false alarms due to non-human events (animals, shadows, branches) below 2%.”
C. The commercial team consults with clients to find out if they would be willing to pay for an additional thermal analysis module on the drone that would improve its performance.
D. An improvement is implemented in the detection model using a neural network trained with labels distinguishing humans from other moving objects.
E. The annotated data set is analyzed, and it is determined that the current model is biased towards movement rather than shape or behavior.
F. A real-time tracking system is installed that reports the false alarm rate weekly and blocks deployments if the threshold is exceeded.
