Looking for Automation System to Differentiate Distinct Sounds

Click For Summary

Discussion Overview

The discussion revolves around finding an automated system capable of differentiating distinct sounds, specifically in the context of detecting metal on metal contact in a processing environment. Participants explore various technological approaches and systems that could potentially replace a human operator's auditory monitoring.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested

Main Points Raised

  • One participant notes the challenge of finding a suitable automation system, mentioning that a previous attempt by a company to use a sensor was unsuccessful.
  • Another participant questions the predictability of the sound, suggesting that if it is simple enough, a microphone could be used with real-time FFT analysis to detect frequency and magnitude.
  • A different participant proposes looking into machine predictive maintenance systems, which utilize FFT sampling and monitoring, indicating they could alert users based on set limits regardless of the input sound.
  • Another suggestion involves exploring computer-based voice recognition systems that can be trained to recognize specific sounds.
  • The original poster clarifies that the sound in question is metal on metal contact, which is repetitive but not predictable due to variations in the size of the metal pieces. They describe additional interference factors affecting sound detection.
  • The original poster expresses hope that a suitable system exists, despite previous unsuccessful attempts, and indicates interest in real-time FFT systems based on the feedback received.

Areas of Agreement / Disagreement

Participants present multiple competing views and approaches without reaching a consensus on a specific solution or system. The discussion remains unresolved regarding the best method for sound differentiation.

Contextual Notes

Limitations include the variability of the sound due to different metal sizes and the interference caused by the material's drop height and composition. The effectiveness of proposed systems may depend on these factors.

lollerskate
Messages
2
Reaction score
0
I'm currently looking for a system which can automate a part of my process. Currently, there is an operator listening for sound, which is distinct enough for the human ear. Upon hearing this sound, he shuts down the system. I would imagine that if an ear can easily differentiate the sound, a sensor should be capable of doing so as well. However, I've only had the luck of ONE company coming out so far, and he was unsuccessful at getting his sensor to do so (it was a system originally meant for sounds in a grinding process). Can someone please direct me to a company or a system which might serve my purpose? I have been using google, but my search terms must not be up to par to find what I need.
 
Engineering news on Phys.org
How predictable is the sound. Is it like a tone or a more complex sound? Would you say that the frequency of the sound remains fairly constant?

If its simple enough, it seems like you should be able to simply mic it up and run the data through some sort of real-time FFT. Observe for the frequency + magnitude within some tolerance.
 
Along what Minger said about the microphone set up, you may want to look at machine predictive maintenance systems. While they are meant for measuring machine vibrations, they include FFT sampling and monitoring based on your limits and will alert you if certian limits are met. They shouldn't care what the real input is. If the sound is a pure tone it is even easier.
 
One other thought would be to look into computer-based voice recognition systems. They can often be "taught" to recognize words or phrases, so maybe you can teach it to recognize the sound.
 
The sound is metal on metal contact. It is repetitive since it is made by metal bouncing on a metal shaker table. The sound is not very predictable since the size of the metal coming through can vary. It is only liberated metal we wish to detect, which makes the task harder. The table normally has chunks of rubber going through it, with beads of metal implanted in the rubber. While this metal can hit the table, it doesn't make the same sound as items such as bolts, or chunks of hardened steel.

Problems being observed:
The material originally drops onto the table from a height of ~1.5', causing interference
Since the material sometimes has metal that is only mostly embedded within it (which is fine), there is also interference when from it hitting.

There was a system that was applied to our process, but it didn't look promising. However, with the sound being detectable by ear, there should be a system available.

Thanks for the comments so far. I will look into real-time FFT systems.
 

Similar threads

Non magnetic linear displacement system for moving a Hall Sensor
  • · Replies 11 ·
Replies
11
Views
2K
Why does a series of pulses generate a pitch?
  • · Replies 27 ·
Replies
27
Views
5K
Very preliminary thoughts on a recoil reducing gun
  • · Replies 5 ·
Replies
5
Views
2K
Graduate Integrability of Systems of Differential Equations
  • · Replies 5 ·
Replies
5
Views
4K
Automotive Spark length of a points ignition compared to an HEI system
  • · Replies 7 ·
Replies
7
Views
3K
Automation Ethics: Should your car serve you or serve society?
  • · Replies 46 ·
2
Replies
46
Views
8K
Tabular Method for a Unique Spur Gear Differential System
  • · Replies 2 ·
Replies
2
Views
3K
Physics of Water in the Closed System of the Great Pyramid
  • · Replies 15 ·
Replies
15
Views
4K
What's the best way to break bad speech habits?
  • · Replies 1 ·
Replies
1
Views
1K
Chemistry Entropy in isolated composite system for irreversible process
  • · Replies 3 ·
Replies
3
Views
3K