AI

A repairman fixing a washing machine. / Photo by: Andriy Popov via 123RF

 

Computer scientists and engineers are tasked to ensure the performance of machines. They are also assigned to analyze data to know machine problems. While that works in today's generation, a company in Finland and another in the Netherlands are looking into an AI system that can translate sound into something that can identify troubled machines.

Noiseless Acoustics in Helsinki, Finland, and OneWatt in Amsterdam, the Netherlands, are using AI to understand problems of machines using patterns of sound. The AI enables the research team to make sound visible, whether it can be heard or not. To do this, they applied non-invasive sensors, machine learning algorithms, and predictive maintenance solutions.

OneWatt employed its Embedded Acoustic Recognition Sensors or EARS device and combined it with frequency analysis and machine learning to detect and predict faults. Correct predictions can reveal what the problem will be, when will it occur, and where will it happen.

“Audio is the most apparent sign of mechanical failure… Visible light is not a good indicator since it is not able to see through the motor, and could not tell if the bearings are degrading,” explained Paolo Samontañez, chief technology officer at OneWatt.

Meanwhile, Noiseless Acoustics is using a combined set of hardware, software, and analytic technologies to listen to sounds. The NL Camera enables them to detect any problem using sounds in the same manner as thermal imaging. The heat signals detected by the camera are translated as noise on the screen.

The NL Sense allowed them to locate where the problem exactly is. The compact wireless sensor hub and its sensors placed on any surface will send a signal if issues are found. Through cloud service, the data will be analyzed and processed.

“Sound describes things, it’s completely its own world,” said Kai Sakesla, CEO at Noiseless Acoustics.

Audio-based technical diagnostics of machines and robots could be a beneficial tool for modern-day devices such as gas pumps, power grids, factories, and more.

 
 

 

Computer scientists and engineers are tasked to ensure the performance of machines. They are also assigned to analyze data to know machine problems. While that works in today's generation, a company in Finland and another in the Netherlands are looking into an AI system that can translate sound into something that can identify troubled machines.

Noiseless Acoustics in Helsinki, Finland, and OneWatt in Amsterdam, the Netherlands, are using AI to understand problems of machines using patterns of sound. The AI enables the research team to make sound visible, whether it can be heard or not. To do this, they applied non-invasive sensors, machine learning algorithms, and predictive maintenance solutions.

OneWatt employed its Embedded Acoustic Recognition Sensors or EARS device and combined it with frequency analysis and machine learning to detect and predict faults. Correct predictions can reveal what the problem will be, when will it occur, and where will it happen.

“Audio is the most apparent sign of mechanical failure… Visible light is not a good indicator since it is not able to see through the motor, and could not tell if the bearings are degrading,” explained Paolo Samontañez, chief technology officer at OneWatt.

Meanwhile, Noiseless Acoustics is using a combined set of hardware, software, and analytic technologies to listen to sounds. The NL Camera enables them to detect any problem using sounds in the same manner as thermal imaging. The heat signals detected by the camera are translated as noise on the screen.

The NL Sense allowed them to locate where the problem exactly is. The compact wireless sensor hub and its sensors placed on any surface will send a signal if issues are found. Through cloud service, the data will be analyzed and processed.

“Sound describes things, it’s completely its own world,” said Kai Sakesla, CEO at Noiseless Acoustics.

Audio-based technical diagnostics of machines and robots could be a beneficial tool for modern-day devices such as gas pumps, power grids, factories, and more.