Predicitve Maintenance Project mit Arduino

[dannyg]

Hi everyone,
I’m planning a small project to simulate predictive maintenance using Arduino and Python.

The idea is: I’m working with a small motor and want to monitor temperature and current. For example, 500 mA and 30 °C would be considered normal operating values. If the temperature stays above 30 °C and the current rises above 700 mA for more than 2 minutes, the system should issue a warning like: “Motor might fail in 4 days if these conditions continue.”

I know this is very basic and not meant to give perfectly accurate predictions – my goal is just to create a simple and working prototype to demonstrate the idea of predictive maintenance. The sensor data will be collected by the Arduino and sent to a PC via serial connection. On the PC, I want to analyze the data using Python (e.g., with scikit-learn) and let the system decide whether a potential fault is developing.

Has anyone here done something similar or have ideas on how to implement this in a simple but effective way? I’d also be grateful for links to related projects or tutorials! I didnt started yet I am searching for Information. Thank you.

[Larz60+]

you can find some temperature monitor software here
and some arduino current monitor software here

[DeaD_EyE]

Operating resources break down after a certain period of time, which is sometimes specified in the corresponding data sheets. Then there are factors that accelerate the aging of operating resources.


Possible influencing factors:

• Operating temperature
  
• Ambient temperature
  
• Power consumption
  
• Rotational speed
  
• Vibrations
  
• Mechanical stress
  
• Humidity
  
• Radiation: radioactive, electromagnetic
  
• ... list is incomplete
  

The following rule applies to semiconductors, for example:
For every 10°C increase in operating temperature, the failure rate doubles (or, in other words, the service life is halved).

In motors, for example, the operating temperature also plays a role, as it causes the insulation of the winding to age more quickly.

A simple class to do the calculation for a motor:

class MotorLifeMonitor:
    def __init__(self, rated_lifetime_hours, reference_temp):
        self.remaining_life_hours = rated_lifetime_hours
        self.reference_temp = reference_temp

    def update(self, current_temp, elapsed_hours):
        """
        Update the remaining life based on current temperature and elapsed real time.

        :param current_temp: Current temperature in °C
        :param elapsed_hours: Past time in hours until last call
        :return: Remaing life time hours
        """
        delta_t = current_temp - self.reference_temp
        aging_factor = 2 ** (delta_t / 10)
        life_used = elapsed_hours * aging_factor

        self.remaining_life_hours -= min(life_used, self.remaining_life_hours)
        return self.remaining_life_hours


motor = MotorLifeMonitor(20_000, 60)
motor.update(60, 1)
motor.update(60, 1)

It is also important to store the remaining hours (remaining_life_hours) on the microcontroller and to initialize the object with remaining_life_hours from non-volatile storage (NVS) when starting the microcontroller.

It is also possible to calculate how much a motor degrades when the speed is higher than the maximum speed specified by the manufacturer. However, this requires knowledge of the bearings and the design of the motor. I assume that this is not as significant as an increased operating temperature.

Vibration sensors on the motor can be used to determine whether the motor bearing is about to fail.