Self-Protected Virtual Sensor Network for Microcontroller Fault Detection.

Area of Science:

• Electronics Engineering
• Artificial Intelligence
• Machine Learning

Background:

• Comparing functional behavior of individual electronic hardware units is essential for ensuring device integrity.
• Existing methods may not adequately capture the time-dependent nature of electronic hardware signals.
• The virtual sensor network (VSN) approach offers a framework for modeling physical system outputs.

Purpose of the Study:

• To introduce a procedure for comparing the functional behavior of individual electronic hardware units of the same type.
• To estimate the functional integrity of an unknown device unit by comparing it to a known reference unit.
• To extend the VSN approach using neural networks to model time-dependent electronic hardware behavior.

Main Methods:

• Utilized a virtual sensor network (VSN) approach.
• Employed a neural network (NN) with Long-Short-Term-Memory (LSTM) layers to model time-dependent functional behavior.
• Validated the method on measurements from a remote-controlled drone with reference and malfunctioning controller units.

Main Results:

• The proposed method successfully identified and described unexpected behavior in the malfunctioning controller unit.
• Achieved a signal sample prediction time of 0.14 ms.
• Obtained a reconstruction accuracy with a root mean square error (RMSE) below 0.04 relative to the data range.

Conclusions:

• The developed procedure effectively assesses the functional integrity of electronic hardware units.
• The integration of LSTM-based neural networks enhances the modeling of time-dependent signal behavior.
• Three self-protection features were introduced to quantify the certainty of the virtual sensor network model output.