Self-Calibration Algorithm for a Pressure Sensor with a Real-Time Approach Based on an Artificial Neural Network.

Area of Science:

• Sensor Technology
• Artificial Intelligence
• Metrology

Background:

• Traditional pressure sensor calibration is manual, time-consuming, and computationally inadequate for nonlinear sensors.
• Existing methods struggle with sensor issues like hysteresis, gain variation, and lack of linearity.
• Accurate pressure measurement is critical for applications such as advanced grasping mechanisms.

Purpose of the Study:

• To develop and validate a novel self-calibration methodology for nonlinear pressure sensors.
• To address limitations of traditional calibration methods using an advanced artificial neural network.
• To improve the accuracy and reliability of pressure sensor readings over time.

Main Methods:

• Implementation of a Levenberg Marquardt Back Propagation Artificial Neural Network (LMBP-ANN) model.
• Utilized a real-time dataset collected from pressure sensors, with a load cell as a reference.
• Validated the model by comparing predicted pressure outputs against experimental target pressures.

Main Results:

• The LMBP-ANN model demonstrated superior performance compared to traditional methods, achieving a maximum mean square error of 0.17325.
• The model achieved an R-value over 0.99 for training, testing, and validation, indicating high accuracy.
• Successfully predicted pressure over time, even accounting for material creep and sensor uncertainties.

Conclusions:

• The proposed LMBP-ANN model effectively overcomes hysteresis, gain variation, and linearity issues in pressure sensors.
• This self-calibration approach enhances sensor durability and can improve the performance of related mechanisms, like robotic graspers.
• The analysis methodology provides a valuable tool for evaluating real-time measurement system performance.