Research on Decoupling Model of Six-Component Force Sensor Based on Artificial Neural Network and Polynomial...

Area of Science:

• * Engineering
• * Artificial Intelligence
• * Sensor Technology

Background:

• * Multidimensional mixed loading poses challenges for six-component force sensors.
• * Decoupling load measurements is crucial for accurate sensor data.
• * Existing methods may struggle with complex loading scenarios.

Purpose of the Study:

• * To propose a novel two-stage decoupling model for six-component force sensors.
• * To address the challenge of multidimensional mixed loading.
• * To improve the accuracy and reliability of force and moment measurements.

Main Methods:

• * Development of a two-stage artificial neural network model.
• * Stage 1: Six-dimensional load categorization using a deep BP neural network and 63 load category labels.
• * Stage 2: Six-dimensional load regression combining polynomial regression with a BP neural network.
• * Design of a six-component force sensor utilizing Fiber Bragg Grating (FBG) sensors.
• * Establishment of elastomer simulation and experimental datasets for validation.

Main Results:

• * Simulation data: 93.65% accuracy in the classification stage.
• * Simulation data: Mean Absolute Percentage Error (MAPE) of 6.29% for force and 3.24% for moment in the regression stage.
• * Experimental data: 87.80% accuracy in the classification stage.
• * Experimental data: MAPE of 5.63% for force and 4.82% for moment in the regression stage.

Conclusions:

• * The proposed two-stage decoupling model effectively addresses the load decoupling problem for six-component force sensors under multidimensional mixed loading.
• * The model demonstrates high accuracy in both load categorization and regression stages, validated by both simulation and experimental data.
• * The use of FBG sensors in conjunction with the developed model shows promise for enhanced sensor performance.