A Sensor Dynamic Measurement Error Prediction Model Based on NAPSO-SVM.

Area of Science:

• Engineering
• Computer Science
• Data Science

Background:

• Dynamic measurement error correction is crucial for enhancing sensor precision.
• Support Vector Machine (SVM) is commonly used for predicting sensor dynamic measurement errors, but manual parameter tuning limits performance.

Purpose of the Study:

• To propose and evaluate an improved Support Vector Machine (SVM) method, termed NAPSO-SVM, for predicting dynamic measurement errors in sensors.
• To enhance the SVM's predictive accuracy by optimizing its parameters using a novel algorithm.

Main Methods:

• Developed a novel particle swarm optimization (PSO) algorithm, named NAPSO, incorporating natural selection and simulated annealing to improve local optima avoidance.
• Applied the NAPSO algorithm to optimize SVM parameters for dynamic measurement error prediction.
• Compared the performance of NAPSO-SVM against standard PSO-SVM and Glowworm Swarm Optimization-SVM (GSO-SVM) using sensor error data.

Main Results:

• The NAPSO-SVM method achieved higher prediction precision and lower prediction errors compared to PSO-SVM and GSO-SVM.
• Evaluation metrics included Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE).
• Experimental results validated the effectiveness of NAPSO-SVM in dynamic measurement error prediction.

Conclusions:

• The proposed NAPSO-SVM method is an effective approach for predicting dynamic measurement errors in sensors.
• The integration of natural selection and simulated annealing in PSO significantly improves SVM parameter optimization.
• NAPSO-SVM offers a more precise and reliable solution for sensor error prediction compared to existing methods.