MATLAB Live Scripts for measured-reference Broadband active vibration control of a voice-coil-actuated SDOF platform
Description
This software dataset provides three self-contained MATLAB Live Scripts accompanying the manuscript "Broadband FxLMS Active Vibration Control of an SDOF Mass-Spring Platform Using Measured Reference Acceleration and In-Run Secondary-Path Identification." Part 1 presents direct LMS and filtered-x LMS demonstrations, step-size and filter-length studies, single-tone and multi-component control, a moving-frequency simulation, and numerical convergence diagnostics. Part 2 contains six optional synthetic protocol studies covering parameter variation, a local frequency sweep and hammer disturbances. The third Live Script implements a reduced electromechanical model for converting acceleration-domain commands to voice-coil actuator voltage. The scripts include their helper functions and generate figures and numerical exports during execution. No external input data or separate MATLAB source files are required. The implementation was tested in MATLAB R2020a without add-on toolboxes. The dataset contains numerical demonstrations, not experimental acquisition records or the LabVIEW controller. The optional studies use selected model parameters and do not reproduce measured traces. Very large attenuation values in noise-free convergence diagnostics represent numerical cancellation and must not be interpreted as physical performance. The accompanying README documents execution, output locations, model scope and interpretation.
Files
Steps to reproduce
Download the three MATLAB Live Scripts to a writable local folder. Open `BFxLMS_Convergence_Study_Part_1.mlx` in MATLAB R2020a or newer and select Run All. Run `BFxLMS_Optional_Protocol_Surrogates_Part_2.mlx` independently for the optional synthetic studies, and run `VCM_Acceleration_to_Voltage_Mapping_Live.mlx` independently for the actuator-command calculations. No add-on toolboxes or external input data are required. Parameters are defined in the opening sections. The scripts generate their output directories automatically and export figures and numerical results. The adaptive studies use a fixed random seed of 4 with documented case-specific offsets. See README.md for output locations and interpretation limits.
Institutions
- Beihang UniversityBeijing, Beijing
Categories
Funders
- China Scholarship CouncilMinistry of Education of the People's Republic of ChinaBeijing, BeijingGrant ID: 2024GSP007924