Study of blind source separation algorithm based on particle swarm optimization in nuclear power background

In nuclear power plant or marine nuclear power plant, mechanical devices that carry the important functions such as power transmission tends to age due to high-speed operation of components. If they cause a malfunction, it will have serious consequences. However, these devices and components usually generate vibration during operation, and there is often a close relationship between the signals generated by the vibration and the operating conditions. The effective monitoring and analysis of the relevant vibration signals enables the timely detection and elimination of equipment failures and other factors that endanger safety, then we can ensure the normal operation of equipment and facilities. The vibration signal measured in reality is a complex mixed signal from multiple unknown vibration sources, and sources of vibration signals need to be differentiated before signal analysis. In this context, blind source separation can be used. Simulation results show that the blind source separation algorithm based on particle swarm optimization can effectively separate signals, the convergence speed of the algorithm is fast. After adding the gradient information to optimize, the convergence speed slightly improves. (author)