Fault Diagnosis of Active Phase Change Control Device based on SGSSA-BP Neural Network

The great advantage of stratospheric lighter-than-air vehicle in earth observation and communication applications, active phase change control device as the core of overpressure regulation, must ensure long flight time health and reliable operation. To this end, this paper uses the sparrow search algorithm to optimize and adjust the initial weights and thresholds of the BP neural network, and improve the problems of slow convergence, easy to fall into local optimum and poor global search ability of the sparrow search algorithm. Chaotic Maps are used to improve the diversity of the initial population, and a nonlinear sine learning factor and golden sine algorithm are introduced in the position update of discoverers and joiners as a way to improve the global search and jump out of local optimum ability of the algorithm and increase the convergence speed. The experimental results show that the performance of the improved sparrow search algorithm is significantly better, and the classification prediction accuracy of active phase change control device fault diagnosis is increased from 53.33% to 96.66%, which proves the practicality and effectiveness of the algorithm and is of great significance for the stable operation of stratospheric lighter-than-air vehicles.