A BP-IPSO Algorithm Suitable for Centralized Thermoelectric Generation System Power Tracking under Nonuniform Temperature Distribution

The power-voltage (P-V) characteristic curve of the centralized thermoelectric generation (TEG) system under nonuniform temperature distribution (NTD) exhibits multiple extreme point characteristics, and traditional maximum power tracking (MPPT) algorithm is prone to fall into local (LMPP) takes a long time track. This paper designs BP-IPSO based on back propagation neural network (BPNN) improved particle swarm optimization (IPSO) for MPPT. firstly utilizes good nonlinear function fitting ability BPNN obtain relationship between control input output establish TEG array prediction model. Then, dynamic learning factor weight coefficient are introduced (PSO) search model realize MPPT control. MATLAB/Simulink experiment results show that can effectively avoid falling LMPP, quickly accurately track global (GMPP), suppress oscillation voltage during process. Especially in start-up test experiment, compared with perturb observe (P&O), PSO, grey wolf optimizer (GWO), energy generated by increased 12.84%, 3.18%, 4.75%, respectively, which efficiency.