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

Procedure 1

The execution procedure of BP-IPSO for MPPT of TEG system.

1. Initialize the training samples and the parameters of each particle;

2. Initialize the power prediction model structure by Eq. (7)–(9);

3. FOR to 10.

4.  Input the duty cycle of the th training sample into the boost converter;

5.  Collect the real-time voltage and current signals of the TEG array, and calculate the output power;

6. END FOR.

7. Train the BPNN to fit the control input-power output (I/O) curve using Eq. (17) as the objective function;

8. FOR to 3.

9.  Perform the th IPSO search on the I/O curve, and update each particle flight position and the global optimal value by Eq. (10)–(13);

10.  Input the global optimal value obtained by IPSO th search into the boost converter;

11.  Collect the real-time voltage and current signals of the TEG array, and calculate the output power;

12.  Add new training samples to the neural network;

13. END FOR.

14. Use the new training samples to train the BPNN again and fit the I/O curve;

15. Perform IPSO search on the I/O curve and output the optimal duty cycle;

16. When the TEG array output power variation , the algorithm restarts.