Programmable Matrix-Array Thermoelectric System With Tri-State Energy Conversion

This article presents a matrix-arranged thermoelectric programmable control system. The system flexibly reconfigures the functions of each thermoelectric module in the matrix, overcoming the limitations of unidirectional energy flow and fixed functionality in traditional thermoelectric systems. By constructing a three-state (cooling/heating/energy harvesting) reconfigurable topology, distributed cooperative control of temperature management and energy recovery in nonuniform thermal fields is realized. A dynamic temperature management strategy based on the fuzzy-PID algorithm is adopted, allowing the region's temperature to converge rapidly. Experimental results indicate that, compared to conventional methods, the proposed system achieves a temperature error of only 1.72 °C, reduces convergence time by 50.88%, and improves temperature conversion efficiency by up to 92.57%. In energy harvesting mode, the system achieves a maximum captured voltage of 320 mV, with a ripple rate below 1.6% after boosting to 5 V, and a peak power density of 7.24 × 10⁻⁵ W/cm²K. This work provides an efficient and energy-consistent integrated solution for semiconductor cooling, industrial thermal management, and other applications.