Abstract: Combined Heat and Power Economic Dispatch (CHPED) is an important problem in the energy field, and it is beneficial for improving the utilization efficiency of power and heat energies. This paper proposes a Modified Genetic Algorithm (MGA) to determine the power and heat outputs of three kinds of units for CHPED. First, MGA replaces the simulated binary crossover by a new one based on the uniform and guassian distributions, and its convergence can be enhanced. Second, MGA modifies the mutation operator by introducing a disturbance coefficient based on guassian distribution, which can decrease the risk of being trapped into local optima. Eight instances with or without prohibited operating zones are used to investigate the efficiencies of MGA and other four genetic algorithms for CHPED. In comparison with the other algorithms, MGA has reduced generation costs by at least 562.73$, 1068.7$, 522.68$ and 1016.24$, respectively, for instances 3, 4, 7 and 8, and it has reduced generation costs by at most 848.22$, 3642.85$, 897.63$ and 3812.65$, respectively, for instances 3, 4, 7 and 8. Therefore, MGA has desirable convergence and stability for CHPED in comparison with the other four genetic algorithms.

Key words: Modified genetic algorithm , · Combined heat and power economic dispatch , · Uniform distribution , · Guassian distribution , · Disturbance coefficient , · Prohibited operating zone