The application of fuel cell hybrid construction equipment (FCHCE) represents an attractive option for future industrial development. Construction equipment has highly repetitive modes of operation, which can be used for predictive control. However, the loads usually change markedly and frequently, leading to adverse scenarios for energy management. In this paper, we focus on alleviating this problem by presenting a predictive energy management strategy based on a model predictive control (MPC) framework for FCHCE. The supervisory-level energy management objectives are to respond to rapid variations in the load of the construction equipment while minimizing the fuel consumption, improving fuel cell durability, and maintaining the state of charge (SoC) of supercapacitors within the allowable bounds. To improve the performance and practicality of the predictive controllers, we present two power demand prediction methodologies based on Markov chains and neural networks. Simulations are carried out on the representative driving cycles of a wheel loader. Simulation results validate the feasibility and effectiveness of the proposed MPC. Neural network model made better predictions than the Markov chain model, which is preferable when modeling comprehensive driving cycles. (C) 2018 Elsevier Ltd. All rights reserved.