Non-orthogonal multiple access (NOMA) in communication system has recently received significant attention since it provides a superior spectral efficiency and supports massive connectivity. We study a resource allocation problem in multi-user downlink NOMA systems to achieve the maximum energy efficiency. The original optimization problem is a non-convex problem that is difficult to solve. Therefore, the original optimization problem is decomposed into two sub-problems, i.e., the user assignment and the power allocation. A user grouping method based on greedy algorithm is firstly used to reduce the computational complexity of the exhaustive method. Secondly, for the given user grouping method, the power allocation coefficient expression of the multiplexed user on each sub-channel is obtained. In order to further improve the energy efficiency, a non-equal power allocation scheme is investigated. The problem of power allocation on all sub-channels is formulated as a nonlinear and non-convex problem of the sum of the ratios, and a Dinkelbach-like algorithm is proposed to obtain a suboptimal solution. Simulation results show that the NOMA system equipped with the proposed algorithms perform well in terms of sum rate and energy efficiency.