Communication, illumination, and energy saving are three basic objectives of visible light communication (VLC) system. However, it's difficult to simultaneously optimize these performances. Therefore, this paper jointly optimizes energy consumption and illumination while communication and other constraints are satisfied. The problem is formulated using multi-objective and multi-control-variable optimization. A novel p-optimality based clonal selection algorithm (CLONALG) is developed to solve it. Compared with conventional Pareto optimality method, our algorithm exerts prior preference on optimization objectives, enabling multiple search directions towards the optimal solution set. Simulation results prove that illumination uniformity is improved while LED power consumption is reduced. Also, advantages of p-optimality are identified and validated by comparing with Pareto optimality method.