Relay nodes are used to improve the throughput, delay and reliability performance of energy harvesting networks by assisting both energy and information transfer between information nodes and access point. Previous studies on radio frequency energy harvesting networks are limited to single source single/multiple relay networks. In this paper, a novel joint relay selection, scheduling and power control problem for multiple source multiple relay network is formulated with the objective of minimizing the total duration of wireless power and information transfer. The formulated problem is non-convex mixed-integer non-linear programming problem, and proven to be NP-hard. We first formulate a subproblem on scheduling and power control for a given relay selection. We propose an efficient optimal algorithm based on a bi-level optimization over power transfer time allocation. Then, for optimal relay selection, we present optimal exponential-time Branch-and-Bound (BB) based algorithm where the nodes are pruned with problem specific lower and upper bounds. We also provide two BB-based heuristic approaches limiting the number of branches generated from a BB-node, and a relay criterion based lower complexity heuristic algorithm. The performance of the proposed algorithms are demonstrated to outperform conventional harvest-then-cooperate approaches with up to 88% lower schedule length for various network settings.