Optimal　foraging　algorithm　(OFA)　was　presented　as　a　stochastic　search　algorithm　to　solve　global　optimization　problems　in　2017.　As　an　emerging　algorithm,　there　are　many　excellent　potentials　to　be　explored.　To　enhance　the　performance　of　OFA,　a　novel　optimal　foraging　algorithm　with　direction　prediction　is　presented　in　this　paper,　named　OFA/DP.　During　the　iterations,　the　population　information　can　be　fully　utilized　by　OFA/DP.　Once　a　new　optimal　solution　is　found,　the　evolutionary　direction　prediction　strategy　is　applied　to　generate　more　potential　candidates.　In　addition,　considering　the　situation　that　the　population　does　not　evolve　for　a　long　time,　which　means　that　the　algorithm　has　achieved　the　global　optima　or　trapped　into　local　optima.　In　this　case,　the　Gaussian　oscillation　strategy　is　adopted　to　attempt　to　find　a　better　solution,　and　escaping　from　local　optima.　To　validate　the　efficiency　of　the　proposed　algorithm,　the　numerical　experiments　on　20　benchmark　functions,　30　CEC　test　functions,　6　large　scale　functions,　28　CEC2017　constrained　problems,　3　engineering　problems,　6　unconstrained　multi-objective　functions　and　10　constrained　multi-objective　functions　are　executed.　The　simulation　results　and　the　statistical　test　demonstrate　that　OFA/DP　has　a　superior　performance　in　most　of　functions　with　faster　convergence　speed.　(C)　2021　Elsevier　B.V.　All　rights　reserved.