The renewable energy based distributed generation draws much attention because of its environmentally-friendly feature. At the same time, these kind of generation is volatile and uncertain. So it’s important to make good placement in the distribution network to make full use of the profit. But it must be noted that distribution networks due to the unequal phase loads and mutual impedance between various phases of lines are virtually unbalanced. So in this paper, determination of the appropriate size and place of distributed generations is done in unbalanced distribution network that distributed generations’ appropriate size obtained with respect to probabilistic methods and their convenient location is selected using weighted multi-objective method. With regard to impact of exact load flow in calculations, linear three-phase unbalanced load flow method considering the loads model and loads connection type is used which is the fast and accurate unbalanced load flow method. To solve this model, an improved krill herd algorithm is proposed, which can bring a rapid convergence. Simulations are done in IEEE 37 bus unbalanced network in Matlab software. The simulation results indicate that network power losses and voltage unbalance factor are reduced, voltage profile of each phase is improved and significant profit is obtained for distribution companies.