Wind power integration with high penetration in a power system is indispensable. However, wind power integration, especially with high level, raises the power system instability problems due to its natural variability and unpredictability, which increases system uncertainties. Thus, uncertainties and correlations amongst wind farms should be considered in a power system operation and planning. One of the best solutions for facilitating the wind power integration is the installation of an energy storage system (ESS). However, the location and sizing of ESSs should be optimally planned to achieve maximum benefits such as minimising total cost, time shifting, reliability and power quality enhancement, minimising power loss, improving the power factor and providing environmental support. In this paper, a new probabilistic discretising method is derived and developed to discretise the continuous joint power distribution of correlated wind farms. Combining the new probabilistic discretising method with a multi-objective hybrid particle swarm optimisation (MOPSO) and non-dominated sorting genetic algorithm (NSGAII), a new hybrid probabilistic optimisation algorithm is proposed. The proposed hybrid algorithm aims to search for the best location and size of energy storage system (ESSs) and considers the power uncertainties of multi-correlated wind farms. The objective functions to be minimised include a system's total expected cost restricted by investment budget, total expected voltage deviation and total expected carbon emission. IEEE 30-bus and IEEE 57-bus systems are adopted to perform the case studies using the proposed hybrid probabilistic optimisation algorithm. The simulation results demonstrate the effectiveness of the proposed hybrid method in solving the optimal allocation problem of ESSs and considering the uncertainties of wind farms’ output power and the correlation amongst them.