<p>The Beluga Whale Optimization (BWO) is a meta-heuristic algorithm that simulates the life behavior of beluga whales. Aiming at the shortcomings of the BWO, such as poor solution accuracy, insufficient robust performance, and weak ability to jump out of local trap, this paper proposes an improved multi-strategy BWO (IMS-BWO). To prevent excessive tracking of individual with the best fitness value during the exploitation stage, a roulette-based fitness distance balancing strategy is used to provide scientific guidance on their hunting behavior. Utilizing a random differential restart strategy, the beluga whale’s position is periodically reset to aid in escaping local optimal solution. A non-monopoly search method incorporating Levy motion strategy is proposed to adjust the dimensional imbalance of the optimal individual. In this paper, corresponding experiments are conducted on IMS-BWO using CEC2017 and CEC2022 standard test sets, including motion analysis experiment of the non-monopoly search method, performance experiment of strategy, comparison algorithm experiment and portability experiment of strategy. The experimental results show that under the CEC2017 test set (CEC2022 test set), IMS-BWO outperforms BWO, HHO, HPHHO, MELGWO, QMESSA, GWO and RIME with 100% (100%), 99.1% (100%), 99.1% (100%), 98.3% (87.5%), 84.5% (79.2%), 83.6% (87.5%), and 69.8% (62.5%) respectively. Therefore, IMS-BWO has higher optimization accuracy and stronger ability to jump out of local trap. In addition, IMS-BWO has superior portability and excellent robustness. Finally, the IMS-BWO is used to solve five engineering optimization problems, and the results are all ranked first, proving the ability of IMS-BWO to solve real-life applications.</p>