As the cloud computing market continues to grow, there is an increasing need for research into task scheduling strategies in cloud computing. This paper focuses on the limitations of traditional scheduling algorithms and heuristic intelligent scheduling algorithms, and the emphasis is on investigating swarm intelligence search algorithms to enhance the efficiency and reliability of cloud computing task scheduling systems. Through research and practical application, the ant colony algorithm, gray wolf algorithm, whale fish optimization algorithm, and Beluga whale optimization algorithm are being studied and implemented， and designing task and virtual machine sorting mechanism, and conducting experimental comparison on Cloudsim platform, it is concluded that no need to sort task and virtual machine in advance and enhancing the performance of swarm intelligence algorithms can be achieved by incorporating a more judicious level of randomness. In addition, to solve the problem of uneven load and low early search efficiency of ant colony algorithm, this paper introduces a novel algorithm called the Immune-Ant Colony Optimization Algorithm (IMACO), and improved Beluga whale optimization algorithm (IBWO), which introduce load monitoring heurist, dynamic volatility and dual reward and penalty pheromone update strategy, and make full use of the global search and diversity advantages of immune algorithm. Effectively improve algorithm performance. The simulated annealing and quasi-reverse learning strategies are combined to solve the local optimal problem and achieve better completion time, system load balance and cost control. The outcomes indicate that these methods can enhance the efficiency of task scheduling. These researches provide practical value for improving the performance and stability of cloud computing system.