The aims is to solve the problem of difficult parameter selection caused by the coupling of plasma spraying process parameters to improve the mechanical properties of the AlCoCrNiFe high entropy alloy coating. An algorithm which seagull optimization algorithm based on global chaotic and Gaussian fusion (CGSOA) is proposed to optimize the weights and thresholds so that BP (Back Propagation) neural network training outputs ideal control parameters. The improved logistic chaotic sequence is used to realize the global search of the initial population of network parameters, and the initial quality of weights and thresholds is improved; the improved logistic mapping is introduced to jump out of the local optimum, and the local search capability is strengthened to improve the accuracy of the algorithm convergence; the introduction of Gaussian mutation increases the diversity of the population, improve the global search capability; select 6 benchmark functions to test the BAS, PSO, ACO, SOA and CGSOA algorithms. The simulation results show that the proposed algorithm had faster convergence speed, higher optimization accuracy and stability. CGSOA algorithm optimizes the BP neural network to obtain the best control amount:spraying distance 99.7 mm, spraying current 649.6 A, spraying voltage 56.3 V, powder feeding carrier gas 203.1 L/h, powder feeding voltage 5.1 V. The spraying test with this parameter shows that the bonding strength and microhardness of the coating were 25.2 MPa and 616.8HV, respectively, and the relative errors with the predicted value of the model were 3.02% and 2.91%, respectively. This result verifies the feasibility of applying the CGSOA algorithm to actual projects. CGSOA-BPNN has a certain guiding significance for optimizing the plasma spraying process parameters of AlCoCrNiFe high-entropy alloy coating, thereby improving the coating performance.