In order to reduce the fluctuation range of output speed and torque of permanent magnet synchronous motor at high speed, this paper proposes an improved flux-weakening control method of single current regulator of permanent magnet synchronous motor. Based on chaotic mapping and Monarch Butterfly Optimization, an adaptive chaos monarch butterfly optimization is proposed to use for online parameter identification. This algorithm generates a chaotic sequence distributed according to a specific pattern through a single chaotic mapping, and combines it with the results of the previous parameter identification to generate an initialized population. Secondly, the impact mechanism of different load curve positions on the smoothness and efficiency of motor operation was analyzed, and the influence of motor parameters on voltage limited elliptical distortion was analyzed. Finally, a method of giving the cross-axis voltage considering motor parameters is designed. The simulation and experimental results show that the parameters of the motor can be accurately identified online based on ACMBO. When the load is 85 N･m, the range of motor torque output error is effectively reduced.