Abstract: The controller models employed for various motors also differ, with motor parameters exhibiting diversity. Under multiple operating conditions, parameters may change during operation and load disturbances may occur, leading to issues such as low motor parameter identification accuracy and motor-controller model mismatch. An improved recursive least squares (RLS）algorithm for diagnosing multi-motor parameter mismatch under diverse operating conditions was proposed. Traditional RLS methods suffer from slow tracking speeds and poor disturbance rejection during online parameter identification due to fixed forgetting factors. A variable forgetting factor that adapts to changing operating conditions were introduced, which enhances tracking speed and load disturbance rejection. To validate the reliability, robustness, and generalization capability of the improved RLS, five hypothetical operating conditions were established and multiple experimental comparisons were conducted. Analysis of motor speed response, d-q axis currents, and quantitative assessment of torque tracking and R_s parameter identification accuracy confirmed the enhanced algorithm's robust performance and broad applicability. Furthermore, analysis of performance metric data, primarily average speed and average q-axis current, confirms the validity of the improved algorithm's analytical results.