In response to the complex and variable real-world operating conditions of electric vehicles and the technical prololems of traditional motor systems relying on calibration based on single standard operating conditions leading to low efficiency across the entire operating range, this paper proposes an adaptive optimization method integrating real-time operating condition perception, multi-parameter collaborative control, and thermal management linkage. This method collects multi-dimensional operational data such as motor speed, torque, coolant temperature/flow rate, constructs a three-dimensional feature space (speed-torque-thermal state) with dynamic partitioning, and collaboratively optimizes power device driving parameters (gate resistance, driving voltage), Pulse Width Modulation Switch (PWM) switching frequency, and cooling system operation strategy based on partition characteristics. Compared to the traditional fixed redundancy design for the most severe operating conditions, the proposed method achieves precise loss control across the entire operating range through "hardware-software collaboration+thermal-electric coupling" while satisfying system safety and reliability requirements. Bench test verification shows that this method improves the motor system's CLTC comprehensive cycle efficiency by approximately 0.34~0.67 percentage points, effectively reducing the vehicle's electricity consumption per 100 km and significantly extending its driving range.