To address the problems of severe low-speed disturbances, low control accuracy, and poor reliability of traditional sensors in brushed DC motors, this paper designs a high-precision disturbance compensation and intelligent switching-type non-invasive detection control system. The system employs a combination of friction feedforward, secondorder Extended State Observer (ESO), and adaptive sliding mode control to suppress multi-source disturbances. Combined with the least squares-adaptive Kalman filtering composite algorithm and the intelligent switching of operating conditions, it achieves non-invasive speed measurement. Experiments were conducted on the dSPACE platform, and the results show that the steady-state error of this system is 46% of that of traditional PI control and 63% of that of traditional sliding mode control. The response time is ≤ 0.15 s. Under stable operating conditions, the non-invasive detection does not exceed 1.2%, and the dynamic accuracy is on par with full-order Kalman filtering. The computational resource consumption is reduced by 82.7%, enabling full-condition sensorless high-precision starting and speed regulation.