The scope of this research is to develop an adaptive control system of a DC electric drive implemented on the basis of industrial DС converter Siemens Sinamics DCM. A neural tuner adjusting speed PI-controller parameters is chosen as an adaptation method. Contrary to classical methods of adaptive control, this tuner does not require an accurate nonlinear model of the drive. Instead of this, it evaluates transients quality in a speed loop and, if it does not follow the requirements, adjusts the corresponding controller parameters. This assessment is made by a developed rule base, which calculates the value of a learning rate for the neural network online training. The network output is the controller parameters. The experiments with a real DC motor are conducted under the following conditions. The DC motor inertia moment is changed by 50% from its nominal value. As a result, the neural tuner adjusts the parameters of the speed PI-controller and achieves the required quality of transients. At the same time, the overshoot obtained with the help of the classic PI-controller is 11% higher than required.