Abstract: Abstract: There are a variety of disturbances wide existing in the agricultural devices, including uncertain parameters, input voltage variations and load changes and these perturbations heavily affect the output voltage of the Buck converter. The traditional proportional integral (PI) control methods are difficult to obtain satisfactory performance. To solve this problem, many nonlinear theories have been applied to the Buck converter for obtaining better control performance. However, the implementation of these nonlinear controllers, such as adaptive controller and fuzzy-neural controller, generally requires complex hardware circuit or advanced processor, which increases the cost of the production and reduces the reliability of the control circuit. In this paper, based on the disturbance observation theory, a new control method for disturbance attenuation is proposed. First of all, the conventional PI controller is replaced with the variable-parameter PI (VAPI) controller, which not only possesses the easy-implementation characteristic the conventional PI holds, but also guarantees that the closed loop system will have a better performance in different control stages by online tuning the PI parameters. Then, a linear disturbance observer (DOB) is constructed to estimate the exact value of the disturbance which is caused by input voltage variations and load changes. The disturbance is estimated through the information of output voltage and the structure of control system, and the estimated value is used to compensate the disturbance in forward channel, which will significantly improve the convergence and disturbance rejection property of the closed loop system. With the above improvements, a new composite controller can thus be obtained. The composite controller in the paper is implemented by using a semi-physical experimental platform. The platform is mainly based on a data acquisition card and a buck circuit. When the system of the Buck converter is running, the output voltage can be collected by the data acquisition card. By calculating these voltage data, the control output signal can be obtained directly. Later, the data acquisition card will transmit the obtained control signal to the buck converter circuit. The advantages of this semi-physical experimental platform are making the implementation of the algorithm more flexibly and the modification of controller parameters more conveniently. Based on the experiment platform, the implementation of the composite controller can be summarized as three steps. Firstly, the continuous composite controller is constructed in frequency domain. Secondly, the continuous composite control strategy is discretized so that the data can be processed by computer. Finally, the discrete controller is implemented directly by using the LabVIEW program. Experimental results obtained from the semi-physical experiment platform show that after replacing PI controller with the composite controller, when the load varies, the recovery time of the Buck converter can be shortened by 71.4% and the output voltage error can be reduced by 20.8%. When the input voltage varies, the recovery time can be shortened by 58.3% and the output voltage error can be reduced by 30.0%. According to these comparative results, it is obvious that the proposed new composite controller can effectively improve the convergence and anti-disturbance performance of the Buck converter. Meanwhile, considering the property of easy-implementation, the composite control method has potential applications in engineering area.