Abstract: Abstract: In order to improve the processing quality and reliability of the cleaning screen of grain combine harvester, it is necessary to eliminate the manufacturing defects in the design and pilot stage, which causing the additional load, , and to locate and eliminate the processing defects by measuring in the production stage. In this paper, a method can locate and calculate the exciting forces act on the connection points of cleaning screen and thresher body by classical transfer path analysis (TPA) were developed. By measuring and comparing the vibrations of the connection points, it is found that the vibration frequency components were the same and were strongly correlated, so it is impossible to find the main excitation source by analyzing the frequency to locate the manufacturing defects. According to the relationship between exciting force and defect, it is pointed out that there should be major manufacturing defects near the maximum excitation force. The vibration of connection points and a certain observation point of the test bench were measured by using triaxial accelerometers and dynamic signal analyzer , and the characteristics of time domain and time-frequency domain were analyzed. The results showed that although there was no change in the frequency components in the spectrum, the amplitude changed greatly, and the phase changed greatly at different times of each frequencies. The transfer functions of the connection points to each measurement points were detected by using a modal force hammer and vibration measuring devices. Since the phase was unstable, according to the principle that the product of each excitation force and the transfer function (ie, the contribution of the excitation force to the vibration) had the largest vibration response in the same direction, the maximum acceleration amplitude was introduced into the inverse matrix method formula, and the influence of the phase was ignored, and the problem that the ill-conditioned matrix of the transfer function matrix could not be inverted was solved by increasing the measurement point, the generalized inverse matrix of transfer function was calculated by singular value decomposition. The calculated excitation force was optimized by least squares method, and finally the practical incentive calculation formula was derived. In order to verify the validity of the method, the magnitude of the excitation force during normal operation and the contribution to the vibration in Ydirection of the observation point were calculated. The results showed that the sum of contributions was only slightly smaller than the measured acceleration, which was about 84.7%-94.6% of the measured value, the excitation force calculated by this method was basically correct. The keyway clearance was used as a typical defect for the location verification test. It was found that the amplitude of each frequency in the excitation force spectrum near the defect increased significantly (71%-3 271%), while the amplitude of the slot excitation force away from the defect was only slightly added, the effectiveness of the positioning method was verified. The defect localization method proposed in this paper only adds one accelerometer and one measuring instrument to the original cleaning screening assembly quality inspection platform, the vibration excitation force at the connection points between the cleaning screen and the rack mounting could also be roughly calculated by measuring 4-8 acceleration responses.