The fault signals of rolling bearings in mechanical systems are usually nonlinear, non-stationary, and accompanied by strong background noise, which makes it difficult to realize early warning of bearing fault. An intelligent early-warning method for rolling bearings based on beta distribution and filter denoising algorithm is proposed in this article. First, the early warning threshold value interval of the monitoring data is calculated by the threshold determination method based on the beta distribution. Then, the average filtering algorithm is used to reduce the noise of the collected data to eliminate the data monitoring noise. Meanwhile, the noise reduction effects of moving average filtering, H-P filtering, and morphological filtering are compared and analyzed. Finally, the calculated warning threshold interval is compared with the filtered data, and the early warning is made according to whether the monitoring data exceed the threshold interval. The XJTU-SY dataset and bearing experimental data are used to evaluate the accuracy of the algorithm. The results show that the proposed method can accurately calculate the early warning threshold interval of rolling bearings in smooth operation and effectively warn the bearings of early failures. The fastest early warning response time is 56. 76 s and the slowest early warning response time is 778. 20 s. Furthermore, the comparative analysis results show that the effect of moving average filtering is better than those of H-P filtering and morphological filtering when filtering the original data.