In the manufacturing process of sealed electronic equipment, it is very important to detect and locate loose particles. Aiming at the problem of the large size of the equipment and the difficulty of determining the location of loose particles, parameter optimization Support Vector machines is used to locate the loose particle inside equipment. By designing a signal conditioning circuit and a multichannel signal synchronization acquisition circuit, the weak loose particle signal is processed and collected. By designing a two-stage dual-threshold pulse extraction algorithm and a multi-channel pulse matching algorithm, the signals are preprocessed to obtain effective signal data. By extracting and selecting the time domain and frequency domain features with excellent performance, to construct a locating data set. Comparing the performance of different classification algorithms on the data set, optimizing the inherent parameters of better-performed support vector machine. And finally using the optimized support vector machine locating model for physical testing. The test results show that the optimized support vector machine locating model has an average accuracy of 82. 58% in the loose particle locating test inside the aerospace power supply. The generalization ability of the locating model is good and meets the accuracy requirements of aerospace system engineering. Theoretically, this method can be extended to the research on the location of collision signals with similar generation mechanism.