Popis: |
In the process of studying the interconnection reliability of electronic products, due to the small size of the solder joints, the solder joint stress value of the chip cannot be directly measured by conventional methods, so this paper proposes a method based on BP neural network to predict the maximum bending stress of BGA solder joints. First, the orthogonal analysis method is used to establish an orthogonal combination of three factors and two levels of solder ball diameter, solder joint height, and solder joint spacing. Then the ANSYS simulation software was used to establish a model of the orthogonal analysis results, and the stress at the center position of the chip in the X and Y directions and the maximum stress of the solder joint under the condition of applied load were measured. Finally, the BP neural network built by TensorFlow was trained and tested with the data obtained from ANSYS simulation, and the rationality of the method was verified according to the test results. |