Injection molded direct joining is a promising technique to fabricate metal-plastic direct joints. The successful joining is due to the infiltration of melted plastic into surface structures of metal plates. Thus, the injection parameters, which control the plastic flow, have great influence on the joining strength. It is desirable to test every combination of injection parameters to get a full glimpse of their influence on the joining strength. However, this desire would require several hundreds of experiments, which is practically difficult. Here, we tried to achieve this desire by using Taguchi method and machine learning. Metal-plastic joints of a 30% glass fiber reinforced polybutylene terephtalate (PBT) and hot water treated aluminum alloy (A5052) plates are used. At first, experiments were designed with Taguchi method and performed to prepare data for machine learning. Then, the data were used to train a back propagation neural network model. The model was then used to predict the joining strength of every combination of injection parameters. The results show that injection parameters have strong interaction with each other. A condition of high packing pressure and low injection speed will produce higher joining strength.
