Abstract: The spot welding quality between copper wire and phosphor-copper sheet in relay manufacturing was monitored by using artificial neural networks(ANN)technology.Based on BP(backpropagation) ANN model and algorithm, the model took welding voltage and current as input parameters, shear load of welded joint as output parameter. Models with different hidden layer node number and transfer function were simulated in M atlab so as to select appropriate hidden layer node number and transfer function.Finally the model was tested on inverted resistance spot welder.The results show that the quality monitor model' s precision by ANN estimation method is able to satisfy engineering application demand.