Research on weight prediction method based on IJSOA-BPNN for cigarette filter rod

The weight of filter rods is an important indicator for monitoring the production quality of cigarette filter rods. This paper proposes a prediction method for filter rod weight, which is based on IJSOA-BPNN. The improved jumping spider optimization algorithm (IJSOA) is used to adjust the threshold value of the BP neural network to compensate for its susceptibility to local optimum. The trained BP is applied to the prediction model to predict the weight of the next moment of the filter rod forming machine. The results show that the proposed algorithm achieves the specifications including RMSE of 1.6159, MAPE of 0.0018, and R
²
of 93.35%, which are superior to BPNN and JSO-BPNN in terms of prediction accuracy and stability.