Intelligent prediction method for electromagnet quality based on optimized long short-term memory network

In the realm of automation control, the electromagnet stands as a crucial element. With the continuous expansion of market demand and production scale, the quality control criteria for electromagnets have been steadily enhanced. This article puts forward a data-driven intelligent prediction and control approach for electromagnet quality. Initially, a novel prediction algorithm is formulated by fine-tuning the key hyperparameters of the long short-term memory (LSTM) neural network via the particle swarm optimization (PSO) algorithm. Subsequently, the proposed prediction method is implemented in an actual case concerning a round-tube electromagnet utilized in low-voltage circuit breakers. Thereafter, comparative experiments are carried out on the identical dataset to contrast this method with BP and LSTM. Ultimately, the outcomes demonstrate that the validated method attains a higher level of accuracy in forecasting electromagnet quality.