Artificial Neural Network-Based Optimization of Shielding Cylinder

Time-varying magnetic fields with a wide frequency band and large variations exist in the ground environment at all times, including the geomagnetic field with an amplitude of about S0μT and the magnetic fields generated by human activities. While these magnetic fields protect humans and bring convenience to life, they also drown out some weak magnetic signals that are rich in content, such as neural currents and other biomagnetic fields with high research significance. These signals are of high scientific research value, and the design of magnetic field shielding devices is a very important aspect in order to study these signals. In the design of shielding devices, in addition to the size limitation requirements of the relevant area, there are also different optimization requirements for the selection of shielding materials, quality, structural performance and cost of the shielding solution, which is a typical multi-objective or super multi-objective optimisation problem. The main objective of this paper is to investigate the optimisation of shielding cylinders (SC) based on artificial neural networks. A cylindrical shape is chosen as the final shape of the MSC. This paper analyses the material selection of the MSD and the shape of the MSC, and investigates the shielding coefficient of the multi-layer MSC; the analysis of the shielding effectiveness of different shapes of the SC shows that the rectangular body is slightly better than the cylindrical shape. However, the rectangular shape takes up more space, has more consumables and is costlier to manufacture. Therefore, the cylindrical shape was chosen as the final shape of the MSC. A three-layer MSC with the optimised dimensions in Table 4 was designed and simulated, and the centre point shielding coefficient of the three-layer MSC was 144.085dB with a remanence of less than 0.01nT, meeting the design requirements.