Optimization of the radiation pattern of the array antenna

Array antennas play an important role in processing signals coming from multiple directions. Array antennas are used to create a beam of radiation that is focused at a specific angle, concentrating the power at that angle. Instead of using one large antenna, which is difficult to build and operate, we use multiple arrays, each with a specific power feed spacing and size, to achieve the desired radiation pattern. If we can achieve the same desired results in this type of antenna with a smaller number of elements, i.e., by removing some elements that are no longer the same distance apart and at half the wavelength, we can offer an antenna that is lighter and less expensive in array mode. This type of array antenna is called a scattered array antenna. In fact, in a scattered array antenna, we try to achieve the same result with a smaller number of elements, which results in lower cost, lower weight, etc. The radiation beam equation of the array has an equation that in this paper we want to optimize using reinforcement learning and specifically the deep learning model using the PPO algorithm, which has a learning advantage over other methods in optimization and can handle problems with stochastic transitions and rewards without requiring adaptations and also has fewer iterations and has a higher speed and which will provide an antenna with less number of elements and the desired signal performance.

کلیدواژه ها:

Reinforcement learning ، Q learning ، Beam shaping ، Array antenna ، Deep learning