Optimal Design Method of Filter Based on Deep Learning Algorithm

Abstract

The invention belongs to the technical field of filters, and specifically relates to a filter optimization design method based on a deep learning algorithm. The optimal design method of the present invention is carried out for the structural parameters of the filter, and the structural parameters of the filter are reflected by the filter response curve of the filter; in the design, reverse neural network, forward neural network and genetic algorithm are used to carry out deep learning: filter The filter response curve of is obtained by synthesis of Chebyshev polynomials; the target filter response curve is used as the input of the reverse neural network to obtain the initial value of the structural parameters; the initial value is input to the genetic algorithm and the forward neural network for iterative optimization; the optimization target The difference between the filter response curve output for the forward neural network and the basis filter response curve is the smallest, and finally the optimized filter response curve is output, and the structural parameters of the final filter are obtained.

Description

technical field [0001] The invention belongs to the technical field of filters, and in particular relates to a design method of filters. Background technique [0002] In modern radio frequency transceiver systems, filters are often required at the back end of the antenna to control the transmission and reception of electromagnetic waves in the required frequency band. Today, a large number of portable devices, handheld devices and wearable devices are appearing, especially with the concept of 5G "Internet of Everything", a variety of terminal devices will inevitably appear in large numbers. environment, the role of filters is becoming more and more important. The traditional filter design often needs to select the corresponding filter function according to the performance index, use the circuit theory to construct the corresponding circuit model, and then use the simulation software to extract the coupling coefficient, establish the connection between the actual model and t...

AI Technical Summary

Problems solved by technology

The existing design based on evolutionary algorithms such as particle swarm optimization algorithm has a strong dependence on the initial value, and needs to use simulation software continuously during the iterative process, which is time-consuming, while the trained neural network can be completed in a few seconds The design of primary filters has obvious advantages compared with traditional design methods and optimization algorithms

Images