Assembly line ADC calibration hardware system based on GABP neural network and implementation method

By designing a pipeline ADC calibration system based on GABP neural network, the shortcomings in calibration effects and hardware implementation of traditional BP neural network algorithms are solved, and efficient calibration and low-power hardware implementation are achieved.

CN119696582APending Publication Date: 2025-03-25NANJING MODULUS ZHIXIN MICROELECTRONICS TECH CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202411757739.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-03
Publication Date
2025-03-25

AI Technical Summary

Technical Problem

In pipeline ADC calibration, traditional BP neural network algorithms have problems such as slow network convergence speed, easy to fall into local minimum values ​​and poor calibration results, and the resource consumption and power consumption are large when hardware is implemented.

Method used

A pipeline ADC calibration system based on GABP neural network is designed, and the calibration function of hardware circuits is realized through data preprocessing module, parameter storage module, GABP neural network calibration module and calibration data output module. The system adopts partial parallel structure and quantization processing to reduce hardware resource consumption and power consumption.

Benefits of technology

It improves the neural network calibration effect, reduces the number of network training times, reduces hardware resource consumption and power consumption, and realizes real-time calibration function.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention requests to protect hardware circuit implementation of an assembly line ADC calibration algorithm based on a GABP neural network. The system is mainly composed of a data preprocessing module, a parameter storage module, a GABP neural network calibration module and a calibration data output module. Wherein the data preprocessing module carries out data normalization and cache sorting operation on input to-be-calibrated data, the parameter storage module is used for storing neural network parameters which are trained and quantized, and the GABP neural network calibration module carries out operation on the preprocessed to-be-calibrated data and the parameters in the parameter storage module. And the calibration data output module processes and outputs finally calibrated data. After the design is completed, a verilog code is deployed on an ALINX AX7020 development board, calibrated data are derived and output is subjected to fast Fourier transform (FFT) to obtain a spectrogram and performance indexes, and calibration performance evaluation is carried out.
Need to check novelty before this filing date? Find Prior Art