Deep learning dynamic model clipping reasoning system and method based on FPGA

Abstract

The invention discloses a deep learning dynamic model clipping reasoning system and method based on an FPGA, and belongs to the technical field of the Internet of Things, the system comprises an FPGA cloud, a distributed base station, an FPGA Internet of Things terminal, a UPS power supply, a fire-fighting power supply and a commercial power supply, the distributed base station and the FPGA cloud respectively comprise a PLC development board A and a PLC development board B, the PLC development board A is in communication connection with the FPGA Internet of Things terminal through the downstream wireless transceiver module A. The signal output end of the PLC development board A is connected with an upstream wireless transceiver module. According to the deep learning dynamic model clipping inference system and method based on the FPGA, the intermediate transmission mechanism is arranged, the data processing pressure of the FPGA cloud end is reduced, the FPGA cloud end and the intermediate transmission mechanism can conveniently encrypt data, the safety effect is good, the FPGA cloud end and the intermediate transmission mechanism are provided with multiple sets of switchable power sources, uninterrupted work can be ensured, the reliability during operation is good, and the FPGA cloud has local and network alarm functions.

Description

technical field [0001] The invention belongs to the technical field of the Internet of Things, and in particular relates to an FPGA-based deep learning dynamic model tailoring reasoning system and method. Background technique [0002] In the Internet of Things, with the continuous development of technology, the hardware configuration of FPGA IoT terminals is getting better and better. In order to continuously improve and optimize the experience of using FPGA IoT terminals, it is necessary to transmit data to the FPGA cloud and perform deep machine learning on the data Finally, the optimized data is transmitted to the FPGA IoT terminal, and the software and algorithm of the FPGA IoT terminal are updated to realize the optimization of the FPGA IoT terminal. [0003] However, the existing inference system does not have an intermediate transmission mechanism, and directly transmits the data of the FPGA IoT terminal to the same FPGA cloud, which increases the data processing pres...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide an FPGA-based deep learning dynamic model tailoring reasoning system and method, to solve the problem that the existing reasoning system proposed in the background technology does not have an intermediate transmission mechanism, the data processing pressure on the FPGA cloud is high, and the FPGA The cloud and the intermediate transmission mechanism cannot easily encrypt data, and the security effect is poor. The FPGA cloud and the intermediate transmission mechanism do not have multiple sets of switchable power supplies, which cannot ensure uninterrupted work, and the reliability is not good. The FPGA cloud does not have local and network alarms. function problem

Images