Subway inspection AGV fault detection method

Abstract

The invention provides a subway inspection AGV fault detection method comprising the following steps: 1) establishing a first deep neural network used for identifying camera information collected by asubway inspection AGV, extracting subway complete information, shielding other image noises, and establishing a corresponding activation function and a loss function; 2) establishing a second deep neural network for performing fault detection on the subway information obtained by the first neural network in the step 1); and 3) enabling the inspection robot to perform self-adaptive fault detectionthrough improved DDPG algorithm deep reinforcement learning. The method is a subway inspection AGV fault detection method based on a deep neural network and reinforcement learning. By means of supervised learning and unsupervised learning, the inspection robot can conduct self-adaptive learning on the collected subway fault types and severity degrees, the fault recognition rate in the subway inspection process is continuously increased, intelligent analysis is conducted on the fault severity degrees, and the accuracy and reliability of fault detection are greatly improved.

Description

technical field [0001] The invention relates to subway inspection technology, in particular to a subway inspection AGV fault detection method. Background technique [0002] In Chinese cities, there will be faults and maintenance problems in the long-term operation of the track lines of subway stations. Many of the existing problems require manual participation. Especially for the fault detection and maintenance of subway track lines, a large number of track maintenance personnel are required every day. When the subway is out of service, walking into the subway tunnel for maintenance is not only affected by the environment, but the dark environment leads to low work efficiency, and there are many defects such as personal safety hazards and high labor costs. Manual subway fault detection vehicles also have high costs and cannot meet the problems of complex environment detection, and it is difficult to effectively solve the safety of urban rail transit. The automatic subway in...

AI Technical Summary

Problems solved by technology

[0002] In Chinese cities, there will be faults and maintenance problems in the long-term operation of the track lines of subway stations. Many of the existing problems require manual participation. Especially for the fault detection and maintenance of subway track lines, a large number of track maintenance personnel are required every day. When the subway is out of service, walking into the subway tunnel for maintenance is not only affected by the environment, but the dark environment leads to low work efficiency, and there are many defects such as personal safety hazards and high labor costs.

Manual subway fault detection vehicles also have high cost and cannot meet the problems of complex environment detection, and it is difficult to effectively solve the safety of urban rail transit

Images