Method and system for processing abnormal work orders

Abstract

The embodiment of the invention provides a method for processing abnormal work orders, and the abnormal work orders are other work orders except a work order with a qualified quality inspection conclusion. The method for processing the abnormal work orders comprises the steps: extracting feature attributes from the abnormal work orders through a deep neural network model, the feature attributes have m dimensions, and m is larger than 200; converting the characteristic attributes into a state matrix; inputting the state matrix into an abnormal work order decision model; wherein the abnormal work order decision-making model is obtained by training through a deep reinforcement learning algorithm by taking an information record of an abnormal work order subjected to manual processing as sampledata, the input quantity of the abnormal work order decision-making model is the state matrix, and the output quantity of the abnormal work order decision-making model is a weight vector representinga processing scheme; and calculating a processing scheme for the abnormal work order through the abnormal work order decision model.

Description

technical field [0001] The embodiments of the present invention relate to the field of computer technology, and in particular to a method and system for processing abnormal work orders. Background technique [0002] The traditional way of analyzing work order quality inspection results is to manually review unqualified and unrecognizable work order pictures, and summarize and feed back the same type of problematic work orders to the installation and maintenance personnel, and guide them to make improvements, or feed back to the quality inspection system development people and let them optimize the system. However, the cost of manual review is high, and it is required to be familiar with the project and understand the algorithm model. After manual review, a lot of work is required to summarize and give feedback. Therefore, technicians thought of using machines to complete the review process of work orders. [0003] One existing technology is to use a computer to extract the ...

AI Technical Summary

Problems solved by technology

In this way, the computer can automatically correspond to the processing plan according to the extracted labels; however, the artificially set one-to-one matching rules have poor generalization ability, and do not have the ability to learn new feature labels.

[0004] In addition, since the feature information in the work order picture is high-dimensional, the existing reinforcement learning model will perform dimensionality reduction processing on high-dimensional feature attributes, or artificially set low-dimensional features to represent the feature labels of the same attribute. Feature attributes. Although this method reduces the complexity of calculation, it is difficult for low-dimensional features to represent the real feature attributes of work order pictures, which will affect the accuracy of the processing plan.

Images