Container deployment system and method for industrial gateway machine learning model based on pmml

Abstract

The PMML-based industrial gateway machine learning model container deployment system and method of the present invention export the cloud-trained machine learning model into PMML format, and send it to the industrial gateway through a message, and the industrial gateway is based on the purpose and binding of the machine learning model The machine learning model control object is dynamically created at the equipment measurement point, and the predictive analysis process of the machine learning model is controlled. Each device data collection of the industrial gateway triggers the predictive analysis of the corresponding machine learning model for early warning. Once the abnormality of the device is deduced, the early warning information is pushed to the industrial site and the cloud, and the machine used for fault diagnosis under the same measurement point is triggered Learn the model, and then push the reasoned fault diagnosis results to the industrial site and the cloud. Compared with the state monitoring mode of predictive analysis of data in the cloud, the predictive analysis of data in the gateway is realized, which reduces the bandwidth pressure caused by uploading a large amount of data and the transmission delay of information in the cloud and the industrial site, and ensures the real-time nature of the information.

Description

technical field [0001] The invention belongs to the technical field of equipment state monitoring, and in particular relates to a PMML (predictive model markup language)-based container-type deployment method of an industrial gateway machine learning model. Background technique [0002] Equipment status monitoring technology is an important technology in the current industrial digital transformation. The status data of the equipment is collected through various sensors on the industrial site, and then the data is forwarded to the cloud after protocol conversion through the industrial gateway. The cloud uses big data technology and Machine learning technology analyzes and predicts the data to judge the health status of the equipment. However, this cloud-centric condition monitoring has the following problems. [0003] 1. The data collected by sensors in the industrial field has a large amount of high-frequency waveform data, so the data volume is very large. Uploading it to ...

AI Technical Summary

Problems solved by technology

[0003] 1. The data collected by sensors in the industrial site has a large amount of high-frequency waveform data, so the data volume is very large. Uploading it to the cloud through the gateway requires a large amount of bandwidth, which will cause huge pressure on bandwidth and a large amount of bandwidth cost.

[0004] 2. Some data collected on the industrial site involves the company's process secrets and cannot be uploaded to the cloud for data analysis and prediction

[0005] 3. The process of uploading the data of the industrial site to the cloud and the process of pushing the analysis results to the site from the cloud will cause a certain delay, which cannot meet the condition monitoring application with high real-time requirements

Images