MES system based on microservice architecture: The system takes an automatic bar processing workshop as the basic object, and adopts the development method of separating the front and back end for system development. The front end relies on technology stacks such as Vue.js, Elementui, Echarts to realize the front-end data interaction and page design, and the back end relies on components such as Spring Cloud Alibaba, Spring Cloud Gateway, Nacos, RPC open source framework, Sentinel to realize the construction of micro-service framework. Moreover, it relies on Spring, Spring Boot, Mybatis, YOLOv3 target detection algorithm to complete the realization of basic functions. Finally, the real-time perception of hardware device data is realized by dynamic link library, Redis cache database and WebSocket communication protocol. Each layer of services is isolated from each other, and the microservices in the layer also follow the design principle of high cohesion and low coupling to minimize the coupling degree of the system, improve its scalability and versatility, and make the system have better reliability, faster response time and appropriate throughput.

Behavior recognition detection system: The behavior recognition detection system based on machine vision can be applied to tooling detection, personnel detection in restricted areas, mobile phone detection, access cover detection and other fields. The detection process of recognition system is divided into four stages. Firstly, data is obtained, and multi-channel camera data is obtained through an RTSP protocol. Secondly, data caching, creating a storage container, establishing a parallel processing mechanism, storing the data obtained by the multi-channel camera into it and taking out the data for further processing at the same time. Then, the data is input into the pre-training model and the result is output, and the result is judged and processed to judge whether an alarm is needed. If not, an image is directly output, and if necessary, the fourth stage process is performed. Finally, the character and picture information to be alarmed are respectively transmitted to the target server through the MQTT protocol and the POST request, and the image to be alarmed is then post-processed and displayed

Amphibious robots: Amphibious robots are robots designed to inspect, clean, and maintain sewer systems. They use high-resolution cameras and sensors to detect pipe damage, blockage and corrosion in real time. The robots can clean the deposits and dirt in the pipeline, and remove the blockage to restore the smoothness of the pipeline. In addition, they can also perform pipeline maintenance, such as repairing damaged parts and replacing aging parts. The robots are capable of autonomous navigation, using various navigation technologies for internal exploration and movement. They record and store data from pipeline inspections and maintenance, and generate detailed reports to help operators make decisions and plans. Amphibious robots increase work efficiency, reduce manpower requirements, and reduce safety risks, ensuring the normal operation and reliability of pipeline systems.