Smart Factory: Thermal Image Recognition Systems
The latest development in MtM+’s Smart Factory solution is the incorporation of Thermal Image Recognition into the existing Data Collection Station (DCS). The main targets of this new technology are factories that store their chemicals in tanks. Whether the chemical is hazardous or not, the storage of these substances must be taken seriously.
Heat, is a particularly important matrix in terms of storing condition. Certain chemicals when heated up will release harmful by-products that are a risk to health. Unanticipated reaction of certain chemical might release a lot of heat and may result in explosions. This calls for improved real-time monitor systems that can control water spray systems or inform an evacuation when necessary. The following Figure-1 block diagram provides an overview of such systems.
Data Collection Station (DCS) is equipped with various wireless connectivity options and various I/O ports for smart factory implementation. See Specifications here (https://www.mtmtech.com.tw/DCSV1.html).
A DCS implemented with an IR camera generates real-time image data and transmit such data to the Cloud or the Local WAN through a LoRa network. LoRa is a digital wireless network that is capable for long range transmissions up to 10 km in rural areas. The DCS’s LoRa system is implemented using MtM+’s M908 module. See specification here (https://www.mtmtech.com.tw/M908.html). With this data securely on the Cloud or the Local WAN, it can be accessed by a Personal Computer (PC) through a Wifi Access Point (AP) for further analysis. With these data available on the PC, reactionary systems can be implemented based on the incoming data.
The IR camera selected for this project is the Radiometric Lepton 2.5 manufactured by FLIR Systems. It features a small size of 8.5x11.7x5.6mm, 80hx60v pixel resolution, 9Hz frame rate with I2C/SPI interfaces. The most important feature of the Lepton 2.5 is the radiometric capability compared to its predecessors. Lepton 2.5 can output a temperature value for each pixel in a frame irrespective of the camera temperature with an accuracy of +/-5˚C.
With the 14-bit pixel temperature value (in Kelvin) successfully collected into the PC, they are divided into different blocks for analysis, see Figure-2 image below.
This system has been implemented and tested with a Petrochemical company to provide a real-time monitor system for the safety of their chemical storage (see Figure-3 images below).
“We need your Feedback!”
MtM+’s smart factory solution on particle monitoring has been deployed in a semiconductor manufacturing company with great results. Big data generated with these PSN boxes are vital for analysis in improving factory flow, floor plan and process. The keys to success are strategic placement of each sensors along with MtM+’s calibration algorithm, and we will cover more about our calibration process and benchmark details in the next article.
article source: https://blog.mtmtech.com.tw/2018/12/17/smart-factory-thermal-image-recognition-systems/
