On the morning of July 4, 2022, the Guilin Science and Technology plan project "Online assessment and intelligent monitoring system for purification and sterilization environment of Food Production and deep Processing" undertaken by Xiaocao successfully passed the review of the expert team organized by the Guilin Science and Technology Bureau, and the project was successfully completed and accepted.

The project relies on cloud big data for decision analysis in software, takes artificial intelligence algorithm and expert system idea as the core for overall design, uses ARM architecture to integrate artificial intelligence algorithm in hardware, and integrates intelligent chip of grass clean environment intelligent expert system in system. It can complete self-cooling, strong cooling, waiting and inoculation environment (XC-CO-CE-2101-Z401) after wet heat sterilization of edible fungi. Bacteria storage, cultivation, fermentation environment (XC-CO-CE-2101-Z402), product deep processing clean environment (XC-CO-CE-2101-Z403) for real-time data acquisition and intelligent control. According to the latest standard GB25915-2022 of "Clean Room and Related Controlled Environment" issued by the National Standards Committee (officially implemented from March 1, 2022), the system carries out normative design of the number of clean room sampling points, area, particle sampling time, sampling amount and sampling sensor. The current environmental parameters of the product can be sampled are: National standard PM1.0, National standard PM2.5, National standard PM10, US standard PM1.0, US standard PM2.5, US standard PM10, Total suspended particulate matter (TSP) μm0.3, total suspended particulate matter (TSP) μm0.5, total suspended particulate matter (TSP) μm1.0, total suspended particulate matter (TSP) μm2.5, Total suspended particulate matter (TSP) μm5, total suspended particulate matter (TSP) μm10, fresh air outlet, supply air outlet, return air outlet, exhaust outlet, wind speed, air volume, laminar Reynolds number Re=pvd/g, static pressure difference between clean area and non-clean area, pressure difference between clean area and clean area, ozone concentration, UV intensity and other data. The comprehensive scheduling and intelligent control of sterilization equipment are carried out by establishing mathematical models in different production and processing environments, and the effect is constantly evaluated in the actual regulation process. The results are evaluated by convolutional neural network algorithm (CNN) and reinforcement learning (RL) : reinforcement learning algorithm continuously optimizes the control model to achieve accurate, economical and efficient environmental control of purification and sterilization.

The project passed the evaluation of China Software Evaluation Center and obtained one national invention patent and four software copyright certificates. The project can fill the gap of online monitoring of the clean environment of edible fungi. Compared with the traditional way of clean environment detection, it has the outstanding advantages of real-time, rapid and accurate, and plays a good role in preventing the infection caused by the unqualified clean environment in the production and processing of edible fungi.