In order to realize  the highvalue utilization of straw biogas residue and livestock manure, the effect of cocomposting corn straw biogas residue and cow manure was studied, and the environmental benefits of mixed compost were analyzed by life cycle method. And the effects of different mixing ratios and composting time on the water content, pH value, total carbon, total nitrogen, volatile solid content and electrical conductivity of the pile were analyzed, the environmental benefits of cocomposting was also calculated. The results showed that the mixing ratio had a significant effect on the characteristics of the heap. With the ratio of cow dung and corn stover biogas residue increased from 2∶1 to 6∶1, the moisture content of the heap increased significantly, but when the ratio exceeded 4∶1, the effect of mixed composting on the moisture content of the pile gradually decreased. The straw biogas residue had a great influence on the pH value of the pile at the initial stage, and the change rule was not obvious, but when the fermentation time exceeded 40 days, the pH of the pile body showed a significant decreasing trend; increasing the proportion of cow dung helped to increase the pH value of the pile after composting. With the addition ratio of cow manure increasing from 2∶1 to 6∶1, the pH value of the reactor increased from 8.2 to 8.4, the total carbon content of the reactor increased from 35.34% to 36.81% after 60 days of decomposition, the volatile solids content decreased significantly from 46.96% to 3766%, and the electrical conductivity increased from 8.3ms/cm to 9.4ms/cm. At the same time, cocomposting of cow manure and biogas residues significantly reduced the emissions of CO2, CH4, N2O, NOX, NH3, SO2, PO43- and other pollutants in the compost substrate compared with direct emissions.

In order to improve the problems of low phosphorus and low potassium content in biogas residue and cow dung compost, biomass fly ash was added to compost heap, and the physical and chemical characteristics of compost heap were analyzed. The fly ash supplemental levels of 8%, 12%, 20% and 25% were selected, and the changes of water content, electrical conductivity, total carbon, total nitrogen, total phosphorus and total potassium were recorded every 10 days according to the quartering method, and the effects of fly ash supplemental levels on biogas residue and cow dung composting were analyzed and studied. The results showed that when the addition of fly ash increased from 8% to 25%, the moisture content of the compost increased, but when the addition amount exceeded 12%, it had little effect on the moisture content, which was maintained at about 58% in the end; the existence of alkali metals and other substances in the fly ash neutralized the acid produced in the aerobic fermentation process, increased the pH value of the heap, and also increased the conductivity of the compost improved the fermentation environment of the compost; the biochar and potassium in the fly ash increased the total carbon and total potassium content of the compost. After 60 days of decomposing, the total carbon content of the compost increased by 37.37% to 47.81%, the maximum value of total potassium content reached 35.74 g/kg, but the effect on total nitrogen and total phosphorus content was not obvious. However, excessive fly ash addition would inhibit the growth of microorganisms, thereby reducing the fermentation rate and product quality.

In order to obtain high accuracy of plant disease classification with only a few training samples, a few-shot learning model was used as the disease classifier, and five kinds of shallow networks, Conv4, Conv6, ResNet10, ResNet18, and ResNet34, were used as feature extraction networks under the framework of three typical few-shot learning algorithms, including MatchingNet, ProtoNet, and RelationNet. Their performances were compared on the plant disease data set of PlantVillage. Under the condition of 5way、1shot, the average accuracies of MatchingNet, ProtoNet, and RelationNet were 72.29%, 72.43%, and 69.45%, respectively. ProtoNet+ResNet34 was the optimal combinational mode, and the accuracy reached 77.60%. Under the condition of 5way、5shot, the average accuracies of MatchingNet, ProtoNet, and RelationNet were 87.11%, 87.50%, and 82.92%, respectively. The accuracies were significantly improved compared to that of the 1shot condition. ProtoNet+ResNet34 was still the optimal one with an accuracy of 89.66%. The above test results show that by optimizing the combination of a few-shot learning framework and feature extraction network, the recognition model can achieve good effects for diseases with a small number of samples.