云南省水利水电勘测设计研究院

To meet the intelligent control requirements for curtain grouting in limestone dam foundations,this study proposes a dynamic prediction method that integrates Grouting Energy Intensity(GEI)with machine learning,aiming to enhance process precision and adaptability in complex karst strata.Utilizing field data from the Dehou Reservoir and the Dongzhuang Water Control Project,a grouting database was established,incorporating key parameters such as grouting pressure(P),flow rate(Q),and Lugeon value(q).A P-Q time-series prediction model was developed using machine learning algorithms to forecast future grouting parameters dynamically.The results indicate that the incorporation of GEI significantly enhances the model's coefficient of determination(R2).Additionally,a four-level rock mass permeability classification system based on Lugeon values and unit cement consumption was constructed,facilitating intelligent parameter matching and adaptive process control.This research offers a practical technical framework for intelligent construction and quality assurance in curtain grouting within karst regions.

The efficient and accurate acquisition of structural plane information in rock masses is critical for classifying tunnel surrounding rock.Due to the fragmented nature of sandy dolomite rock masses,traditional geological mapping methods face challenges in precisely obtaining structural plane data.This study focuses on a typical sandy dolomite tunnel from the Water Diversion Project in Central Yunnan.By employing three-dimensional laser scanning technology,the geometric parameters of structural planes are accurately characterized,and these parameters are subsequently applied to the engineering geological evaluation of the surrounding rock.The results indicate that:(1)Utilizing 3D laser point cloud data,key geometric parameters such as the occurrence of structural planes,the dominant structural plane group,structural plane spacing,roughness,and ductility of dolomite with varying degrees of sandification are quantitatively characterized.The error ranges for dip direction,dip angle,and joint spacing are within±7°,±5°,and±6 cm,respectively.(2)The structural plane spacing derived from point cloud data enables the indirect calculation of the rock mass integrity coefficient(Kv)and the RQD index for classifying dolomite sandification grades.This effectively addresses the quantitative characterization of the rock mass integrity index Kv and RQD value.(3)Based on the established classification system for sandy dolomite tunnel surrounding rock,the rock mass integrity index Kv,RQD value,and structural plane roughness score index are quantitatively refined by correlating them with the volume joint number(Jv).Additionally,the structural plane extension length score is enhanced through a continuous scoring method,resulting in more accurate surrounding rock classification outcomes.The findings of this research provide a novel approach for the detailed evaluation of the engineering geological characteristics of surrounding rock in sandy dolomite tunnels and contribute to the foundational theory of disaster prevention and control in sandy dolomite underground engineering.

Aiming at the problem of determining the drought-limited water level of emergency drought-resistant backup reservoirs,a model for determining the drought-limited water level was established based on the principle of wide and shallow damage with the goal of minimizing the time period average water shortage rate and the number of time periods with severe water shortage.A method and operational rules for determining the drought-limited water level of emergency drought-resistant backup reservoirs under extreme drought conditions were proposed.The proposed method was applied to the Deze Reservoir,a backup water source for Kunming City,and the degree of water supply guarantee under different schemes.The results indicate that the method can realize the optimal design of phased drought-limited water level in emergency drought-resistant backup reservoirs,and effectively reduce the adverse impact of extreme drought on socio-economic water use.By setting the drought-limited water level of the Deze Reservoir,the number of months with severe water shortage for emergency water supply in Kunming City can be reduced from 229 to 34,and the average monthly water shortage rate can be reduced from 27.58％to 12.74％,thereby mitigating the occurrence of extreme water shortages.Under extreme drought conditions,the combined use of the drought-limited water level and dead storage capacity can further significantly enhance the level of water supply guarantee,reducing the number of months with severe water shortages to 21 and the average monthly water shortage rate to 2.52％.

Soil microbes at the rhizosphere are one of the key factors in maintaining soil health.However,it is unclear on the effect of different water-biochar coupling on the microbial environment of soil at the rhizosphere.Particularly,asparagus is one type of perennial herbaceous plant.There is a seriously continuous cropping disorder in asparagus,with the growth of years,leading to weak growth,low yield,and quality.The crop failure can be attributed to the soil's physicochemical properties and the unbalance of the rhizosphere soil microbial community.This study aims to explore the effects of water-biochar coupling on the structure of soil microbial communities in the asparagus rhizosphere.The mechanisms were also given to regulate the rhizosphere ecosystem.The field positioning experiment of asparagus planting was carried out on four irrigation levels(W0:severe deficit irrigation,I=320 mm;W1:middle deficit irrigation,I=375 mm;W2:mild deficit irrigation,I=430 mm;W3:adequate irrigation,I=480 mm,I represents the amount of irrigation)and four biochar application levels(0 t/hm2(B0),10 t/hm2(B10),20 t/hm2(B20),and 30 t/hm2(B30)).Illumina next-generation sequencing(NGS)technology was used to investigate the changes in the structure and abundance of the soil microbial community in the asparagus rhizosphere using water-biochar coupling treatments.The results showed that moderate deficit irrigation was promoted to form the bacterial abundance and diversity of asparagus rhizosphere soil.The relative abundance of soil fungi increased by 4.48％to 22.05％in the asparagus rhizosphere,with the increasing soil moisture;Biochar application increased the abundance and diversity of rhizosphere soil microorganisms.Furthermore,the number of rhizosphere bacterial communities increased,whereas,the number of fungal communities decreased under the B20(20 t/hm2)treatment,compared with the B0 and B10 treatments.There was a significant inhibitory in the application of 30 t/hm2 biochar.In addition,there was no significant effect on the structure of bacterial and fungal communities under different water-biochar coupling.The biochar application reduced the growth and reproduction of the root rot causal fungus Fusarium,indicating a better soil microenvironment asparagus rhizosphere.Optimal deficit irrigation was maintained on the microbial diversity and community stability.The higher numbers of soil bacterial communities were observed in the medium and mild deficit irrigation than those in the rest.There was the greatest community richness and diversity of rhizosphere soil bacteria at the medium deficit irrigation level.Therefore,the W1B20 treatment was more favorable to regulating the asparagus rhizosphere ecosystem,where the growth of the bacterial community was promoted to suppress the abundance of harmful fungal communities.This finding can also provide the theoretical basis to regulate the soil microcosm of the asparagus rhizosphere in continuous cropping,particularly for sustainable agriculture in the seasonal arid areas of Southwest China.

Deep learning has emerged as a cutting-edge approach for land cover classification using hyperspectral remote sensing images.However,hyperspectral remote sensing images suffer from two inherent limitations:spectral similarity between different objects and the absence of ground height information.To address these issues,this study proposes a multimodal land cover classification algorithm based on cross-feature granularity fusion network(CFCGNet),which establishes feature fusion between LiDAR data and hyperspectral remote sensing images.CFCGNet incorporates a multimodal cross-feature fusion module designed to extract high-order semantic features and complementary information from multisource data.Moreover,a composite granularity feature integration module is designed to integrate multiscale semantic information from various data sources.To mitigate the issue of uneven distribution of land use type samples,a weighted loss function is developed to optimize the model training process.Experimental results on the publicly available MUUFL,Houston 2018,and Trento datasets demonstrate that the proposed CFCGNet algorithm improves classification efficiency and enhances theoretical interpretability.