Accurate power prediction of clean energy generation is the key to realize the efficient utilization of clean energy. However, a variety of factors lead to drastic fluctuations in clean energy power generation, which brings great challenges to power prediction. In view of the current demand for large-scale grid connection of clean energy, the research significance of clean energy power generation prediction models and their classification are discussed from multiple perspectives, and the latest applications of artificial intelligence technology in the field of clean energy power generation prediction, including traditional machine learning, deep learning and combinatorial methods, are reviewed, compared and summarized. In addition, the key factors affecting clean energy access are analyzed in depth, and the technical development of clean energy consumption analysis models is discussed, with different focuses and development directions from those of prediction models elaborated. Finally, the future development trend of clean energy prediction and consumption analysis models and their increasing significance are discussed.

Cancer is seriously threatening human health. Chemotherapy, as a common treatment for cancer, often causes serious toxic side effects due to its indiscriminate killing of normal cells. Photo-activated chemotherapy (PACT) is one of the effective methods to reduce the toxic and side effects of drugs. It combines photoremovable groups (PPGs) with inhibitor prodrugs to form photoactivated inhibitor prodrugs, which can be activated by light at specific sites and release inhibitor, so as to achieve accurate targeting of cancer tissues and minimize the toxic and side effects of anti-tumor agents on normal tissues. This review comprehensively covers the basic structures and strategies of photoactivated prodrugs, common PPGs and the important advances for anti-tumor research in recent years. According to the different mechanism of action of these prodrugs, this review introduces cytotoxic, molecular targeting, immune and hormone anti-tumor prodrugs in turn, aiming to provide a new scheme for the precision treatment of cancer.

Electric load forecasting is critical for power dispatch and system security. A combined forecasting model is proposed for ultra-short-term load forecasting, integrating Complementary Ensemble Empirical Mode Decomposition (CEEMD) with machine learning and intelligent optimization algorithms. The model first decomposes the original data using CEEMD, followed by the use of permutation entropy (PE) thresholds to classify the load components for separate forecasting methods. Bidirectional Long Short-Term Memory (BiLSTM) is applied to predict high-frequency components, while low-frequency components are predicted using Hybrid Kernel Extreme Learning Machine (HKELM) optimized by the Snow Ablation Optimizer (SAO). The final forecast is obtained by summing the predicted components. Experimental results show that the model achieves root mean square error of 61.61 kW, mean absolute error of 43.91 kW, and mean absolute percentage error of 0.38%, significantly outperforming traditional models. These results demonstrate that the model effectively captures the inherent features of the data and combines the advantages of various forecasting methods, providing high accuracy and generalizability for ultra-short-term load forecasting.

With the widespread adoption of electric vehicles (EVs) and the implementation of environmental policies, efficiently scheduling EV charging behaviors has become increasingly important for meeting user demands and can ensure grid stability. Addressing the different charging needs of EV owners in smart grids and the curse of dimensionality problem in traditional machine learning algorithms, this paper proposes an EV charging scheduling algorithm based on Deep Q-Network (DQN). This algorithm leverages the advantages of deep reinforcement learning, combining real-time electricity prices and vehicle status information to dynamically adjust EV charging and discharging behaviors, aiming to maximize economic benefits and optimize grid load. Experimental results show that compared with uncontrolled charging strategies, the proposed algorithm can reduce charging costs by approximately 65.6% over a 30-day test period. It effectively adapts to real-time grid price fluctuations and changes in user demands, reduces peak charging requirements, effectively lowers charging costs, and improves grid stability.

In the transient stability prevention and control of power system, the time domain simulation calculation is complex, and there is a problem of sample class imbalance in the system, which affects the performance of machine learning model. To solve these problems, a transient stability prevention and control method based on deep extreme learning machine (DELM) is proposed. Firstly, the oversampling technique is used to deal with the unbalance of sample class. Then, DELM is used to discover the potential information of the balanced data set, and a mapping model between the operating parameters of the power system and the transient stability index is established. The transient stability prediction model based on DELM is introduced in preventive control to replace the transient stability constraint optimal power flow (TSCOPF) model containing differential algebraic equations with transient stability constraints, the computational complexity is reduced, and the model is solved by firefly algorithm to obtain the final strategy. Finally, the IEEE 39-node system is simulated and verified. The results show that the proposed preventive control method can improve the transient stability of the system at an optimal adjustment cost of \$2 042, adjust the transient instability to stability, and the calculation time of the firefly algorithm can be controlled within 20 s. This indicates that the transient stability prevention and control method based on DELM proposed in this paper can effectively improve the transient stability of the system, and has a fast calculation speed and good economy.

This paper addresses the need for high timeliness and reliability in industrial Internet of things (IIoT) data acquisition systems. It considers the coexistence of traffic with different priority levels, the scarcity of resources, and the randomness of channels. A method based on a stochastic hybrid system is proposed to construct a timeliness characterization model suitable for such networks. The model is designed to describe the timeliness features of data acquisition networks under conditions involving a large volume of traffic with varying priorities. Based on this model, the paper explores the impact of coupling parameters within the system on the information timeliness. The research results indicate that by optimizing the link sensing strategy, the negative impact of transmission error probability on information age can be effectively reduced. In networks with the same priority, the information age of the link gradually stabilizes towards a limiting value as the data generation rate increases. In networks with different priorities, higher-priority links improve their timeliness by sacrificing the timeliness of other links. As the data generation rate increases, the system gradually trends towards a degradation into a single-link mode. Therefore, to ensure the overall timeliness of the system, it is necessary to optimize the sensing resource allocation for low-priority links to maintain efficient system operation.

To address the issue of low fault detection accuracy in intermittent processes due to complex dynamic characteristics and multi-stage features, a fault detection method for multi-stage intermittent processes based on Bottom-Up Derivative Dynamic Time Warping (BU-DDTW) and Semi-Positive Definite Sparse Principal Component Analysis (SPCA-PSD) is proposed. Initially, an encoder-decoder model is utilized to capture the dynamic characteristics of the time series after batching. Subsequently, by integrating the BU-DDTW merging strategy, the dynamic structural similarity between different subsequences is measured, achieving precise phase division. Then, the SPCA-PSD method introduced sparsity and semi-positive definite constraints to accurately identify key variables representing the characteristics of each stage, constructing a multi-stage fault detection model. Experiments are conducted on data from the penicillin supplementation batch fermentation process, and the results demonstrate that the proposed method achieves an average fault detection rate of 95.2% for six different types of faults, significantly outperforming other methods. The results not only prove the effectiveness of the proposed method in phase division of intermittent processes, but also enhance the accuracy and interpretability of the multi-stage fault detection model.

Occlusion and pose variations are the main distractors that affect facial expression recognition in natural scenes. Most existing methods use attention to enhance expression-related information and reduce the impact of occlusion and pose variations on expression recognition performance. However, these methods use the same attention mechanism at different locations in the network, ignoring the differences between shallow and deep feature tensors in spatial and channel dimensions, which affects the accuracy of feature expression. For this reason, a Granularity-Aware Multi-Dimensional Adaptive Attention network (GA-MDA) is proposed. Firstly, a Cross-granularity Spatial Aware Attention module (CSA) is designed for enhancing the feature expression ability of the shallow network. Then, a Multi-Dimensional Adaptive Attention module (MAA) is introduced to adaptively optimize the spatial and channel feature representations in different dimensions to further enhance the feature expression ability of the model. The results show that GA-MDA achieves recognition accuracy of 92.01% and 90.36% on RAF-DB and FERPlus datasets, improves the recognition performance by 0.09% and 0.43%, and reduces the number of model parameters by 2.963×10⁷ and 6.341×10⁷, respectively, compared with the current state-of-the-art methods HANet and GE-LA.

Automated facial acne detection is crucial for precise clinical diagnosis and treatment. However, existing methods still suffer from significant issues of missed detection and false detection of small acne targets. For more accurate acne detection, an enhanced YOLOv8s method with two key modifications is proposed in this paper. Firstly, the original backbone network of YOLOv8s is improved into a hybrid backbone network that integrates with Transformer. This improvement effectively combines the advantages of convolutional neural network capturing local detail information and Transformers maintaining global feature information, significantly enhancing the feature extraction and representation capabilities for small acne objects. Secondly, a multi-scale channel attention module is integrated into the neck network, enabling adaptive feature weight adjustment through cross-scale context aggregation, thereby mitigating semantic-scale inconsistency. Experiments on both public and self-built facial acne datasets demonstrate that, compared with the current state-of-the-art DSDH method, the proposed method achieves mAP improvements of 1.20% and 5.24% on respective datasets, with corresponding detection speed increases of 46.3 and 47.6 frame/s.

In the field of 3D human pose estimation, occlusion leads to inaccurate extraction of human joint points. To address this problem, this paper proposes a 3D human pose estimation algorithm that combines spatio-temporal attention and channel attention. Firstly, a feature filtering module is proposed, which further captures the feature information of human joint points by introducing the position embedding module. Then, a mobile vision transformer temporal attention module is proposed, which can obtain more details of pose features by introducing the SiLU activation function. Finally, a channel attention module is proposed, which adjusts the weights of the output channel features by introducing a parallel branch processing architecture and adding normalization layers, so that the algorithm can focus on human pose features while reducing the influence of background features. Experiments are conducted on the Human3.6M dataset. Compared with the baseline model Strided Transformer, the mean per joint position error (MPJPE) and the procrustes-aligned mean per joint position error (P-MPJPE) decrease by 2.5% and 2.3%, respectively, when the 2D joint points extracted from the cascaded pyramid network (CPN) are used as input. The MPJPE decrease by 6.7% when the annotated 2D joint points of the Human3.6M dataset are used as input. Experimental results show that the proposed algorithm has high accuracy.

Existing text-based person re-identification methods are limited by issues of feature alignment and semantic ambiguity. To address these challenges, a cross-modal semantic collaboration framework is proposed. Shared semantic information between images and text is learned, and local visual-text correspondence constraints are established to improve the matching efficiency between images and text. Specifically, a text semantic clustering module is introduced to automatically extract text related to local visual semantics, while image self-supervised learning is applied to enhance the learning of local features. A common semantic collaboration module is then built to capture both the differences and commonalities between the image and its description, establishing a semantic consistency mapping in the embedding space. Finally, a semantic constraint reasoning module is incorporated to perform retrieval by combining the semantic consistency scores of images and text, thereby improving retrieval efficiency. Experiments on three benchmark datasets show that the proposed method effectively enhances the performance of the model.For Rank-1 indicators, it has improves 0.75%, 1.43%, 0.88%, respectively, and the precision rises by 0.64%, 2.56%, 3.96%, respectively.

The boundedness and attractiveness of a class of nonlinear neutral delay differential equations are studied by employing Krasnoselskii's fixed point theorem. By introducing auxiliary functions and using an integral factor, the differential equation is transformed into an integral equation. Then, the Krasnoselskii's fixed point theorem is used to discuss the boundedness and attractiveness of the transformed equation. Finally, the properties of auxiliary functions are used to discuss the boundedness and attractiveness of the original equation. The results obtained in this paper improve the corresponding results in the existing literatures, and an example is provided to illustrate the effectiveness of the obtained results.

In this paper, the problem of solving limit cycles for a class of bivariate chemical oscillation reaction equations is investigated. To solve these equations, firstly, the bivariate chemical oscillation reaction equations are simplified into a general form of Liénard equation and the perturbation increment method is used to solve limit cycles. Starting from the zero-order perturbation solution obtained during the perturbation phase, the increment phase iteratively approximates step by step until the final limit cycle is achieved. Through examples of continuous stirred tank reactor reactions and glycolysis reactions, and compared with numerical integration method, the results show a high degree of consistency, demonstrating the effectiveness of the perturbation-increment method in handling limit cycle problems of such nonlinear dynamic equations.

In longitudinal data analysis, the normality of model errors is a common assumption; however, this assumption may contradict the true characteristics of the data. Additionally, overlooking the correlation between longitudinal data and survival data can lead to biased analytical results. To address these issues, this paper proposes a Bayesian joint model: the longitudinal process is modeled using a linear mixed-effects model with error terms following a Skew-t distribution, while the survival process employs a Cox proportional hazards model. Bayesian estimation of the unknown parameters in the joint model is conducted using the Metropolis-Hastings (MH) algorithm and Gibbs sampling. Numerical simulation results indicate that the Skew-t method demonstrates superior performance in data fitting compared with the traditional estimation methods. Finally, this methodology is applied to the analysis of AIDS data, and validation confirms that it provides good fitting results and accurate parameter estimates.

In order to reveal the effects of soil indicators in different producing areas on the effective ingredients and activities of Bletilla striata, this study took B. striata and its cultivated soil from 8 producing areas in Guizhou as the research objects. 10 soil indicators, 5 functional components, 4 trace elements, tyrosinase inhibitory activity and anticoagulation activity of B. striata were determined. Redundancy analysis was used to investigate the correlation between soil indicators, functional components and activities, so as to reflect the effects of different soil indicators on the quality and activity of B. striata. The results showed that the contents of total ash, polysaccharide, total phenol, militarine and BTS from 8 different producing areas were 27.1~41.5 mg/g, 36.2%~56.4%, 1.8~4.1 mg/g, 35.4~82.6 mg/g and 8.2~12.7 mg/g, respectively. The contents of Cu, Zn, Fe and Mn were 6.21~16.51 mg/kg, 6.38~13.77 mg/kg, 124.49~253.4 mg/kg, 6.51~42.14 mg/kg, respectively. The inhibitory activity of tyrosinase was 40.2%~66.5%, and the clotting time was 236.1~306.4 s. Soil indicators had a great influence on the effective components, trace elements and activity of B. striata, among which the six soil indicators with significant influence were available zinc, pH, total nitrogen, available copper, available phosphorus and organic matter in order of contribution degree. The results of this study revealed the key soil factors affecting the quality difference of B. striata, and laid the research foundation for the standardization and scale cultivation of B. striata.

Excessive use of chemical fertilisers will change the physical properties of the soil and destroy its structure, and coupled with the effects of continuous cropping barriers, soil nutrient effectiveness will be severely reduced. In this paper, the effects of microbial agents on the physicochemical properties and microbial community structure of soil planted with Ophiopogon japonicus and its soil, which is widely planted in Santai County, Mianyang City, Sichuan Province, the largest Ophiopogon japonicus planting base in the country, were investigated by using the third-generation high-throughput sequencing Illumina MiSeq. The results of the study showed that the yield of Ophiopogon japonicus increased from 5 344.05±987.00 kg/hm² to 6 501.45±1 154.10 kg/hm² by 21.63% (P<0.05) after the application of microbial agents. Soil pH significantly increased from 6.50±0.24 to 6.86±0.11 (P<0.05). At the same time, soil organic matter, total nitrogen and total phosphorus content increased to varying degrees. Correlation analysis showed that soil pH and total nitrogen profoundly affected the yield of Ophiopogon japonicus with significant (P<0.05) and highly significant (P<0.01) correlations, respectively. The use of fungicides changed the soil microbiological environment, and the relative abundance of bacterial Acidobacteria, Chloroflexi, and Firmicutes in the soil was significant increased after fungicide application (P<0.05), and that of fungal Basidiomycota was significantly increased (P<0.05); among them, the relative abundance of Bacillus and unclassified_Chaetomiaceae, which had good resistance to plant pathogens, also increased significantly (P<0.05). Correlation analysis showed that the yield of Ophiopogon japonicus was significant positive correlated (P<0.05) with the relative abundance of Bacillus, Nitrosomonadaceae_MND1, Rokubacteriales, Nitrospira, and unclassified_Chaetomiaceae significantly positive correlated (P<0.05) and significantly negative correlated (P<0.05) with the relative abundance of Pseudogulbenkiania. This study showed that the application of microbial fungicides increased the pH of the soil planted with Ophiopogon japonicus, alleviated acidification, and increased the total organic matter, total nitrogen and total phosphorus contents of the soil. Meanwhile, the application of microbial fungicides improved the structure of soil microbial community. The results provide a theoretical basis for the application of microbial agents in the ecological cultivation of Ophiopogon japonicus.

Semisulcospira gredleri, Viviparus tricinctus and Sinotaia limnophila are all important economic freshwater snail species, which play significant ecological roles in the restoration of water ecology. The gut microbes of aquatic animals play a crucial role in food digestion and nutrient absorption. Therefore, in this study, 16S rDNA high-throughput sequencing was use to compare the composition and function of gut microbiota in three species of freshwater snails from the Lijiang River. The results showed that at the phylum level, the common dominant microorganisms in the three groups were Proteobacteria, Firmicutes, Bacteroidetes, Actinobacteria. Proteobacteria was the most abundant bacteria in both groups. At the genus level, the common dominant bacteria were Aeromonas, Acinetobacter, Pseudomonas. In terms of community diversity, there was no significant difference between Semisulcospira gredleri and Viviparus tricinctus (P>0.05), while there was a highly significant difference between Sinotaia limnophila and both Semisulcospira gredleri and Viviparus tricinctus (P<0.01). Based on PICRUSt analysis, the intestinal microbiota of three species of snails was predicted to be mainly involved in metabolic processes, including amino acid metabolism, metabolism of cofactors and vitamins, xenobiotics biodegradation and metabolism, metabolism of terpenoids and polyketides, lipid metabolism, metabolism of other amino acids.

Vegetation gross primary productivity (GPP) is a key parameter for measuring the carbon cycle in terrestrial ecosystems. The study of growing season gross primary productivity (GPP_GS) of vegetation in Yunnan Province can help to understand the vegetation dynamics and carbon cycling pattern of terrestrial ecosystems, which is of great significance to the sustainable development of regional ecosystems. Using Theil-Sen Median trend analysis and Mann-Kendall significance test, the spatial and temporal characteristics of vegetation GPP_GS changes in Yunnan Province were analysed, and then the direct, indirect and combined effects of climatic factors on the changes of vegetation GPP_GS were revealed by using the pathway analysis. The results showed that: ① the vegetation GPP_GS in Yunnan Province from 2001 to 2020 showed a fluctuating upward trend, with an increasing rate of 1.216 g·(m²·a)^-1; most of the vegetation GPP_GS showed an upward trend, among which the meadows had the highest rate of increase, which was 1.674 g·(m²·a)^-1. ② The area with an upward trend in vegetation GPP_GS in Yunnan Province accounted for 73.83%, with the area with an upward trend in alpine vegetation, meadow, coniferous forest, shrubland, herbaceous vegetation, cultivated plants, and broad-leaved forest being 85.21%, 84.64%, 79.11%, 75.85%, 73.58%, 71.76%, and 58.67%, respectively. ③ As a result of the through-trail results, changes in the GPP_GS of both meadow and coniferous forests were most strongly influenced by temperature fluctuations, with average temperature playing a central role, precipitation was the main factor causing the change in vegetation GPP_GS of herbaceous vegetation and cultivated plants, and the primary influence on the variation of vegetation GPP_GS in alpine vegetation, shrublands, and broadleaf forests was solar radiation. ④ The dominant factors that directly influenced vegetation GPP_GS in Yunnan Province accounted for average temperature (54.87%), precipitation (7.86%), and solar radiation (9.08%).

Watershed, as the most fundamental geomorphological unit, serves as a crucial natural geographic element for documenting the genesis and evolution of watersheds. The development and spatial traits of geoheritage within this specific domain can be manifested, and the interaction between watersheds and various natural geographic elements, with geoheritage being a part of it, has emerged as a cutting-edge focus in geoscientific research. In this study, the entire Guangxi Hongshui River Basin was selected as a paradigmatic research zone. Grounded on comprehensive and detailed field investigations, a spatial characterization of geological relics was conducted on a large basin scale, encompassing multiple types and a holistic perspective in terms of their distribution patterns, taxonomies, grades, as well as protection and utilization modalities. A quantitative analysis of the spatial characteristics of the geological relics within the Guangxi Hongshui River Basin was carried out, thereby furnishing a scientific underpinning for exploring the interaction between the evolution of the Hongshui River and the natural geological elements within the basin. This also contributes to promoting the construction of ecological civilization within the watershed and lays a foundation for rural revitalization. The research findings indicate that: 1) There are 93 exemplary geological relics within the basin, with the middle reaches harboring the highest number of such sites, followed by the lower reaches. The majority of the geological relics in the upper, middle, and lower reaches are situated above or in the vicinity of the main tributaries of the Hongshui River. Even those geological relics that are relatively distant from the main tributaries are often interconnected via underground rivers. 2) The geological relics within the basin can be classified into 6 major categories, 12 classes, and 16 subclasses. Among them, the geomorphic landscape major category is the most prolific, trailed by the water landscape major category. The rocky geomorphic landscape category (with karst landscape as the principal subcategory) and the river landscape category are concentrated in the middle and lower reaches, whereas the fluvial geomorphic landscape category and the lake landscape category are preponderant in the upper reaches and middle and lower reaches, respectively. 3) Nearly 60% of the basin's geological relics have attained the provincial level or above. The world-class and national-level geological relics are predominantly clustered in the middle reaches and exhibit a close spatial correlation with the main streams and tributaries of the Hongshui River. Concurrently, the higher the grade of the geological relics, the more pronounced the non-uniformity of their spatial characteristics. 4) The conditions for the protection and utilization of geological relics within the basin are generally favorable. The downstream geological relics possess the shortest average highway distance to the surrounding significant towns, the upstream exhibits the highest degree of compatibility with other tourism resources in the vicinity, and the middle and lower reaches have a relatively higher proportion of protected utilization.

Aiming to solve the problems of low juice yield and poor quality in the processing of Rhodomyrtus tomentosa fruit juice, the process conditions of ultrasonic-assisted enzymatic hydrolysis were optimized by response surface method, and then the quality of R. tomentosa fruit juice was evaluated. The results showed that the optimal processing conditions for R. tomentosa fruit juice were as follows: pectinase∶cellulase 1∶3, enzyme dosage 0.45%, enzymolysis temperature 50 ℃, enzymolysistime 1.5 h, ultrasonic power density 16 W/L. Under these conditions, the juice yield of R. tomentosa fruit was 75.13%, and the light transmission rate was 74.20%. Among the fruit juices prepared by four methods, namely the treatment without ultrasound wave and enzymes, enzymolysis treatment alone, ultrasound wave treatment alone, ultrasonic-assisted enzymatic hydrolysis treatment, the fruit juice prepared by ultrasonic-assisted enzymatic hydrolysis treatment had the highest juice yield, light transmission, soluble solids and nutrients, while the fruit juice prepared by enzymolysis treatment alone had the strongest antioxidant activity and the highest sensory score.