Joint Network Model Research Articles

An Attention-Based Spatial-Spectral Joint Network for Maize Hyperspectral Images Disease Detection

Maize is susceptible to pest disease, and the production of maize would suffer a significant decline without precise early detection. Hyperspectral imaging is well-suited for the precise detection of diseases due to its ability to capture the internal chemical characteristics of vegetation. However, the abundance of redundant information in hyperspectral data poses challenges in extracting significant features. To overcome the above problems, in this study we proposed an attention-based spatial-spectral joint network model for hyperspectral detection of pest-infected maize. The model contains 3D and 2D convolutional layers that extract features from both spatial and spectral domains to improve the identification capability of hyperspectral images. Moreover, the model is embedded with an attention mechanism that improves feature representation by focusing on important spatial and spectral-wise information and enhances the feature extraction ability of the model. Experimental results demonstrate the effectiveness of the proposed model across different field scenarios, achieving overall accuracies (OAs) of 99.24% and 97.4% on close-up hyperspectral images and middle-shot hyperspectral images, respectively. Even under the condition of a lack of training data, the proposed model performs a superior performance relative to other models and achieves OAs of 98.29% and 92.18%. These results proved the validity of the proposed model, and it is accomplished efficiently for pest-infected maize detection. The proposed model is believed to have the potential to be applied to mobile devices such as field robots in order to monitor and detect infected maize automatically.

A Joint Network for Low-Light Image Enhancement Based on Retinex

Methods based on the physical Retinex model are effective in enhancing low-light images, adeptly handling the challenges posed by low signal-to-noise ratios and high noise in images captured under weak lighting conditions. However, traditional models based on manually designed Retinex priors do not adapt well to complex and varying degradation environments. DEANet (Jiang et al., Tsinghua Sci Technol. 2023;28(4):743–53 2023) combines frequency and Retinex to address the interference of high-frequency noise in low-light image restoration. Nonetheless, low-frequency noise still significantly impacts the restoration of low-light images. To overcome this issue, this paper integrates the physical Retinex model with deep learning to propose a joint network model, DEANet++, for enhancing low-light images. The model is divided into three modules: decomposition, enhancement, and adjustment. The decomposition module employs a data-driven approach based on Retinex theory to split the image; the enhancement module restores degradation and adjusts brightness in the decomposed images; and the adjustment module restores details and adjusts complex features in the enhanced images. Trained on the publicly available LOL dataset, DEANet++ not only surpasses the control group in both visual and quantitative aspects but also achieves superior results compared to other Retinex-based enhancement methods. Ablation studies and additional experiments highlight the importance of each component in this method.

Numerical Investigation of Stratified Slope Failure Containing Rough Non-Persistent Joints Based on Distinct Element Method

To address the critical issue of slope stability in jointed rock masses with complex and rough structural planes, a rough joint network model using the Fourier transform was proposed and applied to the Shilu open pit mine. The on-site structural plane survey results were combined with MATLAB and PFC2D to establish numerical models for slope stability analysis considering both linear-jointed and rough-jointed rock slopes. By comparing the slip body and fracture distribution, it was found that the rough-jointed slope was stabler than the linear-jointed slope. In addition, the fracture patterns and slip displacement were significantly influenced by the inclination and undulation of the bedding planes. Slip failure patterns occurred when the angle of inclination was set at 60°. The joints played a crucial role in inducing the shear strength of slope rock masses, and the slide area was mainly observed in the shallow slope surface for inclination angles of 0° and 45°, and in the middle deep area for 60° and 90°. These results demonstrated the importance of considering rough non-persistent fractures when developing a new numerical model for slope failure modes.

Rock Mass Structure Classification of Caves Based on the 3D Rock Block Index

In large-scale water conservancy and hydropower projects, complex rock structures are considered to be the main factor controlling the stability of hydraulic structures. The classification of rock mass structure plays an important role in the safety of all kinds of large buildings, especially underground engineering buildings. As a quantitative classification index of rock mass, the rock block index is very common in the classification of borehole and dam foundation rock mass structures. However, there are few studies on the classification of underground engineering rock masses. Moreover, their classification criteria have disadvantages in spatial dimension. Therefore, this paper takes the long exploratory cave CPD1 in the water transmission and power generation system of the Qingtian pumped storage power station in Zhejiang Province as the research object and launches a study on the structural classification of the rock mass of a flat cave based on the 3D rock block index. According to the group distribution of joints, the sections are statistically homogeneous. Additionally, the Monte Carlo method is used to carry out random simulations to generate a three-dimensional joint network model. The virtual survey lines are arranged along the center of the shape of the three different orthogonal planes of the 3D joint network model to represent the boreholes, and the RBI values of the virtual survey lines on each orthogonal plane are counted to classify the rock mass structure of the flat cave in a refined manner using the rock block index of the rock mass in 3D. The above method realizes the application of the 3D rock block index in underground engineering and overcomes the limitations of traditional rock mass classification methods in terms of classification index and dimension. The results show that: (1) Three-dimensional joint network simulations built on statistical and probabilistic foundations can visualize the structure of the rock mass and more accurately reflect the structural characteristics of the actual rock mass. (2) Based on the 3D rock block index, the rock mass structure of the long-tunnel CPD1 is classified, from that of a continuous structure to a blocky structure, corresponding to the integrity of the rock mass from complete to relatively complete. The classification results are consistent with the evaluation results of horizontal tunnel seismic wave geophysical exploration. (3) Based on the 3D joint network model, it is reasonable and feasible to use the 3D rock block index as a quantitative evaluation index to determine the structure type of flat cave rock masses. The above method is helpful and significant in the classification of underground engineering rock mass structures.

AVERAGE TRAPPING TIME ON THE 3-DIMENSIONAL 3-LEVEL SIERPINSKI GASKET NETWORK WITH A SET OF TRAP NODES

As a basic dynamic feature on complex networks, the property of random walk has received a lot of attention in recent years. In this paper, we first studied the analytical expression of the mean global first passage time (MGFPT) on the 3-dimensional 3-level Sierpinski gasket network. Based on the self-similar structure of the network, the correlation between the MGFPT and the average trapping time (ATT) is found, and then the analytical expression of the ATT is obtained. Finally, by establishing a joint network model, we further give the standard process of solving the analytical expression of the ATT when there is a set of trap nodes in the network. By illustrating examples and numerical simulations, it can be proved that when the trap node sets are different, the ATT will be quite different, but the super-linear relationship with the number of iterations will not be changed.

A Sentence-Level Joint Relation Classification Model Based on Reinforcement Learning.

Relation classification is an important semantic processing task in the field of natural language processing (NLP). Data sources generally adopt remote monitoring strategies to automatically generate large-scale training data, which inevitably causes label noise problems. At the same time, another challenge is that important information can appear at any place in the sentence. This paper presents a sentence-level joint relation classification model. The model has two modules: a reinforcement learning (RL) agent and a joint network model. In particular, we combine bidirectional long short-term memory (Bi-LSTM) and attention mechanism as a joint model to process the text features of sentences and classify the relation between two entities. At the same time, we introduce an attention mechanism to discover hidden information in sentences. The joint training of the two modules solves the noise problem in relation extraction, sentence-level information extraction, and relation classification. Experimental results demonstrate that the model can effectively deal with data noise and achieve better relation classification performance at the sentence level.

Representative Sample Sampling Method for Size Effect Experiment of Jointed Rock Mass

The influence of size effect on the hydromechanical behavior of rock mass has long been recognized. As a result, analysis of the rock mass size effect has been developed. However, when conducting size effect studies, the representativeness of the sample is less considered. Therefore, combining the existing statistical methods and sampling methods, a comprehensive representative sampling method for rock mass is present. In this method, a large number of sample statistics of different sizes are provided in the progressive coverage method; then, the plane density of the track length is defined as the value of stratification to perform stratified sampling for representative samples. Furthermore, it is applied to a joint network model generated in accordance with the actual situation and compared with those of other methods. The sampling results show that the proposed method can provide a certain reference value for studying the scale-dependent behavior of rock mass mechanical parameters.

A supervised deep convolutional based bidirectional long short term memory video hashing for large scale video retrieval applications

Recently, large scale video content retrieval has gained more attention due to a large amount of user-generated images and video content available over the internet. Hashing is one of the effective techniques that encode the high dimensional features vectors into compact binary codes. The aim of hashing is to generate the short binary codes and map the similar hash code values to retrieve similar video from the database with minimum distance measure. Deep learning-based hashing networks are employed to learn the representative video feature to estimate the hash functions. However, the existing hashing approaches fail due to the frame-level feature representation and does not well exploits the effective temporal features in visual search. Furthermore, significant loss of features during the dimensionality reduction step causes low accuracy. Hence, it is essential to develop a deep learning-based hashing framework that should exploits both the strong spatial and temporal features for the scalable video search. The main objective is to learn the high dimensional features from the entire video and derive compact binary codes to retrieve the similar videos for the input query sequence. In this paper, we propose a joint network model of supervised stacked heterogeneous convolutional multi-kernel (Stacked HetConv-MK)-bidirectional Long Short Term Memory (BiDLSTM) network model that effectively encodes the rich structural as well as the discriminative features from the video sequence to estimate the compact binary codes. Initially, the video frames are passed to the stacked convolution networks with heterogeneous convolutional kernel size and residual learning to extract the spatial features at different views from the video sequences and to improve the learning efficiency. Then, the bidirectional network computes the sequence in both forward and backward directions and obtains the series of hidden state output. Finally, the fully connected structure with an activation unit performs hashing to learn the multiple codes for each video. Experimental analysis is performed on three datasets and the result shows a better accuracy measure than the other state-of-art approaches.

Investigating the Failure Mechanism of Jointed Rock Slopes Based on Discrete Element Method

This paper presents a comprehensive engineering method to investigate the failure mechanism of the jointed rock slopes. The field geology survey is first carried out to obtain the slope joint data. A joint network model considering the structural complexity of rock mass is generated in the PFC software. The synthetic rock mass (SRM) approach for simulating the mechanical behavior of jointed rock mass is employed, in which the flat‐jointed bonded‐particle model (FJM) and smooth joint contact model (SJM) represent intact rock and joints, respectively. Subsequently, the effect of microparameters on macromechanical properties of rock is investigated for parameter calibration. Moreover, the scale effect is analyzed by multiscale numerical tests, and the representative elementary volume (REV) size in the selected research area is found as 16 m × 16 m × 16 m. The microparameters of the SRM model are calibrated to match the mechanical properties of the engineering rock mass. Finally, an engineering case from Shuichang open‐pit mine is analyzed and the failure process of the slope during the excavation process from micro‐ to macroscale is obtained. It has been found that failure occurs at the bottom of the slope and gradually develops upwards. The overall failure of the slope is dominated by the shallow local tension fracture and wedge failure.

Numerical investigations of rock bridge effect on open pit slope stability

Abstract In this study, the effect of rock bridges on rock slope stability was investigated by incorporating non-persistent joint networks in numerical models, and the critical profiles of an open pit mine were analysed. Parallel deterministic networks of infinite and finite lengths, ubiquitous joint network model and Veneziano joint network model were used in order to simulate the rock fractures. Materials were modelled based on the generalised Hoek―Brown and equivalent Mohr–Coulomb failure criteria. The parallel deterministic infinite and the ubiquitous joint network models produced lower safety factors. The introduction of rock bridges along discontinuity planes in the parallel deterministic network and Veneziano joint network models significantly contributed to the stability and strain distribution, which should be considered in stability analysis of rock mass in open pit by rock slope practitioners. The results show the significance of joints in hard rock behaviour and the joints should be included in order to attain practical and realistic simulations.

Integration of homogeneous structural region identification and rock mass quality classification.

In rock engineering projects, professionals assess the overall rock mass qualities using a sole value. However, the true qualities of partial rock masses are incompatible with such a value. To address this problem, the idea of regionally classifying rock mass qualities is proposed and the associated procedure presented. To achieve this goal, the probabilistic and deterministic joints within the study area were determined, and a three-dimensional joint network model was created. Then, the three-dimensional joint network model was discretized into interlocking subdomains, and the modified blockiness index (MBi) was used to finely identify the homogeneous structural regions, together with the k-means algorithm and the sum of squared errors (SSE). A synthetic model comprising homogeneous structural regions was developed and validated with respect to the extracted cross-sections. Next, an improved rock mass rating system (RMRmbi) was introduced, and the viability of RMRmbi was supported through a significant amount of theoretical cases and several real cases. Finally, visualization of regional RMRmbi classification results was performed. Results show that: (i) the homogeneous structural regions are finely demarcated in three dimensions, and (ii) the proposed idea can overcome the problem of rock mass quality classification using the conventional approach often leading to ‘overgeneralization’.

Multi-scale modeling of shale laminas and fracture networks in the Yanchang formation, Southern Ordos Basin, China

Outcrops and core observations show that sandy and tuffaceous laminas and natural fractures are well developed in the continental shale gas reservoir. This study reports the distribution of laminas and natural fracture networks at different scales, in cases of outcrops, wells and cores from Yanchang Formation, southern Ordos Basin. Firstly, based on two-dimensional fracture field investigations, a 3-D joint network model of the study area was built using the Monte Carlo simulation method. Then the laminations were identified from macroscopic scales to microscopic scales with multiple probe techniques. Statistical analysis of multi-scale layer thickness suggested that the average thickness of layers at multiple scales showed a fractal feature, and that the layer thickness at each scale followed with an exponential probability distribution, whose exponential coefficient was −3.38, −1.91, −1.65, −1.21, corresponding to meter scale, decimeter scale, centimeter scale and millimeter scale respectively. The fitting curves indicated that the majority of laminas at each scale were relatively thin, and the thicker the layer, the less the number of layers contained in shale. Finally, the lamination models were constructed and superposed over the joint network model to generate 3-D geological structure models at various scales. The models were validated by reproducing fracture and lamina parameters fairly close to those obtained from realistic geological bodies in the study area. The findings of this work could provide a more reliable geological model for the numerical simulation and physical model test of hydraulic fracturing in shale gas reservoir.

A joint network design and multi-echelon inventory optimisation approach for supply chain segmentation

Segmenting large supply chains into lean and agile segments has become a powerful strategy allowing companies to manage different market demands effectively. A current stream of research into supply chain segmentation proposes demand volume and variability as the key segmentation criteria. This literature adequately justifies these criteria and analyses the benefits of segmentation. However, current work fails to provide approaches for allocating products to segments which go beyond simple rules of thumb, such as 80-20 Pareto rules. We propose a joint network and safety stock optimisation model which optimally allocates Stock Keeping Units (SKUs) to segments. We use this model, populated both with synthetic data and data from a real case study and demonstrate that this approach significantly improves cost when compared to using simple rules of thumb alone.

English

Nowadays, modeling is increasingly used as a method for determination of the mechanical and hydraulic behavior of the rock mass. In this paper, with regard to the high importance of joint persistence characteristic on the mechanical and hydraulic behavior of the rock mass, geometric-stochastic joint network model has been developed by considering the statistical characteristics of joint size based on Veneziano model. With the use of surveyed data in the right bank of Rudbar Lorestan dam plant and estimation of the best probability distribution function on geometric characteristics of existing joint sets in this region, the three-dimensional geometric model of joint network has been developed. In order to model implementation, a computer code written in C++, called FRAC3D, has been developed that is able to represent the joint network in different directions and to generate text outputs. The results of this model can be a useful input for numerical stability analysis and hydraulic behavior studies of rock mass. Key words: Geometrical-Stochastic modeling, joint network, joint surveying, statistical studies.

Estimation of Hydraulic Property of Jointed Rock Mass Considering Excavation-induced Change in Permeability of Each Joints

Underground excavation can induce significant deformation of pre-existing joints in rock mass. Due to the sliding and opening of the joints, consequent changes in the flow characteristics of a jointed rock mass would be anticipated. In the present work, the finite element excavation analysis for a jointed rock mass and the finite element flow and transport analysis using the discrete joint network model are employed and combined so as to make it possible to evaluate excavation induced changes in flow properties of a jointed rock mass. A separate computational module connecting these two codes is incorporated to modify initially imposed transmissivities into the excavation induced ones considering the deformation of joints during an excavation. The effect of excavation on the flow properties is evaluated by including these excavation induced transmissivity changes of individual joints. The excavation induced transmissivities are calculated using the analytical approximation solution developed for flow through a single rock joint. The calculation of the excavation induced transmissivity involves the distributions of shear displacements and normal stresses around the excavation, and explicitly takes into account of the effect of surface geometric roughness. A test simulation for a deep underground repository is performed to demonstrate the typical results predicted in the proposed approach.