Narrow Results

With the maturity of the smart meter embedded system integration mode, the power consumption information can be fed back to the management center by the intelligent terminal through the power Internet of Things. The smart terminal adopts the narrowband data communication mode. However, the narrowband channel interval is hindered by interference signals, resulting in serious data packet loss. Therefore, a stable and efficient acquisition scheme is needed to ensure its smooth interaction. Based on the application of smart meter data communication, we analyze the power consumption optimization and carrier transmission optimization of the power Internet of Things and propose a data analysis method based on self-learning and external characteristics, which is used to calculate the offline execution strategy of smart meters when they encounter factors such as power outage or intrusion. A production task-driving rule is designed. The external characteristics of deep learning data of the power Internet of Things are collected. The accuracy of data classification is judged according to the parallel Internet of Things of a convolutional neural network and a long-term and short-term memory deep neural network. When dealing with high-dimensional original data, the data network is graded to reduce the work of feature selection and the difficulty of training. Moreover, we improve the autonomous execution ability of an embedded terminal device in an offline state. We also strengthen the ability of abnormal data detection and judgment learning to meet the requirements of efficient and stable operation of the power grid. Simulation results verify that compared with the current mainstream schemes, the performance of the proposed scheme is improved by more than 17%. Besides, the fault tolerance rate is enhanced by 29%, and the power consumption after optimization is reduced by 26.3% and 42.8%. These effects highlight the role of data processing in embedded systems. It solves the problem of abnormal data and energy loss affecting the accuracy of error estimation of electric energy meters, opens up a new way for data acquisition and processing of large-scale smart meters, and further improves the service level of information management of power systems.

Each local feature in the appearance image of cigarette packs is a key element to reflect the corresponding brand information. If only a single convolutional neural network is used, the context information of sequence data may be lost, resulting in an insufficient grasp of the overall information. In order to realize the deep-level feature extraction of the appearance of cigarette packs and realize the appearance detection of cigarette packs with higher accuracy and speed, a new method for the appearance detection of cigarette packs was proposed by combining the convolutional neural network and the cyclic neural network methods in the deep learning algorithm. The image acquisition card is used to collect the appearance images of cigarette packs, and contrast enhancement and rotation correction are performed on the collected images to effectively improve their quality and provide a good guarantee for subsequent feature extraction and detection. The preprocessed cigarette package image is input into the convolutional neural network to realize the deep-level feature extraction of the cigarette package appearance image. The cigarette package appearance features’ output from the convolutional neural network is input into the short-term and long-term memory unit, as well as the gate-controlled cyclic unit, of the corresponding recurrent neural network, in order to process time sequence information based on the efficient extraction of details from the image, retain the sequence and context information of the input data, and ultimately achieve accurate detection of the cigarette package appearance. Through experimental analysis, this method can effectively identify the appearance defects of cigarette packs, mark them immediately, and present them in an intuitive way, so that staff can quickly locate the problem and take corresponding measures. The method can detect appearance defects larger than $1.59\mathrm{\text{mm}}\times 1.59\mathrm{\text{mm}}$ with high accuracy. For various appearance defects, the detection rate can be guaranteed to be over 98%, providing strong support for quality control and product upgrading in the tobacco industry.

A ground-breaking solution that combines solar thermal energy and lithium-bromide vapor absorption technology to produce energy-efficient cooling and heating is the Intelligent Solar Assist 1kW Lithium Bromide Vapor Absorption system. This cutting-edge system uses the sun’s energy to power the absorption cycle, offering environmentally friendly and economically viable thermal management. Solar thermal collectors, a lithium bromide absorption chiller, a thermal energy storage device, and sophisticated control algorithms comprise the system’s main parts. Sunlight is captured and converted by solar thermal collectors into thermal energy, which is then used to produce the necessary heat for the lithium bromide absorption chiller. This chiller uses the absorption refrigeration cycle to efficiently cool or heat the specified area or process. When intelligent control algorithms are incorporated, the system performs and operates more effectively and efficiently. These algorithms regulate the thermal energy storage unit and optimize the use of solar energy, delivering a constant and dependable supply of cooling or heating as needed. Advanced monitoring and diagnostics features are also built into the system, allowing for remote control and in-the-moment performance evaluation. Disadvantages are ethical issues, lack of generalization, interpretability and complexity, scalability and processing resources, and scientific agreement. A novel Chimp-based recurrent model (CbRM) has been planned to be designed to predict the desired efficiency from the Evacuated Tube Collector (ETC) to overcome this issue. Comparing the Intelligent Solar Assist system to conventional heating and cooling systems, several benefits must be had. It minimizes greenhouse gas emissions, lessens reliance on traditional energy sources, and promotes a more sustainable future. The system also saves money using solar energy, lowering power costs and enhancing energy efficiency. Moreover, the proposed system implementation is done in Matlab. The method achieves the high efficiency of ETC in the range of about 0.9% which increases by 0.3% and the higher rate of COP was about 9.5% which increases up to 6%, as the increased concentration level of the strong solution was about 6.5% it was nearly 5% increase. The parameters in the suggested model were compared to the current parameters for the comparison analysis, and it was discovered that the proposed model had superior presenting efficiency.

In the field of structural dynamic differential equations, although traditional numerical solution methods are mature, they generally have problems with high computational costs and numerous constraints. These constraints include whether the system is linear, sensitivity to time step size, limited ability to handle high-frequency responses, and limited applicability to nonlinear systems and complex dynamic environments. Therefore, this paper attempts to use a recurrent neural network (RNNs) to solve dynamic differential equations. Unlike intelligent models that require a large amount of sample data, this solver does not need samples. Instead, it constrains the loss function using the physics-informed neural network method and provides high-precision predictive solutions based on input parameters and memory parameters. The solver is applied to single-degree-of-freedom and multi-degree-of-freedom systems, including undamped and damped free vibrations as well as undamped and damped harmonic load problems. The feasibility of the model is validated based on the average relative error, with results showing that the average relative error between the model predictions and theoretical values is $10^{-4}$, with a few cases reaching 1%. The relationship between network structure, learning rate, and model performance is also explored. Additionally, common activation functions perform poorly when applied to this problem, so this paper constructs a new activation function, verifying its effectiveness and efficiency through the solver and analyzing the factors influencing the model’s performance.

A three-dimensional model of the ankle system was constructed based on CT scan image data with MIMICS software. This model was refined through surface smoothing and fitting operations in Geomagic software. Finite element method (FEM) was applied to simulate the forces acting on the ankle system during standing in order to generate plantar pressure nephogram. FEM results were compared with the experimental results, which illustrated a strong correlation regarding pressure peaks, thus indicating the effectiveness of the proposed model. Moreover, by changing the interaction force between the ankle system and the ground at different angles during the gait cycle, the FEM was utilized to obtain the curve of peak plantar pressure throughout the gait cycle in order to provide data for the training and testing of the AI algorithm model. The Long Short-Term Memory (LSTM) and Recurrent Neural Network (RNN) models were developed to predict peak plantar pressure during the gait cycle. After training, the mean squared error (MSE) of the LSTM model converged to $2.2518\times 10^{-4}$, and the coefficient of determination ($R^2)$ on the test set reached 99.47%. Meanwhile, the MSE of the RNN model converged to $2.4245\times 10^{-4}$, and the $R^2$ on the test set reached 99.20%. Additionally, the prediction results were compared with experimental data to evaluate the validity of both models. The prediction accuracy of the two algorithms was evaluated based on the $R^2$value. The results show that both models’ predictions effectively reflect variations in peak plantar pressure, with the LSTM model proving to be more accurate for predicting peak plantar pressure throughout the gait cycle. Moreover, when taking the overall computation time into account, the RNN model demonstrates significantly higher efficiency in general.

Speech Emotion Recognition (SER) stands as a crucial field within Human–Computer Interaction (HCI), drawing significant attention from researchers because of its broad applications. The ability to recognize emotions conveyed through speech is essential for effective two-way communication, making it a pivotal aspect in various domains. Several existing researches have addressed certain drawbacks such as dataset issues, lack of computational efficiency, interpretability, and so on. To overcome the issues and to perform the efficient SER, a combined Triplet and Single-headed Attention-enabled Recurrent Long Gated Network (TriSiH-RLG2) model is proposed in the research, which is designed through the integration of triplet attention, single-headed attention, and recurrent networks that includes Recurrent Neural Network (RNN), Long Short-term Memory (LSTM), and Gated Recurrent Unit (GRU), which eliminates the drawbacks of each model and enhances the overall efficacy of the research. Further to elevate the accuracy of the research, the Gradient Boosting Machine (GBM) is utilized as the classifier that performs the efficient SER through the ensemble learning approach. The contributed TriSiH attention is the incorporation of the two standard attention mechanisms, which act as the spotlight to highlight the relevant features from the input, thus making the research model highly proficient. The experimentation is done using RAVDESS emotional speech dataset and the performance of the provided model is revealed through metrics such as Precision, Recall, Accuracy, and $F$1-score that achieve 94.338%, 98.344%, 96.875%, and 96.299%, respectively.

This paper defines the truncated normalized max product operation for the transformation ofstates of a network and provides a method for solving a set of equations based on this operation. The operation serves as the transformation for the set of fully connected units in a recurrent network that otherwise might consist of linear threshold units. Component values of the state vector and ouputs of the units take on the values in the set {0, 0.1, …, 0.9, 1}. The result is a much larger state space given a particular number of units and size of connection matrix than for a network based on threshold units. Since the operation defined here can form the basis of transformations in a recurrent network with a finite number of states, fixed points or cycles are possible and the network based on this operation for transformations can be used as an associative memory or pattern classifier with fixed points taking on the role of prototypes. Discrete fully recurrent networks have proven themselves to be very useful as associative memories and as classifiers. However they are often based on units that have binary states. The effect of this is that the data to be processed consisting of vectors in ℜⁿ have to be converted to vectors in {0, 1}^m with m much larger than n since binary encoding based on positional notation is not feasible. This implies a large increase in the number of components. The effect can be lessened by allowing more states for each unit in our network. The network proposed demonstrates those properties that are desirable in an associative memory very well as the simulations show.

This paper proposes a non-recurrent training algorithm, resilient propagation, for the Simultaneous Recurrent Neural network operating in relaxation-mode for computing high quality solutions of static optimization problems. Implementation details related to adaptation of the recurrent neural network weights through the non-recurrent training algorithm, resilient backpropagation, are formulated throughan algebraic approach. Performance of the proposed neuro-optimizer on a well-known static combinatorial optimization problem, the Traveling Salesman Problem, is evaluated on the basis of computational complexity measures and, subsequently, compared to performance of the Simultaneous Recurrent Neural network trained with the standard backpropagation, and recurrent backpropagation for the same static optimization problem. Simulation results indicate that the Simultaneous Recurrent Neural network trained with the resilient backpropagation algorithm is able to locate superior quality solutions through comparable amount of computational effort for the Traveling Salesman Problem.

Connections play a crucial role in neural network (NN) learning because they determine how information flows in NNs. Suitable connection mechanisms may extensively enlarge the learning capability and reduce the negative effect of gradient problems. In this paper, a new delay connection is proposed for Long Short-Term Memory (LSTM) unit to develop a more sophisticated recurrent unit, called Delay Connected LSTM (DCLSTM). The proposed delay connection brings two main merits to DCLSTM with introducing no extra parameters. First, it allows the output of the DCLSTM unit to maintain LSTM, which is absent in the LSTM unit. Second, the proposed delay connection helps to bridge the error signals to previous time steps and allows it to be back-propagated across several layers without vanishing too quickly. To evaluate the performance of the proposed delay connections, the DCLSTM model with and without peephole connections was compared with four state-of-the-art recurrent model on two sequence classification tasks. DCLSTM model outperformed the other models with higher accuracy and F1$\text{\_}$score. Furthermore, the networks with multiple stacked DCLSTM layers and the standard LSTM layer were evaluated on Penn Treebank (PTB) language modeling. The DCLSTM model achieved lower perplexity (PPL)/bit-per-character (BPC) than the standard LSTM model. The experiments demonstrate that the learning of the DCLSTM models is more stable and efficient.

In this paper, an autonomous ensemble approach of improved quantum inspired gravitational search algorithm (IQI-GSA) and hybrid deep neural networks (HDNN) is proposed for the optimization of computational problems. The IQI-GSA is a combinational variant of gravitational search algorithm (GSA) and quantum computing (QC). The improved variant enhances the diversity of mass collection for retaining the stochastic attributes and handling the local trapping of mass agents. Further, the hybrid deep neural network encompasses the convolutional and recurrent neural networks (HDCR-NN) which analyze the relational & temporal dependencies among the different computational components for optimization. The proposed ensemble approach is evaluated for the application of facial expression recognition by experimentation on Karolinska Directed Emotional Faces (KDEF) and Japanese Female Facial Expression (JAFFE) datasets. The experimentation evaluations evidently exhibit the outperformed recognition rate of the proposed ensemble approach in comparison with state-of-the-art techniques.

Time series data can be used to predict the dynamical behaviors without knowing equation model of a system. In this study, long-short term memory (LSTM) neural network is implemented to construct a complex dynamical system from data series. The network is trained through minimizing the loss function to obtain the optimal weight matrices of LSTM cells. We find that the LSTM network can well ”learn” the information of the complex system. The data series generated from periodic orbits of a nonlinear system can be exactly predicted by comparing the output of neural networks with the real complex system. For the chaotic data series, the time evolution of trajectories could exactly match the actual system in the short-term data. Moreover, the long-term ergodic behavior of the complex system remains in our prediction, although such chaotic data series are quite sensitive to the initial conditions and the ensuing increase in uncertainty.

For click-through rate (CTR) prediction tasks, a good prediction performance can be obtained by full explorations of both user behavior and item behavior. Since user’s interests have a great influence on user’s behaviors, it is very important to learn users’ intrinsic interests according to their behaviors. User interests are not only diverse but also in dynamic change. However, the dynamics of user interests’ change are not fully taken into account by the majority of current CTR models. The latest sequential recommendation algorithm ignores the subjectivity of users when it uses a two-layer recurrent neural network to model the item behavior from the perspective of the evolution of items. In this work, we propose a recurrent neural network model called DTIAN (Deep Time-Aware Interest Attention Network) to address these issues. By leveraging the user behaviors and the corresponding temporal information, DTIAN captures user interests and intent changes to the target item. Therefore, the users’ recent interests are enhanced compared to early interests with the attention mechanism. In addition, each module of the proposed model can be plugged into other mainstream models to improve the performance of current models. The experimental results show that the proposed DTIAN can outperform the current popular CTR prediction models slightly and significantly reduce the training time, which makes it possible to implement lightweight models.

A new car-following model is proposed based on recurrent neural network (RNN) to effectively describe the state change and road traffic congestion while the vehicle is moving. The model firstly gives a full velocity difference car-following model according to the driver’s reaction sensitivity and relative velocity, and then takes the vehicle position and velocity as the input parameters to optimize the safe distance between the front and rear vehicles in the car-following model based on RNN model. Finally, the effectiveness of the above model is validated by building a simulation experiment platform, and an in-depth analysis is conducted on the relationship among influencing factors, e.g., relative velocity, reaction sensitivity, headway, etc. The results reveal that, compared with traditional car-following models, the model can quickly analyze the relationship between initial position and velocity of the vehicle in a shorter time and thus obtain a smaller safe distance. In the case of small velocity difference between the front and rear vehicles, the running velocity of the front and rear vehicles is relatively stable, which is conducive to maintaining the headway.

Electrical engineering models often rely on complex circuit configurations that facilitate the dynamic flow of electrically charged particles within a closed conductive network. These circuits serve as essential tools for simulating and analyzing diverse electrical systems and components. This paper introduces a study on nonlinear fractional circuits modeling through the development of a stochastic neuro-computational artificial intelligent-based solver to address mathematical models governing the Fractional order Caputo–Fabrizio stiff electric circuit model (FO-CFSECM) by manipulating the knacks of layered recurrent neural networks (LRNNs) trained with Gradient-based local search algorithm (GLA). In fractional calculus, the Caputo–Fabrizio (CF) fractional order derivative (FOD) emerges as a powerful instrument, binding its capabilities to deliver remarkably accurate solutions for fractional stiff systems. The objective of this work is to exploit the numerical treatment comprehensively for the dynamics of fractal Resistor–Capacitor (RC) and fractal Resistor–Inductor (RL) circuit models by introducing the CF fractional operator. Through the application of artificial intelligence-based soft computing and advanced back-propagative deep neural networks, a deeper understanding of the behavior and distinctive characteristics inherent in these models is sought. The Levenberg–Marquardt optimizer serves as an efficient training GLA tool for learning of LRNNs weights of fractal RL/RC circuit models. The comparative studies on variants of FO-CFSECM demonstrate that LRNNs achieve an impressive mean square error (MSE) ranging from ${10}^{-9}$ to ${10}^{-19}$ and absolute error (AE) within ${10}^{-6}$ to ${10}^{-8}$. The accuracy, reliability, and efficiency of LRNNs for solving the FO-CFSECM were further validated through MSE, AE, controlling parameters of state transitions, error histograms, and correlation measures.

The main contribution of this paper is the development of an Integer Recurrent Artificial Neural Network (IRANN) for classification of feature vectors. The network consists both of threshold units or perceptrons and of counters, which are non-threshold units with binary input and integer output. Input and output of the network consists of vectors of natural numbers that may be used to represent feature vectors. For classification purposes, representatives of sets are stored by calculating a connection matrix such that all the elements in a training set are attracted to members of the same training set. The class of its attractor then classifies an arbitrary element if the attractor is a member of one of the original training sets. The network is successfully applied to the classification of sugar diabetes data, credit application data, and the iris data set.

Aiming at the “bottleneck” problems of the traditional network security situation awareness model, such as large equipment limitations, single data source and poor integration ability, weak level of autonomous learning and data mining, a network security situation awareness framework suitable for big data is constructed. A gate recurrent unit (GRU) model is established to effectively extract features from the situation data set through the deep learning algorithm of big data. It is a method to automatically mine and analyze the hidden relationship and change trend of network security situation, realize the high-speed acquisition and fusion of massive multi-source heterogeneous data, and perceive the network security situation from an all-round perspective. The experimental results show that this method has a good awareness effect on network threats, and has strong representation ability in the face of network threats. It can effectively perceive the network threat situation without relying on data labels, which verifies that this method can effectively improve the efficiency and accuracy of security situation awareness.

It is important to locate the fault distance and identify the fault types quickly, take effective measures to maintain line stability, and minimize the losses timely when there are short-circuit faults in transmission lines. For this purpose, a method based on deep learning is proposed for short-circuit faults identification in the transmission line. According to the similarity of samples in the reconstruction phase, a minimum neighborhood sample set is selected from the massive samples firstly, and then, the samples are trained using the back propagation algorithm along time in a recurrent neural network (RNN) with long-short term memory (LSTM) units. Compared with existing algorithms, the experimental results show that this algorithm meets the requirements of rapid fault diagnosis in the case of variable parameters, and higher fault type recognition accuracy and lower fault distance error can be obtained.

This paper proposes a novel hybrid embedding to enhance scope of word embeddings by augmenting these with natural language processing operations. We primarily focus on the proposal of new hybrid word embedding generated by augmenting BERT embedding vectors with polarity score. The paper further proposes a new deep learning architecture inspired by the use of convolutional neural network for feature extraction and a bidirectional recurrent network for contextual and temporal feature exploitation. Use of CNN with hybrid embedding allowed the network to extract even the higher-level styles in writing, while bidirectional RNN helped in understanding context. The paper justifies that the proposed architecture and hybrid embedding improves performance of sentiment classification system by performing a large number of experiments and testing on a number of deep learning architectures. The architecture on new hybrid embeddings incurred an accuracy of 96%, which is a significant improvement when compared with recent studies in the literature.

In this paper, we construct a deep framework for full-reference image quality assessment (FR-IQA) by combining convolution and self-attention features; this approach effectively uses multiscale features to mimic the image evaluation in human eyes. We achieve the integration of information from local to global. First, convolutional neural network (CNN) and encoder of Swin transformer are used to extract multiscale paired features. Second, for each type of features, we fuse them after converting them to a fixed number of channels. For the fused features, we use the square of the difference between two pixel values at the corresponding channel positions to represent the features of the distortion degree. Then, we introduce a recurrent neural network (RNN) to capture the global features. Finally, for the two types of features, we use the full connection (FC) layer to regress two scores and then add the weights to compute the ultimate perceived score. By training and testing on publicly available FR-IQA datasets, experimental results further validate the superiority of our approach.

Exponential stability of reaction–diffusion fuzzy recurrent neural networks (RDFRNNs) with time-varying delays are considered. By using the method of variational parameters, M-matrix properties and inequality technique, some delay-independent or delay-dependent sufficient conditions for guaranteeing the exponential stability of an equilibrium solution are obtained. One example is given to demonstrate the theoretical results.