Layer Nodes Research Articles

Stall Warning Algorithm of Axial Compressor Based on SSA-DBN

<p>To solve the problem of stall warning for axial compressors, this paper proposes a stall warning algorithm based on the sparrow search algorithm, which optimizes the deep belief network. The deep belief network is trained by using deep learning to extract the FFT spectrum of compressor stall experiment data directly as the feature vector. To improve the accuracy of DBN classification, parameters of hidden layer nodes n and initial weights w of DBN were optimized by SSA algorithm, and stall warning algorithm model of SSA-DBN axial-flow compressor was established. The experimental results of the algorithms show that the stall data at each speed is at least 0.1～0.3s in advance for early warning. This method is 0.075~0.281s ahead of the various models from the past to the present, and noise is superimposed on the experimental data to verify the Robustness of the way, better surge warning margin performance, and engineering practicability. </p> <p> </p>

Pfizer Stock Price Prediction Based on Extreme Learning Machine

Nowadays, an Extreme learning machine (ELM) is known to be a fast learning algorithm of single-hidden layer feedforward neural network (SLFNs), and overcomes the disadvantages of the classical learning algorithm in neural network methods multiple iterations, huge search space and a large number of calculations, only needs to set the appropriate numbers of hidden layer nodes, assigns the weight of input and deviation of hidden layers without iteration. Research shows that the stock market is a very complex nonlinear system, which requires artificial intelligence theory, statistics theory and economic theory to study the stock price forecast. In this paper, ELM is introduced in predicting the stock price of Pfizer company, and by comparing it with SVM and BP, we analyze its feasibility and advantage in stock price prediction. The experiment results show that ELM is of high accuracy of prediction and apparent advantages in parameter selection and learning speed．

Efficient Learning of Quadratic Variance Function Directed Acyclic Graphs via Topological Layers

Directed acyclic graph (DAG) models are widely used to represent casual relationships among random variables in many application domains. This article studies a special class of non-Gaussian DAG models, where the conditional variance of each node given its parents is a quadratic function of its conditional mean. Such a class of non-Gaussian DAG models are fairly flexible and admit many popular distributions as special cases, including Poisson, Binomial, Geometric, Exponential, and Gamma. To facilitate learning, we introduce a novel concept of topological layers, and develop an efficient DAG learning algorithm. It first reconstructs the topological layers in a hierarchical fashion and then recovers the directed edges between nodes in different layers, which requires much less computational cost than most existing algorithms in literature. Its advantage is also demonstrated in a number of simulated examples, as well as its applications to two real-life datasets, including an NBA player statistics data and a cosmetic sales data collected by Alibaba. Supplementary materials for this article are available online.

Modeling the Mechanical Properties of Heat-Treated Mg-Zn-RE-Zr-Ca-Sr Alloys with the Artificial Neural Network and the Regression Model

In this study, an artificial neural network approach and a regression model are adopted to predict the mechanical properties of heat-treated Mg-Zn-RE-Zr-Ca-Sr magnesium alloys. The dataset for artificial neural network (ANN) modeling is generated by investigating the microhardness of heat-treated Mg-Zn-RE-Zr-Ca-Sr alloys using Vickers hardness tests. A back-propagation (BP) neural network is established using experimental data that enable the prediction of mechanical properties as a function of the composition and heat treatment process. The input variables for the BP network model are Ca and Sr contents, ageing temperature and ageing time. The output variable corresponds to the microhardness. The optimal BP network model is acquired by optimizing the number of the hidden layer nodes. The results indicate that a reliable correlation coefficient is above 0.95 for architecture (4-8-1), which has a high level of accuracy for prediction. In addition, a second-order polynomial regression model is developed based on the least squares method. The results of determination coefficients and Fisher’s criterion indicate that the regression model is capable of modeling mechanical properties as a function of composition and the ageing process. Furthermore, supplemental experiments are conducted to check the accuracy of the BP model and the regression model, suggesting that the model predictions are well in accordance with experimental results. Therefore, both the BP network and regression models have high accuracy in modeling and predicting mechanical properties of heat-treated Mg-Zn-RE-Zr-Ca-Sr alloys.

Piano Information Teaching Mode Based on Deep Learning Algorithm

In order to improve the effect of piano information teaching, a piano information teaching mode based on deep learning algorithm is proposed. The teaching objectives are divided into three levels: classroom teaching objectives, curriculum objectives, and education and training objectives. A piano information classroom integrating cloud application platform, teaching platform, resource platform, learning space, and interactive classroom is built. The previous teaching mode is optimized to build an innovative teaching mode of piano information classroom. The evaluation index system of piano informatization classroom teaching quality is constructed, and the hierarchical structure model of each evaluation index is established by using the analytic hierarchy process. The hierarchical analysis method is used to establish a hierarchical structure model of each evaluation index. The judgment matrix is determined by the nine-digit scale method. After the consistency verification of the judgment matrix, the weight of the quality evaluation of piano information classroom teaching is calculated. The new mode optimizes the weight and threshold of BP neural network in deep learning algorithm by genetic algorithm (GA). The weight of each classroom teaching quality evaluation index is input into the GA-BP neural network, and the network output result is the piano information classroom teaching quality evaluation score. The test results show that the optimal number of hidden layer nodes for the BP neural network is 7, when the GA-BP neural network iterations are 95. This method can evaluate the quality of piano information classroom teaching, with high evaluation accuracy and strong practical application.

Mechanisms of Scaling Effect for Emerging Nanoscale Interconnect Materials.

The resistivity of Cu interconnects increases rapidly with continuously scaling down due to scatterings, causing a major challenge for future nodes in M0 and M1 layers. Here, A Boltzmann-transport-equation-based Monte Carlo simulator, including all the major scattering mechanisms of interconnects, is developed for the evaluation of electron transport behaviors. Good agreements between our simulation and the experimental results are achieved for Cu, Ru, Co, and W, from bulk down to 10 nm interconnects. The line resistance values of the four materials with the inclusion of liner and barrier thicknesses are calculated in the same footprint for a fair comparison. The impact of high aspect ratio on resistivity is analyzed for promising buried power rail materials, such as Ru and W. Our results show that grain boundary scattering plays the most important role in nano-scale interconnects, followed by surface roughness and plasma excimer scattering. Surface roughness scattering is the origin of the resistivity decrease for high-aspect-ratio conductive rails. In addition, the grain sizes for the technical nodes of different materials are extracted and the impact of grain size on resistivity is analyzed.

Numerical Simulation of Erosion Characteristics and Residual Life Prediction of Defective Pipelines Based on Extreme Learning Machine

Aiming to solve the problem that the residual life of defective elbows is difficult to predict and the prediction accuracy of a traditional extreme learning machine (ELM) is unsatisfactory, a genetic algorithm optimization neural network extreme learning machine method (GA-ELM) that can effectively predict erosion rate and residual life is proposed. In this method, the input weight and hidden layer node threshold of the hidden layer node is mapped to GA, and the input weight and threshold of the ELM network error is selected by GA, which improves the generalization performance of the ELM. Firstly, the effects of solid particle velocity, particle size, and mass flow rate on the erosion of elbow are studied, and the erosion rates under the conditions of point erosion defect, groove defect, and double groove erosion defect are calculated. On this basis, the optimized GA-ELM network model is used to predict the residual life of the pipelines and then compared with the traditional ELM network model. The results show that the maximum erosion rate of defect free elbow is linearly correlated with solid particle velocity, particle size, and mass flow rate; The maximum erosion rate of defective bend is higher than that of nondefective bends, and the maximum erosion rate of defective bend is linearly related to mass flow rate, but nonlinear to solid particle flow rate and particle size; the GA-ELM model can effectively predict the erosion residual life of a defective elbow. The prediction accuracy and generalization ability of the GA-ELM model are better than those of the traditional ELM model.

The Application of Artificial Intelligence Technology in the Asset Management of Start-Ups in the Context of Deep Learning

With the coninuous improvement and development of artificial intelligence (AI) technology, this technology has been used in the asset management of companies. To improve the asset management level of Chinese start-ups, firstly, back-propagation neural network (BPNN) has been studied in depth, and an evaluation system of the company's asset quality has been established. Secondly, the BPNN is integrated with the evaluation indicators of asset quality, and an evaluation model of asset quality based on BPNN is constructed. Next, start-up A is taken as the experimental object; the evaluation score of the asset quality of A company is input into the model, which proves that there is still a certain gap between the asset management level of start-ups and mature companies. Finally, to find out the problems of the company's asset quality, the traditional financial analysis method is used to carry out a specific microanalysis of the evaluation indicators of its asset quality. In view of the existing problems, suggestions are put forward for prudent investment, improve inventory operation efficiency, increase investment in R&D and innovation, improve the quality of sales outlets, and increase the proportion of high-quality intangible assets. The asset quality evaluation system for start-ups established here includes 19 evaluation indicators. The BPNN-based asset quality evaluation model selects 5 mature companies in the same industry as sample companies. The scores of the evaluation indicators of asset quality of the 5 sample companies in the past three years are normalized and input into the model. The model contains 19 nodes of the input layer, 39 nodes of the hidden layer, and 1 node of the output layer. The target error rate is 0.001, the learning rate is 0.1, the number of training times is 1000, and the training function is the trainlm function. This research has a certain reference for the application of AI technology in the asset management of start-ups.

Using Improved BPNN to Predict the Time of Social Poverty Reduction under the Background of Internet

The world has been overturned by the power of the Internet, and all walks of life are constantly changing with the trend of the times. Therefore, with the help of the effective technology and means of the Internet, people can start from a more complex perspective. Analyze the social environment and economic situation, and focus on poverty. By actively monitoring the number of poor people in society, we can grasp the actual situation of poverty. Starting from the depths of the problem, aiming at the characteristics of poor groups, this paper excavates the scale data of poor people. Finally, the improved BP neural network model is used to predict the time when the society gets rid of poverty and reduces poverty. The results show the following: (1) After P value judgment and T value test, the experiment determined 12 indicators and proportion of poverty. The most prominent indicators of contribution rate are T2, T4, and T5. The most significant changes in T test results were T8 and T12. (2) The reformed BP network has less iteration times, shorter training time, and higher prediction accuracy. The network converges 30 times, and the fitting accuracy reaches the best, and the ideal state is 0.000936578. Training the poverty alleviation data from 2016 to 2020 with the model, the final poverty alleviation work has achieved remarkable results. (3) According to the hidden layer nodes of neural network structure, the final number of hidden layer nodes is determined to be 11. (4) Comparing the absolute error and APE error of the prediction results. The predicted value is very close to the actual value, which can correctly reflect the mapping relationship of poverty reduction and the result is reliable. (5) Compare the predicted values of BPNN model before and after modification; it is found that the new BPNN model is more accurate in prediction. The predicted population is close to the actual population, and the error level curve approaches 0%. After empirical test, the improved BPNN model is accurate and effective. Through the poverty reduction prediction of the model, people can observe the actual situation more intuitively. By evaluating the effectiveness of social poverty alleviation, we can optimize the next poverty alleviation strategy and adjust the strategic policy.

An efficient resource allocation strategy for three-layer networks

Ranging from computer network to traffic network, complex networks are ubiquitous in our lives. How to control congestion to make network operate more effectively is one of the most essential issues in complex network research. At present, a large number of studies on traffic dynamics mainly focus on single-layer networks. However, in fact, many complex systems are coupled by multiple networks. In this paper, we propose a resource allocation strategy for three-layer network with which the limited total delivery capacity can be reasonably allocated to each node of physical layer based on the degree of nodes in two logical layers. Compared with the average allocation of delivery capacity, the traffic capacity is improved and the network has better transmission performance in average traveling time and average throughput with our strategy.

Quantitative Inversion Modeling Method for Grading Deerni Copper Deposits Based on Visible and Near-Infrared Hyperspectral Data

Quantitative metal grade inversion based on hyperspectral data is an effective approach to achieve the real-time in situ determination of ore body grades and has the advantages of low cost compared with traditional chemical analysis methods. However, the redundant nature of hyperspectral data and the parameter-limiting nature of machine learning algorithms reduce the modeling accuracy and precision, resulting in severe limitations on the application of hyperspectral techniques for the grade inversion of Deerni copper ore bodies. In this paper, we first obtained visible-NIR hyperspectral data for 190 ore samples using a spectrometer and determined the copper content of the sample set using chemical analysis; then, we processed the raw hyperspectral data using three dimensionality reduction algorithms and optimized a BP neural network based on an evolutionary algorithm. Finally, a Deerni copper grade inversion model was established using the hyperspectral data before and after dimensionality reduction, and the inversion accuracy and precision was compared and analyzed with that obtained by the BP neural network, the random forest and the variable hidden layer nodes models. The combination of the LLE dimensionality reduction algorithm and the optimized BP neural network algorithm achieves the highest modeling precision, with an R 2 of 0.950.

An Ensemble-Learning Approach To Predict the Coke Yield of Commercial FCC Unit

This work proposes an ensemble learning-based catalytic cracking coke yield prediction model called the harmonic-ensembled extreme learning machine (HEELM). The model integrates extreme learning machine (ELM) base learners with different activation functions to improve the overall prediction effect. An overfitting index is proposed, and the optimal number of hidden layer nodes of ELM-base learners is determined with it. By examining the influence of different activation functions on the prediction results, the best activation function structure of the ELM-base learner has been determined. Besides, a harmonic layer is established to determine the weight of each base learner in real-time. The proposed model is validated using 1.5 years of historical data from China’s commercial fluidized catalytic cracking (FCC) plant. The results show that the proposed model has outperformed most other ELM-base learners. The relative prediction error is further reduced by 10.97% after introducing the harmonic layer. The proposed model exhibits stable performance with good generalization in three segments of industrial data, and it has guiding significance for stable operation and CO2 emission reduction of the FCC plant.

Web Service Recommendation Based on QoS Analysis

Abstract To correctly recommend the web services and help users choose more appropriate web services, a QoS analysis and recommendation method based on radial basis function (RBF) neural network is proposed (this method is referred to as WSR-RBF). This method constructs an optimized RBF neural network to analyze the correlation between qualities of service attributes and uses a set of irrelevant attributes to accurately predict quality of service. To find the number of hidden layer nodes in our proposed model and determine the center value of the hidden layer in a short time, we propose a clustering algorithm combining Affinity Propagation and K-means. The experimental results show that the proposed method has low error values on three data sets, which verifies its good recommendation effect.

Fast stochastic configuration network based on an improved sparrow search algorithm for fire flame recognition

Flame image recognition is of great significance in the fire detection and prevention. In this paper, in order to improve the accuracy of fire recognition, a fast stochastic configuration network (FSCN) method based on an improved sparrow search algorithm (ISSA) is proposed. In the design of fast stochastic configuration network (FSCN), the gradual increase of hidden layer nodes in the original stochastic configuration network (SCN) is canceled, and the number of them is set directly. An improved sparrow search algorithm (ISSA) is used to generate the input weights and biases of hidden layer nodes. At the same time, the supervisory mechanism is retained to judge the weights and biases of all hidden layer nodes, and ISSA is used to regenerate corresponding weights and biases for the nodes that do not meet the constraints in the supervisory mechanism. In the ISSA, sine map, adaptive adjustment of hyper-parameters and mutation strategy are used to improve the optimization ability of the original sparrow search algorithm (SSA). Some parameters in FSCN are optimized by ISSA to make it have better classification performance. Finally, the image processing technology is used to extract features from the flame images and the interference images, and then the feature vectors are obtained to train the ISSA-FSCN. Several simulation experiments have been carried out to verify effectiveness of the proposed ISSA-FSCN method. In the performance verification of ISSA on CEC test suit, ISSA averagely outperforms other algorithms by 33.6% in the average results of 20 functions. In the performance verification of FSCN, the average results of accuracy, precision, recall, F1 and auc are compared. In the experiment 1, ISSA-FSCN averagely outperforms other algorithms by 19.7%, 14.7%, 12.8%, 14.5% and 23.0%. In the experiment 2, ISSA-FSCN averagely outperforms other algorithms by 2.3%, 2.1%, 2.5%, 6.0%, 3.2%. In the experiment 3, ISSA-FSCN averagely outperforms other algorithms by 5.9%, 4.0%, 4.1%, 4.1%, 6.8%.

Deep convolutional neural network-based signal quality assessment for photoplethysmogram

Quality assessment of bio-signals is important to prevent clinical misdiagnosis. With the introduction of mobile and wearable health care, it is becoming increasingly important to distinguish available signals from noise. The goal of this study was to develop a signal quality assessment technology for photoplethysmogram (PPG) widely used in wearable healthcare. In this study, we developed and verified a deep neural network (DNN)-based signal quality assessment model using about 1.6 million 5-s segment length PPG big data of about 29 GB from the MIMIC III PPG waveform database. The DNN model was implemented through a 1D convolutional neural network (CNN). The number of CNN layers, number of fully connected nodes, dropout rate, batch size, and learning rate of the model were optimized through Bayesian optimization. As a result, 6 CNN layers, 1,546 fully connected layer nodes, 825 batch size, 0.2 dropout rate, and 0.002 learning rate were needed for an optimal model. Performance metrics of the result of classifying waveform quality into ‘Good’ and ‘Bad’, the accuracy, specificity, sensitivity, area under the receiver operating curve, and area under the precision–recall curve were 0.978, 0.948, 0.993, 0.985, 0.980, and 0.969, respectively. Additionally, in the case of simulated real-time application, it was confirmed that the proposed signal quality score tracked the decrease in pulse quality well.

Design of Fault Prediction System for Electromechanical Sensor Equipment Based on Deep Learning.

With the increasing complexity, scale, and intelligentization of modern equipment, the maintenance cost of equipment is increasing day by day. Moreover, once an unexpected major failure occurs, it will cause loss and damage to production, economy, and safety. Based on the considerations of system reliability and safety, fault prediction has gradually become a hot topic in the field of reliability. As a new branch of machine learning, deep learning realizes deep abstract feature extraction and expression of complex nonlinear relations by stacking deep neural networks and makes its methods solve bad problems in many traditional machine learning fields. The improvement and excellent results have been achieved. This article first introduces the model structure and working principle of the classic deep learning model noise reduction autoencoder and combines the feature extraction results of the experimental data of electromechanical sensor equipment and the model characteristics to analyze that this type of model can be trained using only normal samples. Under the restriction, the reason for abnormal features can also be correctly filtered. Then, in order to suppress the overfitting of training, the dropout layer is used in this article. The dropout layer will make the hidden layer nodes to be dropped with probability p according to the set probability. Because the lost nodes are random, each training is equivalent to training a new model. It achieves an effect similar to independently training the model and then superimposing it. Experiments prove that the dropout layer is very effective in solving overfitting. Finally, this paper conducts experimental verification on the improved algorithm. The results show that the improved model has a certain improvement in accuracy under the limited training algebra. Under the same training parameters, the accuracy is increased by approximately 2.44%, and the improved model has a better training effect and can be used for electromechanical effective prediction of sensor equipment failure.

Disaster propagation in interdependent networks with different link patterns.

Disaster propagation in complex, interdependent, and multilayered networks has attracted considerable research interest in recent years. In this paper, we propose a model that combines two dynamic mechanisms, i.e., the spreading of failure in layer-dependent networks, where each node in a layer depends on one in another layer. We first investigate the robustness of the Erdős-Rényi (ER)-ER, scale-free (sf)-ER, and sf-sf pattern of interdependent networks against cascading failure with different probabilities of triggering, and then use the random link, assortative link, and disassortative link patterns between the networks to analyze the scope of propagation of failure. The numerical results show that with increasing probability of triggering, the number of damaged nodes in both layers increased and the robustness of the scale-free network to random failures decreased due to the interdependence. Regardless of the topological structure, the two layers eventually tended to have similar failure characteristics due to their interdependence. In addition, the different link patterns had a significant effect on enhancing disaster propagation in interdependent networks.

Digital Computer Modulation Signal Classification Based on Neural Network

Computer is one of the indispensable tools in the human world, and human needs for it are increasing, so the emergence and application of more advanced computers are needed. The current computers do not respond intelligently, and it is difficult to meet people’s needs for information processing in the era of big data. In order to solve these problems, this paper proposes the application of a neural network-based data classification algorithm in computers, aiming to study the practical application of the algorithm in computers. The research method of this paper is to introduce the BP neural network, select the appropriate method of classification features, and then study the data classification algorithm. The function of the research method is to compare the classification error and convergence speed of the BP network composed of different hidden layer nodes, to study whether a certain feature item of the data exists and the difference in the amount of information classification of the entire document, and to select high efficiency, accuracy, and scalability algorithm. This paper compares the forward reasoning time of the model before and after cutting through experiments based on neural network model design, algorithm design, and man-machine dialogue model design. The results show that, in terms of computing speed, the adaptive model compression method based on the accuracy and redundancy ratio compresses the model after the forward reasoning time is greatly reduced, and the reasoning time becomes 35% of the original, and in terms of calculation accuracy, the absolute error after using the SOM method in this article has not reached 0.5.

Broad stochastic configuration network for regression

Stochastic configuration network (SCN) is an incremental learning approach with universal approximation property for analyzing high-dimensional and large-scale data. It can start with a small structure and gradually increase hidden layer nodes through the supervision mechanism. Broad learning system (BLS), as a novel structure, has been widely used in various fields because of its high efficiency, generalization performance and scalability. Unlike deep learning models, BLS is proposed based on random vector functional link network (RVFLN) which can map input features to a more appropriate feature space and the mapped features are enhanced as enhancement nodes. Therefore, to enhance the feature learning ability and improve the effectiveness of SCN, inspired by BLS architecture, a novel broad stochastic configuration network (BSCN) is proposed, then we give the structure and theory of BSCN. In this model, the original features are transformed into mapping features in the feature layer and the mapping features are enhanced in the enhancement layer, then the input weights and biases of enhancement nodes are determined according to the supervision mechanism and the output weight matrix can be calculated through the standard least squares. The experimental results on function approximation problems and real-world datasets indicate that BSCN improves the performance of SCN, and it has higher regression accuracy and stability.

Study on intelligent prediction method of rock drillability based on Bayesian lithology classification and optimized BP neural network

With the rapid development of shale gas in northeastern Sichuan province, problems such as low drilling rate, long drilling cycle and serious bit wear are becoming more and more prominent in shale gas drilling process. This paper proposes new prediction theory with two steps based on logging data sensitive to lithology by introducing Bayesian classifier and Self Adaptive Double Chain Quantum Genetic Algorithm (S-DC-QGA) which could optimize BP neural network prediction model. In the first step, the Bayesian network classifier achieved the classification and recognition of lithology, improved the correlation of model sample data, and enhanced the accuracy and speed of lithology classification. In the second step, S-DC-QGA characterized as the global optimization is used to optimize the number of hidden layer nodes, weights and thresholds of the BP neural network. Results showed that Bayesian lithologies classification enhanced the correlation of samples, moreover optimized BP neural network solved overfitting, random initial weights of BP itself and oscillation of fitting and generalization ability with slight change of network parameters, improving the prediction accuracy of rock drillability parameters and the generalization ability of the model. The findings of this study can help for better drilling design and improvement of mechanical drilling speed.