Nonlinear Network Research Articles

Prediction approach on pedestrian outdoor activity preference under factors of public open space integrated microclimate

This study conducted measurements of microclimates and investigations of spatial environments and human activities in six typical waterfront public spaces in Shanghai. The aim was to explore the synergistic influence mechanism of public open space (POS) integrated microclimate (POSIM) on pedestrians’ activities to predict the utilization of public space. The measurement period was during the spring and early summer when outdoor activities were more diversified. 121 observation points were set up in urban waterfront public spaces, and an outdoor activity characteristic database of 25,583 people was established. In addition, multiple linear regression and nonlinear neural network models were introduced to analyze data from different types of activities to calculate the fitness of the models and the influence weights of the variables. The results indicated that the neural network had stronger predictive ability for the spatial integration of microclimate demands of different activities. The prediction degree for strolling and sitting activities was the highest, with R2 (goodness of fit) of 0.704 and 0.844, respectively, while the prediction degree for viewing and sports activities was lower (R2 < 0.5). This study integrated the synergistic influence of urban waterfront public open spaces and microclimates factors on pedestrian outdoor activities to predict the preference patterns of different activities for spatial types. Focus on the requirements of space occupants, this study analyzed the public space environment from the bottom up and provided reference and inspiration for subsequent design optimization of urban waterfront areas.

Bidirectionally self-normalizing neural networks

The problem of vanishing and exploding gradients has been a long-standing obstacle that hinders the effective training of neural networks. Despite various tricks and techniques that have been employed to alleviate the problem in practice, there still lacks satisfactory theories or provable solutions. In this paper, we address the problem from the perspective of high-dimensional probability theory. We provide a rigorous result that shows, under mild conditions, how the vanishing/exploding gradients problem disappears with high probability if the neural networks have sufficient width. Our main idea is to constrain both forward and backward signal propagation in a nonlinear neural network through a new class of activation functions, namely Gaussian–Poincaré normalized functions, and orthogonal weight matrices. Experiments on both synthetic and real-world data validate our theory and confirm its effectiveness on very deep neural networks when applied in practice.

A persistent major mutation in canonical jasmonate signaling is embedded in an herbivory-elicited gene network

When insect herbivores attack plants, elicitors from oral secretions and regurgitants (OS) enter wounds during feeding, eliciting defense responses. These generally require plant jasmonate (JA) signaling, specifically, a jasmonoyl-L-isoleucine (JA-Ile) burst, for their activation and are well studied in the native tobacco Nicotiana attenuata. We used intraspecific diversity captured in a 26-parent MAGIC population planted in nature and an updated genome assembly to impute natural variation in the OS-elicited JA-Ile burst linked to a mutation in the JA-Ile biosynthetic gene NaJAR4. Experiments revealed that NaJAR4 variants were associated with higher fitness in the absence of herbivores but compromised foliar defenses, with two NaJAR homologues (4 and 6) complementing each other spatially and temporally. From decade-long seed collections of natural populations, we uncovered enzymatically inactive variants occurring at variable frequencies, consistent with a balancing selection regime maintaining variants. Integrative analyses of OS-induced transcriptomes and metabolomes of natural accessions revealed that NaJAR4 is embedded in a nonlinear complex gene coexpression network orchestrating responses to OS, which we tested by silencing four hub genes in two connected coexpressed networks and examining their OS-elicited metabolic responses. Lines silenced in two hub genes (NaGLR and NaFB67) co-occurring in the NaJAR4/6 module showed responses proportional to JA-Ile accumulations; two from an adjacent module (NaERF and NaFB61) had constitutively expressed defenses with high resistance. We infer that mutations with large fitness consequences can persist in natural populations due to compensatory responses from gene networks, which allow for diversification in conserved signaling pathways and are generally consistent with predictions of an omnigene model.

Transmission line planning using global best artificial bee colony method

Introduction. Network expansion, substation planning, generating expansion planning, and load forecasting are all aspects of modern power system planning. The aim of this work is to solve network planning considering both future demand and all equality and inequality constraints. The transmission network design problem for the 6-bus system is considered and addressed using the Global Best Artificial Bee Colony (GABC) method in this research. The program is written in the Matrix Laboratory in MATLAB environment using the proposed methodology. Novelty of the work consist in considering the behavior of bees to find food source in most optimized way in nature with feature of user based accuracy selection and speed of execution selection on any scale of the system to solve Transmission Lines Expansion Problem (TLEP). The proposed method is implemented on nonlinear mathematical function and TLEP function. When demand grows, the program output optimally distributes new links between new generation buses and old buses, determines the overall minimum cost of those links, and determines if those linkages should meet power system limits. Originality of the proposed method is that it eliminated the need of load shedding while planning the future demand with GABC method. Results are validated using load flow analysis in electrical transient analyzer program, demonstrating that artificial intelligence approaches are accurate and particularly effective in non-linear transmission network planning challenges. Practical value of the program is that it can use to execute cost oriented complex transmission planning decision.

Encrypted Model Predictive Control of a Nonlinear Chemical Process Network

This work focuses on developing and applying Encrypted Lyapunov-based Model Predictive Control (LMPC) in a nonlinear chemical process network for Ethylbenzene production. The network, governed by a nonlinear dynamic model, comprises two continuously stirred tank reactors that are connected in series and is simulated using Aspen Plus Dynamics. For enhancing system cybersecurity, the Paillier cryptosystem is employed for encryption–decryption operations in the communication channels between the sensor–controller and controller–actuator, establishing a secure network infrastructure. Cryptosystems generally require integer inputs, necessitating a quantization parameter d, for quantization of real-valued signals. We utilize the quantization parameter to quantize process measurements and control inputs before encryption. Through closed-loop simulations under the encrypted LMPC scheme, where the LMPC uses a first-principles nonlinear dynamical model, we examine the effect of the quantization parameter on the performance of the controller and the overall encryption to control the input calculation time. We illustrate that the impact of quantization can outweigh those of plant/model mismatch, showcasing this phenomenon through the implementation of a first-principles-based LMPC on an Aspen Plus Dynamics process model. Based on the findings, we propose a strategy to mitigate the quantization effect on controller performance while maintaining a manageable computational burden on the control input calculation time.

The extreme risk connectedness of the global financial system: G7 and BRICS evidence

Using daily money, stock, bond, foreign exchange, and credit markets data in the G7 and BRICS between 2006 and 2022, this paper investigates the extreme risk interconnectedness across countries and markets. Specifically, we propose a multilayer nonlinear extreme risk spillover network based on the CAViaR model and nonlinear Granger causality test to capture extreme risk spillovers across and within layers from static and dynamic perspectives, respectively. We find that the extreme risks of the G7 countries are higher than those of the BRICS countries. Simultaneously, extreme risks in the stock and foreign exchange markets are significantly higher than those in other markets. The stock market tends to be the net emitter of extreme risks, and the bond and credit markets tend to be the net recipients. During special event periods, BRICS countries (except Russia) tend to be net recipients of extreme risks. Our study provides new evidence on the interconnectedness of extreme risk across markets and countries, which has several practical implications for managing financial risks and maintaining the financial system’s stability.

CO2 emission forecasting based on nonlinear grey Bernoulli and BP neural network combined model

CO2 emission forecasting based on nonlinear grey Bernoulli and BP neural network combined model

Simulation Approximators Using Linear and Nonlinear Integration Neural Networks

Simulation Approximators Using Linear and Nonlinear Integration Neural Networks

Microcrack Localization and Evaluation Using Nonlinear Lamb Wave-Mixing Technique and Lamb Transformer Network

Microcrack Localization and Evaluation Using Nonlinear Lamb Wave-Mixing Technique and Lamb Transformer Network

The Role of Gas Path Diagnostics in the Changing Gas Turbine After-market

Aero gas turbine maintenance strategies are likely to see profound changes in the future. The remarkable size of the aftermarket for both military and civil aero gas turbines has already changed the nature of the business case. The aftermarket is large, more lucrative and not subject to the cyclical downturns of the original equipment sales. The competition which this considerably profitable after market creates will spur advances in diagnostic techniques. Of particular importance is advanced gas path diagnostics where, in recent years, engineering science research has opened pathways using somewhat surprising technologies. The paper will present work covering linear and non-linear gas path analysis, genetic algorithms, neural networks, fuzzy logic and hybrid systems. Much of the costs associated with the gas turbine after market relates to the gas path and these techniques are seen to play an important role

Finite Time Stability Results for Neural Networks Described by Variable-Order Fractional Difference Equations

Variable-order fractional discrete calculus is a new and unexplored part of calculus that provides extraordinary capabilities for simulating multidisciplinary processes. Recognizing this incredible potential, the scientific community has been researching variable-order fractional discrete calculus applications to the modeling of engineering and physical systems. This research makes a contribution to the topic by describing and establishing the first generalized discrete fractional variable order Gronwall inequality that we employ to examine the finite time stability of nonlinear Nabla fractional variable-order discrete neural networks. This is followed by a specific version of a generalized variable-order fractional discrete Gronwall inequality described using discrete Mittag–Leffler functions. A specific version of a generalized variable-order fractional discrete Gronwall inequality represented using discrete Mittag–Leffler functions is shown. As an application, utilizing the contracting mapping principle and inequality approaches, sufficient conditions are developed to assure the existence, uniqueness, and finite-time stability of the equilibrium point of the suggested neural networks. Numerical examples, as well as simulations, are provided to show how the key findings can be applied.

Wholesale Food Price Index Forecasts with the Neural Network

Food price forecasts in the agricultural sector have always been a vital matter to a wide variety of market participants. In this work, we approach this forecast problem for the weekly wholesale food price index in the Chinese market during a 10-year period of January 1, 2010–January 3, 2020. To facilitate the analysis, we propose the use of the nonlinear auto-regressive neural network. Technically, we investigate forecast performance, based upon the relative root mean square error (RRMSE) as the evaluation metrics, corresponding to one hundred and twenty settings that cover different algorithms for model estimations, numbers of hidden neurons and delays, and ratios for splitting the data. Our experimental result suggests the construction of the neural network with three delays and 10 hidden neurons, which is trained through the Levenberg–Marquardt algorithm, as the forecast model. It leads to high accuracy and stabilities with the RRMSEs of 1.93% for the training phase, 2.16% for the validation phase, and 1.95% for the testing phase. Comparisons of forecast accuracy between the proposed model and some other machine learning models, as well as traditional time-series econometric models, suggest that our proposed model leads to statistically significant better performance. Our results could benefit different forecast users, such as policymakers and various market participants, in policy analysis and market assessments.

River flow modeling for flood prediction using machine learning techniques in Godavari river, India

Floods are highly impactful natural calamities, inflicting significant damage to infrastructure and causing numerous fatalities. These devastating events occur when rivers exceed their capacity or breach their banks due to intense precipitation.Forecasting river flow in the minimum Godavari river basin of eastern India allowed researchers to examine the potential of four data-driven techniques, including the artificial neural network (ANN), support vector machine (SVM), nonlinear autoregressive network with exogenous input (NARX) and Gaussian process regression (GPR), and compare the outcomes to those of the proposed neuro-tree method. By combining values of the antecedent river flow from two gauging stations, various models were built utilizing the methodologies, and the results were compared to see which models had the best match. The performances of the generated models were examined using mean square error, coefficient of correlation (R), and Nash-Sutcliffe coefficient, three widely used statistical performance assessment metrics (NS).An extensive analysis of the overall performance indicators revealed that the proposed neuro-tree algorithm models were more effective in flood prediction than the other four techniques employed in this study.

Event-Based Fuzzy Adaptive Consensus Tracking for Stochastic High-Order Nonlinear Multiagent Networks With Specified-Time Convergence

In this paper, a specified-time event-triggered fuzzy adaptive control algorithm is developed to solve the consensus tracking control problem for stochastic high-order nonlinear multi-agent networks, which is intrinsically challenging due to the existence of stochasticity and high-order (positive odd integers greater than one) terms. More precisely, a novel specified-time performance function (STPF) is incorporated into time-varying high-order tan-type barrier Lyapunov function to guarantee that the tracking errors remain under time-varying constraints within specified time. Combining fuzzy logic systems with the adding on power integrator technique, an adaptive approximation policy is introduced to handle the system uncertainties. Moreover, a new switching threshold event-triggered mechanism is devised to determine the control signals updating instants, which reduces the transmission and computation burden, while resizing the triggering threshold in real-time. The Zeno phenomenon is excluded by guaranteeing that the triggering intervals is lower bounded by a positive constant. Two simulation examples are provided to demonstrate the effectiveness of the designed algorithm.

Comparative analysis of deep learning and classical time series methods to forecast natural gas demand during COVID-19 pandemic

ABSTRACT The lockdown measures implemented to contain the COVID-19 pandemic have had a considerable effect on the consumption of natural gas, which is closely linked to the economic growth of countries. Accurately forecasting natural gas demand is critical for making informed decisions in unprecedented and unexpected situations. This study aims to compare artificial learning-based algorithms and classical statistical time series models in predicting natural gas demand during the pandemic, using Turkey as a case study. Common time series prediction methods, including Autoregressive Integrated Moving Average (ARIMA), Nonlinear Autoregression Neural Network (NARNN), Support Vector Regression (SVR), and Long Short-Term Memory (LSTM), were utilized for this purpose. The impact of the pandemic on natural gas demand was analyzed by including 2-year natural gas consumption data since its onset. Root mean square error (RMSE), correlation coefficient (R), and mean absolute error (MAE) criteria were used as performance evaluation metrics to select the best model. The results confirmed that the deep-learning-based LSTM model provided better prediction accuracy than time-series benchmark models, with the lowest RMSE (9.442) and the highest R (0.997) values in the test dataset. Furthermore, the results were validated by statistical analysis using the Diebold-Mariano and Nemenyi tests.

BEENE: deep learning-based nonlinear embedding improves batch effect estimation.

Analyzing large-scale single-cell transcriptomic datasets generated using different technologies is challenging due to the presence of batch-specific systematic variations known as batch effects. Since biological and technological differences are often interspersed, detecting and accounting for batch effects in RNA-seq datasets are critical for effective data integration and interpretation. Low-dimensional embeddings, such as principal component analysis (PCA) are widely used in visual inspection and estimation of batch effects. Linear dimensionality reduction methods like PCA are effective in assessing the presence of batch effects, especially when batch effects exhibit linear patterns. However, batch effects are inherently complex and existing linear dimensionality reduction methods could be inadequate and imprecise in the presence of sophisticated nonlinear batch effects. We present Batch Effect Estimation using Nonlinear Embedding (BEENE), a deep nonlinear auto-encoder network which is specially tailored to generate an alternative lower dimensional embedding suitable for both linear and nonlinear batch effects. BEENE simultaneously learns the batch and biological variables from RNA-seq data, resulting in an embedding that is more robust and sensitive than PCA embedding in terms of detecting and quantifying batch effects. BEENE was assessed on a collection of carefully controlled simulated datasets as well as biological datasets, including two technical replicates of mouse embryogenesis cells, peripheral blood mononuclear cells from three largely different experiments and five studies of pancreatic islet cells. BEENE is freely available as an open source project at https://github.com/ashiq24/BEENE.

Multiple description encoding-decoding-based resilient filtering for complex networks under the round-Robin protocol

The problem of distributed filtering for a class of discrete-time nonlinear complex networks subject to the network bandwidth limitation is investigated. In order to reduce the network transmission burden, the round-robin protocol is employed. Under which, a quantization-based encoding-decoding mechanism is introduced to protect the security of measurements. Taking into account the quality of transmission data packets, the multiple description coding scheme is proposed. In order to describe the behavior of the packet dropouts induced by DoS attackers, two independent Bernoulli random variables are presented. Based on the round-robin protocol and multiple description encoding-decoding scheme, a distributed recursive filtering is derived to estimate the states of complex networks. Furthermore, the sufficient conditions to obtain the upper bound of estimation error covariance are given. Finally, an example is presented to certify the effectiveness of the proposed distributed recursive filtering.

Rogue waves dynamics of cubic–quintic nonlinear Schrödinger equation with an external linear potential through a modified Noguchi electrical transmission network

In this paper, we have considered a modified Noguchi nonlinear transmission network with a dispersive element. Using the reductive perturbation method in the semi-discrete limit, we show how wave dynamics in the network can be governed by a cubic–quintic nonlinear Schrödinger equation (CQ-NLSE) with an external linear potential. Through the modified Darboux transformation, we constructed the first, second and third order rogue wave solutions. Although the linear potential function Γ(τ) does not influence the instability criterion of waves propagating in the network, it has a significant effect on the dynamics of the amplitude, phase and shirp of rogue waves. Our results also reveal that it is possible both to predict the position where the amplitude of rogue waves reaches its maximum value and to reduce their spatial width.

Theory of localization-hindered thermalization in nonlinear multimode photonics

Our society’s appetite for ultra-high bandwidth communication networks and high-power optical sources, together with recent breakthroughs in mode multiplexing/demultiplexing schemes, forced the photonics community to reconsider the deployment of nonlinear multimode systems. These developments pose fundamental challenges stemming from the complexity of nonlinear mode-mode mixing by which they exchange energy in the process towards an equilibrium Rayleigh-Jeans (RJ) distribution. Here we develop a universal one-parameter scaling theory for the relaxation rates of out-of-equilibrium excitations towards their RJ thermal state. The theory predicts an exponential suppression of the rates with increasing disorder due to the formation of stable localization clusters resisting the nonlinear mode-mode interactions that tend to separate them. For low optical temperatures, the rates experience a crossover from linear to nonlinear temperature dependence which reflects a disorder-induced reorganization of the low frequency eigenmodes. Our theory will guide the design of nonlinear multimode photonic networks with tailored relaxation-scales.

Deep learning framework for day-ahead optimal charging scheduling of electric vehicles in parking lot

The increase of electric vehicles (EVs) in the power system, in addition to environmental benefits, also has economic advantages. Charging parking lots can perform better economically through aggregators the benefits of participating in energy markets and developing different charging programs. The main challenge for participating in the electricity market for parking lots is the uncertainties in the behavior of EV owners and the day-ahead market (DAM) price. Therefore, this paper proposes a deep learning-based framework for day-ahead optimal charging scheduling, which is in three stages. In the first stage, the parameters of charging sessions are modeled using the proposed Copula generative adversarial network (CopulaGAN). In the second stage, the DAM price is forecasted using the time series multi-input nonlinear autoregressive neural network for 24 h ahead. In addition, in the proposed framework, the balancing energy market (BEM) is used to keep the charging power balance on the implementation day of the optimal charging program. In the third stage, optimal day-ahead EV scheduling is performed. The simulation is carried out using Grey Wolf Optimization (GWO) based on the Caltech charging sessions dataset and the Germany energy market dataset. The simulation results show that using the proposed deep learning framework reduces the charging cost for the parking lots compared to the fixed price approaches, the use of the BEM, the time of use (TOU) program, and using normal distribution modeling about 19.94 €, 159.97 €, 14.27 €, and 114.03 € respectively.