Modular Artificial Neural Network Research Articles

Long-term power forecasting of photovoltaic plants using artificial neural networks

Increasing the robustness of forecasting energy production models from a photovoltaic plant is crucial in investment decisions, while profitability is closely linked to its location since its energy production depends on the local climate conditions. This study aimed to mitigate uncertainty regarding the installation of photovoltaic systems by addressing the challenge of forecasting electrical energy production over the long term. Artificial Neural Network models were used for this purpose, predicting the power output of a photovoltaic plant based on the ambient temperature, cell temperature, and solar irradiance. Data recorded every minute over one year at an experimental photovoltaic plant revealed a strong correlation between energy production and the input variables. This research compared the performance of multilayer perceptron, feedforward, long short-term memory, and modular artificial neural networks architectures. Validation was carried out using mean squared error, mean absolute error, mean absolute percentage error, root mean squared error, and coefficient of determination as key metrics. Notably, the long short-term memory network demonstrated the highest accuracy in energy forecasting achieving a mean squared error of 0.0089 p.u, a mean absolute error of 0.0527 p.u, a root mean squared error of 0.0944 p.u, and a mean absolute percentage error of 49.0124 %. Nonetheless, the modular model came close to long short-term memory accuracy but with lower computational cost.

Damage localization in composite structures based on Lamb wave and modular artificial neural network

Lamb wave-based technology for damage diagnosis in composite structures has emerged as a promising tool for structural health monitoring (SHM). However, traditional SHM faces challenges in multi-sensor localization, including complex data processing, high costs, and cumbersome maintenance management. The current trend is to employ data-driven approaches, but achieving precise damage localization across the entire structure with fewer sensors still presents difficulties. To address this issue, a damage localization method based on Lamb wave envelope characteristics and modular artificial neural network (M-ANN) is developed. In this approach, the envelope characteristics are employed to characterize the damage information and reduce the data dimension. Subsequently, the M-ANN model is innovatively designed with dropout layers in each hidden layer, significantly enhancing performance on random datasets. In the composite laminate experiment using only four sensors, the proposed method achieves an average relative error of 2.96 % for random damage samples, with 95 % falling within 6.17 %, outperforming other advanced damage localization approaches. Additionally, the method is utilized to investigate the requirements for the relative density of data acquisition under different damage localization needs in composite structures.

Long-term and short-term rainfall forecasting using deep neural network optimized with flamingo search optimization algorithm

Rainfall forecasting is essential because heavy and irregular rainfall creates many impacts like destruction of crops and farms. Here, the occurrence of rainfall is highly related to atmospheric parameters. Thus, a better forecasting model is essential for an early warning that can minimize risks and manage the agricultural farms in a better way. In this manuscript, Deep Neural Network (DNN) optimized with Flamingo Search Optimization Algorithm (FSOA) is proposed for Long-term and Short-term Rainfall forecasting. Here, the rainfall data is obtained from the standard dataset as Sudheerachary India Rainfall Analysis (IRA). Moreover, the Morphological filtering and Extended Empirical wavelet transformation (MFEEWT) approach is utilized for pre-processing process. Also, the deep neural network is utilized for performing rainfall prediction and classification. Additionally, the parameters of the DNN model is optimizing by Flamingo Search Optimization Algorithm. Finally, the proposed MFEEWT-DNN- FSOA approach has effectively predict the rainfall in different locations around India. The proposed model is implemented in Python tool and the performance metrics are calculated. The proposed MFEEWT-DNN- FSOA approach has achieved 25%, 26%, 25.5% high accuracy and 35.8%, 24.7%, 15.9% lower error rate for forecasting rainfall in Cannur at Kerala than the existing Map-Reduce based Exponential Smoothing Technology for rainfall prediction (MR-EST-RP), modular artificial neural networks with support vector regression for rainfall prediction (MANN-SVR-RP), and biogeography-based extreme learning machine (BBO-ELM) (BBO-ELM-RP) methods respectively.

Explaining the Neuroevolution of Fighting Creatures Through Virtual fMRI

While interest in artificial neural networks (ANNs) has been renewed by the ubiquitous use of deep learning to solve high-dimensional problems, we are still far from general artificial intelligence. In this article, we address the problem of emergent cognitive capabilities and, more crucially, of their detection, by relying on co-evolving creatures with mutable morphology and neural structure. The former is implemented via both static and mobile structures whose shapes are controlled by cubic splines. The latter uses ESHyperNEAT to discover not only appropriate combinations of connections and weights but also to extrapolate hidden neuron distribution. The creatures integrate low-level perceptions (touch/pain proprioceptors, retina-based vision, frequency-based hearing) to inform their actions. By discovering a functional mapping between individual neurons and specific stimuli, we extract a high-level module-based abstraction of a creature's brain. This drastically simplifies the discovery of relationships between naturally occurring events and their neural implementation. Applying this methodology to creatures resulting from solitary and tag-team co-evolution showed remarkable dynamics such as range-finding and structured communication. Such discovery was made possible by the abstraction provided by the modular ANN which allowed groups of neurons to be viewed as functionally enclosed entities.

Inference of affordances and active motor control in simulated agents.

Flexible, goal-directed behavior is a fundamental aspect of human life. Based on the free energy minimization principle, the theory of active inference formalizes the generation of such behavior from a computational neuroscience perspective. Based on the theory, we introduce an output-probabilistic, temporally predictive, modular artificial neural network architecture, which processes sensorimotor information, infers behavior-relevant aspects of its world, and invokes highly flexible, goal-directed behavior. We show that our architecture, which is trained end-to-end to minimize an approximation of free energy, develops latent states that can be interpreted as affordance maps. That is, the emerging latent states signal which actions lead to which effects dependent on the local context. In combination with active inference, we show that flexible, goal-directed behavior can be invoked, incorporating the emerging affordance maps. As a result, our simulated agent flexibly steers through continuous spaces, avoids collisions with obstacles, and prefers pathways that lead to the goal with high certainty. Additionally, we show that the learned agent is highly suitable for zero-shot generalization across environments: After training the agent in a handful of fixed environments with obstacles and other terrains affecting its behavior, it performs similarly well in procedurally generated environments containing different amounts of obstacles and terrains of various sizes at different locations.

Vibration analysis for predictive maintenance and improved reliability of rotating machines in ETRR-2 research reactor

Abstract In this work, both hardware and software modifications in a typical research reactor protection system (RPS) is proposed. The reactor cooling pumps are tripped based on vibrations safety signals of the pumps while the reactor SCRAM signal is initiated based on low flow rate and pressure drop across the reactor core which is a direct result of pumps trip. The main objective of this work is to develop reactor SCRAM signal based of core cooling pumps vibration signals. The early shutdown of the reactor based on pumps vibration signals is of significant importance not only in cooling the decay power of the reactor core after shutdown but also to prevent pumps failure. In the hardware model, the core cooling pumps vibration signals are feed to RPS to initiate reactor SCRAM signal. In the software model, a modular artificial neural network (ANN) is used in modeling the vibration monitoring of the research reactor (ETRR-2). The input and the output signals of the vibration transducer are used as a source data for training the neural network model. The type of the network used in this methodology is the supervised Multilayer Feed-Forward Neural Networks with the back-propagation (BP) algorithm. Vibration analysis programs are used in research reactors (RRs) to identify faults in machinery, plan machinery repairs, and keep machinery functioning for as long as possible without failure. The vibration severity limits are determined based on the International Organization for Standardization (ISO) 10816. The ANNs were designed using two different methods; one is by using hardware application contained two out of three voting and dynamic modules for trip signal by using ANNs. The current model classifies the vibration signals into five ranges low, good, satisfactory, unsatisfactory, and unacceptable vibration. The ANN is trained to detect the signal and vote to take the correct and safe action. The results demonstrate that the ANN can help in taking predictive actions for the safe core coolant pumps operation.

Continuous real-time prediction of surgical case duration using a modular artificial neural network

BackgroundReal-time prediction of surgical duration can inform perioperative decisions and reduce surgical costs. We developed a machine learning approach that continuously incorporates preoperative and intraoperative information for forecasting surgical duration. MethodsPreoperative (e.g. procedure name) and intraoperative (e.g. medications and vital signs) variables were retrieved from anaesthetic records of surgeries performed between March 1, 2019 and October 31, 2019. A modular artificial neural network was developed and compared with a Bayesian approach and the scheduled surgical duration. Continuous ranked probability score (CRPS) was used as a measure of time error to assess model accuracy. For evaluating clinical performance, accuracy for each approach was assessed in identifying cases that ran beyond 15:00 (commonly scheduled end of shift), thus identifying opportunities to avoid overtime labour costs. ResultsThe analysis included 70 826 cases performed at eight hospitals. The modular artificial neural network had the lowest time error (CRPS: mean=13.8; standard deviation=35.4 min), which was significantly better (mean difference=6.4 min [95% confidence interval: 6.3–6.5]; P<0.001) than the Bayesian approach. The modular artificial neural network also had the highest accuracy in identifying operating theatres that would overrun 15:00 (accuracy at 1 h prior=89%) compared with the Bayesian approach (80%) and a naïve approach using the scheduled duration (78%). ConclusionsA real-time neural network model using preoperative and intraoperative data had significantly better performance than a Bayesian approach or scheduled duration, offering opportunities to avoid overtime labour costs and reduce the cost of surgery by providing superior real-time information for perioperative decision support.

Hill Climb Modular Assembler Encoding: Evolving Modular Neural Networks of fixed modular architecture

The paper presents a novel generative Neuro-Evolutionary (NE) method called Hill Climb Modular Assembler Encoding (HCMAE). The target application of the HCMAE is to evolve modular Artificial Neural Networks (ANNs) whose modular structure is known in advance.Different variants of HCMAE were tested on two well-known ANN benchmarks, i.e. the Two-Spiral problem (feed-forward ANNs), and the Inverted-Pendulum problem (recurrent ANNs), for four different modular neural architectures. Particle Swarm Optimization and Differential Evolution were selected as rivals for HCMAE. Both rival methods were tested in two variants, i.e. a classical one-population variant and cooperative co-evolutionary multi-population variant. The paper presents the proposed method and reports all the experiments.

A new modular neural network approach with fuzzy response integration for lung disease classification based on multiple objective feature optimization in chest X-ray images

This paper describes a new hybrid approach, based on modular artificial neural networks with fuzzy logic integration, for the diagnosis of pulmonary diseases such as pneumonia and lung nodules. In particular, the proposed approach analyzes medical images, which are digitized chest X-rays, focusing on a classification method based on descriptors, such as grayscale histogram features, gray-level co-occurrence matrix (GLCM) texture-based features, and local binary pattern texture features. Then, to perform feature reduction, a multi-objective genetic algorithm is used to obtain an optimized neuro-fuzzy classifier, which is able to classify the pathology found in the analyzed chest X-ray. The main contribution of this paper is the proposed modular neural network approach, which divides features to achieve specialized analysis in the modules for digital image analysis and classification. The proposed approach achieves high classification accuracy after evaluating the neuro-fuzzy model with three large datasets of chest X-rays.

Adaptive Distance Protection Scheme for Mutually Coupled Line

The availability of zero-sequence current, under normal circumstances, determines the accuracy of the operation of a distance relay which is connected to a mutually coupled parallel line. When this is not available, the system adopts a different compensation factor which if, not properly calculated introduces errors in the relay operation. The proposed adaptive protection scheme, described in this paper, consists of three modular artificial neural networks model (ANN). This is developed using the feed-forward nonlinear backpropagation Levenberg–Marquardt algorithm that determines the actual status of the mutually coupled lines. The remote terminal units connected to the current and voltage transformers are used to acquire the appropriate data. The proposed scheme also carefully determines the ground distance element reach settings by calculating the apparent impedance while considering mutual coupling for all practical system configurations from the ANN; this eliminates the need for a compensation factor. The results of the apparent impedance (R + jX) calculated by the proposed adaptive and the conventional schemes, showed an average percentage error of (0.06% and 0.02%) and (15% and 41.5%) respectively. Having obtained this result, the performance of the proposed adaptive scheme showed the exact fault location with a higher accuracy when compared with a compensated conventional scheme.

Study of the fatigue behavior of composites using modular ANN with the incorporation of a posteriori failure probability

To study the fatigue behavior of composites, a large number of experimental assays for both deterministic and probabilistic analysis are needed. Recent studies have demonstrated mathematical and methodological models that are applied to determine the deterministic fatigue of composite materials, but to date, no reliable model has been developed to analyze probabilistic fatigue behavior using a small amount of experimental data and considering failure probability in analysis. As such, this study aimed to develop an artificial neural network (ANN) with modular architecture in order to model probabilistic fatigue behavior, using only three S-N curves in training and applying a posteriori failure probability (after ANN training). To that end, two methodologies were used to obtain Weibull distribution parameters and this result was incorporated into deterministic ANN architecture The advantage of this strategy is that training and using a deterministic ANN has demonstrated repeatability (robustness) in the training and generalization of the final result, while attempts to train an ANN with probabilistic data do not always obtain satisfactory results after training.

Crop yield prediction using aggregated rainfall-based modular artificial neural networks and support vector regression

At the present time, one of the most important sources of survival as well as the most crucial factor in the growth of Indian economy is agriculture. More than 70% of the Indian population is involved in agricultural activities. The crop yield prediction is one of the most desirable yet challenging tasks for every nation. Nowadays, due to the unpredictable climatic changes, farmers are struggling to obtain a good amount of yield from the crops. To feed the increasing population of India, there is a need to incorporate the latest technology and tools in the agricultural sector. This study focuses on the prediction of major kharif crops in Andhra Pradesh’s one of the largest costal districts: Visakhapatnam. As rainfall is the main factor in determining amount of kharif crop production, in this study, first we predict the amount of monsoon rainfall by using modular artificial neural networks (MANNs), and then, we predict the amount of major kharif crops that can be yielded by using the rainfall data and area given to that particular crop by using support vector regression (SVR). By using the methodology of MANNs-SVR, proper agricultural strategies can be made in order to increase the yield of the crops. Comparison with other machine learning algorithms has been done which shows that the proposed methodology outperforms in predicting the instances for kharif crop production.

Reconstruction of 3D Object Shape Using Hybrid Modular Neural Network Architecture Trained on 3D Models from ShapeNetCore Dataset.

Depth-based reconstruction of three-dimensional (3D) shape of objects is one of core problems in computer vision with a lot of commercial applications. However, the 3D scanning for point cloud-based video streaming is expensive and is generally unattainable to an average user due to required setup of multiple depth sensors. We propose a novel hybrid modular artificial neural network (ANN) architecture, which can reconstruct smooth polygonal meshes from a single depth frame, using a priori knowledge. The architecture of neural network consists of separate nodes for recognition of object type and reconstruction thus allowing for easy retraining and extension for new object types. We performed recognition of nine real-world objects using the neural network trained on the ShapeNetCore model dataset. The results evaluated quantitatively using the Intersection-over-Union (IoU), Completeness, Correctness and Quality metrics, and qualitative evaluation by visual inspection demonstrate the robustness of the proposed architecture with respect to different viewing angles and illumination conditions.

Wide area monitoring and protection of microgrid with DGs using modular artificial neural networks

The prominence of incorporating the renewable energy resources in a power system via microgrids has increased in the recent years, which impose a caution on conventional protection schemes. Protection schemes proposed earlier use local measurements, but fault classification for selective phase tripping using wide area measurements for microgrid has not been reported so far. This paper presents a wide area monitoring and protection of microgrid with distributed generations (DGs) using modular artificial neural networks (MANNs) for the fault detection and classification without affecting the relays in non-faulty or healthy sections of the microgrid. The distinct characteristics of the microgrid sort the proposed methodology into two stages. In stage 1, ANN 1 is developed to identify the operating mode of microgrid, whether it is operating in grid-connected mode (GCM) or islanded mode (IM). In stage 2, there are two MANNs corresponding to GCM and IM. Each MANNs consists of three separate ANNs for fault detection, classification, and section identification. A standard IEC 61850-7-420 microgrid with DGs (wind and photovoltaic) penetration is modeled in MATLAB/Simulink. The three-phase voltages and currents are measured with time synchronization considering the microphasor measurement units located at each bus. The extensive study includes different simulation scenarios such as shunt faults, high impedance fault, and dynamic situations like connection/disconnection of DGs/distribution lines. The results confirm the efficacy of the proposed methodology.

FT-IR Hyperspectral Imaging and Artificial Neural Network Analysis for Identification of Pathogenic Bacteria.

Identification of microorganisms by Fourier transform infrared (FT-IR) spectroscopy is known as a promising alternative to conventional identification techniques in clinical, food, and environmental microbiology. In this study we demonstrate the application of FT-IR hyperspectral imaging for rapid, objective, and cost-effective diagnosis of pathogenic bacteria. The proposed method involves a relatively short cultivation step under standardized conditions, transfer of the microbial material onto suitable IR windows by a replica method, FT-IR hyperspectral imaging measurements, and image segmentation by machine learning classifiers, a hierarchy of specifically optimized artificial neural networks (ANN). For cultivation, aliquots of the initial microbial cell suspension were diluted to guarantee single-colony growth on solid agar plates. After a short incubation period when microbial microcolonies achieved diameters between 50 and 300 μm, microcolony imprints were produced by using a specifically developed stamping device which allowed spatially accurate transfer of the microcolonies' upper cell layers onto IR-transparent CaF2 windows. Dry microcolony imprints were subsequently characterized using a mid-IR microspectroscopic imaging system equipped with a focal plane array (FPA) detector. Spectral data analysis involved preprocessing, quality tests, and the application of supervised modular ANN classifiers for hyperspectral image segmentation. The resulting easily interpretable segmentation maps suggest a taxonomic resolution below the species level.

A novel modular ANN architecture for efficient monitoring of gases/odours in real-time

Data pre-processing is tremendously used for enhanced classification of gases. However, it suppresses the concentration variances of different gas samples. A classical solution of using single artificial neural network (ANN) architecture is also inefficient and renders degraded quantification. In this paper, a novel modular ANN design has been proposed to provide an efficient and scalable solution in real–time. Here, two separate ANN blocks viz. classifier block and quantifier block have been used to provide efficient and scalable gas monitoring in real—time. The classifier ANN consists of two stages. In the first stage, the Net1-NDSRT has been trained to transform raw sensor responses into corresponding virtual multi-sensor responses using normalized difference sensor response transformation (NDSRT). These responses have been fed to the second stage (i.e., Net2-classifier). The Net2-classifier has been trained to classify various gas samples to their respective class. Further, the quantifier block has parallel ANN modules, multiplexed to quantify each gas. Therefore, the classifier ANN decides class and quantifier ANN decides the exact quantity of the gas/odor present in the respective sample of that class.

Comparison of single and modular ANN based fault detector and classifier for double circuit transmission lines

Comparison of single and modular artificial neural network based techniques for shunt faults detection and classification in double end fed double circuit transmission line is presented in this paper. The proposed method uses the voltages and currents signals available at the local end of line. The model of the power system is developed using MATLAB® software. Effects of variations in pre-fault power flow angle, fault inception angle, source strength, fault resistance, fault type, fault distance and CT saturation have been investigated extensively on the performance of the ANN based protection scheme. Additionally, the effects of network changes: double circuit and single circuit operation with other line switched out and grounded (/ungrounded), intercircuit and cross-country faults have also been investigated. Thus, the present work encompasses practically the entire range of possible operating conditions, which has not been reported earlier. The simulation results are presented to validate the effectiveness of the proposed approach. The main advantage of the proposed technique is that the variation in power system parameters under a variety of operating conditions and single/double circuit operations does not affect its performance. And this technique correctly detects and classifies the intercircuit and cross-country faults which have not been reported so-far.Keywords: Artificial neural network, cross country fault; transmission line protection; intercircuit faults and single circuit operation.

Neural prediction of concrete compressive strength

In the past, a few researchers predicted the compressive strength of concrete from its ingredients by employing artificial neural network (ANN). Evaluation of the models with different performance metrics, such as correlation coefficient and errors of estimation, indicated that there was scope for improvement in the prediction performance of ANN. In this paper development of prediction models has been carried out for superior performance with two concepts: single ANN and modular ANN. Experimental data from literature have been utilised for the study. Better overall performance of the developed ANN models than reported in the literature was ascertained by higher correlation, less root mean square error and mean absolute error. The performance of the modular ANN concept was superior to that of single ANN concept for the present application.

Neural prediction of concrete compressive strength

In the past, a few researchers predicted the compressive strength of concrete from its ingredients by employing artificial neural network (ANN). Evaluation of the models with different performance metrics, such as correlation coefficient and errors of estimation, indicated that there was scope for improvement in the prediction performance of ANN. In this paper development of prediction models has been carried out for superior performance with two concepts: single ANN and modular ANN. Experimental data from literature have been utilised for the study. Better overall performance of the developed ANN models than reported in the literature was ascertained by higher correlation, less root mean square error and mean absolute error. The performance of the modular ANN concept was superior to that of single ANN concept for the present application.

Diffusion-based neuromodulation can eliminate catastrophic forgetting in simple neural networks.

A long-term goal of AI is to produce agents that can learn a diversity of skills throughout their lifetimes and continuously improve those skills via experience. A longstanding obstacle towards that goal is catastrophic forgetting, which is when learning new information erases previously learned information. Catastrophic forgetting occurs in artificial neural networks (ANNs), which have fueled most recent advances in AI. A recent paper proposed that catastrophic forgetting in ANNs can be reduced by promoting modularity, which can limit forgetting by isolating task information to specific clusters of nodes and connections (functional modules). While the prior work did show that modular ANNs suffered less from catastrophic forgetting, it was not able to produce ANNs that possessed task-specific functional modules, thereby leaving the main theory regarding modularity and forgetting untested. We introduce diffusion-based neuromodulation, which simulates the release of diffusing, neuromodulatory chemicals within an ANN that can modulate (i.e. up or down regulate) learning in a spatial region. On the simple diagnostic problem from the prior work, diffusion-based neuromodulation 1) induces task-specific learning in groups of nodes and connections (task-specific localized learning), which 2) produces functional modules for each subtask, and 3) yields higher performance by eliminating catastrophic forgetting. Overall, our results suggest that diffusion-based neuromodulation promotes task-specific localized learning and functional modularity, which can help solve the challenging, but important problem of catastrophic forgetting.