Multiple Artificial Neural Network Models Research Articles

Artificial neural network based prediction of ship speed under operating conditions for operational optimization

Ship speed is one of the most fundamental parameters which influences ship design, the energy efficiency of its operation, and safety. Therefore, ship speed selection and prediction under various environmental and operational conditions are of great concern recently for optimizing ship design and operational performance. Among the different approaches that address the ship speed topic, data-driven methodologies and Artificial Neural Network (ANN) techniques are attracting widespread interest due to its efficiency, accuracy, robustness, flexibility, and fault tolerance. Consequently, this study investigates multiple ANN model sizes and architectures to determine the suitable network parameters for ship speed prediction. Thus, we have a good balance between the model’s prediction accuracy and computational complexity. For this study, a publicly-available high-quality operational dataset suitable for benchmarking the results is utilized. This analysis also includes the effect of the data quantity and sampling duration on the data correlation and the ANN performance. The results indicate that the proposed ANN model can accurately predict ship speed under real operational conditions with an error of less than 1 knot. Furthermore, it has been shown that the proposed model can help with the decision-making and optimization processes of voyages planning and execution.

A New Cloud-Based IoT Solution for Soiling Ratio Measurement of PV Systems Using Artificial Neural Network

Solar energy is considered the most abundant form of energy available on earth. However, the efficiency of photovoltaic (PV) panels is greatly reduced due to the accumulation of dust particles on the surface of PV panels. The optimization of the cleaning cycles of a PV power plant through condition monitoring of PV panels is crucial for its optimal performance. Specialized equipment and weather stations are deployed for large-scale PV plants to monitor the amount of soil accumulated on panel surface. However, not much focus is given to small- and medium-scale PV plants, where the costs associated with specialized weather stations cannot be justified. To overcome this hurdle, a cost-effective and scalable solution is required. Therefore, a new centralized cloud-based solar conversion recovery system (SCRS) is proposed in this research work. The proposed system utilizes the Internet of Things (IoT) and cloud-based centralized architecture, which allows users to remotely monitor the amount of soiling on PV panels, regardless of the scale. To improve scalability and cost-effectiveness, the proposed system uses low-cost sensors and an artificial neural network (ANN) to reduce the amount of hardware required for a soiling station. Multiple ANN models with different numbers of neurons in hidden layers were tested and compared to determine the most suitable model. The selected ANN model was trained using the data collected from an experimental setup. After training the ANN model, the mean squared error (MSE) value of 0.0117 was achieved. Additionally, the adjusted R-squared (R2) value of 0.905 was attained on the test data. Furthermore, data is transmitted from soiling station to the cloud server wirelessly using a message queuing telemetry transport (MQTT) lightweight communication protocol over Wi-Fi network. Therefore, SCRS depicts a complete wireless sensor network eliminating the need for extra wiring. The average percentage error in the soiling ratio estimation was found to be 4.33%.

Predicting the Optimum Performance of a Vertical-Axis Savonius Wind Rotor With Parametric Modeling Using Artificial Neural Network and Golden Section Method

Abstract This paper explores the function approximation characteristics of Artificial Neural Network (ANN) by implementing it on the vertical-axis Savonius wind rotor technology. In this regard, a suitable experimental dataset documented in literature is exploited to train the ANN comprising the rotor performance as output and 11 different design and operating parameters as input with the help of matlab R2020b software. Multiple ANN models are trained by varying the number of hidden neurons which are then evaluated based on their estimation error and correlation coefficient (R) as decision criteria. The optimum ANN architecture demonstrates R ≈ 0.96 and 0.98 for the testing and training datasets, respectively. Further, in the quest of finding the optimum performance from the entire power curve of the rotor, the Golden Section Method (GSM) is linked with the trained ANN model. Using these soft computing techniques, a parametric study is carried out to understand the dependency of rotor performance on their design and operating parameters. At the end, a graphical interface is developed as a product so as to allow the user to predict the performance of the new rotor designs intuitively.

Effect of input-target datasets on sediment transport modelling in alluvial rivers using artificial neural network

ABSTRACT Numerical simulation of sediment transport in alluvial rivers often becomes untenable due to the absence of requisite data. Over the years, artificial neural network (ANN) models have been successfully employed for tackling such scenarios. However, the effect of input and target datasets on the simulation of sediment transport using ANN has not yet been adequately addressed in the literature so far. To study the effect of input-target dataset on the performance of ANN models, the present study employs seven input-target datasets for the development of several ANN models and assessment of their performance. The limited study carried out reveals that there are multiple ANN models – having different combinations of training algorithms, transfer functions and input/target datasets – that can reliably be employed to estimate bed elevation changes in alluvial rivers.

Days-ahead water level forecasting using artificial neural networks for watersheds.

Watersheds of tropical countries having only dry and wet seasons exhibit contrasting water level behaviour compared to countries having four seasons. With the changing climate, the ability to forecast the water level in watersheds enables decision-makers to come up with sound resource management interventions. This study presents a strategy for days-ahead water level forecasting models using an Artificial Neural Network (ANN) for watersheds by conducting data preparation of water level data captured from a Water Level Monitoring Station (WLMS) and two Automatic Rain Gauge (ARG) sensors divided into the two major seasons in the Philippines being implemented into multiple ANN models with different combinations of training algorithms, activation functions, and a number of hidden neurons. The implemented ANN model for the rainy season which is RPROP-Leaky ReLU produced a MAPE and RMSE of 6.731 and 0.00918, respectively, while the implemented ANN model for the dry season which is SCG-Leaky ReLU produced a MAPE and RMSE of 7.871 and 0.01045, respectively. By conducting appropriate water level data correction, data transformation, and ANN model implementation, the results of error computation and assessment shows the promising performance of ANN in days-ahead water level forecasting of watersheds among tropical countries.

RESPECTING PATIENT PRIVACY WITH FEDERATED ARTIFICIAL INTELLIGENCE

Multiple research has shown that deep artificial neural networks (ANN) can assist physicians in diagnosing a patient with greater accuracy and sensitivity. ANN applications are not limited to just the classification of diseases but also include image segmentation, tumor localization, and mortality rate predictions. Nonetheless, the great march of success by ANN is only possible by the availability of an open medical dataset. Only with such open datasets can developers build, train and test ANN models to obtain higher accuracy. However, there is yet an open medical dataset in Malaysia that can be used to validate ANN performance on local patients. This may be due to local medical institution's hesitance to release any medical images and records to respect patient's confidentiality. One way around this is to adopt the Ensemble or Federated Learning system. In federated learning, medical institutes share their locally trained ANN model's weight and bias to create a single federated level ANN model. While in ensemble learning, multiple ANN models are used to vote on the most probable class for each data. In both systems, no sharing of patient's data is required. In our experiment, we tested the capability of 25 ANN models to classify chest radiograph images into three classes; normal, bacterial pneumonia, and viral pneumonia. Each ANN model is given a training dataset that is random in size and class ratio. The result obtained from the experiment shows that the federated system obtains the highest score in all measured metrics. It obtains a score of 0.76, 0.72, and 0.72 for average weighted precision, weight sensitivity, and F1, respectively. It also has the lowest standard deviation in all performance metrics compare to other learning systems. The result obtains here further strengthens the notion that if Malaysia wants to adopt a national-level artificial intelligent system for medical purposes, it should utilize the federated learning system at its core. It ensures Malaysia has an artificial intelligence system that respects patient's privacy while maintains its robustness.

Predicting GPR Signals from Concrete Structures Using Artificial Intelligence-Based Method

This paper presents the application of an Artificial Intelligence-based method in analyzing the effects of environmental conditions, chloride contamination in concrete, and surface corrosion of rebars on the amplitude of Ground Penetrating Radar (GPR) signals. Six reinforced concrete slabs with different chloride contamination mixtures were fabricated and tested. GPR data were collected under various temperature and ambient humidity combinations. A total of 288 rebar picks were used for training, validation, and testing the proposed Artificial Neural Network (ANN) model. Multiple ANN model configurations with a variation in learning algorithms and the number of nodes in the hidden layer were explored to obtain the optimal model for the nondestructive data. It is shown that the “trainlm” learning algorithm produced the high accuracy prediction of the reflection amplitude of GPR signals. The sensitivity analysis was also conducted with the ANN model to investigate the effects of the input on the output parameters. Results from the sensitivity analysis revealed that the GPR reflection amplitudes were more sensitive to the changes of temperature parameter (TEM) and chloride contamination level (CCL), while they were less sensitive to the variation of ambient relative humidity (ARH) and rust condition on the rebar surface (CSR).

Investigation of sewage sludge and peanut shells co‐combustion using thermogravimetric analysis and artificial neural network

To solve the problems of energy shortage and waste accumulation, the method of co-combustion of sewage sludge (SS) and peanut shells (PS) was proposed. SS-PS co-combustion characteristics in air were investigated using artificial neural networks (ANN) and thermogravimetric analyses (TGA). The proportion of PS in the mixture was 10%-50%. The temperature range of PS combustion (160-560°C) is lower than that of PS combustion (170-600°C). Activation energy was estimated from two non-isothermal kinetic analysis methods: Kissinger-Akahira-Sunose (KAS) and Flynn-Wall-Ozawa (FWO). The kinetic mechanism of the combustion process was determined by using the master-plots method. Multiple ANN models were established to predict TG data of SS-PS co-combustion. The best prediction model (ANN21) was obtained. The results showed a good overlap between the predicted and experimental TG data. The ANN model and the master-plots method are the main innovations of this study. This study can promote the utilization and reduction of solid waste, and give guidance for the large-scale application of SS-PS co-combustion.

Sediment transport modelling in an alluvial river with artificial neural network

Simulation of flow and transport of sediment in alluvial rivers, particularly during release of water through dams, is a complex problem. Such simulations are generally carried out using various numerical models, subject to availability of requisite morphological and sediment data sets. In addition, various simplifying assumptions, quite often, are part and parcel of most of the available numerical sediment transport models. As such, non-compliance of any of the assumptions or non-availability of all the requite data sets restricts meaningful use of any numerical simulator in estimation of morphological changes in alluvial rives. This study attempts to resolve such issues associated with numerical models by using Artificial Neural Networks (ANN). The primary objective of the study is to explore and identify best performing ANN models in estimating morphological changes in alluvial rivers. In the present study, the one dimensional hydrodynamic modelling system, HEC-RAS (Hydrologic Engineering Centre’s River Analysis System), is employed for generating database to train multilayer feedforward networks (MLFN). Performance evaluations of all the developed MLFN models are made by using three statistical parameters, i.e. Mean Square Error (MSE), Mean Absolute Error (MAE) and Coefficient of Determination (R2). The study reveals that there are multiple ANN models, stretching across different training algorithms and transfer functions, which can be reliably used in developing sediment transport models in alluvial rivers.

Exploring Copula-based Bayesian Model Averaging with multiple ANNs for PM2.5 ensemble forecasts

Quantifying predictive uncertainty of ensemble air quality forecast is very crucial and challenging. This study integrated a Copula-based Bayesian Model Averaging (CBMA) and multiple deterministic artificial neural networks (ANNs) to make accurate ensemble probabilistic PM2.5 forecasts. The new approach (CBMA), has a flexible structure that grants the posterior distribution to have any shape owing to the Copula function. The CBMA approach could remove the data transformation and bias correction procedures as it is done in the original BMA, which was taken as the benchmark. The air quality in Taipei City of Taiwan was selected as a study case to evaluate the applicability and reliability of the proposed approach. Three kinds of air quality monitoring stations denoted heavy traffic loads, intensive commercial trading and human intervention, and a natural circumstance with fewer human activities respectively. The forecasts of PM2.5 concentrations were regarded as a math function involving meteorological and air quality variables, using long-term (2010–2018) hourly observational datasets. Firstly, four deterministic ANN models were established and evaluated to provide inputs for ensemble forecasting. Then, the two post-processing techniques (i.e. CBMA and BMA) were employed to produce ensemble probabilistic forecasts based on the forecasts obtained from multiple ANN models. The results demonstrated that the CBMA not only could outperform the BMA but also could provide a practical and reliable approach as a complement to multiple deterministic ANN models to create ensemble probabilistic forecasts. From horizons t+1 up to t+4, the CBMA approach could drive up the Containing Ratio (CR) values by 3.12% ˗ 9.58% as well as reduce the average Relative Band-width (RB) values by 8.63% ˗ 34.48% and the Continuous Ranked Probability Score (CRPS) values by 7.62% ˗ 32.89%, in comparison with the BMA one. Consequently, the predictive uncertainty could be alleviated while model reliability and PM2.5 forecast accuracy could be considerably increased.

Joint bagged-boosted artificial neural networks: Using ensemble machine learning to improve short-term electricity load forecasting

This paper uses artificial neural networks (ANNs) based ensemble machine learning for improving short-term electricity load forecasting. Unlike existing methods, the proposed technique combines both bagging and boosting to train bagged-boosted ANNs. The combination consists of training an ensemble of multiple ANN models in parallel and averaging the forecasted loads from these models to get the final forecasted load. The models are trained on data obtained by bootstrap sampling the original training data. Each of these models further consists of an ensemble of ANNs that are trained sequentially, such that the prediction error between the training output and the forecasted output from the preceding ANN in the sequence is minimized. We use real data to demonstrate that the proposed technique enables reduction of both bias and variance compared to a single ANN, bagged ANN and boosted ANN. We further show that it reduces the forecasting error compared to existing load forecasting methods.

Fuel Cell Hybrid Model for Predicting Hydrogen Inflow through Energy Demand

Hydrogen-based energy storage and generation is an increasingly used technology, especially in renewable systems because they are non-polluting devices. Fuel cells are complex nonlinear systems, so a good model is required to establish efficient control strategies. This paper presents a hybrid model to predict the variation of H2 flow of a hydrogen fuel cell. This model combining clusters’ techniques to get multiple Artificial Neural Networks models whose results are merged by Polynomial Regression algorithms to obtain a more accurate estimate. The model proposed in this article use the power generated by the fuel cell, the hydrogen inlet flow, and the desired power variation, to predict the necessary variation of the hydrogen flow that allows the stack to reach the desired working point. The proposed algorithm has been tested on a real proton exchange membrane fuel cell, and the results show a great precision of the model, so that it can be very useful to improve the efficiency of the fuel cell system.

Modeling and simulation of river discharge using artificial neural networks

The study developed multiple artificial neural network models with the aim of establishing the most suitable non-linear discharge perdition model of Ibu River. A 12-year daily discharge of River Ibu gauged near Sagamu was obtained from the Ogun-Oshun River Basin Development Authority (OORBDA), Abeokuta Nigeria to model and simulate daily discharge. The back-propagation method was used in developing the artificial neural network model. The study revealed that only three artificial neural network (ANN) models out of fifteen developed, have overall results that are satisfactory for prediction, out of these, the model with the least error was used for validation. The results obtained with ANNs based on two hidden layers for 1-day ahead are better than those obtained by models with single layers. It was concluded that the general performance of ANN models depends solely on the data used. While it was recommended that additional basin characteristics such as slope, geology, morphology and surface roughness features should be included to obtain more robust river discharge models. Keywords: Artificial Neural Network, Hydrologic record, Prediction

Exploring geometrical patterns in streamflow time series: utility for forecasting?

Abstract Streamflow time series often provide valuable insights into the underlying physical processes that govern responses of any watershed to storm events. Patterns derived from time series based on repeated structures within these series can be beneficial for developing new or improved data-driven forecasting models. Data-driven models, artificial neural networks (ANN), are developed in the current study for streamflow prediction using input structures that are classified by geometrically similar patterns. A new modular and integrated ANN architecture that combines multiple ANN models, referred to as pattern-classified neural network (PCNN), is proposed, developed and investigated in this study. The PCNN relies on the development of several independent local models instead of one global data-driven prediction model. The PCNN models are evaluated for one step-ahead prediction of daily streamflows for Reed Creek and Little River, Virginia, and Elkhorn Creek, Kentucky in the United States. Results obtained from this study suggest that the use of these patterns has improved the performance of the neural networks in prediction. The improved performance of the PCNN models can be attributed to prior classification of data benefiting generalization abilities. PCNN model outputs can also provide an ensemble of forecasts that help quantify forecast uncertainty.

Surface roughness estimation from RADARSAT-2 data in a High Arctic environment

Synthetic aperture radar (SAR) data are often used to determine the physical properties of the soil surface, such as soil moisture and surface roughness. Although these analyses are commonly applied in agricultural environments, there has been limited application in more natural environments, particularly at high latitudes. For the research reported here, an artificial neural network (ANN) is developed to model surface roughness in the Canadian High Arctic. This research represents the first phase of the overall goal of developing an operational methodology for estimating surface roughness, vegetation cover and soil moisture using SAR and limited field measurements. Multiple incidence angle data and fully polarimetric data from RADARSAT-2 are combined with long and short profile in situ surface roughness measurements from 134 sample locations located across two distinct High Arctic study sites. Multiple ANN models were developed using various backscatter, textural, and polarimetric variables. The ANN models exhibited a moderate to strong agreement to field-measured surface roughness. This study demonstrates that operational surface roughness modeling in the Canadian High Arctic is feasible with RADARSAT-2 polarimetric data.

Crack identification in curvilinear beams by using ANN and ANFIS based on natural frequencies and frequency response functions

This paper presents different artificial intelligence (AI) techniques for crack identification in curvilinear beams based on changes in vibration characteristics. Vibration analysis has been performed by applying the finite element method (FEM) to compute natural frequencies and frequency response functions (FRFs) for intact and damaged beams. The analysis reveals the changes in natural frequencies and amplitudes of FRFs of the beams for cracks of different sizes at different locations. These changes are used as input data for single and multiple artificial neural networks (ANN) and multiple adaptive neuro-fuzzy inference systems (ANFIS) in order to predict the size of the crack and its location. To avoid large models, the principal component analysis (PCA) approach has been carried out to reduce the computed FRFs data. The analysis of different techniques shows that the average prediction errors in the multiple ANN models is less than those in the single ANN model and in the multiple ANFIS. It is shown that the cracks longer than 5 mm can be located with satisfactory accuracy, even if the input data are corrupted with various level of noise. Multiple ANFIS is adopted to construct a more reliable and less sensitive model for noise excitation.

River flow forecasting with artificial neural networks using satellite observed precipitation pre-processed with flow length and travel time information: case study of the Ganges river basin

Abstract. This paper explores the use of flow length and travel time as a pre-processing step for incorporating spatial precipitation information into Artificial Neural Network (ANN) models used for river flow forecasting. Spatially distributed precipitation is commonly required when modelling large basins, and it is usually incorporated in distributed physically-based hydrological modelling approaches. However, these modelling approaches are recognised to be quite complex and expensive, especially due to the data collection of multiple inputs and parameters, which vary in space and time. On the other hand, ANN models for flow forecasting are frequently developed only with precipitation and discharge as inputs, usually without taking into consideration the spatial variability of precipitation. Full inclusion of spatially distributed inputs into ANN models still leads to a complex computational process that may not give acceptable results. Therefore, here we present an analysis of the flow length and travel time as a basis for pre-processing remotely sensed (satellite) rainfall data. This pre-processed rainfall is used together with local stream flow measurements of previous days as input to ANN models. The case study for this modelling approach is the Ganges river basin. A comparative analysis of multiple ANN models with different hydrological pre-processing is presented. The ANN showed its ability to forecast discharges 3-days ahead with an acceptable accuracy. Within this forecast horizon, the influence of the pre-processed rainfall is marginal, because of dominant influence of strongly auto-correlated discharge inputs. For forecast horizons of 7 to 10 days, the influence of the pre-processed rainfall is noticeable, although the overall model performance deteriorates. The incorporation of remote sensing data of spatially distributed precipitation information as pre-processing step showed to be a promising alternative for the setting-up of ANN models for river flow forecasting.