Low Error Values Research Articles

Bio-diesel synthesis from waste sunflower oil: Experimental and multilayer perceptron neural network modeling

Transesterification being the most widely utilized way for biodiesel synthesis, the present work focuses on a single-step transesterification reaction between second-used sunflower oil and methanol as a reactant mixture with potassium hydroxide as a base catalyst. The parameter study exhibits encouraging results in terms of biodiesel/ ester conversion of the second used sunflower oil. Optimum parameters like temperature (62.3 °C), the molar ratio of methanol/ oil (1:5), and potassium hydroxide (2.85 wt%) were estimated for maximum biodiesel conversion (97.06 %). An artificial intelligence (AI) based formalism known as multilayer perceptron neural network (MLPNN) is utilized to develop the prediction model for biodiesel conversion (ml) (y0). The input space of the MLPNN-based model comprises of four operating parameters, viz. temperature (oC), second used sunflower oil (ml), methanol (ml), and reaction time (min). The proposed model is trained with the error-back propagation methodology. The high correlation coefficient (CCTraining=0.978;CCTest=0.999) and low root mean square error (RMSETraining=80.392;RMSETest=27.891) values indicate the excellent accuracy of prediction and generalization of the developed model. As a result, the proposed model can be conveniently engaged for further designing of biodiesel synthesis experimentation.

Forecasting exchange rates in Sri Lanka: a comparison of the double seasonal autoregressive integrated moving average models (DSARIMA) and SARIMA models

Exchange rates serve as a medium for currency trading in the financial market. The variations and the uncertainty movements in exchange rates have a potential effect on the performance of a country. The objective of this study is to forecast daily exchange rates in Sri Lanka using Double Seasonal Autoregressive Integrated Moving Average (DSARIMA) and Seasonal Autoregressive Integrated Moving Average (SARIMA) with Autoregressive Conditional Heteroscedasticity (ARCH)/ Generalized ARCH (GARCH) models. The study collected a few daily exchange rates from the Yahoo finance website in terms of LKR from 1st January 2008 to 28th February 2022. The DSARIMA and SARIMA models were incorporated with the ARCH/ GARCH specifications of normal, skew-normal, student-t and skew-t due to the accurate specification of the proper error distribution led to an increase in the accuracy of the fitted model. The model comparisons were carried out considering different performance measures. The overall results from the actual and fitted graphs and lower error values of the fitted models suggested a SARIMA model for CHF/LKR, a SARIMA model with ARCH/GARCH for USD, EURO, JPY, GBP and AUD against LKR and a DSARIMA model with ARCH/GARCH for CAD and SGD against LKR were suitable to forecast the respective exchange rate. Overall, the results from this study will support government, investors, corporate, financial and managerial sectors in their future decisions to accomplish their objectives. The originality of this study concerns the application of DSARIMA models in exchange rates due to the availability of double seasonality in data.

An experimental investigation and machine learning-based prediction for seismic performance of steel tubular column filled with recycled aggregate concrete

Abstract This work presents the design and application of a low-cycle reciprocating loading test on 23 recycled aggregate concrete-filled steel tube columns and 3 ordinary concrete-filled steel tube columns. Additionally, a systematic study on the influence of various parameters (e.g., slenderness ratio, axial compression ratio, etc.) was conducted on the seismic performance of the specimens. The results show that all the specimens have good hysteresis performance and a similar development trend of skeleton curve. The influence of slenderness ratio on the seismic index of the specimens is more significant than that of the axial compression ratio and the steel pipe wall thickness. Furthermore, artificial intelligence was applied to estimate the influence of parameter variation on the seismic performance of concrete columns. Specifically, Random Forest with hyperparameters tuned by Firefly Algorithm was chosen. The high correlation coefficients (R) and low root mean square error values from the prediction results showed acceptable accuracy. In addition, sensitivity analysis was applied to rank the influence of the aforementioned input variables on the seismic performance of the specimens. The research results can provide experimental reference for the application of steel tube recycled concrete in earthquake areas.

Terminal voltage prediction of Li-Ion batteries using Combined Neural Network and Teaching Learning Based Optimization algorithm

The static and dynamic model parameters are critical parameters for the accurate estimation of open-circuit voltage and the terminal voltage of a Lithium-Ion (Li-Ion) battery. This work mainly focuses on the investigation of an accurate method for predicting the open-circuit voltage and terminal voltage of Li-Ion batteries. This work proposes an Enhanced Neural Network and Teaching Learning-Based Optimization (ENN-TLBO) algorithm to predict the terminal voltage of a Li-Ion battery used in electric vehicle applications. In this work, the static model is obtained by using a neural network (NN) while the dynamic model parameters are predicted by using the Teaching Learning Based Optimization technique (TLBO). In the NN method, different network parameters are varied to predict the accurate State of Charge (SoC)-Open Circuit Voltage (OCV) relation of the battery. The proposed static models are validated by the performance metrics such as RMSE and R2, the best static model is identified by low Root Mean Square Error (RMSE) and high R2 value of the model. In recent times many optimization algorithms are reported in the literature to predict the dynamic model, which requires certain algorithm-specific tuning parameters that affects the performance of the algorithm. The proposed TLBO algorithm is implemented with lesser tuning parameters to predict the hysteresis constant of the dynamic model. The RMSE value of the predicted terminal voltage at different temperature profiles is calculated to validate the proposed method. The proposed method has several advantages such as fewer tuning parameters, simple to implement, and high accuracy.

Assessing Waste Marble Powder Impact on Concrete Flexural Strength Using Gaussian Process, SVM, and ANFIS

The study’s goal is to assess the flexural strength of concrete that includes waste marble powder using machine learning methods, i.e., ANFIS, Support vector machines, and Gaussian processes approaches. Flexural strength has also been studied by using the most reliable approach of sensitivity analysis in order to determine the influential independent variable to predict the dependent variable. The entire dataset consists of 202 observations, of which 120 were experimental and 82 were readings from previous research projects. The dataset was then arbitrarily split into two subsets, referred to as the training dataset and the testing dataset, each of which contained a weighted percentage of the total observations (70–30). Output was concrete mix flexural strength, whereas inputs comprised cement, fine and coarse aggregates, water, waste marble powder, and curing days. Using statistical criteria, an evaluation of the efficacy of the approaches was carried out. In comparison to other algorithms, the results demonstrate that the Gaussian process technique has a lower error bandwidth, which contributes to its superior performance. The Gaussian process is capable of producing more accurate predictions of the results of an experiment due to the fact that it has a higher coefficient of correlation (0.7476), a lower mean absolute error value (1.0884), and a smaller root mean square error value (1.5621). The number of curing days was identified as a significant predictor, in addition to a number of other factors, by sensitivity analysis.

Innovative biosynthesis, artificial intelligence-based optimization, and characterization of chitosan nanoparticles by Streptomyces microflavus and their inhibitory potential against Pectobacterium carotovorum

Microbial-based strategy in nanotechnology offers economic, eco-friendly, and biosafety advantages over traditional chemical and physical protocols. The current study describes a novel biosynthesis protocol for chitosan nanoparticles (CNPs), employing a pioneer Streptomyces sp. strain NEAE-83, which exhibited a significant potential for CNPs biosynthesis. It was identified as Streptomyces microflavus strain NEAE-83 based on morphological, and physiological properties as well as the 16S rRNA sequence (GenBank accession number: MG384964). CNPs were characterized by SEM, TEM, EDXS, zeta potential, FTIR, XRD, TGA, and DSC. CNPs biosynthesis was maximized using a mathematical model, face-centered central composite design (CCFCD). The highest yield of CNPs (9.41 mg/mL) was obtained in run no. 27, using an initial pH of 5.5, 1% chitosan, 40 °C, and a 12 h incubation period. Innovatively, the artificial neural network (ANN), was used for validating and predicting CNPs biosynthesis based on the trials data of CCFCD. Despite the high precision degree of both models, ANN was supreme in the prediction of CNPs biosynthesis compared to CCFCD. ANN had a higher prediction efficacy and, lower error values (RMSE, MDA, and SSE). CNPs biosynthesized by Streptomyces microflavus strain NEAE-83 showed in-vitro antibacterial activity against Pectobacterium carotovorum, which causes the potato soft rot. These results suggested its potential application for controlling the destructive potato soft rot diseases. This is the first report on the biosynthesis of CNPs using a newly isolated; Streptomyces microflavus strain NEAE-83 as an eco-friendly approach and optimization of the biosynthesis process by artificial intelligence.

Marshall Stability Prediction with Glass and Carbon Fiber Modified Asphalt Mix Using Machine Learning Techniques.

Pavement design is a long-term structural analysis that is required to distribute traffic loads throughout all road levels. To construct roads for rising traffic volumes while preserving natural resources and materials, a better knowledge of road paving materials is required. The current study focused on the prediction of Marshall stability of asphalt mixes constituted of glass, carbon, and glass-carbon combination fibers to exploit the best potential of the hybrid asphalt mix by applying five machine learning models, i.e., artificial neural networks, Gaussian processes, M5P, random tree, and multiple linear regression model and further determined the optimum model suitable for prediction of the Marshall stability in hybrid asphalt mixes. It was equally important to determine the suitability of each mix for flexible pavements. Five types of asphalt mixes, i.e., glass fiber asphalt mix, carbon fiber asphalt mix, and three modified asphalt mixes of glass-carbon fiber combination in the proportions of 75:25, 50:50, and 25:75 were utilized in the investigation. To measure the efficiency of the applied models, five statistical indices, i.e., coefficient of correlation, mean absolute error, root mean square error, relative absolute error, and root relative squared error were used in machine learning models. The results indicated that the artificial neural network outperformed other models in predicting the Marshall stability of modified asphalt mix with a higher value of the coefficient of correlation (0.8392), R2 (0.7042), a lower mean absolute error value (1.4996), and root mean square error value (1.8315) in the testing stage with small error band and provided the best optimal fit. Results of the feature importance analysis showed that the first five input variables, i.e., carbon fiber diameter, bitumen content, hybrid asphalt mix of glass-carbon fiber at 75:25 percent, carbon fiber content, and hybrid asphalt mix of glass-carbon fiber at 50:50 percent, are highly sensitive parameters which influence the Marshall strength of the modified asphalt mixes to a greater extent.

A heteroscedastic Bayesian model for method comparison data

Abstract When implementing newly proposed methods on measurements taken from a human body in clinical trials, the researchers carefully consider whether the measurements have the maximum accuracy. Further, they verified the validity of the new method before being implemented in society. Method comparison evaluates the agreement between two continuous variables to determine whether those measurements agree on enough to interchange the methods. Special consideration of our work is a variation of the measurements with the magnitude of the measurement. We propose a method to evaluate the agreement of two methods when those are heteroscedastic using Bayesian inference since this method offers a more accurate, flexible, clear, and direct inference model using all available information. A simulation study was carried out to verify the characteristics and accuracy of the proposed model using different settings with different sample sizes. A gold particle dataset was analyzed to examine the practical viewpoint of the proposed model. This study shows that the coverage probabilities of all parameters are greater than 0.95. Moreover, all parameters have relatively low error values, and the simulation study implies the proposed model deals with the higher heteroscedasticity data with higher accuracy than others. In each setting, the model performs best when the sample size is 500.

A novel application of artificial neural network for classifying agarwood essential oil quality

<p><span lang="EN-US">This work studies the agarwood oil classification into high and low quality by using two different techniques. Initially, the Forest Research Institute Malaysia (FRIM) and Universiti Malaysia Pahang (UMP) are where the sample preparation and compound extraction of agarwood oil is collected. The data collections were done from the previous researcher consists of 96 samples from seven significant agarwood oil compounds. The artificial neural network (ANN) and the proposed stepwise regression technique were used in this study. The stepwise regression was done the feature selection and successfully reduced agarwood oil compounds from seven to four. Then, the ANN technique was used to classify agarwood oil into high and low using input from seven and four compounds separately. The performance of ANN with different inputs is compared (ANN with seven inputs compared with ANN with four inputs). All the experimental work was performed using the MATLAB R2017b using the “patternet” implemented Levenberg Marquardt algorithm and ten hidden neurons. It was found that the ANN technique using seven compounds obtained the best performance according to high accuracy and lower mean square error (MSE) value. Finally, 1 hidden neuron in ANN with seven inputs selected as the best neuron for grading the agarwood oil compounds.</span></p>

Estimation of PM10 and SO2 Substances with Extreme Learning Methods for Analysis of Air Pollution in Amasya Province

Air pollution can lead to various diseases. Especially, Sulfur dioxide (SO2) and Particulate Matter (PM10 ) concentrations in the air directly affect air pollution. Therefore, the estimation of and substances in the air plays an important role in air pollution prediction. In this study, for the data received from the Ministry of Environment and Urbanization air monitoring center, the and pollutants are estimated based on past data with the extreme-learning-adaptive fuzzy extraction system method. Excessive learning algorithms in prediction problems both work fast and can reach low error values. On the other hand, the adaptive neural-fuzzy inference system offers an effective convergence by handling uncertain and incomplete data. Within the scope of the study, estimated and actual values are visualized with figures. The results of extreme learning algorithms are graphed with the error values they reach. Extreme learning-adaptive neural-fuzzy inference system; It is compared with ELM, KELM, CSELM, CDELM, and SELM algorithms. As a result of the comparison, KELM and ELM-ANFIS in terms of RMSE metric for give the best results close to each other with 0.1151 and 0.1155 values, respectively. For , it has been determined that the ELM-ANFIS showed the best performance with 0.0842 RMSE and 0.8171 R2 metric values.

Quantitative evaluation of several chemical mechanisms for gasoline surrogate-ethanol blends and study of dominant chemistry under super-knock relevant conditions

A comprehensive database that covers wide ranges of temperature, pressure and equivalence ratio conditions was established for gasoline surrogate fuels composed of n-heptane, iso-octane, toluene with and without ethanol addition. The database includes ignition delay time measurements in shock tubes and laminar flame velocity measurements which are relevant combustion properties under super-knock relevant conditions. Due to the lack of reaction pathways for low-temperature iso-octane combustion and ethanol oxidation in Caltech mechanism, an updated mechanism has been constructed. The update has been carried out as follows: (i) changing the iso-octane sub-mechanism, (ii) adding an ethanol sub-mechanism, and (iii) modifying thermodynamic and kinetic parameters for few species and reactions for removing collision limit violation. The predictive capabilities of the modified Caltech mechanism and of three reaction models developed for n-heptane/iso-octane/toluene/ethanol based mixtures have been quantified using the model evaluation method of Olm. Significant improvement of the combustion properties predictions by the modified Caltech model has been demonstrated by the lower error value. For the prediction of ignition delay time, the updated Caltech mechanism shows better performance than the other models, especially in the high-temperature regime. Error maps were drawn over wide ranges of thermodynamic conditions to show the overall performance of all the evaluated mechanisms. For the prediction of laminar burning velocity, the modified Caltech mechanism performs better for most cases. Laminar burning velocity is always over-estimated by Livermore’s mechanism. The new mechanism was employed to determine the most important chemical pathways under super-knock relevant conditions corresponding to the following thermodynamic states: (630 K, 2.3 MPa), (750 K, 6.0 MPa), (720 K, 2.8 MPa), and (1000 K, 12.0 MPa). The modified Caltech model, or reduced versions of it, could be employed in future work to investigate super-knock for realistic gasoline surrogate.

Modelling the effect of tourism disturbance on hatching rate of the Chinese giant salamander (Andrias davidianus) by using artificial neural network

The endangered Chinese giant salamander (Andrias davidianus) is an endangered species among the conserved Chinese amphibians. Tourism-related pressures have increased for this species recently. The effect of tourism on the hatching rate of the target species was determined by experimentally observing the influences of different intensities of tourism disturbance on salamander in the Zhangjiajie Chinese Giant Salamander National Nature Reserve. Water quality factors (e.g. total nitrogen, total phosphorus, dissolved oxygen and Escherichia coli abundance) were analysed, and hatching rate was estimated. Results showed that high levels of tourism disturbance (500,000–1,200,000 visitors per year) had active effects on the hatching time and negative effects on the hatching rate. The prediction performance of artificial neural network models was validated by the low root mean square error values of 2.2539 and 3.2612 for the training and testing data and high determination coefficient values of 0.9732 and 0.9508 for the training and testing data, respectively. The potential for positive or negative feedback mechanisms in such relationships between tourists and wildlife highlights the importance of considering both sides of the complex interaction to find a balance between the development of tourism and wild animal protection.

Evaluation of machine learning approaches for prediction of pigeon pea yield based on weather parameters in India.

Pigeon pea is the second most important grain legume in India, primarily grown under rainfed conditions. Any changes in agro-climatic conditions will have a profound influence on the productivity of pigeon pea (Cajanus cajan) yield and, as a result, the total pulse production of the country. In this context, weather-based crop yield prediction will enable farmers, decision-makers, and administrators in dealing with hardships. The current study examines the application of the stepwise linear regression method, supervised machine learning algorithms (support vector machines (SVM) and random forest (RF)), shrinkage regression approaches (least absolute shrinkage and selection operator (LASSO) or elastic net (ENET)), and artificial neural network (ANN) model for pigeon pea yield prediction using long-term weather data. Among the approaches, ANN resulted in a higher coefficient of determination (R2 = 0.88-0.99), model efficiency (0.88-1.00) with subsequent lower normalised root mean square error (nRMSE) during calibration (1.13-12.55%), and validation (0.33-21.20%) over others. The temperature alone or its interaction with other weather parameters was identified as the most influencing variables in the study area. The Pearson correlation coefficients were also determined for the observed and predicted yield. Those values also showed ANN as the best model with correlation values ranging from 0.939 to 0.999 followed by RF (0.955-0.982) and LASSO (0.880-0.982). However, all the approaches adopted in the study were outperformed the statistical method, i.e. stepwise linear regression with lower error values and higher model efficiency. Thus, these approaches can be effectively used for precise yield prediction of pigeon pea over different districts of Karnataka in India.

Effects of Waste Powder, Fine and Coarse Marble Aggregates on Concrete Compressive Strength

The use of marble wastes in concrete mixtures, causing air and water pollution, has been promoted in the academic and practical spheres of the construction industry. Although the effects of various forms (powder, fine, coarse and mixed) of this waste on the concrete compressive strength has been subject to a decent number of studies in the literature, the difficulties in reaching specific conclusions on the effect of each test parameter constitute a major restraint for the proliferation of the use of marble wastes in the concrete industry. Most of these studies are far from underscoring all of the parameters affecting the concrete compressive strength. Due to the urgent need in the literature for comprehensive studies on concrete mixtures with marble wastes, the results of the axial compression tests on a total of 429 concrete mixtures with marble aggregates were compiled by paying special attention to reporting all test variables (form and content of marble wastes, water–cement ratio, cement content, proportion of coarse and fine aggregates in all aggregates) affecting the concrete strength. In this context, multivariate regression analyses were carried out on the existing test results. These regression analyses yielded to relationships between the change in concrete compressive strength and the test parameters for each and every form of marble waste (powder, fine and coarse aggregate). The study indicated that independent from the form of marble wastes (as powder, fine aggregate or coarse aggregate), aggregate replacements of up to 50% can yield to significant changes in the concrete compressive strength. In addition, the analytical estimates from the developed equations exhibited a high correlation (a least r value of 0.91) with the experimental results from the previous studies, yielding to rather low error values (RMSE value is 5.06 MPa at max). For this reason, the developed equations can consistently predict the changes in concrete compressive strength with varying amounts and forms of the marble aggregates as well as the other test variables.

Preferences of Acceptance for Gayo Arabica Coffee Based on Sensory Test Using Adaptive Neuro-Fuzzy Inference System (ANFIS)

Gayo Arabica coffee has more than one variety, and each variety has their own taste. The sensory test can be done to see the level of consumer acceptance for Gayo Arabica coffee based on several considered variables. Sensory evaluation through a fuzzy approach is expected to result in several data accumulations from various panelists, making it easier for decision-making. This research is purposed to develop an Adaptive Neuro-Fuzzy Inference System (ANFIS) model, which can be used in predicting panelist acceptance levels towards various varieties of Gayo Arabica coffee products by using different membership functions. Attributes in the sensory test involved fragrance, acidity, body, aftertaste, and flavour as input variables, whereas consumer acceptance level was used as output variables. By 60% of the data were used as training data and 40% of the data were used as testing data with ANFIS model. The result of the research showed that ANFIS model with generalized bell and gaussian membership function has the lowest error value, which is 12.71% and 13.88%. That result indicates that ANFIS model with both membership functions is suitable to be used in estimating the acceptance level for Gayo Arabica coffee by consumers.

An improved MUSLE model incorporating the estimated runoff and peak discharge predicted sediment yield at the watershed scale on the Chinese Loess Plateau

An accurate prediction of sediment yield at the watershed scale is critical not only for sustainable watershed management but also for improving knowledge regarding the relationship between sediment yield and its determinant factors, which often rely on the varying model complexity. In this study, a modified sediment yield formula based on the modified universal soil loss equation (MUSLE) model was developed by introducing a channel factor into the original formula. The reliability of the proposed method was tested using data from 1,341 storm events in 38 watersheds and was applied to 256 storm events in five application watersheds using the optimized parameters. Results indicated that the proposed method is very accurate, as demonstrated by the Nash–Sutcliffe efficiency (NSE) values of 88.18 %, 85.72 %, and 85.51 % during calibration, validation, and application, respectively. The performance of the proposed model was superior to that of the original MUSLE model. Subsequently, the proposed method was used to predict sediment yield from the last five typical watersheds. This prediction utilized the parameters derived from the initial 38 watersheds; the peak discharge predicted by the modified Chemicals, Runoff, and Erosion from Agricultural Management Systems (CREAMS); and the runoff estimated by the modified Soil Conservation Service Curve Number (SCS-CN) method. Elevated NSE (61.88–81.42 %) and low root mean square error values (2.35–11.40 t ha−1) were calculated for the five watersheds. From the results, the proposed sediment yield model, combined with the modified SCS-CN method and CREAMS, was found to accurately predict the event-based sediment yield, peak discharge, and runoff in the Loess Plateau region under varying hydrological and geomorphic conditions.

Data-Driven Techniques for Evaluating the Mechanical Strength and Raw Material Effects of Steel Fiber-Reinforced Concrete.

Estimating concrete properties using soft computing techniques has been shown to be a time and cost-efficient method in the construction industry. Thus, for the prediction of steel fiber-reinforced concrete (SFRC) strength under compressive and flexural loads, the current research employed advanced and effective soft computing techniques. In the current study, a single machine learning method known as multiple-layer perceptron neural network (MLPNN) and ensembled machine learning models known as MLPNN-adaptive boosting and MLPNN-bagging are used for this purpose. Water; cement; fine aggregate (FA); coarse aggregate (CA); super-plasticizer (SP); silica fume; and steel fiber volume percent (Vf SF), length (mm), and diameter were the factors considered (mm). This study also employed statistical analysis such as determination coefficient (R2), root mean square error (RMSE), and mean absolute error (MAE) to assess the performance of the algorithms. It was determined that the MLPNN-AdaBoost method is suitable for forecasting SFRC compressive and flexural strengths. The MLPNN technique’s higher R2, i.e., 0.94 and 0.95 for flexural and compressive strength, respectively, and lower error values result in more precision than other methods with lower R2 values. SHAP analysis demonstrated that the volume of cement and steel fibers have the greatest feature values for SFRC’s compressive and flexural strengths, respectively.

Oxidation kinetics and ANN model for shelf life estimation of pearl millet (Pennisetum glaucum L.) grains during storage

AbstractThis work analyzed the impact of storage period and temperature (5, 25, and 45°C) on the rancidity indicators and oxidation kinetics of pearl millet grains stored for 120 days. The storage period and temperature had a significant (p ≤ 0.05) effect on the free fatty acid (FFA), acid value (AV), and peroxide value (PV). An Arrhenius‐type equation was used to estimate the shelf life of grains derived from the oxidation kinetics data of the rancidity parameters. A pseudo‐zero‐order reaction model fit the kinetics of pearl millet grains better than the first‐order reaction model, and this model was also exploited to anticipate the shelf life of the grains during storage. Higher storage temperatures have a less shelf life as compared to the grains stored at a lower temperature. Similarly, the pearl millet grains stored at 45°C have a shelf life of 8.5 weeks, which was much lower than the shelf life of 15.7 weeks at 5°C, respectively. FFA, AV, and PV of pearl millet grains were estimated using artificial neural network modeling based on Levenberg–Marquardt's backpropagation algorithm. Low root mean square error (0.262) values and high R2 (0.998) values indicate a good fit for the developed model. The developed shelf‐life prediction model has a positive significance for the assessment of quality deterioration of grains during storage.Novelty Impact StatementStorage temperature and time impacted the rancidity indicators, such as FFA, PV, and AV in pearl millet grain. These rancidity indicators can be modeled using oxidation kinetics and on Arrehenius equation and ANN modeling shelf life of the grains can be predicted.

Multilayer perceptron neural network based models for prediction of the rainfall and reference crop evapotranspiration for sub-humid climate of Dapoli, Ratnagiri District, India

Rainfall and reference crop evapotranspiration (ETo) are two important climatic parameters pertaining to agriculture production. Rainfall displays variability whereas reference crop evapotranspiration is nonlinear dynamic and complex phenomena. Reference crop evapotranspiration is an important component in calculation of crop water requirements, climatological and hydrological studies. In the present study, eight independent models are developed by using widely employed artificial intelligence (AI) formalism, namely, multilayer perceptron-artificial neural network (MLP-ANN). The proposed models predict two important weather parameters, viz. rainfall and reference crop evapotranspiration (ETo) using meteorological data of 34 calendar years (1983–2016) at Dapoli, District Ratnagiri, Maharashtra, India. The input space of the models consists of six climatic variables, namely, maximum temperature (Tmax), minimum temperature (Tmin), maximum relative humidity (RHmax), minimum relative humidity (RHmin), wind speed (WS), and sunshine duration (SSD). The high correlation coefficient (CC) values (0.900 to 0.957), and low root mean square error (RMSE) values (0.543 to 22.057) for both training along with test data set indicate very good prediction accuracy and generalization performance of the proposed models. Significant features of this study are (i) it is first instance where in MLP-ANN based models are developed for the rainfall and ETo prediction for meteorological data at Dapoli, Dist. Ratnagiri, Maharashtra, India. (ii) MLP-ANN modeling formalism is found to be an efficient tool for ETo prediction, and the developed models can be gainfully utilized for the further prediction of similar weather parameters in future.

Novel draught resistance sensing elements for measurement of drawbar power of agricultural machinery

Aim of study: To develop an instrumentation system comprising three force sensing elements to measure the draught resistance of any tillage and seeding tools during field operation by connecting one sensing element to each three-point linkage of the tractor.
 Area of study: Department of AgFE, Indian Institute of Technology, Kharagpur, India
 Material and methods: Commercial S-type transducers were packed laterally in between the curved plates perpendicular to the direction of travel in such a way that the magnitude of the imposed force decreased and its nature got reversed consequently during tillage force measurement. Finite element analysis was also performed on the proposed model of the sensing elements. The performance was evaluated on the basis of non-linearity, hysteresis, and non-repeatability. The data were validated with the draught values simultaneously recorded using instrumented three-point linkages of tractor.
 Main results: It offered good sensitivity and linearity during static calibration. The measurement capacity based on maximum applied load during static calibration was 10 kN with accuracy 93.40%. The low values of mean percentage error (9.03%), maximum absolute variation (17.43%), and root mean square error (0.51 kN) revealed good accuracy of the system. Validation was conducted by comparing the data for an offset type disk harrow with the model outputs of previous studies to assess its suitability for other soil working conditions, and the results were satisfactory.
 Research highlights: The advantages of this sensing device in the measurement of drawbar power are fewer changes in the hitching geometry, lower cost, and capability of quick hitching.