Back Propagation Neural Network Method Research Articles

Using ANFIS and BPNN Methods to Predict the Unfrozen Water Content of Saline Soil in Western Jilin, China

Saline soil in seasonally frozen soil areas has caused tremendous damage to engineering and the ecological environment. The unfrozen water is the main factor affecting the properties of saline soil in seasonally frozen soil area and therefore needs to be studied. However, due to the high cost of laboratory measurement of the unfrozen water content, this study focuses on using an adaptive network fuzzy inference system (ANFIS) and a back propagation neural network (BPNN) to predict the unfrozen water content of saline soil in the Zhenlai area, Western Jilin. The data for the constructed model is obtained by nuclear magnetic resonance (NMR) testing. The initial water content (W0), salt content (S), and temperature (T) are used as input parameters for predicting the unfrozen water content (Wu). The results of the ANFIS and BPNN models are compared. The results show that although both methods are suitable for predicting the unfrozen water content of saline soil in the Zhenlai area, western Jilin, the prediction accuracy of the ANFIS model is higher.

An LM-BP Neural Network Approach to Estimate Monthly-Mean Daily Global Solar Radiation Using MODIS Atmospheric Products

Solar energy is one of the most widely used renewable energy sources in the world and its development and utilization are being integrated into people’s lives. Therefore, accurate solar radiation data are of great significance for site-selection of photovoltaic (PV) power generation, design of solar furnaces and energy-efficient buildings. Practically, it is challenging to get accurate solar radiation data because of scarce and uneven distribution of ground-based observation sites throughout the country. Many artificial neural network (ANN) estimation models are therefore developed to estimate solar radiation, but the existing ANN models are mostly based on conventional meteorological data; clouds, aerosols, and water vapor are rarely considered because of a lack of instrumental observations at the conventional meteorological stations. Based on clouds, aerosols, and precipitable water-vapor data from Moderate Resolution Imaging Spectroradiometer (MODIS), along with conventional meteorological data, back-propagation (BP) neural network method was developed in this work with Levenberg-Marquardt (LM) algorithm (referred to as LM-BP) to simulate monthly-mean daily global solar radiation (M-GSR). Comparisons were carried out among three M-GSR estimates, including the one presented in this study, the multiple linear regression (MLR) model, and remotely-sensed radiation products by Cloud and the Earth’s radiation energy system (CERES). The validation results indicate that the accuracy of the ANN model is better than that of the MLR model and CERES radiation products, with a root mean squared error (RMSE) of 1.34 MJ·m−2 (ANN), 2.46 MJ·m−2 (MLR), 2.11 MJ·m−2 (CERES), respectively. Finally, according to the established ANN-based method, the M-GSR of 36 conventional meteorological stations for 12 months was estimated in 2012 in the study area. Solar radiation data based on the LM-BP method of this study can provide some reference for the utilization of solar and heat energy.

Prediksi Volume Penggunaan Air PDAM Menggunakan Metode Jaringan Syaraf Tiruan Backpropagation

As the population growth rate in Duri increases, the need for clean water also increases as needed. In Indonesia, PDAM is an institution that regulates and manages the provision of clean water for the community. So the amount of water produced and distributed should be adjusted to the demand for water. However, the problem arises in the form of waste of water at PT. PDAM Duri. Purpose of this study is to predict the amount of water consumption at PT. PDAM Duri by implementing Backpropagation Artificial Neural Network method. Variables of data taken from customer data were social, general social, household 1, household 2, household 3, commerce 1, commerce 2 and commerce 3. Data used in the prediction process was training data in 2016 and data testing in 2017. Actual amount of data at PT. PDAM Duri City 2016 until 2017 was 2.840.165 when the prediction result using artificial neural network back propagation method was 2.843.388. The number of training epochs was 4595 and the achievement of MSE (Mean Squared Error) on the test was 0,001 and the result of accuracy was 99,99900000%. Final result of this research was artificial neural network using back propagation method could predict the using of water consumption at PT. PDAM Duri for next year.

State of Charge Estimation of Power Battery Using Improved Back Propagation Neural Network

Accurately estimating the state of charge (SOC) of power batteries in electric vehicles is of great significance to the measurement of the endurance mileage of electric vehicles, as well as the safety protection of the power battery. In view of lithium ion batteries’ nonlinear relation between SOC estimation and current, voltage, and temperature, the improved Back Propagation (BP) neural network method is proposed to accurately estimate the SOC of power batteries. To address the inherent limitations of BP neural network, particle swarm algorithm is adopted to modify the relevant weighting coefficients. In this paper, the lithium iron phosphate battery (3.2 V/20 Amper-Hour) was studied. Charge and discharge experiments were conducted under a constant temperature. The training data were used to construct the surrogate model using the improved BP neural network. It is noted that the accuracy of the developed algorithm is increased by 2% as compared to that of conventional BP. Finally, an actual vehicle condition experiment was designed to further verify the accuracy of these two algorithms. The experimental results show that the improved algorithm is more suitable for real vehicle operating conditions than the traditional algorithm, and the estimation accuracy can meet the industry standards to a greater extent.

Investigating the Predictive Performance of Gaussian Process Regression in Evaluating Reservoir Porosity and Permeability

In this paper, a new predictive model based on Gaussian process regression (GPR) that does not require iterative tuning of user-defined model parameters has been proposed to determine reservoir porosity and permeability. For this purpose, the capability of GPR was appraised statistically for predicting porosity and permeability of the southern basin of the South Yellow Sea using petrophysical well log data. Generally, the performance of GPR is deeply reliant on the type covariance function utilized. Therefore, to obtain the optimal GPR model, five different kernel functions were tested. The resulting optimal GPR model consisted of the exponential covariance function, which produced the highest correlation coefficient (R) of 0.85 and the least root mean square error (RMSE) of 0.037 and 6.47 for porosity and permeability, respectively. Comparison was further made with benchmark methods involving a back propagation neural network (BPNN), generalized regression neural network (GRNN), and radial basis function neural network (RBFNN). The statistical findings revealed that the proposed GPR is a powerful technique and can be used as a supplement to the widely used artificial neural network methods. In terms of computational speed, the GPR technique was computationally faster than the BPNN, GRNN, and RBFNN methods in estimating reservoir porosity and permeability.

Identification of key influence factors and an empirical formula for spring snowmelt-runoff: A case study in mid-temperate zone of northeast China

Because of the unique climate characteristics, the runoff law in mid-temperate zone is very different from other regions in spring. Accurate runoff simulation and forecasting is of great importance to spring flood control and efficient use of water resources. Baishan reservoir is located in the upper Second Songhua River Basin in Northeast China, where snowmelt is an important source of runoff that contributes to the water supply. This study utilized long-term hydrometeorological data, in the contributing area of Bashan reservoir, to investigate factors and time-lag effects on spring snowmelt and to establish a snowmelt-runoff model. Daily precipitation, temperature, and wind data were collected from three meteorological stations in this region from 1987–2016. Daily runoff into the Baishan reservoir was selected for the same period. The snowmelt period was identified from March 23 to May 4 through baseflow segmentation with the Eckhardt recursive digital filtering method combined with statistical analyses. A global sensitivity analysis, based on the back propagation neural network method, was used to identify daily radiation, wind speed, mean temperature, and precipitation as the main factors affecting snowmelt runoff. Daily radiation, precipitation, and mean temperature factors had a two-day lag effect. Based on these factors, an empirical snowmelt runoff model was established by genetic algorithm (GAS) to estimate the snowmelt runoff in this area. The model showed an acceptable performance with coefficient of determination (R2) of 73.6%, relative error (Re) of 25.10%, and Nash-Sutcliffe efficiency coefficient (NSE) of 66.2% in the calibration period of 1987–2010, while reasonable performance with R2 of 62.3%, Re of 27.2%, and NSE of 46.0% was also achieved during the 2011–2016 validation period.

Peramalan Jumlah Penumpang Kereta Api di Indonesia dengan Resilient Back-Propagation (Rprop) Neural Network

Train scheduling affects the level of customer satisfaction and profitability of the train service provider. The prediction method of Back-propagation Neural Network (BPNN) has relatively slow convergence. Therefore, this study uses Resilient Back-propagation (Rprop) because it has a more fast convergence and high accuracy. The model produced is a model for Jabodetabek, Java (non-Jabodetabek), Sumatra, and Indonesia. From the results of data analysis conducted, it can be concluded that the performance of neural network model with Resilient Back-propagation (Rprop) formed from training data gives very accurate prediction accuracy level with mean absolute percentage error (MAPE) less than 10% for each model. Then forecasting for the next 12 months conducted and the results compared with the data testing, Rprop provides a very high forecasting accuracy with MAPE value below 10%. The MAPE value for each forecasting the number of rail passengers is 7.50% for Jabodetabek, 5.89% for Java (non-Jabodetabek), 5.36% for Sumatra and 4.80% for Indonesia. That is, four neural network architectures with Rprop can be used for this case with very accurate forecasting results.

Monitoring Method of Total Seed Mass in a Vibrating Tray Using Artificial Neural Network.

To maintain the continuous working performance of a vacuum plate seeder, it is important to monitor the total seed mass in the seed tray in real time and accurately control the pickup position of the suction plate accordingly. Under the excitation of reciprocating vibration varying with time and interference by direction angle, the motion of seeds in a rectangular tray was simulated using the discrete element method (DEM). A measurement method for seed mass in a small area was proposed based on the impulse theorem. The impact force of seeds was monitored with a cantilever force sensor, and the corresponding signal processing circuit was designed. Calibration results indicated that the relative nonlinear error was less than 2.3% with an average seeds-mass-per-unit-area (SMA) of 0.3–2.4 g/cm2. Then, four sets of force sensors were installed symmetrically near the four corners of the vibrating tray which were used to measure the SMA respectively. Back propagation (BP) neural networks which take four SMA measurement results as input parameters were developed to monitor the total seed mass in the tray. Monitoring results using DEM simulation data showed that the general relative error was 3.0%. Experiments were carried out on a test-rig and the results validated that the relative error was reduced to 5.0% by using the BP neural network method.

Comparative Analysis of Risk Assessment for Technical Standards Alliance Based on BP Neural Network and Fuzzy AHP Methods

Establishing unified industrial technical standards for a single enterprise in a highly global integrated market is becoming increasingly difficult. In recent years, leading enterprises have often built technical standards alliances around a key core technology to develop industrial standards cooperatively in order to learn from each other and optimize their resource allocation. Although such technical standards alliances result in huge gains to their members, their internal and external risks threaten both the alliances and their members. As compared to other forms of strategic alliances, the risk of such an alliance has fuzzy characteristics and is difficult to fully and accurately identify. This paper uses a fuzzy pattern-recognition method to evaluate and summarize the risks of technical standards alliances. A fuzzy analytic hierarchy process (AHP) evaluation and back propagation (BP) logic fuzzy neural network methods are used to construct a risk-evaluation model of technical standards alliances while considering an alliance around new-energy automobiles in Zhejiang as an empirical example. The two evaluation models are then contrastively analyzed, and cross validation of the evaluation results is performed in order to provide theoretical guidance and support for the application of two fuzzy evaluation models in practice.

PERAMALAN MENGGUNAKAN METODE BACKPROPAGATION NEURAL NETWORK

The purpose of the study is to forecast the price of rice in the city of Denpasar in 2017 using backpropagation neural network method. Backpropagation neural network is a model of artificial neural network by finding the optimal weight value. Artificial neural networks are information processing systems that have certain performance characteristics similar to that of human neural networks. This analysis uses time series data of rice prices in the city of Denpasar from January 2001 until December 2016. The results of this research, concludes that the lowest rice price is predicted in July 2017 at Rp9791.5 while the highest rice price in April 2017 for Rp9839.4.

BPNN–QSTR Modeling to Develop Isosteres as Sulfur-Free, Anti-Wear Lubricant Additives

The principle of isosterism was employed to design low- or zero-sulfur anti-wear lubricant additives. Thiobenzothiazole compounds and 2-benzothiazole-S-carboxylic acid esters were employed as templates. Sulfur in the thiazole ring or in the branched chain was exchanged with oxygen, CH2, or an NH group. Similarly, the template's benzimidazole ring was replaced with a quinazolinone group. Quantitative structure tribo-ability relationship (QSTR) models by back propagation neural network (BPNN) method were used to study correlations between additive structures and their anti-wear performance. The features of rubbing pairs with different additives were identified by energy dispersive spectrometer-scanning electron microscope analysis. A wide range of samples showed that sulfur substitution in additive molecules was found to be reasonable and feasible. Combined effects of the anti-wear additive and the base oil were able to improve anti-wear performance.

The Development of Javanese Language Teaching Materials Through Introduction of Java Scripts Using Artificial Neural Network

The Java script is a traditional Indonesian scripts known as Hanacaraka or Carakan. Java script becomes less desirable students who have not been introduced by Master to students using interesting digital media. Javanese language teachers in teaching activities do not yet have interactive learning media in making Java script. This research aims to develop digital media recognition Java script using artificial neural network back propagation method as a teaching material of Java language. The sample of research used is basic java script which consist of 20 characters. The method of extraction properties used is Fast Fourier Transform, which is sampled horizontally and vertically. The result of this research showed FFT and ANN can be made of interactive learning media. The effectiveness of system with sensitivity value of 0,046 - 0,085, specification value 0,023 - 0,052, and system reliability is 59,5%. Validation value system that has been built for pattern recognition Java script on the training process is 100%, the testing process with the data -3o rotation, -1o, +1o, +3o reaches 100%, for a rotation -5o testing data, and +5° is 80%, testing of handwritten data is 65%, the test data +10o rotation is 25%, the test with the data translation is 5%, and for testing with the data and the data of rotation +90° zoom in (view) is 0%.

Temperature Compensation of Elasto-Magneto-Electric (EME) Sensors in Cable Force Monitoring Using BP Neural Network.

Techniques based on the elasto-magnetic (EM) effect have been receiving increasing attention for their significant advantages in cable stress/force monitoring of in-service structures. Variations in ambient temperature affect the magnetic behaviors of steel components, causing errors in the sensor and measurement system results. Therefore, temperature compensation is essential. In this paper, the effect of temperature on the force monitoring of steel cables using smart elasto-magneto-electric (EME) sensors was investigated experimentally. A back propagation (BP) neural network method is proposed to obtain a direct readout of the applied force in the engineering environment, involving less computational complexity. On the basis of the data measured in the experiment, an improved BP neural network model was established. The test result shows that, over a temperature range of approximately −10 °C to 60 °C, the maximum relative error in the force measurement is within ±0.9%. A polynomial fitting method was also implemented for comparison. It is concluded that the method based on a BP neural network can be more reliable, effective and robust, and can be extended to temperature compensation of other similar sensors.

Retrieval of leaf area index of Phyllostachys praecox forest based on MODIS reflectance time series data.

Based on the MODIS surface reflectance data, five vegetation indices, including norma-lized difference vegetation index (NDVI), simple ratio index (SR), Gitelson green index (GI), enhanced vegetation index (EVI) and soil adjusted vegetation index (SAVI) were constructed as remote sensing variables, coupled with the seven original spectral reflectance bands of MODIS. Stepwise regression and correlation analysis were used to select the variables, and the stepwise regression and Back Propagation (BP) neural network models were constructed based on the measured LAI to retrieve the LAI time series data of Phyllostachys praecox (Lei bamboo) forest during the period from January 2014 to March 2017. The retrieval results were compared with MOD15A2 LAI products during the same period. The results showed that SR was the single variable selected for the stepwise regression model. The correlations of LAI with bands b1, b2, b3, b7 and five vegetation indices were significant, which could be used as input variables of BP neural network model. There was a significant correlation between the LAI estimated from BP neural network and measured LAI, with the R2 of 0.71, RMSE of 0.34, and RMSEr of 13.6%. R2 was increased by 10.9%, RMSE decreased by 5.6%, and RMSEr decreased by 12.3% compared with LAI estimated from stepwise regression method. R2 was increased by 54.5%, RMSE decreased by 79.3%, and RMSEr decreased by 79.1% compared with MODIS LAI. The LAI of Lei bamboo forest could be accurately retrieved using BP neural network method based on MODIS reflectance time series data, which would be a feasible method for rapid monitoring of LAI in Lei bamboo forest.

PENERAPAN METODE BACKPROPAGATION UNTUK MEMPREDIKSI JUMLAH KUNJUNGAN WISATAWAN BERDASARKAN TINGKAT HUNIAN HOTEL

Tourism is one of the most important factors for the revenue of an area. To attract the interest of tourists takes some supporting factors, one of which is the hotel or the guesthouse. In the process of prediction of tourists visit, required data the number of tourists staying in order to predict for the next time. The prediction method used in this system is a method of Backpropagation Neural Network based on time series data forecasting component variansi random or random process beginning with autocorrelation for determination of input variables. Method of Backpropagation itself is known quite well used in the forecasting of time series data. The result of the method of Backpropagation is a number of predictions that tourists visit can be a reference for related officials in taking decisions for the period ahead.

Classification of Epileptiform Waves Based on Frequency by Using Backpropagation Neural Network

Epilepsi is an abnormal condition of brain activity that can be recorded by using Elecktroencephalography (EEG). On epilepsy patient, most of the recording is interictal wave that in form of spike wave and sharp wave. This study has goal to classify whether the interictal waves are spike wave or sharp wave. The study was conducted in two stages Identification and Classification. Firstly, The epileptogenik wave were identified by shifting the baseline of each wave to select the best baseline that contain all data of the wave, then doing normalization of it to get the features of frequency, amplitude 1 and amplitude 2. Secondly, Back propagation Neural Network method is applied to classify it. Classification is done by using 200 data consisting of 120 training data and 80 testing data. The results show that classification using binary sigmoid activation function give same recognition rate of 91,25 % for all variation of learning rate.

Multidamage Detection of Bridges Using Rough Set Theory and Naive-Bayes Classifier

This paper is intended to introduce a two-stage detection method to solve the multidamage problem in bridges. Vibration analysis is conducted to acquire the dynamic fingerprints which are regarded as information sources. Bayesian fusion is used to integrate these sources and preliminarily locate the damage. Then, the RSNB method which combines rough set theory and Naive-Bayes classifier is proposed to simplify the sample dimensions and fuse the remaining attributes for damage extent detection. A numerical simulation of a real structure, the Sishui Bridge in Shenyang, China, is conducted to validate the effectiveness of the proposed detection method. Data fusion based method is compared with single-valued index method at the damage localization stage. The proposed RSNB method is compared with the Back Propagation Neural Network (BPNN) method at the damage qualification stage. The results show that the proposed two-stage damage detection method has better performances in regard to transparency, accuracy, efficiency, noise robustness, and stability. Furthermore, an ambient excitation modal test was carried out on the bridge to obtain the vibration responses and assess the damage condition with the proposed method. This novel approach is applicable for early damage detection and provides a basis for bridge management and maintenance.

Optimization of a Low-Voltage Load Switch for a Smart Meter Based on a Double Response Surface Model

The existing electric energy meter load switches have issues such as a low electrical life, poor resistance to short-circuit currents, high contact point resistivity and severe heating. Research has thus been conducted on the optimization of the electric energy meter built-in load switch mechanism to establish a double response surface model for the electromagnetic attractive force and reed reaction force. The model introduces a niche fitness sorting strategy and a Gaussian mutation mechanism that are based on the standard particle swarm algorithm and form an improved multi-objective particle swarm optimization algorithm in which the static attractive force is regarded as the optimization objective. The multi-objective optimization of the structural parameters of the electromagnetic system and the reed system was carried out, and the optimization results were applied in the devices’ design and implementation. The new structure improves the electromagnetic suction and reed force at the suction position. And the results from simulation show that the proposed method exceeds that of the back propagation neural network method in achieving the optimization goal and the new structure can effectively improve the performance of meter built-in load switch by avoiding the recoil phenomenon.

The calibration of force offset for rocket engine based on deep belief network

Background Force offset is an important movement and control parameter in rocket motor development process, and its accurate measurement is a vital guarantee of rocket motor reliable operation, so there is an essential significance to achieve accurate force offset calibration. Methods A novel force offset nonlinear calibration method is proposed based on deep belief network. Experimental platform is established and force offset calibration test is completed. Because the Levenberg -Marquardt process has the advantage of both Newton method and gradient descent method, test data are trained with Levenberg -Marquardt, decreasing nonlinear mapping convergence errors and realizing nonlinear calibration of force offset. Results and Conclusions Training results show that the mean deviation rate of force offset after nonlinear calibration is less than 2.7%, better than the back-propagation neural network and least squares method, verifying the reasonableness and practicality of nonlinear compensation calibration method and effectively improving force offset calibration accuracy.

Extraction of Eye and Mouth Features for Drowsiness Face Detection Using Neural Network

Facial feature extraction is the process of searching for features of facial components such as eyes, nose, mouth and other parts of human facial features. Facial feature extraction is essential for initializing processing techniques such as face tracking, facial expression recognition or face shape recognition. Among all facial features, eye area detection is important because of the detection and localization of the eye. The location of all other facial features can be identified. This study describes automated algorithms for feature extraction of eyes and mouth. The data takes form of video, then converted into a sequence of images through frame extraction process. From the sequence of images, feature extraction is based on the morphology of the eyes and mouth using Neural Network Backpropagation method. After feature extraction of the eye and mouth is completed, the result of the feature extraction will later be used to detect a person’s drowsiness, being useful for other research.