LM Algorithm Research Articles

Dynamic Characteristics and Effect on Thrust of Jet Tail Vortex Ring for Pump Jet Propulsion

When a pump is propelled by a propeller, the nozzle flooding of the jet wake area will produce a turbulent quasi-sequence structure and have a certain impact on the outflow field structure and thrust characteristics of the water jet propulsion pump. In this paper, a method that combined numerical simulation with vortex dynamics is adopted, which analyzes the dynamic characteristics and influence on the thrust characteristics of the water jet propulsion pump. A large eddy simulation turbulence model and a dimensionless water jet propulsion pump velocity coefficient were used to reveal flow structure and relation, with the pump operation parameters of the wake vortex ring. The thrust with a trailing vortex ring is 7.0% higher than that without a trailing vortex ring. Vortex dynamics and mathematical statistics are combined to quantitatively analyze the dynamic characteristics of the jet tail vortex ring. Finally, the formation time of the vortex ring is obtained in exponential relation with dimensionless transmission velocity and vorticity coefficient, which has nonlinear relation with vortex intensity coefficient and helicity coefficient. BP Neural Network combined with the LM algorithm is used to establish the mathematical relationship between the thrust and the physical characteristic parameters of the vortex ring.

Research on Multicamera Photography Image Art in BERT Motion Based on Deep Learning Mode.

In order to improve the artistic expression effect of photographic images, this article combines the deep learning model to conduct multicamera photographic image art research in BERT motion. Moreover, this article analyzes the external parameter errors caused in the calibration process and uses the checkerboard in the common field of view to calibrate the spatial coordinates of the corners of the board in multiple camera coordinate systems. In addition, this article aims to match the spatial coordinates of the corresponding points to each other and solve the rotation and translation matrix in the transformation process. Finally, this article uses the LM algorithm to optimize the calibration parameters of the camera and combines the deep learning algorithm to perform image processing. The experimental research results show that the research method of multicamera photography image art in BERT motion based on the deep learning mode proposed in this article can effectively improve the expression effect of image art.

Extraction of individual trees based on Canopy Height Model to monitor the state of the forest

Active remotely sensed data can be used to perform a variety of forestry tasks including stand characterization, inventory, and management of forest and fire behavior modeling. The present work investigates the potential of Airborne Laser Scanning (ALS) derived methods applied in the deciduous forest by processing an individual tree detection (ITD) based on canopy Height model (CHM) and tree segmentation of larger-area point clouds. Different algorithms are tested and their performances are evaluated to show which of them can provide the most adequate number of trees compared with the ground truth. Tree scale information is used in order to determine stand age. The forest height, structure, and density are specified by applying individual tree Detection (ITD) to calculate some forest attributes such as stem volume, forest uniformity, and biomass estimation. The major aim of this post is to examine the state of the forest to monitor it in real-time. We assume that utilizing the LM algorithm, which was originally built for ITD from LiDAR data, trees should be automatically distinguished from the ALS-derived CHM with reasonable accuracy. As a result, the present research work studies the fixed treetop window size (FWS), fixed smoothing window size (SWS), and variable window (VW) effect on ITD performance (RMSE=3.4% and R=0.88). It is obvious, from the obtained results that smaller window sizes result in more trees. In fact, the smallest trees obscured by the largest trees containing the highest points in the neighborhood are often ignored by large windows. Crown delineation is also explored to extract the height of the trees, radius crown and, 3D coordinates and to compare them to those detected by a Low Bluetooth sensor “iBeacon.”.

Clustering assisted artificial neural network for handling noisy big data: An application for estimation of parameter in combined mode conduction and radiation heat transfer

AbstractA pattern net assisted mapping artificial neural network (PAMANN) model for estimation of parameters in problem with large data (1300 × 121 matrix size) is reported. A pattern net‐based multilayer perceptron neural network (MLPNN) model for clustering the data, followed by mapping MLPNN model for mapping the target with the input, is developed as PAMANN model. A heat transfer problem with combined mode conduction and radiation in porous medium is solved numerically, and is called direct model. In the inverse model, a PAMANN model is developed by using data generated through the direct model. The PAMANN model is able to estimate two parameters (extinction coefficient β and convective coupling P2) after taking temperature profile as input. The model is tested for different number of neurons in hidden layer, and different levels of noise in input data. Twelve different algorithms are explored in training of mapping MLPNN, and compared for performance. Levenberg–Marquardt algorithm is found to estimate the parameters with high accuracy, but took high CPU time. Bayesian regularization is found to consume very high CPU time with moderate accuracy in estimation of parameters. Variations in hidden layer neuron number and noise in input data, were done to analyze the performance of mapping MLPNN with different training algorithms. Algorithms O‐Step Secant, conjugate gradient with Polak‐Ribiére updates, and conjugate gradient with Fletcher‐Reeves updates are able to handle all variations of noise and number of neurons in hidden layer, with good accuracy of estimation and low CPU time consumption. Under high computational resource LM algorithm can be used for all cases. Up to 0.99132 value of regression coefficient is obtained in mapping MLPNN model with 15 hidden neurons, indicating the high accuracy of the model. With the help of PAMANN model, highly accurate (absolute error 1.78%) estimation of parameters is obtained. The model can handle upto 1% noise in input data, while giving accurate results.

Research on Calibration Method for On-line Radiation Monitoring of γ-ray Source

In the on-line radiation monitoring of γ -ray source, the active online monitoring module can realize real-time recording of the radiation dose equivalent value and count rate in the detector. Because the amount of radiation at different measuring points is different, their values vary from place to place, so there is no exact value that is completely suitable for all kinds of measurement situations, so the nonlinear relationship between counting rate and dose rate is deduced. For the difference of the dose at the measuring point, the relationship between them will increase with the continuous accumulation of radiation dose, which will have an impact on the iterative algorithm. In this paper, the LM algorithm is used to solve the fitting function model, and the feasibility of the algorithm is verified by MATLAB software programming. The results show that the parameters of the objective function model can be obtained by the LM algorithm, and the relative error of the eighth polynomial can be minimized and a better fitting effect can be achieved.

Evaluation and Analysis of an Industrial Cluster Based on the BP Neural Network and LM Algorithm

With the development of global economy, the evaluation of industrial clusters has become an important method to scientifically analyze the advantages and disadvantages of industrial development. The current research model uses the traditional evaluation method, which leads to the problem of unsatisfactory evaluation results and low effect. This paper proposes an evaluation model based on the BP neural network combined with the LM algorithm, which has the advantages of fast convergence speed and strong application ability. The comprehensive evaluation model of industrial clusters is put forward from the comprehensive application of the scale, benefit, and 27 related evaluation indexes of industrial clusters. This paper takes Beijing, Tianjin, and Hebei industrial clusters for BP-LM evaluation and analysis, which fully illustrates the advantages of this method. The evaluation results show that the evaluation target value and the average error of the overall parity have obvious advantages compared with those of other models, which provides application guidance for local economic development and policy formulation.

Prediction method of energy consumption for high building based on LMBP neural network

With the power consumption of high-rise buildings rising year by year, energy consumption of high-rise buildings has become the main object of energy-saving supervision and transformation, and its prediction has become a research hotspot. Aiming at the problem that the existing total energy consumption forecasting methods cannot accurately distinguish the consumption direction of high-rise building and the forecasting accuracy is low. LM algorithm is used to improve the standard BP neural network, and a prediction model of lighting energy consumption for high-rise buildings based on LMBP neural network is established. According to the prediction results of lighting energy consumption, hourly average outdoor temperature, hourly average outdoor relative humidity, weather characteristics, holidays, hourly average wind speed and 24 hourly hours a day, the energy consumption of air conditioning, power consumption and special energy consumption can be predicted separately. The experimental results show that the MAE and RMSE predicted by the design model for air conditioning energy consumption, power energy consumption and special energy consumption are 5.4707 and 8.6575, 1.7916 and 2.9356, 1.6075 and 2.2178 respectively, which are smaller than the comparison model MAE and RMSE, which proves that the design model can more accurately and effectively predict the sub item energy consumption in the energy consumption of high-rise buildings.

Ligand-based modeling of semicarbazones and thiosemicarbazones derivatives as Cathepsin B, H, and L inhibitors: A multi-target approach

This work describes the development, validation, and comparison of three different multi-target artificial neural network models to explore quantitative structure-activity relationships (mt-QSAR) forin-silico design of compounds containing carbonyl and thiocarbonyl moiety, as inhibitors against the related cysteine proteases, cathepsin B, H, and L. The proposed three models are based on back propagation neural network (trained with LM and SCG algorithm) and radial basis neural network (RBNN). The validation of the model was carried out by using an external test set which showed (r2 (ANN_LM) =0.752, r2 (ANN_SCG)=0.661, r2 (RBNN)=0.45) for these mt-QSAR models. As per results, RBNN is good at the fitting stage but BPNN (LM) outperformed in predicting stage than the rest of the two. The applicability domain of the proposed models is evaluated by the leverage approach. Based on SAR outcome, some new inhibitors are also designed and their activities predicted using the best proposed model. In addition, molecular docking studies of the best selected inhibitors have been performed to determine the optimum binding conformation. Binding energy values are in accordance with inhibitory activity values. For the first time, molecular modeling studies are used to explore the bond rearrangement after ligand-cathepsin interaction to display the interactions with amino acid residues of the receptor. The reported inhibitors also complied with Lipinski's rule of five.

Gaussian Inflection Point Selection for LiDAR Hidden Echo Signal Decomposition

High-quality waveform decomposition, as one of the most critical cores of light detection and ranging (LiDAR) data processing, has become increasingly interesting. However, the current Gaussian decomposition method cannot handle the superimposed waveform with only one peak. Thus, this letter proposes a Gaussian inflection point selection method (GIPS). The method uses the number of inflection points (IPs) near the peak of the echo signal to judge the position of waveform half-width and selects an appropriate waveform half-width to iteratively decompose the echo signal to obtain Gaussian components, which are combined into a Gaussian model. Finally, a global Levenberg–Marquardt least-square algorithm (LM algorithm) is used to optimize the Gaussian model for fitting the echo signal. To verify the accuracy and effectiveness of GIPS, the experiments were conducted using land, vegetation and ice sensor (LVIS) data. The results show that the GIPS method can decompose complex LiDAR echo signals more correctly and efficiently than other methods do with an average <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R^{2}$ </tex-math></inline-formula> of 0.9799.

A Statistical Model to Detect DRG Outliers

This study aims to develop a statistical model to detect both high and low outlier cases in terms of diagnosis-related group (DRG) distributions. A data set containing five DRGs with 458 patient cases was selected for the study. The distributions of DRG cost and length of stay (LOS) are examined firstly, and all the distributions of DRG costs are lognormal whereas all the distributions of LOS are not lognormal or normal. A statistical model referred to as LM is set out for outlier detection in terms of the lognormal distributions of DRG costs. The LM algorithm is compared with the geometric mean (GM), Inter-quartile (IQ) and L3H3 algorithms. LM has the highest statistics for the Accuracy, Kappa coefficient, Sensitivity and Youden’s index. In addition, LM has the largest area under the ROC curve (AUC). We find that LM is a superior method to detect both low and high outliers for DRG costs, thereby improving the efficiency and effectiveness of DRG prospective payment systems and equity of healthcare.

An hFFNN-LM Based Real-Time and High Precision Magnet Localization Method

Magnet tracking is an emerging tracking method for medical instruments and biological tissues, such as endoscopes and catheters. Traditional magnet tracking technology typically uses an optimization algorithm or heuristic algorithm to estimate the pose of a permanent magnet according to the measured magnetic induction intensity. However, this technology is limited by the high delay and the poor quality of the initial solution. This article proposes a new magnetic localization method, i.e., a hybrid feedforward neural network and Levenberg–Marquardt (hFFNN-LM) algorithm. FFNN provides an adjusted initial solution to the LM algorithm for solving the nonlinear inverse magnetic problem. The hFFNN-LM algorithm is promising for real-time magnetic tracking as it provides satisfactory tolerance for initially estimated parameters and fast search speed. The average single localization time is 1.07 ms. The experimental results show that the proposed method has a very low tracking delay compared to some other popular LM-based methods. The proposed hFFNN-LM algorithm achieves superior localization performances, i.e., the mean position and orientation errors are tested to be 0.70 mm and 0.90°, respectively.

Determining diffusion coefficients of coal particles by solving the inverse problem based on the data of methane desorption measurements

The diffusion coefficient is the key parameter been used to evaluate the methane diffusion behavior. Generally, the diffusion coefficient is determined by analytical solutions, including the unipore diffusion model and the bidisperse diffusion model. However, the essential assumptions limit the applicability and accuracy of analytical solutions. Inverse modeling has been proved to be useful in a variety of disparate fields. In this work, the detailed process and principles of using inverse modeling in gas diffusion are introduced. The effectiveness and accuracy of MMA, SNOPT and LM algorithms in determining the diffusion coefficient are analyzed. Results show that the inverse problem is a feasible method and the LM algorithm takes the shortest time and is recommended. The analytical solution (diffusion coefficient) of particles whose sizes are 0.074–0.2 mm, 0.25–0.5 mm and 0.5–1 mm are 13.8% times, 5.0% times and 1.6% times larger than that of the numerical solution, respectively. The numerical method has higher computing accuracy of small coal particles. As the current work is the first to present efficient techniques for the solution of diffusion coefficient-related inverse problems in COMSOL, the applicability of inverse modeling in complicated gas diffusion problems has also been prospected.

Reverse Logistics Enterprise Performance Research Based on Super-Efficiency DEA and LMBP Neural Network

On the basis of considering the characteristics of reverse logistics enterprises, this paper uses stakeholder theory to establish an evaluation system containing 36 indicators in five dimensions, based on the balanced scorecard, and adopts PCA optimization index. In view of the performance evaluation problem, first, the performance data is obtained by using the super-efficiency DEA, then the enterprise performance of A company is fitted by LMBP neural network to obtain the quantitative model of performance evaluation, and finally, according to the resulting weight reference performance benchmark, the direction of reverse logistics enterprise performance improvement is given. The simulation results show that the index system can react to the reverse logistics characteristics very well, and the BP neural network based on LM algorithm can accurately and efficiently evaluate the performance of reverse logistics enterprises which has better generalization, and can find out the key performance indicators and give reasonable optimization direction for enterprises by reverse tracking the weights of the neural network.

A credit risk assessment model of borrowers in P2P lending based on BP neural network.

Peer-to-Peer (P2P) lending provides convenient and efficient financing channels for small and medium-sized enterprises and individuals, and therefore it has developed rapidly since entering the market. However, due to the imperfection of the credit system and the influence of cyberspace restrictions, P2P network lending faces frequent borrower credit risk crises during the transaction process, with a high proportion of borrowers default. This paper first analyzes the basic development of China’s P2P online lending and the credit risks of borrowers in the industry. Then according to the characteristics of P2P network lending and previous studies, a credit risk assessment indicators system for borrowers in P2P lending is formulated with 29 indicators. Finally, on the basis of the credit risk assessment indicators system constructed in this paper, BP neural network is built based on the BP algorithm, which is trained by the LM algorithm (Levenberg-Marquardt), Scaled Conjugate Gradient, and Bayesian Regularization respectively, to complete the credit risk assessment model. By comparing the results of three mentioned training methodologies, the BP neural network trained by the LM algorithm is finally adopted to construct the credit risk assessment model of borrowers in P2P lending, in which the input layer node is 9, the hidden layer node is 11 and output layer node is 1. The model can provide practical guidance for China and other countries’ P2P lending platforms, and therefore to establish and improve an accurate and effective borrower credit risk management system.

Rapid Modeling of the Sound Speed Field in the South China Sea Based on a Comprehensive Optimal LM-BP Artificial Neural Network

Ocean sound speed is an essential foundation for marine scientific research and marine engineering applications. In this article, a model based on a comprehensive optimal back propagation artificial neural network model is developed. The Levenberg–Marquardt algorithm is used to optimize the model, and the momentum term, normalization, and early termination method were used to predict the high precision marine sound speed profile. The sound speed profile was described by five indicators: date, time, latitude, longitude, and depth. The model used data from the CTD observation dataset of scientific investigation over the South China Sea (2009–2012) (108°–120° E, 6°–8° N), which includes comprehensive scientific investigation data from four voyages. The feasibility of modeling the sound speed field in the South China Sea is investigated. The proposed model uses the momentum term, normalization, and early termination in a traditional BP artificial neural network structure and mitigates issues with overtraining and difficulty when determining the BP neural network parameters. With the LM algorithm, a fast-modeling method for the sound field effectively achieves the precision requirement for sound speed prediction. Through the prediction and verification of the data from 2009 to 2012, the newly proposed optimized BP network model is shown to dramatically reduce the training time and improve precision compared to the traditional network model. Results showed that the root mean squared error decreased from 1.7903 m/s to 0.95732 m/s, and the training time decreased from 612.43 s to 4.231 s. Finally, the sound ray tracing simulations confirm that the model meets the accuracy requirements of acoustic sounding and verify the model’s feasibility for the real-time prediction of the vertical sound speed in saltwater bodies.

Modeling and Forecasting Long-Term Records of Mean Sea Level at Grand Isle, Louisiana: SARIMA, NARNN, and Mixed SARIMA-NARNN Models

This study tried to demonstrate the role of time series models in modeling and forecasting process using long-term records of monthly mean sea level from January 1978 to October 2020 at Grand Isle, Louisiana. Following the Box–Jenkins methodology, the ARIMA(1,1,1)(2,0,0)12 with drift model was selected to be the best fit model for the time series, according to its lowest AIC value. Using the LM algorithm, the results revealed that the NARNN model with 9 neurons in the hidden layer and 6 time delays provided the best performance in the nonlinear autoregressive neural network models at its smaller MSE value. The Mixed model, a combination of the SARIMA and NARNN models has both linear and nonlinear modelling capabilities can be a better choice for modelling the time series. The comparative results revealed that the Mixed-LM model with 9 neurons in the hidden layer and 3 time delays yielded higher accuracy than the NARNN-LM model with 9 neurons in the hidden layer and 6 time delays, and the ARIMA(1,1,1)(2,0,0)12 with drift model, according to its lowest MSE in this study. Thus, this study may provide an integrated modelling approach as a decision-making supportive method for formulating local mean sea level forecast in advance. Understanding past sea level is important for the analysis of current and future sea level changes. In order to sustain these observations, research programs utilizing the resulting data should be able to improve significantly our understanding and narrow projections of future sea level rise and variability.

Forecasting of water table fluctuation for Priyadarshini watershed using Ann with different algorithms

Groundwater is an important natural resource essential for sustenance of life. Over 98% of the freshwater on the Earth lies below its surface. The ANN technique is applied as a new approach and an attractive tool to study and predict groundwater levels without applying physically based hydrologic parameters. It is observed that the maximum LM value of R for training and validation are 0.908 and 0.903 shown in well 2 (2-9-1) and testing is 0.949 shown in well 8 (2-9-1), whereas minimum value for training is 0.684 shown in well 1 (4-4-1) for validation and testing are 0.159 and 0.773 well 9 (3-5-1) The maximum observed R value for CG training, validation and testing are 0.76 shown in well 3 (4-5-1), 0.85 shown in well 9(3-5-1) and 0.891 shown in well 7 (2-5-1) whereas the minimum value for training and testing are 0.671 and 0.458 shown in well 1 (4-4-1) for validation the minimum R value is 0.638 well 2 (4-7-1). Considering training, validation and testing period and all the statistics it is difficult to say which algorithm is better among the two selected for study. Because there was a lot of variation in all the statistics among the two selected algorithms for training, validation and testing period. But considering the testing period of all the nine wells it was found that LM algorithm was better than CG for wells i.e., well 1 (2-9-1), well 2 (2-9-1), well 3 (1-8-1), well 4 (1-6-1), well 5 (2-9-1), well 6 (1-9-1), well 8 (2-9-1) while CG algorithm was better than LM for wells i.e., well 7 (2-5-1) and well 9 (3-5-1) So these algorithms for particular well were selected for sensitivity analysis.

Audit credit guarantee risk identification method based on LM algorithm

Aiming at the problem of low recognition accuracy caused by the high probability of training model falling into local extreme value in audit credit guarantee risk identification, an audit credit guarantees risk identification method based on LM algorithm is proposed. The risk identification index system is established and its weight is assigned to provide the index basis of the input layer for the identification model. The BP neural network is improved by LM algorithm to reduce the probability of falling into local extreme value, and the audit credit guarantee risk identification model is established by using the improved network. Using the same audit credit guarantee risk sample to test the identification method, the identification accuracy of this method is 95.15%, which is 4.86% and 4.61% higher than that based on BP neural network and LVQ network respectively. Therefore, the identification results of the design method are more feasible.

A Non-Destructive Electromagnetic Sensing Technique to Determine Chloride Level in Maritime Concrete

Deterioration of concrete due to the corrosion of reinforcement is a serious durability problem faced by the construction industry. This study aims to develop a non-destructive and real-time technique to monitor the chloride level at an early stage to prevent the development of the reinforcement’s corrosion using electromagnetic spectroscopy. The experimental work was performed on 5 concrete specimens with different chloride levels at three different concrete depths (18, 40, and 70mm) using the sensor system (2-12GHz frequency range) and a chlorometer. The LM algorithm was selected to develop a prediction model for the detection of chloride ions. The results demonstrated that the proposed technique can predict the chloride level with R2=0.986709 and RMSE =0.000120 at 5.42GHz. The results demonstrate that the sensor can predict the chloride ion content across the range of the investigated concrete depths with the percentage error of 0.034% with respect to the accuracy of the chlorometer.

A simple-to-use nomogram for reference evapotranspiration assessment using multivariate models: Study on outdoor and greenhouse environments

Experimental methods of reference evapotranspiration (ET 0 ) calculation require extensive environmental data, not readily attainable everywhere. Therefore, two multivariate models using Levenberg–Marquardt nonlinear least-squares algorithm were developed in this research to estimate daily ET 0 by focusing on fewer meteorological variable inputs. Different meteorological parameters were measured for two consecutive years in outdoor and greenhouse environments. The ET 0 models were developed for the aforementioned environments in terms of three major parameters, namely air temperature, relative humidity and pan evaporation. To construct the ET 0 models, three steps were carried out, (i) outdoor wind speed was estimated in terms of the major parameters with appropriate accuracy (R 2 = 0 . 42 , AR 2 = 0 . 4 and RMSE = 0.176 m/s), (ii) net radiation was estimated in terms of the major parameters and estimated wind speed with high accuracy (their R 2 and AR 2 > 0.47 and RMSE < 2.55 MJ/(m 2 -day)), and (iii) ET 0 was accurately estimated in terms of the major parameters, estimated wind speed and net radiation. Our ET 0 models (R. & M. models) were compared with nine common methods for both conditions according to eight statistical indices. The results indicated that in the outdoor condition, FPM and R. & M. were the best models, respectively, while Makkink indicated the poorest estimations. In the greenhouse condition, the R. & M. model was the best, FPM was the third-best and Linacre was the worst method for ET 0 estimation. However, R. & M. models were significantly appropriate for ET 0 assessment, they were multivariate models, so the nomography method was used as a user-friendly method to illustrate the estimations. The difference between the model outputs and nomogram was negligible, indicating that the nomogram was suitable for ET 0 determination and can be used in the weather stations. • Multivariate ET 0 models were developed by focusing on fewer meteorological parameters. • LM algorithm successfully updated the ET 0 model coefficients. • Wind speed and net radiation were modeled as mediator variables for ET 0 modeling. • The R. & M. ET 0 models presented the most accuracy among different reported methods. • ET 0 nomograph provided for outdoor and greenhouse conditions showed good estimation.