Number Of Input Parameters Research Articles

River water quality index prediction and uncertainty analysis: A comparative study of machine learning models

The Water Quality Index (WQI) is the most common indicator to characterize surface water quality. This study introduces a new ensemble machine learning model called Extra Tree Regression (ETR) for predicting monthly WQI values at the Lam Tsuen River in Hong Kong. The ETR model performance is compared with that of the classic standalone models, Support Vector Regression (SVR) and Decision Tree Regression (DTR). The monthly input water quality data including Biochemical Oxygen Demand (BOD), Chemical Oxygen Demand (COD), Dissolved Oxygen (DO), Electrical Conductivity (EC), Nitrate-Nitrogen (NO3 -N), Nitrite-Nitrogen (NO2 -N), Phosphate (PO43-), potential for Hydrogen (pH), Temperature (T) and Turbidity (TUR) are used for building the prediction models. Various input data combinations are investigated and assessed in terms of prediction performance, using numerical indices and graphical comparisons. The analysis shows that the ETR model generally produces more accurate WQI predictions for both training and testing phases. Although including all the ten input variables achieves the highest prediction performance (R2test=0.98, RMSEtest=2.99), a combination of input parameters including only BOD, Turbidity and Phosphate concentration provides the second highest prediction accuracy (R2test=0.97, RMSEtest=3.74). The uncertainty analysis relative to model structure and input parameters highlights a higher sensitivity of the prediction results to the former. In general, the ETR model represents an improvement on previous approaches for WQI prediction, in terms of prediction performance and reduction in the number of input parameters.

Effect of number of neurons and layers in an artificial neural network for generalized concrete mix design

Selection of the number of neurons in different layers of an artificial neural network (ANN) is a key decision-making step involved in its successful training. Although the number of neurons in the input layer is decided by the number of input parameters, and similarly, in the output layer the number of neurons is fixed by the output parameters, however, the number of neurons in the hidden layer is not fixed, whereby the overall efficiency of the ANN depends upon the correct modeling of the realistic scenario in the form of a neural network. Concrete can be mixed in a huge variety of methods and compositions to achieve a specific or a combination of specific results. These results have been observed to repeat and follow patterns. Furthermore, for several types of concrete mixes, only actual physical trial data are available with no mix design method; this makes achieving different required results very complex and expensive task. Having an ANN to be able to predict a concrete mix to handle such variations will save a lot of time and money and will prove to be a universal mix design tool. A thorough analysis for deciding the number of neurons for mix design of concrete has been carried out in this research. The results of neural networks for a varying number of neurons and layers have been correlated. The data for a number of concrete mixes were sorted and normalized. Properties of concrete constituents like specific gravity of cement, coarse and fine aggregates, dry density of coarse and fine aggregates, cement type, type of mineral admixture, water-to-cement ratio, type and temperature of curing and hardened properties like compressive strength, modulus of elasticity and tensile strength were taken as input parameters. The contents of constituents of concrete such as water content, cement content, coarse aggregates content, fine aggregates content and content of mineral admixtures are regarded as output parameters. For 17 inputs and 5 outputs, it has been discovered that simple neural network with single or double hidden layers performed better that 3 or more such layers.

Fast approximations of spectral lineshapes to enable optimization of a filtered Rayleigh scattering experiment

Measuring the properties of scattered light is central to many laser-based gas diagnostic techniques, such as filtered Rayleigh scattering (FRS). Alongside the measurements, a model of the scattered light’s spectral lineshape is often used to extract quantitative information about the flow field like pressure, temperature, and velocity. In particular, the Tenti S6 or S7 model are frequently used to model the lineshape of Rayleigh–Brillouin (RB) scattered light. While accurate, it is well attested in the literature that these models can be computationally expensive when evaluated many times, for example, as part of iterative estimation or optimization routines. To overcome this, approximations of these spectral lineshape models can be used instead. In this paper, we develop a method called support vector spectrum approximation (SVSA). This method uses support vector regression and singular value decomposition to create efficient, accurate, and well-conditioned approximations of any existing spectral lineshape model. The SVSA framework improves upon existing approximation methods by allowing quick calculation of spectral lineshapes for arbitrary flow regimes with any number of input parameters over a wide range of values. We demonstrate the efficacy of SVSA in approximating coherent and spontaneous RB scattering spectra. In application, we use SVSA to optimize the design of a filtered Rayleigh scattering experiment of a complex shock-dominated flow. SVSA allows us to comprehensively minimize expected measurement uncertainty of number density and temperature for this experiment. It does this by enabling a high-resolution design of experiments that is otherwise intractable.

A New Objective Function for the Recovery of Gielis Curves

The superformula generates curves called Gielis curves, which depend on a small number of input parameters. Recovering parameters generating a curve that adapts to a set of points is a non-trivial task, thus methods to accomplish it are still being developed. These curves can represent a great variety of forms, such as living organisms, objects and geometric shapes. In this work we propose a method that uses a genetic algorithm to minimize a combination of three objectives functions: Euclidean distances from the sample points to the curve, from the curve to the sample points and the curve length. Curves generated with the parameters obtained by this method adjust better to real curves in relation to the state of art, according to observational and numeric comparisons.

Method to solve redundant inverse problems based on a Latent Semantic Analysis approach. Application to an aerojet engine

A method for the solution of a certain class of inverse problems is presented. In particular, the method is of interest in cases in which the operation of the engineering system under consideration is extensively monitored, in the sense that the number of output parameters is larger than the number of input parameters. This redundancy of output data often occurs in aerospace engineering applications. The method is based on the Latent Semantic Analysis formulation with a modification specifically adapted to the problem under consideration. The method belongs to the family of the so-called Reduced Order Models since the search for the input parameters is made on spaces whose dimensionality is lower than that of the original space. The main advantage of the method is that most of the computational effort can be shifted off-line, thereby allowing for its use in real time (on-line) applications. The method is described by addressing the inverse problem of input parameter characterization in an aerojet engine.

Machine-Learning-Based Early-Warning System Maintains Stable Production

This article, written by JPT Technology Editor Judy Feder, contains highlights of paper SPE 197365, “A Novel Early Warning System of Oil Production Based on Machine Learning,” by Kang Ma, Hanqiao Jiang, and Junjian Li, China University of Petroleum-Beijing, et al., prepared for the 2019 Abu Dhabi International Petroleum Exhibition and Conference, Abu Dhabi, 11–14 November. The paper has not been peer reviewed. After long-term waterflooding, a number of mature oil fields in China have entered the high-water-cut stage, and abnormal production decline has become the primary problem for stable production. This paper describes an accurate, three-step, machine-learning-based early warning system (EWS) that has been used to monitor production and guide strategy in the Shengli field. Adding artificial samples to the training process improved the system’s prediction accuracy greatly (Fig. 1). Introduction For conventional Chinese oil fields that have entered the high-water-cut stage after decades of waterflooding, stabilizing production has become increasingly difficult. After stimulation treatments throughout the field’s history, abnormal decline rates—that is, exceeding 5%—occurred more frequently. Production declined dramatically in 2004 and has not been maintained since then. To prevent future abnormal production declines, an effective EWS was needed that could release a production alarm to enable engineers to take preventive measures in advance. The complete paper includes a discussion of various early-warning models and their limitations. The paper discusses an EWS based on a neural network method using a previously established data set. Factors that can affect the abnormal decline were selected. The index sets of production composition and injected and produced water obtained from practical statistics were considered as the main assessment indicators. Grey relational analysis was used to evaluate the importance of the different indicators and to eliminate redundant parameters. Machine learning was adopted to build the EWS. Using the degree of deviation from normal as the input data for the prediction model provided the highest accuracy. However, the basic machine-learning model contains many input parameters that cannot be obtained easily. The number of input parameters was optimized on the basis of the variation of accuracy under different input parameter numbers. To improve prediction accuracy, artificial samples were added to the training process. The prediction accuracy of the final optimization model can reach 92%. The result reveals the possibility of the occurrence of anomalous decline in different reservoirs, which can guide oilfield production strategy effectively. The EWS was verified by oilfield production. Fundamentals of Neural-Network Classification Neural networks use existing data to determine the implicit model between input and output data. Classification and prediction are the most common applications. Neural networks are used in the industry to solve classification and regression problems.

Prediction of Nanofluid Temperature Inside the Cavity by Integration of Grid Partition Clustering Categorization of a Learning Structure with the Fuzzy System.

In this study, a quadratic cavity is simulated using computational fluid dynamics (CFD). The simulated cavity includes nanofluids containing copper (Cu) nanoparticles. The L-shaped thermal element exists in this cavity to produce heat distribution along with the domain. Results such as fluid velocity distribution in two dimensions and the fluid temperature field were generated as CFD simulation results. These outputs were evaluated using an adaptive neuro-fuzzy inference system (ANFIS) for learning and then the prediction process. In the training process related to the ANFIS method, x coordinates, y coordinates, and fluid temperature are three inputs, and the fluid velocity in line with Y is the output. During the learning process, the data have been classified using a clustering method called grid clustering. In line with the attempt to rise ANFIS intelligence, the alterations in the number of input parameters and of membership structure have been analyzed. After reaching the highest level of intelligence, the fluid velocity nodes were predicted to be in line with y, especially cavity nodes, which were absent in CFD simulations. The simulation findings indicated that there is a good agreement between CFD and clustering approach, while the total simulation time for learning and prediction is shorter than the time needed for calculation using the CFD method.

Autoclave operation simulation using iTthink software

This research article deals with an issue of temperature regulation in an autoclave. To solve this problem, it is proposed to use the simulation model of autoclave operation which can simulate the autoclave operation process with displaying the operation dynamics in a graphic form hour by hour depending on a number of input parameters as well as to simulate the mutual influence of control loops on each other.

Prediction and modeling of roughness in ball end milling with tool-surface inclination

The quality of mill finishing of shaped surfaces is influenced by a number of input parameters. Current automated manufacturing systems allow adaptation of the machining process aiming at the final surface quality. Despite all the advantages, these systems require a behavioural model, a prediction of the output, based on the input parameters. Some of these models are summarized in this paper, including contemporary evaluated models as well as their functional dependencies; moreover, offers an application of mill finishing with a ball-end cutter incorporating tool axis or surface inclination.

Power-draw prediction by random forest based on operating parameters for an industrial ball mill

Estimation of mill power-draw can play a critical role in economics, operation and control standpoints of the entire mineral processing plants since the cost of milling is the single biggest expense within the process. Thus, several empirical power-draw prediction models have been generated based on a combination of laboratory, pilot and full-scale measurements of different milling conditions. However, they cannot be used in industrial plants, where in full-scale operations, only not few numbers of input parameters used in those models are measured. Moreover, empirical models do not assess the relationship between input features. This investigation is going to introduce random forest (RF), as a predictive model, beside of its associated variable importance measures system, as a sensible means for variable selection, to overcome drawbacks of empirical models. Although RF as a powerful modeling tool has been used in several problem solving systems, it has not comprehensively considered in the powder technology areas. In this investigation, an industrial ball mill database from Chadormalu iron ore processing plant were used to develop a RF model and explore relationships between power-draw and other monitored operating parameters. Modeling results indicated that RF can highly improve the prediction accuracy of power-draw as compared to the regression as a typical method (R2: 0.98 vs. 0.60, respectively) and rank operational milling parameters based on their importance.

Process simulation modeling of phytophagans' influence on the yield of seed fruit crops

Simulation modeling of phytophagans’ influence on the yield of seed fruit crops, in particular apple trees, was carried out. By means of simulation models the importance of phytophagans’ influence at different stages of the vegetation period and the period of fruit ripening was revealed. The software package Matlab was used to build simulation models. As a result, simulation models with nonlinear characteristics were obtained, which maximally reflected the studied processes. The developed models imitate the process of phytophagans’ development. Generation change of pests and all stages of their development are simulated. Their respective numbers are recorded at each stage for all generations. The development process at each stage is modeled by separate subsystems of the simulation model. To simulate the development of one generation of pests, these subsystems are connected by external links. In addition, part of the relationships provides a simulation of generational change. There are a number of input parameters that allow to configure the simulation of the process of changing generations, taking into account the peculiarities of the development of various phytophagans.

New Artificial Neural Networks Model for Predicting Rate of Penetration in Deep Shale Formation

Rate of penetration (ROP) means how fast the drilling bit is drilling through the formations. It is known that in the petroleum industry, most of the well cost is taken by the drilling operations. Therefore, it is very crucial to drill carefully and improve drilling processes. Nevertheless, it is challenging to predict the influence of every single parameter because most of the drilling parameters depend on each other and altering an individual parameter will have an impact on the rest. Due to the complexity of the drilling operations, up to the present time, there is no reliable model that can adequately estimate the ROP. Artificial intelligence (AI) might be capable of building a predictive model from a number of input parameters that correlate to the output parameter. A real field dataset, of shale formation, that contains records of both drilling parameters such as, rotation per minute (RPM), weight on bit (WOB), drilling torque (τ), standpipe pressure (SPP) and flow pump (Q) and mud properties such as, mud weight (MW), funnel and plastic viscosities (FV) (PV), solid (%) and yield point (YP) were used to predict ROP using artificial neural network (ANN). A comparison between the developed ANN-ROP model and the number of selected published ROP models were performed. A novel empirical equation of ROP using the above-mentioned parameters was derived based on ANN technique which is able to estimate ROP with excellent precision (correlation coefficient (R) of 0.996 and average absolute percentage error (AAPE) of 5.776%). The novel ANN-based correlation outperformed three published empirical models and it can be used to predict the ROP without the need for artificial intelligence software.

Effect of Deterministic and Continuous Design Space Resolution on Multiple-Objective Combustor Optimization

Abstract Supervised machine learning is used to classify a continuous and deterministic design space into a nondominated Pareto frontier and dominated design points. The effect of the initial training data quantity on the Pareto frontier and output parameter sensitivity is explored. The study is performed with the optimization of a subsonic small-scale cavity-stabilized combustor. A 3D geometry is created and parameterized using computer aided design (CAD) that is combined with a software for meshing, which automatically transfers grids and boundary conditions to the solver and postprocessing tool. Steady, compressible three-dimensional simulations are conducted employing a multiphase Realizable k–ε Reynolds-averaged Navier–Stokes (RANS) physics with an adiabatic flamelet progress variable (FPV) model. The near-wall turbulence modeling is computed with scalable wall functions (SWFs). For each computational fluid dynamics (CFD) simulation, four levels of adaptive mesh refinement (AMR) are utilized on the original cut-cell grid. There are 15 geometrical input parameters and three output parameters, viz., a pattern factor proxy, a combustion efficiency proxy, and total pressure loss (TPL). Three times the number of input parameters plus one (48) is necessary to yield an optimization independent of the initial sampling. This conclusion is drawn by examining and comparing the Pareto frontiers and global sensitivities. However, the latter provides a better metric. The relative influence of the input parameters on the outputs is assessed by Spearman's order-rank correlation and an active subspace analysis. Some persistent geometric features for nondominated designs are also discussed.

A prediction method for voltage and lifetime of lead–acid battery by using machine learning

Lead–acid battery is the common energy source to support the electric vehicles. During the use of the battery, we need to know when the battery needs to be replaced with the new one. In this research, we proposed a prediction method for voltage and lifetime of lead–acid battery. The prediction models were formed by three kinds mode of four-points consecutive voltage and time index.The first mode was formed by four fixed voltages value during four weeks, namely M1. The second mode was formed by four previous voltage values from prediction time, namely M2. Third mode was formed by the combinations of four previous data with the last predicted data, namely M3. The training data were recorded from 10 lead–acid batteries. We separated between training data and testing data. Data collection for training were recorded in 155 weeks. The examined data for the model was captured in 105 weeks. Three of batteries were selected for prediction. Machine learning methods were used to create the batteries model of voltage and lifetime prediction. Convolutional Neural Network was selected to train and predict the battery model. To compare our model performance, we also performed Multilayer Perceptron with the same data procedure. Based on experiment, M1 model did not achieve the correct prediction besides the linear case. M2 model successfully predicted the battery voltage and lifetime. The M2 curve was almost the same with real-time measurement, but the curve was not fitting smoothly. M3 model achieved the high prediction with smooth curve. According to our research on lead–acid battery voltage prediction, we give the following conclusions and suggestions to be considered. The accuracy of prediction is affected by the number of input parameters is used in prediction. The input parameters need to have time consecutive.

Uncertainty Quantification Using the MAAP5 Code of In-Vessel Hydrogen Generation in a Severe Accident at an Advanced Boiling Water Reactor

MAAP5 is an integral severe accident analysis program that simulates the responses of a light water reactor power plant during a severe accident. This program has been used extensively for probabilistic safety assessments, verification and validation of mitigation actions specified in severe accident management guidelines, and source term quantification. In this study, the uncertainty of in-vessel hydrogen generation predicted by the MAAP5 code was quantified. The surrogate plant that was analyzed is the Lungmen Nuclear Power Station of the Taiwan Power Company. The plant employs an advanced boiling water reactor. We performed sensitivity studies to identify the important model parameters that affect the target output parameters. A range and distribution were assigned to these parameters on the basis of experimental results and expert judgment. The number of input parameters in the analysis was 27. Multiple MAAP5 calculations were performed with an input combination generated from Latin hypercube sampling. The calculation results were analyzed parametrically and nonparametrically to determine the 95th percentile with the 95% confidence level value of the amount of in-vessel hydrogen generation. The Pearson correlation coefficient was used to determine the effect of the model parameters on the target output parameters. The analysis results provide guidance for code applications. The only parameters that pass the threshold of 0.362 for hydrogen generation in the core are FCO and TCLMAX. For hydrogen generation in the lower plenum, FOXBJ is the only input parameter that passes the threshold.

An electronic expansion valve modeling framework development using artificial neural network: A case study on VRF systems

Electronic expansion valves (EEV) are widely used in variable refrigerant flow systems (VRF) to control the mass flow rate of each indoor unit. EEV model is used to predict the mass flow rate through an EEV. While the power-law correlation method has been used to build the EEV model so far, Artificial Neural Network (ANN) methods have been adapted to model the EEV with a fixed speed compressor thanks to its higher accuracy. However, the EEV is typically working with the variable speed compressor in a VRF system. In addition, the parameters used in ANN modeling could be further optimized. The objective of this study is to develop an EEV modeling framework and optimize the input parameter number and hidden neuron number. We presented the framework through an EEV model development for a VRF system. For these, we used the field test data, applied a principal components analysis approach in optimizing the ANN input parameter number, and investigated the proper number of hidden neurons and appropriate transfer function pairs. We found the performance of the ANN model would not improve much as the number of input parameters and the number of hidden neurons reached a threshold. Only three transfer function pairs are suitable for this case in total nine groups we studied. Our ANN model has the average absolute deviation of 2.2%. The framework can be easily applied to EEV modeling in other VRF and air condition systems with necessary data support.

Stochastic and recursive estimation of the hygro-thermo-chemical-mechanical parameters of concrete through Monte Carlo analysis and extended Kalman filter

Hygro-thermo-chemical-mechanical models, used to determine the variations over time of temperature, relative humidity and shrinkage induced deformations in concrete components, are characterised by the presence of a large number of input parameters. Some of these parameters can be evaluated on the basis of the concrete mix specifications or from literature data, while the others present a large variability and, in some cases, do not have a precise physical meaning and, for this reason, require the implementation of proper identification strategies. The experimental work involved for this characterisation can be time-consuming and costly because based on the long-term monitoring of the time evolution of the field quantities in specific positions within concrete components. The aim of this paper is to propose and validate recursive identification strategies that exploit, in a step by step fashion, the information coming from the experimentation for the identification of the model input parameters. The influence of different exposure conditions and of different concrete thicknesses are investigated and, for each scenario considered, the expected identification error of each parameter is estimated, within a stochastic context implemented through Monte Carlo analyses and Kalman Filter, as a function of the monitored time.

Численное моделирование распространения пламени по дискретной совокупности горючих материалов

Introduction. Fires at high-rack storages with a discrete fire load develop dramatically fast, while fire detection and suppression systems might activate too late, which leads to significant property damage. The objective of this work is to demonstrate applicability of the thermal pyrolysis model in predictions of fire development in high-rack storage facilities. Methods. The simulations are performed using FDS 6.6. In the thermal pyrolysis model, the solid material is ex­posed to inert heating until its surface temperature reaches the ignition temperature; combustible material then ignites and burns at a prescribed burning rate. The advantage of this approach is its simplicity and a limited number of input parameters, which include ignition temperature, mass loss rate per unit area, burn-out time and heat of gasification. The model parameters selection procedure is based on literature and experimental data. In this work, FDS simulations are performed for 3- and 5-tier high racks with 2 rows of cardboard boxes (243 = 24 and 245 = 40 boxes in total). Fire suppression systems are not activated. Results and discussions. Simulations results show that high-rack storage fire dynamics can be replicated using thermal pyrolysis model provided that model parameters are properly selected. Fire growth mechanisms include upward and horizontal flame spread over the combustible surfaces. Net heat flux and surface temperature distributions, in-rack gas velocity and temperature are also reported. When number of tiers is increased to 5 the heat ­release rate grows faster compared to the 3-tier case. Conclusions. Thermal pyrolysis model enables reasonable replication of high-rack storage fire dynamics, which is proven by comparison with the full-scale experimental data. The model could be used to simulate fire dynamics in rack storages of different configurations at different ceiling heights, with the purpose of predicting fire detection and the performance of fire suppression systems.

Data-analysis and modelling of the effect of inter-row shading on the power production of photovoltaic plants

The aim of this work is to propose a fast and easily implementable model allowing the consideration of inter-row shading in the simulation of the power production of PV plants. Our objective is to find the simplest model with the least number of input parameters. A formulation of the shaded fraction of the module area dependent on the PV configuration and the sun position is first proposed. An analysis of the measurements of two plants is then carried out to understand the effect of inter-row shading for a complete PV array. Based on this analysis, a model is proposed and validated. It is shown that the consideration of inter-row shading with the proposed model allows a relative improvement of the simulation accuracy (RMSE) ranging between 18 and 32% for time steps affected by inter-row shading and between 7 and 12% for a complete year.

Applicability and comparison of surrogate techniques for modeling of selected heating problems

Possibilities of using surrogate techniques for modeling selected strongly nonlinear coupled problems of heating are evaluated. The main purpose is to significantly reduce the computing time in the case of computations of many variants of a given task by the finite element method on the condition of obtaining results of a still acceptable accuracy. Frequently used surrogate techniques (based on Kriging, neural network etc.) are tested on a particular problem of induction-assisted laser welding that represents a very complicated 3D problem. Here, the most important output quantities are the internal structure of weld (decisive for its mechanical parameters) and its depth that depend on a number of input parameters (power of laser beam, velocity of shift of the welded parts, overall geometry and material properties etc.) and must be known before the process of welding itself. The paper presents both full model of this process and considered surrogate algorithms, and compares the results obtained. It is shown that a careful selection of the surrogate technique together with suitable choice of its input data is very beneficial and may result in high savings in design of the process. Implementation performance and suitability of particular techniques of this kind are also evaluated.