Kind Of Optimization Algorithm Research Articles

Three Steps towards Better Forecasting for Streamflow Deep Learning

Elevating the accuracy of streamflow forecasting has always been a challenge. This paper proposes a three-step artificial intelligence model improvement for streamflow forecasting. Step 1 uses long short-term memory (LSTM), an improvement on the conventional artificial neural network (ANN). Step 2 performs multi-step ahead forecasting while establishing the rates of change as a new approach. Step 3 further improves the accuracy through three different kinds of optimization algorithms. The Stormwater and Road Tunnel project in Kuala Lumpur is the study area. Historical rainfall data of 14 years at 11 telemetry stations are obtained to forecast the flow at the confluence located next to the control center. Step 1 reveals that LSTM is a better model than ANN with R 0.9055, MSE 17,8532, MAE 1.4365, NSE 0.8190 and RMSE 5.3695. Step 2 unveils the rates of change model that outperforms the rest with R = 0.9545, MSE = 8.9746, MAE = 0.5434, NSE = 0.9090 and RMSE = 2.9958. Finally, Stage 3 is a further improvement with R = 0.9757, MSE = 4.7187, MAE = 0.4672, NSE = 0.9514 and RMSE = 2.1723 for the bat-LSTM hybrid algorithm. This study shows that the δQ model has consistently yielded promising results while the metaheuristic algorithms are able to yield additional improvement to the model’s results.

Variational parameter estimation in a two-equation turbulence model: A case study with a 3D primitive-equation ocean model

A three-dimensional and complete adjoint model of the Princeton Ocean Model with a generalized coordinate system (POMgcs) is developed to construct the 4D-Variational data assimilation (4D-Var) algorithm in this study. Uncertain parameters in the Mellor-Yamada 2.5 turbulence submodel (MY-2.5) which is enclosed in POMgcs, are tentatively estimated via the 4D-VAR algorithm within a biased model framework. Here, the control variables in the biased model are set to two uncertain wave-affected parameters (wave energy factor α and Charnock coefficient β ) in the MY-2.5 turbulence model, which play a crucial role in modulating the heat content distribution in the upper coastal sea. First of all, the ocean temperature and salinity in a typical coastal sea, Bohai Sea, are simulated by the model to validate the rationally of the MY-2.5 parameterization scheme for both constructing the “truth model” and generating the “pseudo-observations” in the data assimilation studies. Then, after thoroughly testing the ability of the 4D-Var to optimize the initial state fields of the POMgcs model, a series of parameter estimation experiments are carried out to investigate whether and to what degree the parameters embedded in high-order turbulence models can be significantly optimized. Results of parameter estimation with perfect initial fields show the two estimated parameters in the MY-2.5 submodel can successfully converge to the “truth” value. The local minimum of the cost function can be effectively and efficiently jumped out once two kinds of optimization algorithms, LBFGS and LMBM, are jointly used. In addition, the estimated parameter will converge to the optimal value rather than the truth one to compensate for the initial field error when the state-parameter are estimated simultaneously. Further, the performance of the parameter estimation is also deeply discussed when the observation samples are noised. Finally, prescribing the initial field and parameter as error source, a forecasting experiment for sea temperature is performed. The experiment results indicate that assimilating “pseudo-observations” to the model based on 4D-Var can significantly improve the sea temperature simulation. Moreover, adjusting the initial field and parameter leads to a better result than the only initial field, and this conclusion is more evident at the surface than in the deeper ocean.

Evaluation of Optimization Algorithms and Noise Robustness of Sparsity-Promoting Dynamic Mode Decomposition

In the present study, we organize the existing sparsity-promoting dynamic mode decomposition (DMDsp) in terms of noise robustness, propose faster optimization algorithm for DMDsp, and evaluate its characteristics. Two kinds of DMDsp, namely system-based DMDsp (sDMDsp) and observation-based DMDsp (oDMDsp), combined with three kinds of optimization algorithm, namely the fast iterative shrinkage thresholding algorithm (FISTA), the alternating direction method of multipliers (ADMM), and a greedy algorithm, are investigated. For both sDMDsp and oDMDsp, FISTA yields the shortest processing time. The processing time for sDMDsp with FISTA is shorter than that for oDMDsp with FISTA. The original data reconstruction errors for sDMDsp and oDMDsp are similar among the three optimization algorithms. The noise robustness for sDMDsp and oDMDsp is evaluated. sDMDsp and oDMDsp have similar robustness to observation noise, except for a system with large system and observation noise, for which oDMDsp outperforms sDMDsp.

Research on mechanical optimization methods of cable-driven lower limb rehabilitation robot

AbstractIn order to improve the working performance of the lower limb rehabilitation robot and the safety of the trained object, the mechanical characteristics of a cable-driven lower limb rehabilitation robot (CDLR) are studied. The dynamic model of the designed CDLR was established. Four kinds of cable tension optimization algorithms were proposed to obtain a good rehabilitation training effect, and the quality of the feasible workspace of the CDLR was analyzed. Finally, a real-time evaluation index of the cable tension optimization algorithms was given to measure the calculation speed of the optimization algorithms. The numerical research results were provided to confirm the characteristics of the four kinds of the optimization algorithms. The research results provide a basis for the follow-up research on the safety and compliance control strategy of the CDLR system.

Improving Ant Collaborative Filtering on Sparsity via Dimension Reduction

Recommender systems should be able to handle highly sparse training data that continues to change over time. Among the many solutions, Ant Colony Optimization, as a kind of optimization algorithm modeled on the actions of an ant colony, enjoys the favorable characteristic of being optimal, which has not been easily achieved by other kinds of algorithms. A recent work adopting genetic optimization proposes a collaborative filtering scheme: Ant Collaborative Filtering (ACF), which models the pheromone of ants for a recommender system in two ways: (1) use the pheromone exchange to model the ratings given by users with respect to items; (2) use the evaporation of existing pheromone to model the evolution of users’ preference change over time. This mechanism helps to identify the users and the items most related, even in the case of sparsity, and can capture the drift of user preferences over time. However, it reveals that many users share the same preference over items, which means it is not necessary to initialize each user with a unique type of pheromone, as was done with the ACF. Regarding the sparsity problem, this work takes one step further to improve the Ant Collaborative Filtering’s performance by adding a clustering step in the initialization phase to reduce the dimension of the rate matrix, which leads to the results that K<<#users, where K is the number of clusters, which stands for the maximum number of types of pheromone carried by all users. We call this revised version the Improved Ant Collaborative Filtering (IACF). Experiments are conducted on larger datasets, compared with the previous work, based on three typical recommender systems: (1) movie recommendations, (2) music recommendations, and (3) book recommendations. For movie recommendation, a larger dataset, MoviesLens 10M, was used, instead of MoviesLens 1M. For book recommendation and music recommendation, we used a new dataset that has a much larger size of samples from Douban and NetEase. The results illustrate that our IACF algorithm can better deal with practical recommendation scenarios that handle sparse dataset.

A spatial fractional diffusion model for predicting the characteristics of VOCs emission in porous dry building material

Volatile organic compounds (VOCs) concentrations of dry building materials can significantly affect the quality of indoor air. In this paper, a spatial fractional diffusion model is established by introducing fractional Fick’s law, based on the fact that the pore structure of porous building materials has a strongly impact on VOCs diffusion. Furthermore, the areal porosity of the material is drawn into the convection mass transfer equation. The relevant parameters are estimated by a kind of optimization algorithm. The fractional nonlinear equations are tackled by the finite difference method combined with L2-algorithm. Results indicate that spatial fractional diffusion model agrees better with the experimental data than Deng and Kim’s model. The spatial fractional diffusion model is used to describe the formaldehyde concentration in the particleboard according to the experimental data. Taking formaldehyde as a common substance, the influences of physical parameters on VOCs diffusion are predicted by this model. The results of numerical simulation demonstrate that the diffusion is consistent with the actual situation.

Learning Deep Conditional Neural Network for Image Segmentation

Combining Convolutional Neural Networks (CNNs) with Conditional Random Fields (CRFs) achieves great success among recent object segmentation methods. There are two advantages by such usage. First, CNNs can extract low-level features, which are very similar to the extracted features in primates’ primary visual cortex (V1). Second, CRFs can set up the relationship between input features and output labels in a direct way. In this paper, we extend the first advantage by using CNNs for low-level feature extraction and a Structured Random Forest (SRF)-based border ownership detector for high-level feature extraction, which are similar to the outputs of primates secondary visual cortex (V2). Compared to the CRF model, an improved Conditional Boltzmann Machine (CBM), which has a multi-channel visible layer, is proposed to model the relationship between predicted labels, local and global contexts of objects with multi-scale and multilevel features. Besides, our proposed CBM model is extended for object parsing by using multivisible branches instead of a single visible layer of CBM, which cannot only segment the whole body but also the parts of the body under. These visible branches use each branch for the segmentation of the whole body or one of the body parts. All branches share the same hidden layers of CBM and train the branches under an iterative way. By exploiting object parsing, the whole body segmentation performance of object is improved. To refine the segmentation output, two kinds of optimization algorithms are proposed. The superpixel-based algorithm can re-label the overlapped regions of multiple kinds of objects. The other curve correction algorithm corrects the edges of segmented object parts by using smooth edges under a curve similarity criterion. Experiments demonstrate that our models yield competitive results for object segmentation on the PASCAL VOC 2012 dataset and for object parsing on the PennFudan Pedestrian Parsing dataset, Pedestrian Parsing Surveillance Scenes dataset, Horse-Cow parsing dataset, and PASCAL Quadrupeds dataset.

Green fronthaul allocation and power management in Cloud-RAN

In cloud radio access network (Cloud-RAN), most previous research aims to reduce the energy consumption of the total system and ignores specific requirements of energy-saving in mobile users. For the meantime, some efforts have been made to devise efficient signal quantization techniques for fronthaul capacity restriction to improve the performance of Cloud-RAN. To improve the user-centric energy efficiency (EE), we propose two kinds of optimization algorithms in orthogonal frequency division multiple access-based Cloud-RAN in this paper, whereby Gaussian quantization and uniform scalar quantization methods are utilized. For single-RRH scenario, alternative iterative optimization algorithms are proposed to maximize users’ EE, by jointly allocating resources. Furthermore, in more general scenario for multiple RRHs, authors investigate a modified particle swarm optimization (M-PSO) algorithm to solve more complex problems, which remain non-linear and non-convex. Simulation results demonstrate the proposed schemes in this paper outperform conventional solutions and improve EE of users.

Porcine automation: Robotic abdomen cutting trajectory planning using machine vision techniques based on global optimization algorithm

The purpose of this paper is to provide details on implementation of accurate and intelligent automation solution for porcine abdomen cutting while a pig is hung up by rear legs. The system developed utilized a 6-DOF industrial manipulators, customized tools, 2D camera and PC. Eye-to-hand calibrations built coordinate transformation relations of units in Cartesian space. The porcine abdomen curve was identified and fitted into quintic spline curve from image. Under cavum peritonaei constrains, optimal sectional trajectory was planned based on genetic algorithm (GA) by comparing several kinds of optimization algorithms. The results of experimental replications show that the system was successful both in following the varied position carcass and cutting open abdominal cavity without haslet damage. The system can enhance the quality, hygienic standard and efficiency of the process.

Constitutive parameter identification: An application of inverse analysis to the flow of cement-based suspensions in the fresh state from synthetic data

Rheometers with specific impellers, developed to characterize the behavior of cement-based suspensions in the fresh state, are used to limit the heterogeneities induced during shearing but they make the identification of the rheological parameters less straightforward compared to conventional rheometers. This paper presents the inverse analysis method and discusses the quality of this identification procedure when applied to such materials. The procedure includes a CFD simulation based on the finite element method using a Herschel–Bulkley model. Two kinds of optimization algorithms are used: a deterministic simplex method and a stochastic genetic method. As a first step of a larger study, the procedure reported in this paper used 2D synthetic data, i.e. 2D numerically generated experimental data. Two numerical viscoplastic materials, characterized by shear-thinning and shear-thickening, were selected and studied. The results obtained with the two algorithms are systematically compared to the known parameter solution. Three approaches corresponding to three levels of user’s knowledge about the material under study are considered successively: (1) the user has no a priori knowledge about the material, (2) the user knows whether the material is shear-thinning or shear-thickening and (3) the user is able to estimate the behavior index. The time-consuming genetic method appears to be suitable when the a priori knowledge of the material is slight, whereas the simplex method gives a reliable solution in a few iterations when the level of knowledge is higher. Both algorithms encounter more difficulties with the shear-thinning material than with the shear-thickening one. In this paper, the advantage provided by the genetic method regarding the non-uniqueness of the identification procedure for real experimental data is also highlighted: this method provides a collection of satisfactory solutions among which the user can select the optimal one based on his scientific background and/or on further experimental test results.

Progress and expectation of quantum dots converted light emitting diode package

Quantum dots converted light emitting diode (QCLED) is a new-style lighting device with quantum dots (QDs) as the down conversion material. The QDs with tunable spectrum, remarkable saturation and high quantum yields result in the excellent performance of color rendering and saturation as well as the gamut of QCLED. As a consequence, the QCLED is attracting more and more attentions nowadays. Different from the traditional phosphor powders which are the down conversion material of white LED, the quantum dots are usually stored up in the fluid or solid matrix, and the most common package type of QDs is to be blended with the high polymer and then packaged into the bare LEDs. However, there are 4 crucial problems in the process of packaging: (1) The poor compatibility of QDs and high polymer matrix. Although techniques for incorporating QDs in high polymer films are well developed, the coexistence of QDs and bulk polymer matrix remains to be strengthened, or there will be some problems such as low yield of film, cluster of QDs, fluorescence quenching, and so on. (2) The poor thermal stability of QCLED. When the temperature rises, the organic ligands growing on the surface of QDs will fall off or become inactivated, then the surface defects are exposed to matrix, which will result in the reduction of fluorescence efficiency. (3) The poor stability of QDs against oxygen and moisture. The penetrative oxygen and moisture will corrode the surface ions and ligands on the surface of QDs, consequently resulting in defect trap states and furthermore the degeneration of optical performance of QCLED. (4) The optimization of optical spectrum. There are usually 3 or more spectrum distributions for QCLED light, meaning that 2 or more kinds of QDs are needed to combine with the light extracted from LED chip. In order to obtain the high performance QCLED, the collocation of the QDs and LED chip should be quite well designed. Aiming at these problems, the researchers have tried a good deal of solutions: (1) Ligand modification, surface passivation, QDs microspheres and etc. are employed to enhance the compatibility between the QDs and polymer matrix. (2) Design of QDs films composition structure and optimization of QCLED’s package structure are applied to improve the thermal stability. (3) Improvement of QDs chemical stability and protection of QDs films are conducted to improve the stability of QCLED against oxygen and moisture. (4) Kinds of optimization algorithms are employed to design the spectra distribution, and many packaging experiments are conducted to obtain the QCLED with high performance, and finally the advantages of the QCLED are highlighted at the researchers’ efforts. We summarize the recent developments of the QCLED in this review. The strengths of QCLED in illumination and display are overviewed, and the common packaging processes are introduced. What’s more, the problems mentioned above are explained at length, and the corresponding solutions are discussed one by one. In the last decades, the QCLED has achieved significant developments, and reviewing on these progresses will prompt the QCLED to tend to be mature.

Prediction of Parameters of Soil Stratums and Earthen Dams from Free Field Acceleration Records

This paper is dedicated to the identification of parameters of soil stratums and earthen dams from accelerometer records by inverse analysis using three kinds of optimization algorithms. The first one is the Levenberg-Marquart (L-M) gradient-based method, the second one is based on a genetic algorithm (G-A) optimization process, and the third one combines the two search algorithms to complete the identification task more efficiently. The efficiency of the two first algorithms is studied by identifying the unknown parameters of a multilayer soil profile. Then, the global -local hybrid scheme is applied to identify the soil profile characteristics and determine some information gaps using experimental data recorded within the Adapazari city during the Kocaeli earthquake of August 17, 1999. Finally, the applicability of the proposed identification procedure is verified through comparison of the parameter identification results and experimental ones at Bradbury dam by using ground motions records during the 1978 Santa Barbara earthquake (in USA).

基于MISO折线模糊神经网络的优化算法

In this paper, we proposed a novel multi-input and single-output (MISO) polygonal fuzzy neural network based on polygonal fuzzy numbers and their extension operations. A uniformity analysis of this network is provided for cases where an activation function is nonnegative continuous and monotonous. In order to avoid the complex calculation of the partial derivatives of error functions related to connection weights and threshold values, we use the Hebb rule and the particle swarm algorithm to design two kinds of optimal learning algorithms to handle these two calculations, respectively. Using a simulation example, we analyzed and compared the two kinds of optimization algorithms in detail. The results showed that our optimization algorithms based on MISO fuzzy neural networks exhibited randomness and parameters diversity, but the Hebb algorithm was simple and easy to realize. The particle algorithm was stable and converged quickly

Optimization algorithms for MISO polygonal fuzzy neural networks

In this paper, we proposed a novel multi-input and single-output (MISO) polygonal fuzzy neural network based on polygonal fuzzy numbers and their extension operations. A uniformity analysis of this network is provided for cases where an activation function is nonnegative continuous and monotonous. In order to avoid the complex calculation of the partial derivatives of error functions related to connection weights and threshold values, we use the Hebb rule and the particle swarm algorithm to design two kinds of optimal learning algorithms to handle these two calculations, respectively. Using a simulation example, we analyzed and compared the two kinds of optimization algorithms in detail. The results showed that our optimization algorithms based on MISO fuzzy neural networks exhibited randomness and parameters diversity, but the Hebb algorithm was simple and easy to realize. The particle algorithm was stable and converged quickly

Two-Layer Optimize Algorithm for Microgrid Economic Dispatch

To deal with the super short-time energy management of microgrid, a mixed-integer programming optimization algorithm combined with genetic algorithm is proposed. Firstly, this paper introduced the short-time economic dispatch mathematical model of microgrid. Secondly, a two-layer optimize algorithm is been developed. The lower layer takes no account of power flow constrains, convert the model into a mixed-integer programming problem through linearization techniques. The lower layer gets the generation schedule, and then passes the data to the upper layer. The upper layer takes the power flow constrains into account, optimize the unit output based on genetic algorithm. The simulation result shows that the proposed algorithm achieves a better complementary of the two kinds of optimization algorithm. At the same time, the optimization result also shows the effectiveness of the algorithm.

Comparison of Particle Swarm Algorithm and Ant Colony Algorithm in the Optimization of Uniform Quantizer

In this paper, we focuse on the study of parameter optimization problem of the uniform scalar dead zone quantizer (USDZQ).We used particle swarm optimization (PSO) algorithm and ant colony optimization (ACO) algorithm for the optimization. We used the two kinds of optimization algorithm for ECG signal coding test.Those ECG signals were from the MIT-BIH arrhythmia database. At the end of this paper, the results of compression are compared.

Comparsion of Two Kinds of Optimization Algorithm for Load Model Parameter Identification

Parameter Identification is the key technology in measurement-based load modeling. In order to identify parameters in power system ,the differential method which is based on the multiple curves fitting and interpolated method are compared in the paper. Numerical results illustrate that the differential method can improve the accuracy for load modeling parameter identifications.

Study on Fault Diagnosis of Gear-Box

A BP neural network approach based on differential evolution and particle swarm optimization is developed for fault diagnosis of gear-box. The approach takes a novel kind of optimization algorithm, i.e., differential evolution and particle swarm optimization algorithm, to train BP neural work. The feasibility and effectiveness of this new approach is validated and illustrated by the study cases of fault diagnosis on gear-box. The diagnosis results show that the BP neural network based on differential evolution and particle swarm optimization has a better training performance, better convergence behavior, as well as better diagnosis ability than BP neural network. Owing to the flexibility of this approach, it can be adapted to other problems very easily.

Multi-objective optimization algorithm for flow shop scheduling with family setup times

The objective optimization was to minimize total earliness/tardiness and number of setups at machine.A jointed algorithm to solve problems based on Control Weight Evolutionary Algorithm(CWEA) and optimization algorithm was presented.Firstly,the CWEA was used to determine scheduling sequence preference.Secondly,a kind of optimization algorithm was put forward to adjust the starting time for determined scheduler.The simulation results show that the effectiveness of the proposed algorithm in solving the problem.

PRUNING REDUNDANT SERVICES FOR FAST SERVICE SELECTION

Selecting the optimal service from a mass of functionally equivalent services with low time cost is significant. Previous research addressed this issue excessively relying on several kinds of optimization algorithms. In this work, we propose an approach that prunes redundant services and reduces search space of service selection on the basis of Skyline computing and context inference. We firstly adopt Skyline computing to prune redundant services. Then, we present the notion of context service based on context inference to further reduce the size of service selection problem. Finally, mixed integer programming is employed to find the optimal service from context services according to users' Quality of Service (QoS) requirements. Experimental results on a test bed indicate that our approach can find the optimal service with low time cost.