Search results

Many practical control problems require achieving multiple objectives, and these objectives often conflict with each other. The existing multi-objective evolutionary reinforcement learning algorithms cannot achieve good search results when solving such problems. It is necessary to design a new multi-objective evolutionary reinforcement learning algorithm with a stronger searchability.

The multi-objective reinforcement learning algorithm proposed in this paper is based on the evolutionary computation framework. In each generation, this study uses the long-short-term selection method to select parent policies. The long-term selection is based on the improvement of policy along the predefined optimization direction in the previous generation. The short-term selection uses a prediction model to predict the optimization direction that may have the greatest improvement on overall population performance. In the evolutionary stage, the penalty-based nonlinear scalarization method is used to scalarize the multi-dimensional advantage functions, and the nonlinear multi-objective policy gradient is designed to optimize the parent policies along the predefined directions.

The penalty-based nonlinear scalarization method can force policies to improve along the predefined optimization directions. The long-short-term optimization method can alleviate the exploration-exploitation problem, enabling the algorithm to explore unknown regions while ensuring that potential policies are fully optimized. The combination of these designs can effectively improve the performance of the final population.

A multi-objective evolutionary reinforcement learning algorithm with stronger searchability has been proposed. This algorithm can find a Pareto policy set with better convergence, diversity and density.

It is of a great significance for the health monitoring of a liquid rocket engine to build an accurate and reliable fault prediction model. The thrust of a liquid rocket engine is an important indicator for its health monitoring. By predicting the changing value of the thrust, it can be judged whether the engine will fail at a certain time. However, the thrust is affected by various factors, and it is difficult to establish an accurate mathematical model. Thus, this study uses a mixture non-parametric regression prediction model to establish the model of the thrust for the health monitoring of a liquid rocket engine.

This study analyzes the characteristics of the least squares support vector regression (LS-SVR) machine . LS-SVR is suitable to model on the small samples and high dimensional data, but the performance of LS-SVR is greatly affected by its key parameters. Thus, this study implements the advanced intelligent algorithm, the real double-chain coding target gradient quantum genetic algorithm (DCQGA), to optimize these parameters, and the regression prediction model LSSVRDCQGA is proposed. Then the proposed model is used to model the thrust of a liquid rocket engine.

The simulation results show that: the average relative error (ARE) on the test samples is 0.37% when using LS-SVR, but it is 0.3186% when using LSSVRDCQGA on the same samples.

The proposed model of LSSVRDCQGA in this study is effective to the fault prediction on the small sample and multidimensional data, and has a certain promotion.

The original contribution of this study is to establish a mixture non-parametric regression prediction model of LSSVRDCQGA and properly resolve the problem of the health monitoring of a liquid rocket engine along with modeling the thrust of the engine by using LSSVRDCQGA.

Twin support vector machine (TSVM) is an effective machine learning technique. However, the TSVM model does not consider the influence of different data samples on the optimal hyperplane, which results in its sensitivity to noise. To solve this problem, this study proposes a twin support vector machine model based on fuzzy systems (FSTSVM).

This study designs an effective fuzzy membership assignment strategy based on fuzzy systems. It describes the relationship between the three inputs and the fuzzy membership of the sample by defining fuzzy inference rules and then exports the fuzzy membership of the sample. Combining this strategy with TSVM, the FSTSVM is proposed. Moreover, to speed up the model training, this study employs a coordinate descent strategy with shrinking by active set. To evaluate the performance of FSTSVM, this study conducts experiments designed on artificial data sets and UCI data sets.

The experimental results affirm the effectiveness of FSTSVM in addressing binary classification problems with noise, demonstrating its superior robustness and generalization performance compared to existing learning models. This can be attributed to the proposed fuzzy membership assignment strategy based on fuzzy systems, which effectively mitigates the adverse effects of noise.

This study designs a fuzzy membership assignment strategy based on fuzzy systems that effectively reduces the negative impact caused by noise and then proposes the noise-robust FSTSVM model. Moreover, the model employs a coordinate descent strategy with shrinking by active set to accelerate the training speed of the model.

Unstructured data such as images have defied usage in property valuation for a long time. Instead, structured data in tabular format are commonly employed to estimate property prices. This study attempts to quantify the shape of land lots and uses the resultant output as an input variable for subsequent land valuation models.

Imagery data containing land lot shapes are fed into a convolutional neural network, and the shape of land lots is classified into two categories, regular and irregular-shaped. Then, the intermediate output (regularity score) is utilized in four downstream models to estimate land prices: random forest, gradient boosting, support vector machine and regression models.

Quantification of the land lot shapes and their exploitation in valuation led to an improvement in the predictive accuracy for all subsequent models.

The study findings are expected to promote the adoption of elusive price determinants such as the shape of a land lot, appearance of a house and the landscape of a neighborhood in property appraisal practices.

The purpose of this study is to investigate the thermoforming process of 3D-printed parts made from polylactic acid (PLA) and explore its application in producing wrist-hand orthoses. These orthoses were 3D printed flat, heated and molded to fit the patient’s hand. The advantages of such an approach include reduced production time and cost.

The study used both experimental and numerical methods to analyze the thermoforming process of PLA parts. Thermal and mechanical characteristics were determined at different temperatures and infill densities. An equivalent material model that considers infill within a print is proposed. Its practical use was proven using a coupled finite-element analysis model. The simulation strategy enabled a comparative analysis of the thermoforming behavior of orthoses with two designs by considering the combined impact of natural convection cooling and imposed structural loads.

The experimental results indicated that at 27°C and 35°C, the tensile specimens exhibited brittle failure irrespective of the infill density, whereas ductile behavior was observed at 45°C, 50°C and 55°C. The thermal conductivity of the material was found to be linearly related to the temperature of the specimen. Orthoses with circular open pockets required more time to complete the thermoforming process than those with hexagonal pockets. Hexagonal cutouts have a lower peak stress owing to the reduced reaction forces, resulting in a smoother thermoforming process.

This study contributes to the existing literature by specifically focusing on the thermoforming process of 3D-printed parts made from PLA. Experimental tests were conducted to gather thermal and mechanical data on specimens with two infill densities, and a finite-element model was developed to address the thermoforming process. These findings were applied to a comparative analysis of 3D-printed thermoformed wrist-hand orthoses that included open pockets with different designs, demonstrating the practical implications of this study’s outcomes.

This paper aims to propose an improved version of portfolio optimization model through the prediction of the future behavior of stock returns using a combined wavelet-based long short-term memory (LSTM).

First, data are gathered and divided into two parts, namely, “past data” and “real data.” In the second stage, the wavelet transform is proposed to decompose the stock closing price time series into a set of coefficients. The derived coefficients are taken as an input to the LSTM model to predict the stock closing price time series and the “future data” is created. In the third stage, the mean-variance portfolio optimization problem (MVPOP) has iteratively been run using the “past,” “future” and “real” data sets. The epsilon-constraint method is adapted to generate the Pareto front for all three runes of MVPOP.

The real daily stock closing price time series of six stocks from the FTSE 100 between January 1, 2000, and December 30, 2020, is used to check the applicability and efficacy of the proposed approach. The comparisons of “future,” “past” and “real” Pareto fronts showed that the “future” Pareto front is closer to the “real” Pareto front. This demonstrates the efficacy and applicability of proposed approach.

Most of the classic Markowitz-based portfolio optimization models used past information to estimate the associated parameters of the stocks. This study revealed that the prediction of the future behavior of stock returns using a combined wavelet-based LSTM improved the performance of the portfolio.

This study aims to examine the impact of board characteristics on firm performance while also exploring the influential mechanisms that help Chinese listed companies establish effective boards of directors and strengthen their corporate governance mechanisms.

This paper uses machine learning methods to investigate the predictive ability of the board of directors' characteristics on firm performance based on the data from Chinese A-share listed companies on the Shanghai and Shenzhen stock exchanges in China during 2008–2021. This study further analyzes board characteristics with relatively strong predictive ability and their predictive models on firm performance.

The results show that nonlinear machine learning methods are more effective than traditional linear models in analyzing the impact of board characteristics on Chinese firm performance. Among the series characteristics of the board of directors, the contribution ratio in prediction from directors compensation, director shareholding ratio, the average age of directors and directors' educational level are significant, and these characteristics have a roughly nonlinear correlation to the prediction of firm performance; the improvement of the predictive ability of board characteristics on firm performance in state-owned enterprises in China performs better than that in private enterprises.

The findings of this study provide valuable suggestions for enriching the theory of board governance, strengthening board construction and optimizing the effectiveness of board governance. Furthermore, these impacts can serve as a valuable reference for board construction and selection, aiding in the rational selection of boards to establish an efficient and high-performing board of directors.

The study findings unequivocally demonstrate the superiority of nonlinear machine learning approaches over traditional linear models in examining the relationship between board characteristics and firm performance in China. Within the suite of board characteristics, director compensation, shareholding ratio, average age and educational level are particularly noteworthy, consistently demonstrating strong, nonlinear associations with firm performance. Within the suite of board characteristics, director compensation, shareholding ratio, average age and educational level are particularly noteworthy, consistently demonstrating strong, nonlinear associations with firm performance. The study reveals that the predictive performance of board attributes is generally more robust for state-owned enterprises in China in comparison to their counterparts in the private sector.

This study aims to investigate a new circuitous minichannel cold plate (MCP) design involving flow fragmentation. The overall thermal performance and the temperature uniformity analysis are performed and compared with the traditional serpentine design. The substrate thickness and its thermal conductivity are varied to analyse the effect of axial-back conduction due to the conjugate nature of heat transfer.

The traditional serpentine minichannel is modified into five new fragmented designs with two inlets and two outlets. A three-dimensional numerical model involving the effect of conjugate heat transfer with a single-phase laminar fluid flow subjected to constant heat flux is solved using a finite volume-based computational fluid dynamics solver.

The minimum and maximum temperature differences are observed for the two branch fragmented flow designs. The two-branch and middle channel fragmented design shows better temperature uniformity over other designs while the three-branch fragmented designs exhibited better hydrodynamic performance.

MCPs could be used as an indirect liquid cooling method for battery thermal management of pouch and prismatic cells. Coupling the modified cold plates with a battery module and investigating the effect of different battery parameters and environmental effects in a transient state are the prospects for further research.

The study involves several aspects of evaluation for a conclusive decision on optimum channel design by analysing the performance plot between the temperature uniformity index, average base temperature and overall thermal performance. The new fragmented channels are designed in a way to facilitate the fluid towards the outlet in the minimum possible path thereby reducing the pressure drop, also maximizing the heat transfer and temperature uniformity from the substrate due to two inlets and a reversed-flow pattern. Simplified minichannel designs are proposed in this study for practical deployment and ease of manufacturability.

Inclination angle has been reported to have an enhancing effect on the thermal-hydraulic characteristics and entropy of some thermal systems. Therefore, this paper aims to numerically investigate the effects of inclination angle, volume concentration and Reynolds number on the thermal and hydraulic characteristics and entropy generation rates of water-based Al₂O₃ nanofluids through a smooth circular aluminum pipe in a turbulent flow.

A constant heat flux of 2,000 Watts is applied to the circular surface of the tube. Reynolds number is varied between 4,000 and 20,000 for different volume concentrations of alumina nanoparticles of 0.5%, 1.0% and 2.0% for tube inclination angles of ±90o, ±60o, ±45o, ±30o and 0o, respectively. The simulation is performed in an ANSYS Fluent environment using the realizable kinetic energy–epsilon turbulent model.

Results show that +45o tube orientation possesses the largest thermal deviations of 0.006% for 0.5% and 1.0% vol. concentrations for Reynolds numbers 4,000 and 12,000. −45o gives a maximum pressure deviation of −0.06% for the same condition. The heat transfer coefficient and pressure drop give maximum deviations of −0.35% and −0.39%, respectively, for 2.0% vol. concentration for Reynolds number of 20,000 and angle ±90o. A 95%–99.8% and 95%–98% increase in the heat transfer and total entropy generation rates, respectively, is observed for 2.0% volume concentration as tube orientation changes from the horizontal position upward or downward.

Research investigating the effect of inclination angle on thermal-hydraulic performance and entropy generation rates in-tube turbulent flow of nanofluid is very scarce in the literature.

Because of the various advantages of reinforcement learning (RL) mentioned above, this study uses RL to train unmanned aerial vehicles to perform two tasks: target search and cooperative obstacle avoidance.

This study draws inspiration from the recurrent state-space model and recurrent models (RPM) to propose a simpler yet highly effective model called the unmanned aerial vehicles prediction model (UAVPM). The main objective is to assist in training the UAV representation model with a recurrent neural network, using the soft actor-critic algorithm.

This study proposes a generalized actor-critic framework consisting of three modules: representation, policy and value. This architecture serves as the foundation for training UAVPM. This study proposes the UAVPM, which is designed to aid in training the recurrent representation using the transition model, reward recovery model and observation recovery model. Unlike traditional approaches reliant solely on reward signals, RPM incorporates temporal information. In addition, it allows the inclusion of extra knowledge or information from virtual training environments. This study designs UAV target search and UAV cooperative obstacle avoidance tasks. The algorithm outperforms baselines in these two environments.

It is important to note that UAVPM does not play a role in the inference phase. This means that the representation model and policy remain independent of UAVPM. Consequently, this study can introduce additional “cheating” information from virtual training environments to guide the UAV representation without concerns about its real-world existence. By leveraging historical information more effectively, this study enhances UAVs’ decision-making abilities, thus improving the performance of both tasks at hand.