Abstract:In view of the complicated factors influencing the stock price, we revised the Long Short-Term Memory (LSTM) network, which is commonly used in time series, to predict stock prices under the condition of multivariable.First, the Variance Inflation Factor (VIF) was used to screen variables, and then the adaptive promotion (Adaboost) model was combined to check the importance of characteristic variables.Second, the crawler was used to conduct text analysis of investor sentiment, calculate indicators including sentiment index, and reveal the relationship between them and stock price.Then, prices of three stocks including Gree Electric Appliances, Flyco Electric Appliances and Midea Group were predicted by Multilayer Perceptron (MLP) and LSTM, and the appropriate model was selected as the benchmark model.Finally, indicators of sentiment index and investor concern were added to the benchmark model to construct the LSTM-EM model, and the GM (1, 1) model was used to correct the residual term after considering investor sentiment.The empirical results show that the proposed model can predict the stock price accurately.

Abstract:To address the poor timeliness and simple prediction functions of stock forecasting models, we propose a model abbreviated as BiLSTM-SA-TCN, which combines Bi-directional Long Short-Term Memory (Bi-LSTM) neural network, Self-Attention (SA) and Temporal Convolution Network (TCN).The learning unit and prediction unit in the proposed model can effectively learn important stock data, capture long-term dependency information, and output the predicted next day close price.The experimental results indicate that the BiLSTM-SA-TCN model has more stable prediction results on multiple data sets and has higher modle generalization ability.Furthormore, incomparative experiment, the BiLSTM-SA-TCN model achieves the lowest root mean square error, the lowest mean absolute error, and the best fitting degree of R² on the majority of datasets.

Abstract:To accurately analyze the dynamics and changing trends of KPI (Key Performance Indicator) data in the daily monitoring of cloud computing clusters and predict its subsequent development to achieve high availability of cloud computing clusters, we propose a three-frequency cloud KPI data prediction approach based on combined attention model of EWT-ARIMA-Auto-TPA (EAAT for short).First, low, medium and high frequency Intrinsic Mode Variables (IMFs) of cloud KPI data are obtained via Empirical Wavelet Transform (EWT) to reduce the complexity of data prediction.Second, according to the information characteristics of low, medium and high frequency IMFs obtained from the decomposition, models of ARIMA, Autoformer, and TPA-BiLSTM are used to predict each type of IMFs.Finally, the classification prediction results are combined through the Inverse EWT (IEWT) to obtain the prediction result of the KPI.The proposed prediction approach has been verified on four datasets from Google and Amazon.Whether the data is periodic and stable or not, the proposed approach outperforms comparison models.

Abstract:To improve the poor convergence performance and escape from local optimum of Salp Swarm Algorithm (SSA), a Golden sine SSA with Multi-strategy (MGSSA) is proposed.First, the Selective Opposition-Based Learning (SOBL) strategy is used to improve the population quality by calculating selective opposite solutions for individuals in the population that completely deviate from the optimal individual search direction.Then the optimal individual and elite mean individual are added in the follower position update phase to speed up the convergence of the algorithm.Finally, the golden sine algorithm variation strategy is selected based on the probability to further improve the quality of the solution, and facilitate the algorithm to jump out of the local optimum later.In this study, experiments are conducted on 14 benchmark test functions to compare with other swarm intelligence optimization algorithms and novel improved SSA, and then the proposed approach is applied to test the solution of engineering optimization problems in tension/compression spring design.The results show that the proposed MGSSA has high convergence accuracy and stability, and performs well in solving engineering problems.

Abstract:To address the low efficiency and accuracy of smoke & fire detection due to the small size of target and the confusion of fire feature with actual scene in complex environment, a small scale smoke & fire target detection method based on improved YOLOv5 is proposed.First, a fourth detection layer is added to the third detection layer output in the original YOLOv5 model, so as to obtain a larger feature map for small target detection and strengthen the feature extraction capability of the network model.Second, to solve the easy missing detection of target in shielded scene, DIoU_Loss is used to replace the GIoU_Loss in calculating the regression loss function of the target frame.Finally, TensorRT is used to compress and accelerate the optimization of the model, and then deployed to the Jetson TX2 development board for accelerated inference experiments.In addition, more smoke & fire scene data are constructed by replication enhancement.Experimental results show that the proposed method has fast convergence speed and high accuracy for small scale smoke & fire detection, possessing the prospect for popularization and application.

Abstract:To improve the accuracy and reliability of solar irradiance prediction for photovoltaic power system, we propose a model to forecast short-term solar irradiance based on improved Stacking ensemble learning and error correction.First, the Gradient Boosting Decision Tree (GBDT) is used to perform feature selection and remove redundant characteristics of original data set, thus increase prediction accuracy and computing efficiency.Then, an improved Stacking irradiance prediction model is established.In accordance with the difference in prediction accuracy of prediction models in the primary layer under K-fold cross-validation, the prediction results are weighted, and the Box-Cox is employed to transform and process the training set data input from the first layer to the second layer of Stacking, so as to increase the normality and homoscedasticity of prediction.Finally, the historical prediction error data are extracted, and Random Forest is applied to construct an error model to further improve the prediction accuracy.The experimental results show that, compared with traditional models and classic Stacking models, the proposed method significantly improves the prediction performance on solar irradiance.

Abstract:The flight test data is characterized by multiple types, large data volume and complex algorithms, thus data processing is a key link in flight test.The flight test data processing system needs to have capabilities such as rapid analysis of test data and timely sharing of cross-regional data.Here, on the basis of analyzing current flight test data processing system, we propose new approaches such as data processing center & diversity points, sharing between diversity points, horizontal flight data splitting, and separated processing, then design a new flight test data processing system based on edge cloud collaboration, which realizes functions of data processing, cross-regional data sharing, and performance deduction & analysis for the whole-process subject implementation.

Abstract:Energy system integrated new energies such as wind power and photovoltaic to achieve complementary energy supply of electricity, heat and cold has attracted much concern.When multiple investors are involved in the operation of integrated energy system as independent subjects, it is worthwhile to reasonably allocate the capacity to better absorb new energy and maximize the interests of each investor.Based on the Nash equilibrium principle of game theory, this study establishes a capacity allocation game model for the integrated energy system composed of wind power, photovoltaic, and a combined cooling heating and power system, and uses Particle Swarm Optimization (PSO) algorithm to solve it.The comparative analysis of three scenarios including non-game, non-cooperative game and cooperative game shows that in the cooperative game scenario, the system generates optimal results in investors' return, capacity allocation, and overall system return, thus each participant has obvious possibility to cooperate.This study provides a solution for multi-party participation in the energy supply market.

Abstract:Tripping is a common fault in power transmission and distribution systems.Protection measures against tripping used to be relaying operation and electrical component action, which have hysteresis in handling tripping faults.Therefore, the prediction of tripping faults plays a vital role in dealing with hidden problems and power recovery.Here, a method of power system tripping fault prediction based on multisource time series data is proposed.LSTM is used to extract the time characteristics of multisource data, which alleviates the problem of RNN gradient disappearance on long time series.A peephole connection structure is added to the three-layer grid to enable single units to check the LSTM unit status in the previous stage, thereby strengthening the network timing memory capability.Then L2 regularization measures such as parameter normalization are used to mitigate the impact of over fitting in fault prediction.Finally, support vector machine classifier is introduced to improve the generalization ability and robustness of the overall model.The experimental data were obtained from relevant institutions of the State Grid of China.Experiment results show that the proposed method has higher classification accuracy compared with existing data mining methods.The practical application is discussed for its feasibility in actual scenarios.

Abstract:The tilting of vehicle towards the outside of the curve caused by high-speed turning will lead to a rollover accident in severe cases.To address this problem, the Active Roll Control (ARC) of the vehicle body was studied to improve the vehicle steering stability.A vehicle dynamic model with six Degrees of Freedom (DOFs) was established considering both yaw and roll motions.Then, the desired vehicle roll angle was determined, and an active roll controller was designed to make the actual roll angle approach the desired roll angle.Finally, simulations were carried out to obtain vehicle body roll angles, acceleration perceived by occupants and the lateral load transfer rates, and investigate the power consumption of active suspension for roll control as well as the dynamic deflection of the suspension due to the active roll under different driving conditions.The results show that the ARC can make the actual roll angle rapidly approach the desired roll angle, and still ensure driving stability under complex driving conditions;the ARC reduces the peak value of the suspension dynamic deflection, and decreases the lateral acceleration perceived by occupants and the lateral load transfer to zero;the low power consumption of the active suspension for roll control ensures the vehicle's economic performance.

Abstract:To reduce the influence of system dynamic coupling and external disturbance on the performance of flight control system of fixed-wing UAV thus improve its flight control accuracy, this paper proposes a singular perturbation model for fixed-wing UAV and then designs a sliding mode control approach based on disturbance observer.The velocities and attitudes of the fixed-wing UAV are modeled based on the dynamics of action.Then the dynamic model is transformed into a singular perturbation one and then decomposed to complete decoupling.Two reduced-order uncoupled subsystems are obtained, which is a fast subsystem with angular velocity as fast variable and a slow subsystem with linear velocity and attitude as slow variables.Then anti-disturbance sliding mode controllers are designed for angular velocity loop, and angle & attitude loop.Finally, the feasibility and effectiveness of the sliding mode control approach based on fast and slow decomposition are verified by Simulink simulation.

Abstract:Hydrogen peroxide (H₂O₂) is an environmentally friendly and efficient oxidant, which is widely used in industries like medicine and semiconductor chip.The electrochemical synthesis of H₂O₂ by Oxygen Reduction Reaction (ORR) has great potential to replace traditional anthraquinone method.To commercialize this process, the development of 2e^-ORR electrocatalysts with high activity, high selectivity and long-term stability is imminent.Here, we systematically present the research of currently available metal and non-metal based catalysts, with special emphasis on the control strategy of surface groups, and resolves effects on bond binding strength and electron transfer pathways of intermediates in the reduction process.We focus on key strategies such as electronic and geometric effects, coordination heteroatom doping, and active sites of nonmetal-based materials, highlighting that appropriate meso-structural engineering and kinetic strategies can further optimize the catalytic activity and H₂O₂ selectivity of existing catalysts.Finally, we summarize the challenges in exploring the active centers of non-metallic catalysts, the influence of electrolyte environment on catalysts and industrial equipment design with large output power, and prospect the future development in electrocatalytic synthesis of hydrogen peroxide.

Abstract:To examine the effects of gradual increase of atmospheric CO₂ concentration on soil respiration of winter wheat (Triticum aestivum) field, a gradually increased CO₂ concentration experiment was conducted with automatic control system of CO₂ in open top chambers (OTCs) during 2017-2019 growing seasons.In this study, a gradual increase of atmospheric CO₂ concentration (C₈₀ and C₁₂₀, an increase of 40 μmol·mol^-1 year by year from 2016) was set up based on the ambient atmospheric CO₂ concentration (CK).The soil respiration rate (R_s) was measured by static chamber-gas chromatograph method.The results showed that gradually increased CO₂ did not alter the seasonal patterns of soil respiration, but had significant effect on R_s during winter wheat bloom-growth period.In 2018-2019 growing season, compared to CK, C₁₂₀ treatment significantly increased R_s by 50.2% (P=0.008) at the heading-flowering stage, and significantly increased cumulative amount of CO₂ emissions (CAC) by 25.9% (P=0.044) during the wheat growing season;while in 2017-2018 growing season, compared to CK, C₈₀ treatment had no significant effect on R_s.A positive exponential relationship was found between soil respiration rate and soil temperature.Compared to CK, gradually increased CO₂ concentration reduced the temperature sensitivity coefficient of soil respiration(Q₁₀ values).In summary, a gradual increase of atmospheric CO₂ concentration of 120 μmol·mol^-1 increased CAC during the growing season of winter wheat.