Gongcheng Kexue Yu Jishu/Advanced Engineering Science

Apache Spark is a distributed open source framework for big data processing. The performance of Spark is greatly affected by parameter configuration settings. To get the best performance from Spark is still a big challenge because of a large number of parameters. This parameter is tuned manually by experimentation which is not effective. Besides, these parameters must be re-tuned for various applications. In this work, a method based on machine learning is proposed and developed to effectively self-tune Spark parameters. The results show that the performance is speeded up by 33.4% on an average, compared to the default configuration.

With the increasing contradiction between water supply and demand in large-scale irrigation areas in Northwest China, food production security has gradually become a major threat. The traditional optimized canal water distribution model is mainly based on the objective function to optimize the crop water demand configuration, and the objective function parameters and constraints are more complicated. It is difficult to achieve global optimization of water distribution. This paper takes the Xidong canal system in the Xijun Irrigation District of the Heihe River Basin as an example, adopts the principle of “constant flow, controlled opening”, and establishes a backtracking search algorithm to optimize the water distribution equation based on the principle of minimum remaining flow under the condition of a constant flow of the main canal design. Channel valve opening and closing water distribution time diagram, and further obtain the valve control time point skewness coefficient, and compare the obtained results with the vector evaluation genetic algorithm, particle swarm algorithm water distribution time, and valve time control deviation degree, on this basis The adaptability of the model is evaluated in combination with the abandoned water situation of the canal system of Xiaohe Station. The results show that the water distribution time of the backtracking search algorithm, the vector evaluation genetic algorithm, and the particle swarm algorithm are 12.70, 14.38, and 15.50 d, respectively, and the skewness coefficients at the opening time of the valve are 0.093, 0.328, 0.217, which is obvious compared with the water distribution model of the backtracking search algorithm. The time superiority and stability of the backtracking search algorithm is zero in the Xidong canal system where the canal water utilization rate is low, and the water abandonment phenomenon in the canal system of the Xiaohe station with a high canal system water utilization rate is serious, and the algorithm is common It is suitable for areas with low canal water utilization rate. Using the backtracking search algorithm to optimize the water distribution in the irrigation area, under reasonable applicable conditions, not only can ensure the optimal irrigation time and meet the requirements of the canal irrigation system, but also can maintain the relative stability of fluid transportation to achieve the purpose of optimized water distribution in the canal system.

Nowadays, there is an interest in fully applying on the reaction solution, such as desulfurization by pyrolusite or leaching pyrolusite with sulfur dioxide, for the purpose of controlling air pollution and the recovery of manganese salt. However, some researchers aimed at studying on the technological problem that there was a byproduct of S2O6 2- ,which debase the main product of MnSO4 from the desulfurization solution. In the present work, a large amount of Na2SO4 was prepared by the low temperature pyrolysis of Na2S2O6, which was acquired as the precursor by MnS2O6 that was carbonated by the carbonization reagent (NaHCO3). During the thermal decomposition process of Na2S2O6·2H2O, the reaction products and the temperatures of the key steps of reaction processes were investigated and both of the reaction mechanisms and the kinetic parameters were deduced. The thermal decomposition experiments showed that 255 ℃ was the optimum temperature for the preparation of Na2SO4 by the thermal decomposition of Na2S2O6, which underwent two steps including dehydration and desulfurization. The characterization results of X-ray diffraction (XRD) and Ion chromatography (IC) showed that the decomposition product was the pure and single phase cubic sodium sulfate. The thermal decomposition process of Na2S2O6 was characterized by thermo-gravimetric analysis and differential scanning calorimetry analysis (TG-DSC), in which the activation energies (Ea) of dehydration and desulfurization were calculated by the combining Kissinger differential method and the Coats-Redfern integral method, respectively. The processes of dehydration and desulfurization of Na2S2O6·2H2O were controlled by phase boundary reaction model fitting shrinking sphere equation with the Ea of 14.75～18.11 kJ/mol for dehydration and 132.61～137.18 kJ/mol for desulfurization. SO2 was observed as the main decomposition gas. Finally, the reaction equations were demonstrated combined with the chemical analysis method. Based on realizing the resource recovery, this technology of the thermal decomposition of Na2S2O6·2H2O is feasible and used to avoid the acidic wastewater problem within the liquid-phase method, from which decomposition gas could reacted with sodium bicarbonate again during the process of thermal decomposition.

With the advent of the era of big data, the problem of information overload has become particularly serious. The recommendation system can provide personalized recommendation services for users by analyzing users' basic information and users' behavior information. How to push information accurately and efficiently has become an urgent issue in the era of big data. Based on the Alternating Least Squares (ALS) collaborative filtering recommendation algorithm, this paper reduces the loss of the invisible factor item attribute information by merging the similarity of the item on the loss function. At the same time, the cold start strategy is introduced into the model to complete the recommendation. The algorithm is implemented on the Spark distributed platform and single node by using the Movie Lens dataset published by the GroupLens Lab. The experiment results show that the proposed recommendation algorithm can preferably alleviate the data sparsity problem compared with the traditional recommendation algorithm. Moreover, the algorithm improves the accuracy of recommendation and the efficiency of calculation.

The experimental study investigates the takeoff characteristics of lower jet trajectory from the dovetail-shaped flip bucket under different inlet conditions. The experiment analyzes the influences of velocity and water depth on lower takeoff characteristics, especially the variation of takeoff angle of lower jet trajectory. The following items are addressed 1) relationship between takeoff angle of lower jet trajectory and inlet velocity as well as inlet water depth 2) characteristics of the takeoff angle as a function of changes of inlet Fr0 3) the empirical formulas for calculating takeoff angle of lower jet trajectory. The results indicate that the effects of inlet conditions are significant on lower takeoff characteristics. Additionally, with the established empirical formulas, the calculated values of horizontal distance of lower nappe profile are in good agreement with the experimental data.