基于LSTM的浮选钼精矿品位软测量模型

战略性稀有金属钼矿品位低,组分复杂、嵌布粒度细等特点,其有价金属分离回收难。浮选作为微细粒钼矿分离回收的主要选矿方法之一,其浮选钼精矿品位一直是选厂的关键性产品指标。国内大多数选厂采取轮班制采样,人工化验得到精矿品位结果,但此方式严重滞后于浮选工艺,难以满足对生产过程进行实时监测和操作指导。LSTM是一种特殊的循环神经网络,引入门机制有效的传递或选择性遗忘长时间序列中的信息,解决RNN中的长期依赖、梯度消失和爆炸问题。本文分析整理东坡选厂中各平台源数据,结合选厂浮选工艺及机理,筛选出多个影响浮选钼精矿品位的变量作为模型输入;将输入变量进行异常值判定,缺失值填充和数据降噪等数据预处理,建立高质量浮选钼精矿品位数据库;软测量模型采用PyCharm软件编码,使用BatchNorm批量规范化处理样本数据,加入Dropout正则化防止过拟合,建立基于LSTM的浮选钼精矿品位软测量模型,通过前向传播算法更新神经网络结构参数,并于Linear模型和CNN模型的预测性能指标结果比较。结果表明：基于LSTM的浮选钼精矿品位软测量模型预测准确度高,样本数据误差波动平稳,浮动范围小,模型泛化能力强,模型平均绝对百分比误差MAPE为1.13%,均方根误差RMSE为0.7049%,决定系数R2为0.8763,实现了浮选钼精矿品位的在线预测。     

Avoiding Pitfalls in Comparison of Activity and Selectivity of Solid Catalysts for Electrochemical HMF Oxidation

Electrocatalytic oxidation of 5-hydroxymethylfurfural (HMF) offers a renewable approach to produce the value-added platform chemical 2,5-furandicarboxylic acid (FDCA). The key for the economic viability of this approach is to develop active and selective electrocatalysts. Nevertheless, a reliable catalyst evaluation protocol is still missing, leading to elusive conclusions on criteria for a high-performing catalyst. Herein, we demonstrate that besides the catalyst identity, secondary parameters such as materials of conductive substrates for the working electrode, concentration of the supporting electrolyte, and electrolyzer configurations have profound impact on the catalyst performance and thus need to be optimized before assessing the true activity of a catalyst. Moreover, we highlight the importance of those secondary parameters in suppressing side reactions, which has long been overlooked. The protocol is validated by evaluating the performance of free-standing Cu-foam, and CuCoO modified with NaPO₂H₂ and Ni, which were immobilized on boron-doped diamond (BDD) electrodes. Recommended practices and figure of merits in carefully evaluating the catalyst performance are proposed.     

Multiphase Flow Regime Characterization and Liquid Flow Measurement Using Low-Field Magnetic Resonance Imaging

Multiphase flow metering with operationally robust, low-cost real-time systems that provide accuracy across a broad range of produced volumes and fluid properties, is a requirement across a range of process industries, particularly those concerning petroleum. Especially the wide variety of multiphase flow profiles that can be encountered in the field provides challenges in terms of metering accuracy. Recently, low-field magnetic resonance (MR) measurement technology has been introduced as a feasible solution for the petroleum industry. In this work, we study two phase air-water horizontal flows using MR technology. We show that low-field MR technology applied to multiphase flow has the capability to measure the instantaneous liquid holdup and liquid flow velocity using a constant gradient low flip angle CPMG (LFA-CPMG) pulse sequence. LFA-CPMG allows representative sampling of the correlations between liquid holdup and liquid flow velocity, which allows multiphase flow profiles to be characterized. Flow measurements based on this method allow liquid flow rate determination with an accuracy that is independent of the multiphase flow profile observed in horizontal pipe flow for a wide dynamic range in terms of the average gas and liquid flow rates.     

Vitamin D Inhibits the Early Aggregation of α-Synuclein and Modulates Exocytosis Revealed by Electrochemical Measurements

Alpha-synuclein (α-Syn) localizes at presynaptic terminal and modulates synaptic functions. Increasing evidence demonstrate that α-Syn oligomers, forming at the early of aggregation, are cytotoxic and is thus related to brain neurodegenerative diseases. Herein, we find that vitamin D (VD) can reduce neurocytotoxicity. The reduced neurocytotoxicity might be attributed to the less amount of large-sized α-Syn oligomers inhibited by VD, measured by electrochemical collision at single particle level, which are not observable with traditionally ensembled method. Single-cell amperometry (SCA) results show that VD can recover the amount of neurotransmitter release during exocytosis induced by α-Syn oligomers, further verifying the neuroprotection of VD. Our study reveals the neuroprotective role of VD through inhibiting α-Syn aggregation, which is envisioned to be of great importance in treatment and prevention of the neurodegenerative diseases.     

Quantitative surface-enhanced Raman measurements with embedded internal reference

Analytical applications of SERS are often more associated with qualitative than quantitative analysis, because of the difficulty in obtaining quantitative SERS results. In this paper we introduce a new strategy to quantitatively measure the SERS signals of analytes based on Au-core/Ag-shell nanoparticles with embedded 4-aminothiophenol as the internal reference. Successful detections of two analytes, Toluidine Blue O in aqueous solution (detection limit of 0.1 μM) and melamine in milk (detection limit of ∼5 μM), are demonstrated. The improvement in the linear fitting illustrates that the use of internal reference significantly improves the accuracy of the quantitative SERS measurements. The successful detection of melamine in milk illustrates the versatility of this detection scheme for a wide variety of analytes.     

Sensor–actuator system for dynamic chloride ion determination

Chloride is a crucial anion for various analytical applications from biological to environmental applications. In order to measure the chloride ion concentration, a measurement system is needed which can detect this concentration for prolonged times reliably. Chronopotentiometry is a technique which does not need a long term stable reference electrode and is therefore very suitable for prolonged ion concentration measurements. As the used electrode might be fouled by reaction products, this work focuses on a chronopotentiometric approach with a separated sensing electrode (sensor) and actuating electrode (actuator). Both actuation and sensor electrode are made of Ag/AgCl. A constant current is applied to the actuator and will start the reaction between Ag and Clˉ, while the resulting Clˉ ion concentration change is observed through the sensor, which is placed close to the actuator. The time it takes to locally deplete the Clˉ ions is called transition time. Experiments were performed to verify the feasibility of this approach. The performed experiments show that the sensor detects the local concentration changes resulting from the current applied to the actuator. A linear relation between the Clˉ ion concentration and the square root of the transition time was observed, just as was predicted by theory. The calibration curves for different chips showed that both a larger sensor and a larger distance between sensor and actuator resulted in a larger time delay between the transition time detected at the actuator and the sensor.     

An anion effect on the separation of AgI-containing melts using sound waves

Sound velocities in molten ((LiF + AgI)) and ((LiBr + AgI)) mixtures have been measured to investigate the relationship between the sound velocity and the temperature and the role of the anion in the (liquid + liquid) phase transition. Our results show that the ((LiBr + AgI)) system is biphasic between the melting point and T = 984 K and becomes monophasic above this temperature. We show that the upper consolute critical temperature for the AgI-containing melts increases with decreasing anion size in the series F⁻ > Cl⁻ > Br⁻. The ((LiF + AgI)) melt remains biphasic at all temperatures investigated up to T = 1218 K. The temperature coefficients for the sound velocities in the upper and lower phases of the ((LiBr + AgI)) system have opposite signs because of the superposition of the temperature and composition factors. The difference between the magnitudes of the velocities for the coexisting phases decreases exponentially with increasing temperature and is described by a critical exponent of 0.85 for the ((LiBr + AgI)) melt near the critical temperature. This value is 15% less than that found for alkali halide melts, in which long-range Coulomb forces between ions prevail. This difference may result from the fact that silver halides are intermediate between the typical ionic salts and the fully covalently bonded ones.     

A new device for in-line colorimetric quantification of polypropylene degradation under multiple extrusions

An in-line colorimeter that is able to quantify color changes in real time during extrusion was developed and validated. It is composed of LEDs emitting at three different wavelengths and a photocell that measures the intensity of the light transmitted through the polymer melt flow. The colorimeter was validated at the bench by employing colored aqueous solutions and in-line during the extrusion of a colored polypropylene. Furthermore, it was used to in-line quantify the color changes in a polypropylene as generated over multiple extrusions due to thermo-mechanical degradation. The technique was proved to be fast and suitable to measure color changes in real time during extrusion.     

两相流显微PIV/PTV系统的开发

开发了一个能同时测量两相流中两相速度和细颗粒尺寸分布的显微PIV/PTV系统,其硬件系统包括大功率连续激光器、显微镜、高速摄像机;软件系统由改进的球形颗粒图像识别算法、各种图像处理算法和各种先进的PIV/PTV算法组成。其中改进的圆弧识别算法能够进行更准确地进行曲线分割而能对充满噪音并相互重叠的颗粒图像给出较好的识别结果。应用该PIV系统,可以在微秒和微米数量级上捕获细颗粒/气泡图像,并能较准确地同时得到两相速度、颗粒尺寸和浓度分布。对焚香可吸入颗粒物进行了速度和尺寸的同时测量,得到了较满意的结果。     

基于光学和扫描显微平台的微尺度实验力学检测技术和装置研究

本文基于光学和扫描显微平台,介绍了本研究组在微尺度实验力学检测技术和设备方面的最新研究成果。在检测技术方面涉及显微散斑干涉技术、微标记阵列检测技术、晶粒变形分析技术、光学探针动静态变形分析技术;在检测系统和装置方面介绍了新近开发的双视场薄膜检测系统、散斑微干涉系统、微标记检测平台、AFM和SEM单轴拉伸装置、三维微定位与加载系统、微力传感器及其标定装置、微动平台驱动装置等。探讨了微尺度实验力学检测中的问题和新的检测技术,给出了一些典型的应用和相关装置。