Modelling and Prediction of Diesel Engine Performance using Relevance Vector Machine

Diesel engines are being increasingly adopted by many car manufacturers today, yet no exact mathematical diesel engine model exists due to its highly nonlinear nature. In the current literature, black-box identification has been widely used for diesel engine modelling and many artificial neural network (ANN) based models have been developed. However, ANN has many drawbacks such as multiple local minima, user burden on selection of optimal network structure, large training data size, and over-fitting risk. To overcome these drawbacks, this article proposes to apply an emerging machine learning technique, relevance vector machine (RVM), to model and predict the diesel engine performance. The property of global optimal solution of RVM allows the model to be trained using only a few experimental data sets. In this study, the inputs of the model are engine speed, load, and cooling water temperature, while the output parameters are the brake-specific fuel consumption and the amount of exhaust emissions like nitrogen oxides and carbon dioxide. Experimental results show that the model accuracy is satisfactory even the training data is scarce. Moreover, the model accuracy is compared with that using typical ANN. Evaluation results also show that RVM is superior to typical ANN approach.     

基于BP神经网络优化制备Cu-Ce/TiO_2及其光催化活性研究

采用Cu和Ce对TiO2进行改性,基于正交实验设计和BP神经网络研究了Cu-Ce/TiO2中Cu-Ce对TiO2的摩尔百分数、Cu-Ce/TiO2中CuCe摩尔比及Cu-Ce/TiO2烧结温度对Cu-Ce/TiO2光催化降解甲醛溶液性能的影响.同时,对Cu-Ce/TiO2制备方案进行了优化,并运用X射线衍射仪、扫描电子显微镜和紫外-可见分光光度计对最佳条件下制备的Cu-Ce/TiO2进行表征.结果表明,优化的制备条件为Cu-Ce/TiO2中Cu-Ce对TiO2的摩尔百分数为2.88%,Cu-Ce/TiO2中Cu-Ce的摩尔比为1∶1,Cu-Ce/TiO2的烧结温度为570℃.共掺杂Cu离子和Ce离子能有效避免掺杂TiO2晶格内部表层和近表层产生较多的位错,从而抑制晶格畸变增大;诱导TiO2中锐钛矿型晶体向金红石型晶体转变的能力增强,有效抑制电子-空穴对的复合,产生介电局域效应.     

基于前向神经网络的广义环境系统评价普适模型

为了建立由水环境、空气环境、生态环境、水资源环境、灾害环境、遥感环境、社会经济环境等不同环境系统组成的广义环境系统评价都能普适、通用的神经网络模型,针对BP神经网络因收敛速度慢、易于陷入局部极值而使实用性受限的缺陷,提出以双极性sigmoid函数作为网络隐层节点(神经元)的激活函数,而网络输出为所有隐层节点输出的线性求和的前向神经网络的广义环境系统评价模型.在设置广义环境系统指标参照值和指标值规范变换式,并对指标值进行规范变换的基础上,分别构建了适用于广义环境系统评价的任意2个指标规范值的前向神经网模型(NV-FNN(2)结构)和任意3个指标规范值的前向神经网模型(NV-FNN(3)结构).而对于指标较多的广义环境系统评价,只要将多指标分解为以上2个指标和3个指标的两种简单结构的前向神经网络的广义环境系统评价模型的组合表示即可.理论分析和实例检验结果表明:该模型对任意广义环境系统的规范指标值皆普适、通用,因而使不同环境系统的评价变得简洁、统一.规范变换和优化算法相结合的建模思想和方法对简化广义环境系统评价的多元回归、投影寻踪回归、回归支持向量机和径向基神经网络建模亦有借鉴和启迪作用.     

Prediction of effluent concentration in a wastewater treatment plant using machine learning models

Of growing amount of food waste, the integrated food waste and waste water treatment was regarded as one of the efficient modeling method. However, the load of food waste to the conventional waste treatment process might lead to the high concentration of total nitrogen(T-N) impact on the effluent water quality. The objective of this study is to establish two machine learning models—artificial neural networks(ANNs) and support vector machines(SVMs), in order to predict 1-day interval T-N concentration of effluent from a wastewater treatment plant in Ulsan, Korea. Daily water quality data and meteorological data were used and the performance of both models was evaluated in terms of the coefficient of determination(R~2), Nash–Sutcliff efficiency(NSE), relative efficiency criteria(d rel). Additionally, Latin-Hypercube one-factor-at-a-time(LH-OAT) and a pattern search algorithm were applied to sensitivity analysis and model parameter optimization, respectively. Results showed that both models could be effectively applied to the 1-day interval prediction of T-N concentration of effluent. SVM model showed a higher prediction accuracy in the training stage and similar result in the validation stage.However, the sensitivity analysis demonstrated that the ANN model was a superior model for 1-day interval T-N concentration prediction in terms of the cause-and-effect relationship between T-N concentration and modeling input values to integrated food waste and waste water treatment. This study suggested the efficient and robust nonlinear time-series modeling method for an early prediction of the water quality of integrated food waste and waste water treatment process.     

BP 神经网络模型在环境气象灾害预测中的应用研究--以湖南省为例

依据环境气象数据与自然灾害统计数据,建立BP神经网络模型,对湖南主要气象灾害（洪灾、旱灾、冰冻灾）及受灾经济损失进行实例预测,将在MATLAB7软件中的仿真结果与传统的多元线性回归模型分析结果进行比较和误差分析。结果表明,BP神经网络模型在洪灾、旱灾受灾率方面的预测效果和精度优于多元回归模型,而由于冰灾训练样本不足及经济损失与输入因子的线性相关程度高,在冰灾与受灾经济损失率方面稍逊于多元回归模型。     

基于投入产出分析的虚拟水研究综述及展望

将国内外基于投入产出模型的虚拟水研究进行系统概括,认为按研究层面可以归纳为区域之间的虚拟水贸易和产业间部门用水关联两方面,并指出:(1)目前基于投入产出分析技术的虚拟水研究基本建立在价值型投入产出模型的基础上,实际上将产业虚拟水替代了原本产品虚拟水的概念.(2)目前针对工业品、服务品的虚拟水量化研究不足,并未独立研究非农产品虚拟水量化时的差异性.最后文章提出产业生态学中近年来发展的混合生命周期评价方法(Hybrid life-cycle assessment,HLCA).     

基于百度指数的长江中游城市群城市网络特征研究

随着地理研究迈入大数据时代,运用互联网展开城市网络结构研究逐渐成为经理地理研究的新思路。以长江中游城市群为研究对象,借助百度指数,获取2011年~2014年百度用户关注度,构建百度指数城市信息流网络,从大城市群视角和所辖三大子城市群视角,分别探讨城市网络格局的时空变化。研究发现:长江中游城市群在一体化进程中城市等级日益明晰,差距拉大。三大子城市群表现出极化效应与扩散效应并存,区域非均衡性突显;互联网的普及在一定程度上重塑和改造着城市群,但非完全颠覆传统城市网络格局,地理区位对城市网络格局和城市影响力的影响仍不容忽视。     

A new pre-assessment model for failure-probability-based-planning by neural network

At present, the prediction of failure probability is based on the operation period for laid pipelines, and the method is complicated and time-consuming. If the failure probability can be predicted in the planning stage, the risk assessment system of gas pipeline will be greatly improved. In this paper, the pre-laying assessment model is established to minimize risk of leakage due to piping layout. Firstly, Fault Tree Analysis (FTA) modeling is carried out for urban natural gas pipeline network. According to expert evaluation, 84 failure factors, which can be determined in the planning stage, are selected as the input variables of the training network. Then the FTA model is used to calculate the theoretical failure probability value, and the failure probability prediction model is determined through repeated trial calculation based on BP (Back Propagation Neural Network) and RBF (Radial Basis Function), for obtaining the optimal network parameter combination. Finally, two prediction models are used to calculate the same example. By comparing our pre-assessment model with the theoretical prediction consequences of the fault tree, the results show that the error of RBF prediction model can be close to 3%, which proves the validity and correctness of the method.     

An integrated EDIB model for probabilistic risk analysis of natural gas pipeline leakage accidents

Natural gas pipeline construction is developing rapidly worldwide to meet the needs of international and domestic energy transportation. Meanwhile, leakage accidents occur to natural gas pipelines frequently due to mechanical failure, personal operation errors, etc., and induce huge economic property loss, environmental damages, and even casualties. However, few models have been developed to describe the evolution process of natural gas pipeline leakage accidents (NGPLA) and assess their corresponding consequences and influencing factors quantitatively. Therefore, this study aims to propose a comprehensive risk analysis model, named EDIB (ET-DEMATEL-ISM-BN) model, which can be employed to analyze the accident evolution process of NGPLA and conduct probabilistic risk assessments of NGPLA with the consideration of multiple influencing factors. In the proposed integrated model, event tree analysis (ET) is employed to analyze the evolution process of NGPLA before the influencing factors of accident evolution can be identified with the help of accident reports. Then, the combination of DEMATEL (Decision-making Trial and Evaluation Laboratory) and ISM (Interpretative Structural Modeling) is used to determine the relationship among accident evolution events of NGPLA and obtain a hierarchical network, which can be employed to support the construction of a Bayesian network (BN) model. The prior conditional probabilities of the BN model were determined based on the data analysis of 773 accident reports or expert judgment with the help of the Dempster-Shafer evidence theory. Finally, the developed BN model was used to conduct accident evolution scenario analysis and influencing factor sensitivity analysis with respect to secondary accidents (fire, vapor cloud explosion, and asphyxia or poisoning). The results show that ignition is the most critical influencing factor leading to secondary accidents. The occurrence time and occurrence location of NGPLA mainly affect the efficiency of emergency response and further influence the accident consequence. Meanwhile, the weight ranking of economic loss, environmental influence, and casualties on social influence is determined with respect to NGPLAs.     

Risk assessment of safety level in university laboratories using questionnaire and Bayesian network

Accidents in university laboratories not only create a great threat to students’ safety but bring significant negative social impact. This paper investigates the university laboratory safety in China using questionnaire and Bayesian network (BN) analysis. Sixteen influencing factors for building the Bayesian net were firstly identified. A questionnaire was distributed to graduate students at 60 universities in China to acquire the probability of safe/unsafe conditions for sixteen influencing factors, based on which the conditional probability of four key factors (human, equipment and material, environment, and management) was calculated using the fuzzy triangular theory and expert judgment. The determined conditional probability was used to develop a Bayesian network model for the risk analysis of university laboratory safety and identification of the main reasons behind the accidents. Questionnaire results showed that management problems are prominent due to insufficient safety education training and weak management level of management personnel. The calculated unsafe state probability was found to be 65.2%. In the BN analysis, the human factor was found to play the most important role, followed by equipment and material factor. Sensitive and inferential analysis showed that the most sensitive factors are personnel incorrect operation, illegal operation, and experiment equipment failure. Based on the analysis, countermeasures were proposed to improve the safe management and operation of university laboratories.