Modular artificial neural network for prediction of petrophysicalproperties from well log data

An application of Kohonen's self-organizing map (SOM), learning-vector quantization (LVQ) algorithms, and commonly used backpropagation neural network (BPNN) to predict petrophysical properties obtained from well-log data are presented. A modular, artificial neural network (ANN) comprising a complex network made up from a number of subnetworks is introduced. In this approach, the SOM algorithm is applied first to classify the well-log data into a predefined number of classes, This gives an indication of the lithology in the well. The classes obtained from SOM are then appended back to the training input logs for the training of supervised LVQ. After training, LVQ can be used to classify any unknown input logs. A set of BPNN that corresponds to different classes is then trained. Once the network is trained, it is then used as the classification and prediction model for subsequent input data. Results obtained from example studies using the proposed method have shown to be fast and accurate as compared to a single BPNN network     

Self-adaptive radial basis function neural network for short-term electricity price forecasting

Effective and reliable electricity price forecast is essential for market participants in setting up appropriate risk management plans in an electricity market. A reliable price prediction model based on an advanced self-adaptive radial basis function (RBF) neural network is presented. The proposed RBF neural network model is trained by fuzzy c-means and differential evolution is used to auto-configure the structure of networks and obtain the model parameters. With these techniques, the number of neurons, cluster centres and radii of the hidden layer, and the output weights can be automatically calculated efficiently. Meanwhile, the moving window wavelet de-noising technique is introduced to improve the network performance as well. This learning approach is proven to be effective by applying the RBF neural network in predicting of Mackey-Glass chaos time series and forecasting of the electricity regional reference price from the Queensland electricity market of the Australian National Electricity Market.     

Multivariate outlier detection based on self-organizing map and adaptive nonlinear map and its application

To facilitate visualizing and detecting outliers in high dimensional complex data, a novel method integrating self-organizing map (SOM) with adaptive non-linear map (ANLM) was proposed for multivariate outlier detection. Firstly, the high dimensional complex data are pre-processed by robust scaling. Secondly, SOM is applied to map the pre-processed data onto the SOM plane to obtain the topology of the high dimensional complex data, and then the 2-dimensional projection plane of the trained SOM plane, on which the data distribution can be visualized easily, is obtained via ANLM. In sequel, based on the 2-dimensional plane and the topology, a quasi-3 δ edit rule was proposed to distinguish between the normal data and the outliers in high dimensional complex data. Finally, the proposed multivariate outlier detection was illustrated using synthetic data, two standard benchmark data sets and a real industrial process data. The empirical results show that the outliers in high dimensional complex data are visualized easily on the 2-dimensional plane and effectively detected and eliminated by the quasi-3 δ edit rule, and fully demonstrate its satisfactory ability on dealing with outliers in high dimensional complex data.     

Sensor fusion for online tool condition monitoring in milling

The objective of this paper is to combine a direct sensor (vision) and an indirect sensor (force) to create an intelligent integrated tool condition monitoring (TCM) system for online monitoring of flank wear and breakage in milling, using the complementary strengths of the two types of sensors. For flank wear, images of the tool are captured and processed in-cycle using successive moving-image analysis. Two features of the cutting force, which closely indicate flank wear, are extracted in-process and appropriately pre-processed. A self-organizing map (SOM) network is trained in a batch mode after each cutting pass, using the two features derived from the cutting force, and measured wear values obtained by interpolating the vision-based measurement. The trained SOM network is applied to the succeeding machining pass to estimate the flank wear in-process. The in-cycle and in-process procedures are employed alternatively for the online monitoring of the flank wear. To detect breakage, two features in time domain derived from cutting force are used, and the thresholds for them are determined dynamically. Again, vision is used to verify any breakage identified in-process through the cutting force monitoring. Experimental results show that this sensor fusion scheme is feasible and effective for the implementation of online tool condition monitoring in milling, and is independent of the cutting conditions used.     

Analysis of Volatile Bread Aroma for Evaluation of Bread Freshness Using an Electronic Nose (E-Nose)

This paper presents the use of a tin-oxide sensor array and self-organized map (SOM)-based E-nose for analysis of volatile bread aroma and explores its ability to cluster bread odor data according to the freshness of bread. A low cost tin-oxide sensor array based electronic nose system has been used for the classification of state of freshness of bread. The sensor data was acquired for a period of 3 weeks, and an unsupervised self-organizing map (SOM) model was trained using this data to correlate the sensor response to classify the bread as fresh and stale. A comparative evaluation of 3 week' of bread data was carried out using the SOM. The results suggest that the system developed is able to predict the state of bread as fresh and stale up to 98% accuracy if the test bread data sets are of the same week. The classification accuracy reduces to 75-85% if test bread data sets are from different weeks. The model is also applied on three different brands of bread and similar classification results are obtained.     

Characterization of random rough surfaces from scattered intensities by neural networks

Abstract

Optical scatterometry, a non-invasive characterization method, is used to infer the statistical properties of random rough surfaces. The Gaussian model with rms-roughness [sgrave] and correlation length σ is considered in this paper but the employed technique is applicable to any representation of random rough surfaces. Surfaces with wide ranges of Λ and σ, up to 5 wavelengths (λ), are characterized with neural networks. Two models are used: self-organizing map (SOM) for rough classification and multi-layer perceptron (MLP) for quantitative estimation with nonlinear regression. Models infer Λ and σ from scattering, thus involving the inverse problem. The intensities are calculated with the exact electromagnetic theory, which enables a wide range of parameters. The most widely known neural network model in practise is SOM, which we use to organize samples into discrete classes with resolution ΔΛ = Δσ = 0.5λ. The more advanced MLP model is trained for optimal behaviour by providing it with known parts of input (scattering) and output (surface parameters). We show that a small amount of data is sufficient for an excellent accuracy on the order of 0.3λ and 0.15λ for estimating Λ and σ, respectively.     

An Application of Analytics in Green Electricity Markets With Cap-and-Trade Programs

Abstract:

New cap-and-trade (CT) programs (e.g., Regional Greenhouse Gas Initiative) to mitigate climate change have created the need for decision support in investment, bidding, and oversight in electricity markets. This need is critical for practicing engineering managers in restructured electricity markets operating under market-based competition. In this research we develop a bi-level game-theoretic model to obtain optimal bidding strategies for power generating companies in restructured markets under a CT program. We demonstrate the applicability of our model on data from Northern Illinois electricity market and discuss managerial implications. Detailed sensitivity analyses show how CT programs can impact electricity and allowance prices.     

重庆市某科研教学楼冰蓄冷空调系统设计与分析

对重庆市某科研教学楼进行冰蓄冷空调系统的设计与分析,并与该科研教学楼现在实际运行的常规空调系统进行经济性比较,分析了初投资、运行费用及投资回收期。和常规空调系统相比,冰蓄冷空调系统多投资约97万元,年运行费节省约16.1万元/a,投资回收期为6a;减少装机容量约40.86%;减少配电量约38.58%。该楼采用商业用电政策比采用居民用电政策节约更多的运行费用（居民用电与商业用电峰谷电价比均为3：1,峰谷电价差分别为0.54元/kWh,0.82元/kWh）,并使投资回收期减少一年。     

SOMRuler: a novel interpretable transmembrane helices predictor

Transmembrane helices (TMH) identification is one of the most important steps in membrane protein structure prediction. Existing TMH predictors tend to pursue accurate computational models without carefully considering the interpretability of these models and thus act as a black box. In this paper, a novel TMH predictor called SOMRuler with excellent interpretability while possessing high prediction accuracy is presented. The SOMRuler uses a self-organizing map (SOM) to learn helices distribution knowledge, which is encoded in the codebook vectors of the trained SOM, from the training samples. Human interpretable fuzzy rules are then extracted from the codebook vectors of the trained SOM. By extracting fuzzy rules from the learned knowledge rather than the original training samples, on the one hand, the computational burden of extracting fuzzy rules can be greatly reduced; on the other hand, the reliability of the extracted rules can also be enhanced since noise contained in the original samples can be smoothened by the learning procedure of SOM. The validity of the fuzzy rules extracted by SOMRuler is qualitatively and quantitatively analyzed. Experimental results on the benchmark dataset show that the SOMRuler outperforms most existing popular TMH predictors and is flexible to suite for a wide variety of problems in bioinformatics. The SOMRuler software is implemented by Java and Matlab and is available for academic use at: http://www.csbio.sjtu.edu.cn/bioinf/SOMRuler/.     

冰蓄冷空调系统经济性分析

从初投资和运行费用角度，对深圳某厂房冰蓄冷空调系统与常规空调系统进行经济比较；对不同电价政策下的冰蓄冷空调系统的年运行费用和投资回收期进行比较；提出对于三段分时电价，采用平峰电价的加权平均值代替峰时电价来计算峰谷电价比更为合理。     

Perfect competition vs. strategic behaviour models to derive electricity prices and the influence of renewables on market power

A variety of fundamental modelling approaches exist using different competition concepts with and without strategic behaviour to derive electricity prices. To investigate the quality and practicability of these different approaches in energy economics, a perfect competition model, a Cournot model and a Bilevel model are introduced and applied to different situations in the German electricity market. The three electricity market approaches are analysed with respect to their ability to represent electricity prices and the possibility of market power abuse. Market prices are taken as a benchmark for model validity. As a result, the perfect competition model fits best to today’s market situation in most hours of the year. The Bilevel approach explains prices in high load hours sometimes better than the competition model. But complexity and calculation time increase disproportionately. In addition to the analysis of model quality, we use three scenarios to quantify how a high renewable feed-in influences the ability to abuse market power. Results show that the ability to address market power strongly depends on the amount of installed capacities.     

Dynamic price forecast in a competitive electricity market

Price forecast is a key issue in competitive electricity markets. It provides useful information for the market players and the regulators, in both short and long run. Different approaches have been proposed and implemented. A new dynamic approach for forecasting the market price of electricity in the short term is proposed. The price dates are first clustered according to different types of daily profiles and then, given a proper function representing the trend in price, the set of unknown parameters are identified based on the zeroing of a Lyapunov function. The forecast can be dynamically updated with the latest data available. Higher weight can be attributed to this data in determining the future prices. The proposed approach is validated with reference to real systems in the form of the Italian, New England and New York electricity markets. In addition, an extensive price forecast is provided for the Italian market, an example of a young market that is rather difficult to predict patterns for.     

基于自组织神经网络的声学底质分类研究

研究利用多波束测深系统获取的反向散射强度数据,应用自组织(Self Organizing Map,简称SOM)神经网络分类方法实现了对海底泥、砂、砾石和基岩等底质类型的快速、有效识别。通过实验示例,将SOM神经网络的分类结果与传统海底地质取样获取的真实底质类型进行分析比较,表明该方法是可行和有效的。     

Latent structure modeling underlying theophylline tablet formulations using a Bayesian network based on a self-organizing map clustering

The “quality by design” concept in pharmaceutical formulation development requires the establishment of a science-based rationale and design space. In this article, we integrate thin-plate spline (TPS) interpolation, Kohonen’s self-organizing map (SOM) and a Bayesian network (BN) to visualize the latent structure underlying causal factors and pharmaceutical responses. As a model pharmaceutical product, theophylline tablets were prepared using a standard formulation. We measured the tensile strength and disintegration time as response variables and the compressibility, cohesion and dispersibility of the pretableting blend as latent variables. We predicted these variables quantitatively using nonlinear TPS, generated a large amount of data on pretableting blends and tablets and clustered these data into several clusters using a SOM. Our results show that we are able to predict the experimental values of the latent and response variables with a high degree of accuracy and are able to classify the tablet data into several distinct clusters. In addition, to visualize the latent structure between the causal and latent factors and the response variables, we applied a BN method to the SOM clustering results. We found that despite having inserted latent variables between the causal factors and response variables, their relation is equivalent to the results for the SOM clustering, and thus we are able to explain the underlying latent structure. Consequently, this technique provides a better understanding of the relationships between causal factors and pharmaceutical responses in theophylline tablet formulation.     

上海地区典型数据中心错峰蓄冷方案探讨

介绍了上海地区典型数据中心的错峰蓄冷技术选择,并对蓄冷方案进行了详细的设计,讨论分析了在不同电价下数据中心不同错峰蓄冷方案的经济性。为上海地区类似数据中心项目采用错峰蓄冷技术提供参考。     

An energy-efficient two-stage hybrid flow shop scheduling problem in a glass production

Energy-efficient scheduling is highly necessary for energy-intensive industries, such as glass, mould or chemical production. Inspired by a real-world glass-ceramics production process, this paper investigates a bi-criteria energy-efficient two-stage hybrid flow shop scheduling problem, in which parallel machines with eligibility are at stage 1 and a batch machine is at stage 2. The performance measures considered are makespan and total energy consumption. Time-of-use (TOU) electricity prices and different states of machines (working, idle and turnoff) are integrated. To tackle this problem, a mixed integer programming (MIP) is formulated, based on which an augmented ε-constraint (AUGMECON) method is adopted to obtain the exact Pareto front. A problem-tailored constructive heuristic method with local search strategy, a bi-objective tabu search algorithm and a bi-objective ant colony optimisation algorithm are developed to deal with medium- and large-scale problems. Extensive computational experiments are conducted, and a real-world case is solved. The results show effectiveness of the proposed methods, in particular the bi-objective tabu search.     

Social welfare analysis of the iberian electricity market accounting for carbon emission prices

In this study, the authors analyse the social welfare impact of the integration of Portugal and Spain in the Iberian electricity market (MIBEL), taking into account the CO₂ price for emissions trading. They model the impact of emissions trading on the daily clearing prices and generation scheduling, and its effects on the benefits of integration as a whole. They compare the impact of market integration in Portugal and Spain and show that the welfare impact of the MIBEL is dependent on the CO₂ prices. From their analysis, they conclude high CO₂ prices lead to a change in the merit order. Moreover, natural gas is the generation technology that most benefits from transmission constraints and from high CO₂ prices, as in the base case it is mainly used as a peak technology. The authors have also found that increases in the CO₂ prices do not lead to higher profits. Overall, the introduction of the MIBEL will increase social welfare by reducing generation costs and prices.     

基于独立分量分析特征提取的复合神经网络故障诊断法

首先利用基于固定点迭代的快速算法(FASTICA)提取不同机械状态模式(包括正常、齿轮故障及机座松动)特征．随后以此训练某一典型神经网络(如径向基网络或自组织映射网络)，以实现模式的最终分类。借助独立分量分析(ICA)及基于残余互信息(RMI)的二次特征抽取策略，隐藏于多通道振动观测中的高阶特征得以有效提取，进而实现机械状态模式的准确识别。对照分类实验结果表明，基于无导师学习的自组织映射(ICA-SOM)分类方法不仅具有较好的故障模式分类能力，且实现简单直观．在机器健康状况监测中有较大的应用潜力。     

Dynamic programming approach to the bilateral contract scheduling

Bilateral forward contracts are generally used in electricity markets to stabilise prices and hedge electricity shortage risks. A contract party is able to draw electricity from the contract and resell it to the dayahead wholesale and retail markets. Contract parties schedule electricity deliveries over contract period to obtain the highest profit and estimate the acceptable contract price. Two types of contracts are introduced as a way to coordinate interests of the contract parties. The study formulates optimisation problems for contract scheduling. The stochastic dynamic programming technique is proposed as a numerical method for the problem solving. An algorithm based on preliminary construction of revenue functions is developed. A numerical example demonstrates the efficiency of the algorithm.     

基于条件生成式对抗网络的AFM图像盲重构方法

针对原子力显微镜（AFM）成像过程中针尖展宽效应引起的误差,提出一种基于条件生成式对抗网络（CGAN）的AFM图像盲重构方法。首先,以pix2pixHD模型为基础,通过全局生成网络对仿真样本数据进行对抗训练,引入AFM测量数据采用局部提升网络联合训练;最后,特征匹配损失函数以用于提升栅格边缘横向分辨力。实验结果表明:对于线宽8μm一维矩形栅格在AFM下的测量图像进行盲重构,重构图像标准差为0.33μm×0.45μm,具有较高的成像分辨力,有利于提升AFM图像一维栅格测量的准确度。