Volume measurements in three-dimensional microscopy

A refractive index mismatch between the oil immersion and the microscopic object can lead to a severe over-estimation of the object's size. The cause of this effect is explained and a simple calibration method to compensate for its occurrence is presented. A practical example is discussed. The analysis applies to both conventional three-dimensional, and confocal microscopy.     

Tracing the motion of a large object in a fluidized bed using electrical capacitance tomography

An innovative method using electrical capacitance tomography (ECT) to trace a large object's motion on an air distributor in a fluidized bed is described here. The method fills the large object to be traced with a high permittivity material, and then a recalibration process is applied to reduce the nonlinearity caused by the large permittivity difference between the tracer particle and other fine particles in the measurement zone. The local dynamic threshold selection method is performed on the reconstructed image to locate the tracer particle's position.Static simulations and dynamic experiments testify that tracer particles with a diameter of one ninth of the measured cross-section and a fluidization velocity v/v_cr≤4.2 can be accurately located and traced. Employing this method to trace the motion of a spherical object in a bed shows that the fluidization velocity significantly influences the directional motion of a large, heavy object on an inclined air distributor.     

Neural Networks for Classifying Images of Wood Veneer. Part 2

A decision tree using smaller more specialised modular neural networks for the classification of wood veneer by an automatic visual inspection system was presented in Part 1 [1]. A key process in the design of a modular neural network is the use of "normalised inter-class variation" in the selection of the most appropriate image features to be used for its particular specialised classification task. At the root of the decision tree is a single large (holistic) neural network that initally attempts to classify all of the image classes which include clear wood and 12 possible defects (13 classes). The initial design uses 17 features of the acquired image of the wood veneer as inputs. The selection (or more correctly pruning) of inputs for this large neural network used not only "normalised inter-class variation", but also "normalised intra-class variation" in the features and their "correlation" within the same class. This results in the elimination of 6 inputs. The revised smaller 11 input neural network results in a substantial reduction in classification time, for the computer implementation used here, and at the same time the classification accuracy is improved. This is the root of the decision tree described in the previous paper.     

基于小波--神经网络模拟电路故障诊断方法的研究

对模拟电路提出了一种基于小波神经网络的故障诊断方法 ,该法用小波变换作为故障信号的预处理器 ,大大地减少了神经网络的输入数目 ,从而简化了神经网络的结构和减少了它的训练时间 ,并提高了辨识故障类别的能力。在简介该法故障诊断的基本原理后对实例电路进行了故障划类 ,小波函数及故障特征选择 ,给出了计算故障类别特征的仿真编程及故障类别的识别方法     

Segmentation of 3D microtomographic images of granular materials with the stochastic watershed

Segmentation of 3D images of granular materials obtained by microtomography is not an easy task. Because of the conditions of acquisition and the nature of the media, the available images are not exploitable without a reliable method of extraction of the grains. The high connectivity in the medium, the disparity of the object's shape and the presence of image imperfections make classical segmentation methods (using image gradient and watershed constrained by markers) extremely difficult to perform efficiently. In this paper, we propose a non‐parametric method using the stochastic watershed, allowing to estimate a 3D probability map of contours. Procedures allowing to extract final segmentation from this function are then presented.     

Comments on the shortcomings of predicting the precision of Cavalieri volume estimates based upon assumed measurement functions

It is well known that the estimation of an object's volume by means of serial cross-sections, the so-called Cavalieri method, yields an unbiased estimate. But by itself it provides no means by which to estimate how precise this estimate is unless the shape of the volume is fully known beforehand. This knowledge can only be partially determined from the serial section information that is collected. Methods have been developed that claim to surmount this difficulty by using the serial section data to create a mathematical model of the volume's shape properties. The model then is used to estimate (predict) the precision of the volume estimate (its CE) from the single set of data available. Unfortunately, the theory underlying the model is flawed and so the model itself amounts to no more than an unsubstantiated guess about the shape of the volume. Therefore, the precision of the volume estimates that one obtains from the method is only as good as the model and this cannot be ascertained from the single set of acquired data. In this letter I explain the inadequacies of the modelling method. I suggest that it be used only with caution, if at all. Instead I suggest two alternative ways to predict the CE, one that is based upon a rule-of-thumb approach to the object's shape, and another that is based upon spectral analysis of the measurement function and that is easy to implement with available computer software.     

Prediction of Deformed Configuration and Ductile Fracture for Simple Upsetting Using an Artificial Neural Network

This paper suggests a scheme for simultaneously accomplishing the prediction of fracture initiation and geometrical configuration of deformation in metal forming processes using an artificial neural network. A three-layer neural network is used and a back-propagation algorithm is adapted to train the network. The Cockcroft–Latham criterion is used to estimate whether fracture occurs during the deformation process. The geometrical configuration and the value of ductile fracture are measured by the finite-element method. The predictions of the neural network and the numerical results of simple upsetting are compared. The proposed scheme has predicted the geometrical configuration and fracture initiation successfully.     

故障诊断中基于神经网络的特征提取方法研究

在电路状态检测与故障诊断过程中,恰当地选择特征参数是诊断成败的关键。本文研究了基于神经网络的特征评价和特征提取方法,利用神经网络的训练结果对特征参数进行合理的评价。由于神经网络满足高分辨率信息压缩所需的非线性映射条件,通过特征提取将电路故障模式识别中复杂的分类问题转移到特征处理阶段,利用神经网络有效地实现了特征参数的提取。诊断实例验证了该方法的有效性。     

Quantitative identification of illicit drugs by using SOM neural networks

Qualitative identification of THz spectra of illicit drugs using self-organization feature map (SOM) artificial neural network has been demonstrated. In this paper, investigation results show that SOM has quantitatively identified drug mixtures successfully. Based on Beer’s law THz spectra data of various drug proportions were made for training dates. After analyzing the clustering algorithm of SOM, we introduced a parameter named shortest distance as a quantitative criterion for identification result. By this parameter, an automatic recognition algorithm has been developed and successfully applied to the content identification of experimental samples. Combined with our previous work, the SOM neural network can be an integrated and effective method in the identification the THz spectra of illicit drugs.     

Elman神经网络的红外地球敏感器实时故障诊断

以卫星姿控系统实时仿真信号为诊断依据,设计故障检测Elman神经网络及故障判决,实现系统正常与非正常状态的区分并获取故障发生时刻。提出了基于改进梯度更新策略的故障隔离Elman神经网络方法,对故障时刻点之后时域信号进行故障模式匹配,进一步实现系统故障隔离。运用某卫星姿态控制系统进行在线故障诊断试验的结果表明,本文方法具有较好的实时有效性、输出耦合诊断性能、时域信号诊断泛化性和网络收敛性。     

有限元的神经网络计算方法研究

根据有限元总刚矩阵经修正后具有正定性的特点以及弹性体位能函数的具体形式，提出一种新的神经网络有限元计算模型，即模型中神经网络的能量函数与有限元的优化目标函数相等，从而避免由于神经网络自身结构的原因而带来的计算误差。同时，避免采用基于模拟退火算法等随机神经网络优化计算方法时求解结果的随机性和设定初始退火温度To、内循环次数判据、最终停止判据等人为因素的影响。理论分析和计算机仿真表明，文中提出的方法可靠、有效。     

BACK PROPAGATION NEURAL NETWORK MODEL FOR TEMPERATURE AND HUMIDITY COMPENSATION OF A NON DISPERSIVE INFRARED METHANE SENSOR

The infrared absorption gas sensor detects CH₄, CO, CO₂, and other gases accurately and rapidly. However, temperature and humidity have a great impact on the gas sensor's performance. This article studied the response of an infrared methane gas sensor under different temperatures and humidity conditions. After analyzing the compensation methods, a back propagation neural network was chosen to compensate the nonlinear error caused by temperature and humidity. The optimal parameters of the neural network are reported in this article. After the compensation, the mean error of the gas sensor's output was between 0.02–0.08 vol %, and the maximum relative error dropped to 8.33% of the relative error before compensation. The results demonstrated that the back propagation neural network is an effective method to eliminate the influence of temperature and humidity on infrared methane gas sensors.     

Non-fragile sampled-data robust synchronization of uncertain delayed chaotic Lurie systems with randomly occurring controller gain fluctuation

This paper proposes a new non-fragile stochastic control method to investigate the robust sampled-data synchronization problem for uncertain chaotic Lurie systems (CLSs) with time-varying delays. The controller gain fluctuation and time-varying uncertain parameters are supposed to be random and satisfy certain Bernoulli distributed white noise sequences. Moreover, by choosing an appropriate Lyapunov-Krasovskii functional (LKF), which takes full advantage of the available information about the actual sampling pattern and the nonlinear condition, a novel synchronization criterion is developed for analyzing the corresponding synchronization error system. Furthermore, based on the most powerful free-matrix-based integral inequality (FMBII), the desired non-fragile sampled-data estimator controller is obtained in terms of the solution of linear matrix inequalities. Finally, three numerical simulation examples of Chua's circuit and neural network are provided to show the effectiveness and superiorities of the proposed theoretical results.     

训练样本质量对人工神经网络性能的影响

分析人工神经网络预报中的误差来源。推导人工神经网络预报过程中预报误差和训练样本质量之间的关系，讨论训练样本质量对用于时间序列预报人工神经网络性能的影响，并从统计的观点引入用于评价训练样本质量的数字指标“一致度”（ＤＣＴ），还随新指标给出一个模拟结果和相应的建议，以便在人工神经网络训练中的选择训练样本。     

基于神经网络响应面的疲劳裂纹扩展寿命的可靠性分析

当失效形式的极限状态方程中随机变量个数较多或非线性较高时，其形式很复杂，因此传统的计算失效概率的方法不再适用。针对疲劳裂纹扩展寿命失效概率计算的复杂性，提出基于神经网络响应面的可靠性分析方法。首先建立神经网络响应面模拟疲劳裂纹扩展寿命的极限状态方程，然后使用遗传算法(GA)计算可靠性指标。数值试验表明，本方法可以快速、精确地模拟疲劳裂纹扩展寿命的极限状态函数，进而计算出失效概率和可靠性指标。同其他模拟技术相比，在精度相同的情况下，神经网络响应面法可以大大减少模拟时间。     

基于BP网络判据的工业机器人运动学逆解

以MOTOMAN UP20型六自由度机器人为例,建立了工业机器人的MDH模型。在此基础上提出了一种解决工业机器人运动学逆解中存在多解的问题的方法：利用BP网络算法所得的结果作为判据,进行机器人运动学逆解的自动选择,得出运动学逆解问题的唯一解,以实现从机器人工作空间到关节空间的非线性映射。     

基于ART—2神经网络的故障诊断系统

对ART－2模拟量模式识别神经网络算法的改进能使其有效地适用于故障诊断,神经网络的预处理场F0具有增强对比度,抑制基底噪声的能力,神经网络的F2场根据优先级选择竞争节点。网络模式识别的结果作为专家系统输入的类别及其选取概率。专家系统搜索由知识库构造的关系树并依规则进行推理。本文方法结合神经网络和专家系统的优点,以得到可信的故障诊断结果。本文还给出了验证系统正确性的自动液压机故障诊断实例。     

基于改进灰色神经网络的液压泵寿命预测

改进了GM(1,1)模型,提高了其精度和适应范围;将改进的GM(1,1)模型与神经网络预测模型相结合来构建灰色神经网络组合预测模型;提出了基于支持向量机的液压泵寿命特征启发式搜索策略,以液压泵寿命特征参数特征集的交叉验证错误率为评价指标,从液压泵的特征参数（振动、压力、流量、温度、油液信息等）中选取寿命特征因子;运用小波阈值降噪法进行降噪处理,提取典型的小波包能量特征作为模型的输入。以齿轮泵为例,将改进的灰色神经网络预测模型与原始GM(1,1)模型和改进GM(1,1)模型比较可知,灰色神经网络预测模型预测精度最高,达到98.42%。     

AutoIHC‐scoring: a machine learning framework for automated Allred scoring of molecular expression in ER‐ and PR‐stained breast cancer tissue

In prognostic evaluation of breast cancer Immunohistochemical (IHC) markers namely, oestrogen receptor (ER) and progesterone receptor (PR) are widely used. The expert pathologist investigates qualitatively the stained tissue slide under microscope to provide the Allred score; which is clinically used for therapeutic decision making. Such qualitative judgment is time‐consuming, tedious and more often suffers from interobserver variability. As a result, it leads to imprecise IHC score for ER and PR. To overcome this, there is an urgent need of developing a reliable and efficient IHC quantifier for high throughput decision making. In view of this, our study aims at developing an automated IHC profiler for quantitative assessment of ER and PR molecular expression from stained tissue images. We propose here to use CMYK colour space for positively and negatively stained cell extraction for proportion score. Also colour features are used for quantitative assessment of intensity scoring among the positively stained cells. Five different machine learning models namely artificial neural network, Naïve Bayes, K‐nearest neighbours, decision tree and random forest are considered for learning the colour features using average red, green and blue pixel values of positively stained cell patches. Fifty cases of ER‐ and PR‐stained tissues have been evaluated for validation with the expert pathologist's score. All five models perform adequately where random forest shows the best correlation with the expert's score (Pearson's correlation coefficient = 0.9192). In the proposed approach the average variation of diaminobenzidine (DAB) to nuclear area from the expert's score is found to be 7.58%, as compared to 27.83% for state‐of‐the‐art ImmunoRatio software.     

模糊数据融合的刀具磨损状态智能识别

为实现高速加工时刀具渐变磨损状态的在线准确识别,提出了一种集合多种智能的间接检测刀具磨损状态方法的模糊数据融合方法。尽管这些方法具有算法实现较为简单、处理速度较快的优点,但单一的信号检测及单一的智能建模方法难以获得全面的加工状态信息和准确的识别结果。为此,利用F推理技术对上述方法的冗余和互补信息进行数据融合,应用Makino—Fanuc 74-A20型加工中心的测试数据验证了该方案的可行性,并将刀具后刀面磨损的预测值与基于机器视觉检测的实测值进行比较。实验结果分析表明,多参数模糊融合识别方法能快速获得切削刀具磨损状态更加准确的预测值。