查询词典 neural process

The nitric acid production process optimization question modeling uses the artificial neural network to complete.

硝酸生产过程优化问题的建模采用人工神经网络完成。

The storage capacity and addressability of the neural network after Monte Carlo learning process are both improved considerably.

数值研究表明，经过学习修正后的神经网络模型的寻址能力及存储容量都有较大的改进。

According to the statistic analysis of Prof. Amari on static recognition with neural networks, and introducing the concept of vector position stochastic variable of vector stochastic sequence, the stochastic variable is processed by DRNN. The relationship between input and output variance is analyzed. The stochastic analysis of dynamic identification is given, so as to explain the DRNN characteristics in recognition process.

第二，根据Amari教授对神经网络静态识别时的统计分析，引入矢量随机序列的矢量位置随机变量的概念，以该随机变量经过DRNN处理后，分析其输入与输出方差之间的关系，试给出了DRNN做动态识别时的统计分析，以从理论上说明DRNN在识别过程中的特性。

In the process of design and implementation of Hopfield neural networks, the conditions are very applicable.

在Hopfield神经网络的设计及实现过程中，这些条件是很实用的。

Neural networks is viewed as a universal approximator for nonlinear functions, but there are some fatal bugs with it, so it is necessary to find another method more right for neutralization process.

神经网络是目前中和过程非线性系统建模中常用的一种建模方法，但本身具有一些难以克服的缺陷降低了它的发展和应用效果，因此寻求更适于中和过程非线性系统建立模型的理论及方法是十分必要的。

As to the slow variety of the mill characteristics during the rolling process, adopts self adaption deformation resistance model which accord with the character of temper rolling process, and combined with the neural network model to predict the taupe rolling force.

针对轧制过程中轧件特性发生缓慢变化的特点，采用符合平整轧制过程特点的变形抗力自适应模型，并与神经网络模型结合，预报平整轧制力。

In this paper, the sorting of wear, the mechanics of wear particle formed and the characteristics of wear particles are introduced, and morphology characteristics parameter of wear particle are determined, which is particle brim digital feature. Based on image characters, after pre-process, section and extracting of contour parameter of wear particle using image process technique, four shape character parameters are extracted by fourier series expansion. After analyzing fundamental principle and shortcomings of neural network, current BP algorithm is improved in output optimization, network linearization implication optimization and adding momentum, and then astringe speed of BP algorithm.

本文介绍了磨损的分类以及磨粒的生成机理、磨粒的形态特征，应用摩擦学系统分析的观点，确定磨粒边缘数字特征为磨粒的识别特征参数；根据磨粒图像的特征，利用计算机图像处理技术对磨粒图像进行预处理、分割、轮廓参数的提取，采用傅立叶级数展开式提取磨粒四个形状特征参数：圆形度、散射度、凹度、细长度；针对 BP(Back-Propagation)神经网络收敛速度慢的特点，对现行的 BP 算法进行改进，并通过实验验证改进后的网络收敛速度快；研究设计了磨粒图像采集系统；利用 BP 算法建立磨损磨粒自动识别算法模型 AWPRM(Auto Wear Particle Recognition Model)。

The major achievement of this paper is: Based on characteristics of the traffic data distribution, execute pattern recognition operations on traffic condition on two dimensions by clustering, then use BP neural network to describe and forecast traffic flow aiming at each pattern. Making use of classic flow-occupancy inverse "V" model, implement polynomial fitting using least-squares algorithm and statistics method on flow curves to detect outliers which are proved to be not accord with practice through the actual implement, then use the moving average model to recorrect the outliers and absent. Make correlation analysis on muti-direction flow queues of the intersection and ones of upriver intersections, choose flow queue with high correlation as assistant one to improve the error tolerance of the prediction system, at the same time we can use the method to give an estimation of flow in intersection with out sensors. We design and implement an SOA(Service-Oriented Architecture)-based UTDD(urban traffic data mining development) with high expansibility and performance, which implement unified management and call of the data-mining application though defining a XML-based description of data-mining process and a common interface to call data-mining process, finally we build traffic flow prediction application model on UTDD.

根据交通流量数据分布的特征，提出基于k-means的二次聚类方法，对交通流量在流量大小和时间上进行模式划分，进而对各个交通流模式进行基于BP神经网络的描述和预测，从而提高模型对流量预测的精度； 2）根据流量/时间占有率倒&V&字形曲线分布模型，提出基于最小二乘法的三次多项式曲线拟合和统计方法的异常检测方法，实际应用表明该方法能够有效识别异常数据，然后根据移动平均算法对异常数据进行修正； 3）基于序列相关性分析，分别对预测方向的交通流量数据序列、上游路口相关序列以及预测路口其它各个方向上的交通流量序列进行分析，选择相似性流量序列，作为辅助序列提供其他没有检测器路口的流量估计； 4）设计和实现了基于SOA（Service-Oriented Achitecture）的高性能、可扩展的智能交通数据挖掘系统UTDD，该系统通过定义基于XML的数据挖掘过程描述和通用的过程模型接口，实现数据挖掘应用的统一管理和调用，最后在UTDD上建立了基于路口流量预测的应用模型。

The fuzzy T-S model was much more precise than the neural network model for burning through point based on process parameters. The rise time and regulating time of the fuzzy predictive control method were respectively 9 min and 18 min which were both much shorter than those by using the fuzzy control method for burning through point in lead-zinc sintering process.

结果表明：模糊T-S模型能有效抑制垂直燃烧速度不确定性的影响，具有较高的辨识精度，能够满足工业现场生产指导的需要；此外，烧穿点的模糊预测控制方法比单纯的模糊控制调节时间短、超调量小，能快速有效地响应烧穿点的变化，具有一定的应用价值。

Records from the data warehouse in a coal-fired power plants were used for Data Mining modeling of coal grinding process in a double charge-discharge coal ball mill, the artificial neural network model built to be verified by the actual operation of the ball mill, was employed to simulate the coal grinding process, focusing on the particle filling optimization.

利用火力电厂数据仓库中的128 065条历史记录，对一台双进双出球磨机的煤粉磨制过程进行数据挖掘建模，所建人工神经网络模型得到设备实际运行状态的验证；采用该模型对煤粉制备过程进行模拟分析，着重研究了料位参数的优化问题。

But we don't care about Battlegrounds.

但我们并不在乎沙场中的显露。

Ah! don't mention it, the butcher's shop is a horror.

啊！不用提了。提到肉，真是糟透了。