Search results

The experimental data were collected to obtain training set and testing set.
The results of the ANN model were in good agreement with experimental data.
Finally, 59 groups of data were obtained by experimented.
But, the neural network can be easily over-fitting, which causes the error rate on new unseen data to be much larger than the error rate on the training data.
Fig. 2 shows the root mean square error for both testing data and training data for various numbers of units in the hidden layer by means of trainbr algorithm.

The current CFD models have been validated by the experimental data obtained from the boiler for case study.
These post-combustion control systems are referred to as selective catalytic reduction (SCR) and selective non-catalytic reduction (SNCR).
SNCR typically is limited to lower NOX reduction levels but may be the more economical choice depending on the required NOX reduction or the unique project requirements.
In order to obtain accurate and effective temperature data at the location of the SNCR reagent injection points, a 3-D temperature field reconstruction measuring system [19] using 20 detectors was employed to measure the flame temperatures along the furnace height.
All the data offer a solid validation for the numerical simulations and indicate that the mesh and models adopted in the present study are suitable for investigating the three-fuel combustion of the boiler [21].

The reduction degrees of FeO and Cr2O3 are 68.06% and 52.24% respectively, and comprehensive reduction degree is 58.52%.
However, the data of july to november showed, using the relative lower proportion of fine ore, the greater the number of using pellets, the more technical indicators were attained than fine ore were brought into the furnace directly.
Table 8 The chemical composition of finished pellets Name TCr TFe MCr MFe TC Percent(%) 33.50 22.04 17.50 15.00 2.90 Chrome concentrate pellets reduction used to express the reduction degree of its formula: (10) (11) (12) Formula: MFe——Metallic iron content of reduction kind, %; TFe——Total iron content of reduction kind, %; MCr——Metallic chromium content of reduction kind, %; TCr——Total chromium content of reduction kind, %; ——The reduction degree of iron, %; ——The reduction degree of chromium, %; ——The comprehensive reduction degree , %; Brought the table 5 data, concluded: =68.06%； =52.24%； =58.52%。
Table 9 The mass balance of grate-rotary kiln system Entry Output Item Unit(kg/min) Item Unit(kg/min) Green ball Dehydration loss Ash and bulk materials loss Iron reduction loss Chromium reduction loss Oxidation of coke powder loss Finished pellets 631.31 According to the law of conservation of mass, the material entry should be equal to the output item, then: (16) Brought the data and calculated kg/min, amounted green ball 59.23t/h.
New technology of chrome ore pre-reduction[J].

It can be effective for large-scale incomplete ocean data reduction and it also provides a strong basis for decision making for the marine environment processing and follow-up processing.
The prevalence of incomplete data in marine monitoring and other areas of the internet of things bring tremendous difficulties to data fusion, data mining.
In order to mining knowledge from incomplete data, Literature [6] constructed a new similar relationship.
These studies are for static data, but in reality in many databases are dynamic.
Conclusions The traditional approach to deal with incomplete data is make it completed.

However, there is a need to customize those tools according to the nature of the data type.
In this case, the observed data is defective rate, which is an attribute data.
Thus, relevant literatures regarding the analysis of attribute data need to be reviewed.
This data shows the vital few and trivial many behavior of Pareto’s principle.
Since the measurement is an attribute data, Attribute Agreement Analysis was then performed.

In point cloud data pre-processing, salient outliers are manually removed; Gaussian filter is used for noise suppression; curvature-based sampling is adopted for data reduction.
Data reduction is needed for high efficiency.
In the encircle box reduction, an encircle box which is composed of a quadrangle is used to filtrate data inside or outside the box.
The reduction results of the point cloud data shows in Fig.4.
Then, Gaussian filter, curvature-based sampling data reduction, etc. are used in data pre-processing.

Data in Cleveland database comes from the Cleveland clinic foundation, is provided by Robert Detrano.
Thus, each data sample contains 20 attributes, all data samples are divided into two classes.
Second, run the attribute reduction algorithm based on the rough set theory, take 13 items of properties with better distinguish ability as reduction set, and then construct the corresponding reduction data sets.
References [1] Derchiang Li, Yaohwei Fang, An algorithm to cluster data for efficient classification of support vector machines [J].
Rough sets theoretical aspects of reasoning about data [J].Kluwer Academic Publishers, Dordrecht,1991, 39(1):110-113

This paper describes a dimension reduction method of input vector to improve classification efficiency of LVQ neural network, where GA is used to decrease the redundancy of input data.
Dimension reduction by GA is present in section Ⅲ.
denotes the number of samples in test data set.
· The Origin of Data In this paper, we choose the UCI data sets [7] as our test data, which are considered the standard data sets to compare the capability of various algorithms in data mining domain.
TABLE Ⅰ Experimental Data Sets Data Set Number of Attributes Number of Instances Number of Classes Ionosphere 34 351 2 Vehicle 18 946 4 Sonar 60 208 2 Waveform 40 5000 3 Breast Cancer 32 569 2 Vote 16 436 2 · Evaluative Method For each data set we chose in the table 1, 80 percent data instances are selected at random as training data, and the 20 percent remainder data instances are considered as test data.

Application of a Wavelet Extension De-noising Method In Seismic Data Processing Ye Wua, Bo Zhangb and Jia Weic 1 Institute of Disaster-Prevention Science & Technology, Yanjiao, Sanhe city, 065201 China asun_wuye@163.com, bzhangbo199011@163.com, cwj@163.com Keywords: Wavelet Transform; extension; seismic data; signal de-noising Abstract.
We have removed the high frequency noise in seismic data based on the suppressing detail components method, Fourier transform filtering method, WED method and reconstructing the 5th layer approximate coefficient method respectively, and the results show that the WED method can more effectively restrain noise than the other methods.
[6] Shu Li, Yucheng Shi, Yuankun Huo, Yan Tang, Based on the MATLAB seismic signal wavelet noise reduction, Gansu science and technology , 26(15), (2010), pp:54-55
[9] Junhua Zhang, Youxi Le, The wavelet transform and the fractal properties in the improvement of seismic data of the application of the resolution.
[12] Jingguang Ceng, Yaqin Shu, Yong Zhong, Fractal and chaos characteristics of seismic data, Petroleum geophysical exploration, 30(6), (1995), 743-748.

To use data mining technology, firstly, we must ensure that the historical data adapt to data mining, which needs data preparation, that is data preprocessing.
Data preprocessing includes three parts: 1) Data selection: Data selection is to collect the internal data and external data related to the information of load,and to choose the data applyed to data mining.Data selection, including attribute selection and data sampling is to select data fields and tuples in the data source.
Data integration is to put datas into unique data store[3].
Raw data through data selection, cleaning, integration and conversion generate the data mining library,to prepare for data mining.
It is denoted by IND (P) Reduction and Relative Reduction.