Search results

Drag reduction between 5 and 6% was confirmed through numerical computations by Choi et al. [4].
Fig. 3: Form filling plotted against specific thickness reduction (a) and specific rolling force (b) The calculated simulation data agrees well with that obtained from the experiments.
The dependence between form filling FF and the parameters thickness reduction εh, rolling force Fw and rolling pressure p is investigated by setting the roll gap to different positions simulating different thickness reductions.
Fig. 5 (a) shows form filling plotted against thickness reduction simulated for different materials.
As in the previous chapter, one can state that FF increases with raising thickness reduction.

Finally test data is transmitted though USB interface to upper computer software based on labview, the software’s functions included data reduction, data analysis, data storage etc.
The hardware circuits consist of data conditioning circuit, data acquisition circuit, data transmission and reception circuit and USB communication circuit.
Data Communication.
There are sensor signal acquisition and display, data save, data query etc.
Fig.6 The data acquisition diagram Fig.7 The data save diagram When the “start” button is pressed, the software gets test data, change data into test signals and display test signals.

Data from physical and chemical studies are presented, namely microhardness, porosity, thickness, specific surface area, adhesion and thermal stability of the active layer.
According to the literature data [15], aluminum [16] and titanium [17] were chosen as the substrate metal in advance.
According to data from [23], OT4-1 alloy is the most technologically advanced in terms of processing, and is highly resistant to aggressive environments and catalytic poisons in the production of sulfuric and nitric acids.
RAA method was used to obtain more accurate data, that is X-ray attenuation analysis.
Fig. 6 shows the data that characterize the dependence of the porosity of anodic oxide films on titanium OT4-1 plate on the duration of anodic oxidation.

Therefore, all the test data were modelled as function of strain, temperature and strain rate for precious reflection of material behavior in rigid-plastic finite element model.
It can be found that the simulated force is in a good agreement with the measured data, which demonstrates the reasonable of finite element model.
Figure 2 Cogging Process Model Figure 3 Acquisition data from industry Figure 4 Press force validation Sensitivity Analysis of Forging Process Parameters In this section, the effect of feed, reduction and press speed on workpiece deformation and internal quality was investigated using numerical model combined with Design of Experiment (DoE) method.
Figure 6 Illustration of data acquisition lines The average effective strain data in each data acquisition line is noted as follows, i=0, 1, 2, 3, 4 (1) where i represents the number of data acquisition line, j represents the number of effective strain data in the selected data acquisition line, represents the effect strain data point in the selected data acquisition line.
The normalized standard deviation of average effective strain data in these five lines is defined to reflect the strain variation in the radial direction due to the change of different process parameters.

In order to validate the numerical methods and mesh generation, the experimental data were compared with the CFD results of the tested model without and with spike which has hemispherical nose and cylindrical body from literature [1].
Fig. 6 shows the comparison of the experimental data and CFD results of the lift(CL) and drag(CD) coefficients versus angles of attack(α) at M=1.89.
It is clear to see that the lift coefficient of the present CFD results are in fair agreement with the experimental data.
The drag coefficient of the present CFD results and the experimental data agree well (within 10%) indicating that the numerical methods are accurate enough to capture the main flow characteristics and the aerodynamic dispersions of different spikes.
(a) CL～α (b) CD～α Fig. 6 M=1.89，comparison between the experimental data and CFD results M=1.5 results.

Firstly, this method preprocess original data and the decision table is formed, then attribute reduction and value reduction to delete redundant attributes by Rough Set theory, and finally extract the fault diagnosis decision-making rules. 1 Establishing fault diagnosis decision table based on Rough Set theory (1) Attribute definition In Rough Set theory, in order to better handle data, the data and knowledge we obtained must be formulated.
Because the decision is based on the finite-dimensional discrete data tables, therefore, the data should be normalization processed.
However, because there is a lot of noise in the sample data, some diagnostic rules may only data derived from a small sample, if the sample data is incomplete, the corresponding diagnostic rules will not be reliable.
Coverage is the proportion of data objects which satisfied the antecedent and the consequent.
After normalization, the data was identity to facilitate subsequent processing. 38 groups of failure data was recorded, due to limited space, they are not listed here all, after normalization and attribute reduction, we obtained the decision table as shown in Table 2.

(14) Using experimental data in Fig. 1 and some above-obtained constants, lnε-ln[sinh(ασ)]and ln[sinh(ασ)]-1/Tcurves can be drawn.
The average value of constant P can be obtained from experimental data with different forming conditions.
Generally, the simulation results are coincident with the experimental data for reductions of 20% and 28%.
Generally, the simulation results are coincident with the experimental data within an error of 20%.
The simulated grain sizes generally agreed with the experimental data.

Moreover, it is more suitable for the classification of low dimensional data dimensionally reduced by dimensionality reduction methods, especially those methods obtaining the low dimensional and neighborhood preserving embeddings of high dimensional data.
MNIST data set consists of handwritten digits 0-9 (60,000 training samples, 10,000 test samples).
Recognition rates (%) of several classifiers on MNIST data set (the dimensionality is reduced to 9 by LDA).
Although LPP is a linear transformation method, it doesn’t discover the globally linear structure of data set and preserve it in low dimensional space, but discovers the local neighborhood of data set and preserves it in low dimensional space.
Further, it seems that NN-LSRC performs better than SRC with the samples dimensionally reduced by several dimensionality reduction methods, especially those methods obtaining the low dimensional and neighborhood preserving embeddings of high dimensional data.

This method not only reduces the demand for data, but also ensures accuracy of the results using the sensitivity correction.
The key problem of condition assessment of distribution equipments is to obtain data of the state variables from the equipments.
The more complete and accurate the data is, the more exact the assessment results will be.
Therefore, the state data of the distribution equipments is very small.
Use these data to analysis the sensitivity of non-core status.

One-way ANOVA was used to analyze the data.
Measurement of NH3 Gas Reduction.
Data analysis.
The data showed that the microbes from fermented tobacco leaves could degrade organic matter during fermentation.
The results were indicated by data on the amount of emission per day of ammonia gas produced by excreta (Fig 5).