Search results

The significant reduction in cavity dimensions compared to the finished size of the injected component has to be carefully considered due to the non-uniformity of the expansion found with the foaming EVA composition.
Sample data are: expansion ratio of sphere with diameter of φ10 mm, φ13 mm, φ20 mm and φ22 mm, cylinder with diameter of φ10 mm, φ13 mm, φ17 mm and φ22 mm, tri-prism with side length of 15 mm, 20 mm and 30 mm, quadrangular prism with side length of 15 mm, 25 mm and 30 mm, thin sheet without hole with thickness of 4 mm, 8 mm and 10 mm, thin sheet with hole with thickness of 4 mm, 6 mm and 10 mm
Therefore, considering the situation that distribution range of input data and output data is large, it is necessary to make data normalization, which will not only avoid output distortion and network lull caused by non-uniformity of input parameters, but also reduce the network training time to a certain extent, normalized equation [6] can be show in Eq. 4: (4) where xi and are data before and after normalization respectively, min(xi) and max(xi) are the minimum and maximum of process variables respectively. 2.3 EVA plastic’s expansion ratio predicted by BP neural network Select data obtained from Taguchi experiments as sample data, among them 386 groups of data are used as a neural network’s training sample data, the rest of six as test sample data.
To verify trained network’s prediction performance, six groups of data have been input for trained neural network, as shown in Table 3, where A is molding temperature, B is molding time, C is injection pressure, D is shape and size of test pieces.
Adopt three-layer BP network structure, neurons of input layer, hidden layer and output layer are 4, 4 and 1 respectively, experiment results are set as neural network sample data, among them 386 groups of data are use as training sample, and others are test sample, after neural network is trained, expansion ratio of EVA plastic can be predicted more accurately.

It is yet important to validate this common belief by scientific data.
However, it showed dramatic changes in the later Tang dynasty demonstrating the reduction in forest coverage thus water regulating capability.

The proposed model and constitutive relationships are verified by the experimental data.
But the micro-forming is not just the scale reduction of the traditional one.

There are many well-developed examples and successful technologies for energy reduction and conservation.
To gain consistent results from the investigation data and simulate precisely, the artificial neural network (ANN) is evaluated as a suitable method [5-13].
Results and Discussion The number of RSEC and energy price data set from 1987 to 2012 was totally 26, as shown in Fig. 1 [14].
Fig. 1 Variation of raw data.
[14] Data on http://web3.moeaboe.gov.tw/ECW/populace/home/Home.aspx

The actual data in Fig.4(a) is a smooth curve for it is estimated by integral operation of Fig.4(c), while the actual data in Fig.4(d) fluctuates around the reference data because it is estimated by differential operation of Fig.4(b).
Also, as Fig.4 shows, the actual measured data lags behind the reference data, which causes the tracking error of system.
That is because of a limited communication baud rate of RS232, a time lag between the control operation and the measurement operation, and the inherent accuracy of hardware setup in the experiment system. 3.2 Operating noise performance There are two major reasons for the noises reduction of servo presses machine.

We use a 1:100 reduction gear in front of the motor to magnify its driving torque.
A service data object (SDO) reads from entries or writes to entries of the object dictionary. 
Relative to the SDO, the process data object PDO which is used for fast data transmission with a high priority, do not generate a response for each message. 
Therefore, in this system, the need for frequently interaction data, such as motor’s current, speed and other information, uses PDO mode and the other data uses SDO mode.

The strain distribution in high temperature deformation was confirmed to be uniform in the parallel portion of the specimen and the developed testing system may contribute to improving the accuracy in hot stamping CAE analysis. 1 Introduction In the recent years, there has been an increasing demand for vehicle weight reduction to improve fuel efficiency for the purpose of reducing CO2 emission.
For CAE analysis with higher accuracy, it is important to obtain the high temperature deformation data in the hot stamping thermal history with higher accuracy2) 3).
In the conventional high-temperature tensile test, especially in the higher temperature and strain rate, only the displacement between the crossheads was generally measured, and in that case, the measured data includes the elongation　outside the parallel portion of the specimen.　
The result indicates that the thermal history of the hot stamping can accurately be simulated and the temperature uniformity was maintained withing the parallel portion of 30 mm long during the test. 900℃×100s -10℃/s 750℃ -50℃/s 10℃/s 850℃ -50℃/s Fig.6　Target thermal history assuming hot stamping process Fig.7　Thermal history data of hot stamping simulation with hybrid heating and gas cooling system Fig.8 shows the results of observing the microstructure of the simulation tested specimen.
The result indicates that the high temperature deformation behavior can accurately be evaluated by using the developed hybrid heating system, which may contribute to improving the accuracy in CAE analysis base on high temperature material data.

The arrow shows the data that the specimen did not break, even after 107 cycles of stress application.
Therefore, such data for 900oC and 600oC overlap in Fig. 4.
We can see a clear difference in the data for σw between the material normalized at 900oC and the others.
From the differences between the experimental data and Eq. (1), the fatigue limitσ w can be evaluated by Eq. (1) with high accuracy.
Since the crack might emanate from a lower-hardness area, data closer to Eq. (1) may be obtained when HVmin is used for the evaluation.

Under the control of horizontal Layered data and information, the nozzle does linkage scan in X-Y plane, filling the trajectory in accordance with section profile of the molding parts at the same time.
The application of CAD technology is to realize the part surface or solid modeling, and ensure the precise computation of discrete and complex data conversion; Advanced numerical techniques provides essential basis for two-dimensional scanning speed accurately, which is the premise of accurate and efficient packing materials.
The design of software system The software system includes geometric modeling and information processing, Information processing part, mainly consists of STL file processing, technology processing, numerical control, and figure display, these parts realize the inspection and repair of error data, sheet file generation, filling the route calculation, , NC code generation and the control of forming machine, Among them, processing module based on the STL file is used to determine whether the molding process needs to support, support structure will be designed if needed, In the layered process of STL file: the system extracts geometric information from 3D files, and then reconstructs the CAD models inside the computer, which are displayed by OpenGL graphics library.
Finite element analysis and the tools of MATLAB also play a important role in error reduction.
Rapid prototyping machine is connected with computer through network data lines, and the data is transmitted to rapid prototyping machine, which realizes the integration of design and processing.

Introduction Measuring Internet is to accurately capture the quantitative measurement data of the Internet and their activities.
Measuring results The measuring results in this paper are the router-level Internet topology data measured at 30th, Jan. 2006 from as many as twenty-one CAIDA monitors.
Quantitative analysis of sampling bias 1) CAIDA measurement results For better analysis of the effect of sampling bias on Internet measuring results, we studied all twenty-one measurement data.
Proof: Use of reduction to absurdity.
From Fig. 3, all ten models could fit the observed data well.