Search results

The input data fed to the network was the information from processing the photo images of these elements.
Employing of neural networks for the task allows for data approximation which enables recognition and classification of examples unknown to the network [9,10].
Precisely these data are subsequently used for object identification when employing a neural network for the task The elements under analysis are shown in Figure 2.
Image analysis stages Information obtained in the course of image processing was used as input data for the neural network to carry out the task of element recognition.
The processed data were divided into three sets.

The article presents a method of determining the anisotropic friction model in metal forming using multilayer artificial neural networks based on experimental data obtained from the pin-on-disk tribometer.
The experimental data set used in this research for ANN training is thus optimized by using genetic algorithm (GA).
As depicted in Fig. 1, results of this research show excellent agreement between experimental data and outcomes of neuron model for all range of slip orientation used for the training set.
Furthermore, ANN result is more symmetrical in relation to vertical line 0°-180° by ability to smooth experimental data and find the tendency of output variable variation.
Furthermore, the ANN model was characterized by ability to smooth experimental data noised by system that was recording friction force value.

In order to find out the spatial shape of a granite rock-mass, three kinds of data can be used:1) The exploration data (drilling data, geological profile data, geological boundaries and so on); 2) The ridge extension line data applied to speculate the concealed granite rock-mass based on low magnetic-anomaly zone (T≤50nT) from high-precision magnetic-prospecting and Controlled-Source Audio-Frequency Magnetotelluric (CSAMT) results; 3) The other interpolated data of concealed granite rock-mass boundary on CSAMT comprehensive prospecting profile.
With these data, a simulating 3D model of the granite rock-mass can be established to provide basic data for prospecting deep concealed porphyry-polymetallic ore deposits.
“Digital deposit ” is a kind of 3D geological model on deposit scale, which is established on geological exploration data, including drilling core data, chemical analysis, some measured data on geological sections and so on.
The acquisition of 3D data: 3D spatial database (drilling data, the interpolated data of concealed boundary on the sections, the speculated data of concealed ridge-line extension) of the granite rock-mass is established with Micromine software by its 3D spatial superposition and positioning function.
The classification of the data: the measured and analysis data, speculated data, interpolated data. 3.2.1 The sources of the measured and analysis data of the granite rock-mass.

When this is achieved, there will be reduction in defect, which will result in more profits.
In case of the latter, a new design concept needs to be developed; this new design could be design from scratch (Creative Design) or re-design from a datum design (Incremental Design).
The bench test and field data (design-test-fix-retest) also result in high development cost, longer time to market, lower product quality and marginal competitive edge.
Thus, DFSS strategy subject designs to noise factors through appropriate design parameters; this helps to attack design vulnerabilities and integrate tools and methods for their elimination and reduction.

Based on the result obtained, prediction equations for responses studied are adequate to describe the experimental data on tensile strength, elongation at break, and toughness.
Then, a white calibration was done to get accurate and precise data.
The data that had been taken are the values of the L*, a*, and b* which represent lightness, redness, and yellowness, respectively [8].
A high R2 value, close to 1, is desirable and ensures a satisfactory fit of the quadratic model to the experimental data.
Prediction equations for responses studied are adequate to describe the experimental data on the tensile strength, elongation at break and toughness.

Material parameters were determined step by step at higher strain amplitude of 0.8%, experimental data with lower strain amplitude were used to validate the extrapolation of the model.
Temperature along the gauge length in all the tests were controlled within ±2 °C and experimental data were recorded by software attached to the test system.
The values of the eight model parameters were determined by using experimental data with strain amplitude of 0.8% listed in Table 2.
Good correlation between the simulation and the experimental data was obtained.
It is also worth noting that the material parameters were obtained only from the experimental data with strain amplitude of 0.8%.

Using space probes is a convenient and operable method to acquire data and information from a planet.
Because all of the data about Neptune were acquired by Voyager 2 on Aug 25 1989, so the data obtained are very limited until now, which is also a reason why the aerobraking on Neptune is still difficult.
An estimate of the orbit period change per periapsis passage[4] can be determined by (2) From Eq. 2, we can note that, for the same density, significantly larger period reduction shall occur early in aerobraking because of the dependence on the square of the semi-major axis [4].
Based on there initial value, we can get a new set of data.

The software algorithm accomplishes the data processing on the computer based on the digital signal after converted.
Analysis of the algorithm According to the experimental data acquisition and simulation, the transfer response of the MZM modulator without RF signal is approximation of the cosine curve.
The modulators will appear linear reduction of the signal at the Quad+ or Quad- (shown in the first schematic in Fig.1 (b)), multiplication frequency phenomenon at the Peak （e.g. exporting a RZ pulse signal with the duty ratio of 33%）and Null (shown in the second schematic in Fig.1 (b)，e.g. exporting a CSRZ signal with the duty ratio of 66%), signal distortion at arbitrary (shown in the third schematic in Fig.1 (b)).
(b) Experimental results for output average optical power respect to according to actual experimental data.
The optical signal generated by the laser after MZM was split into two paths, one path was used to observe for the contrast, another path was directed to connect to a photodetector, and the electrical signal after the photodetector was introduced into the AD data card with sampling rate of 1MHz.

is Gray predictive model GM (1,1), among the formula is constant, can get through the least squares fit to the following: , among the formula, , The solution of the differential equations is (5) Formula (5) is the series prediction formula, at the same time, (5) is also the accumulated generating series of predictive value.The reduction predicted values can obtain the original series by (6)
Firstly calculate the value by matlab, =(0.9957,1.0043), which can be calculated by matlab, can use to model GM(1,1), so no data conversion.
Prediction can be applied according to the past data.
But if the data is too much, the calculation will be very complicated.
Therefore, this model is not suitable for large data model to predict.

The results presented in these two works show a good agreement with experimental data, indicating that the numerical models proposed by the authors are capable of capturing the main characteristics of the each process studied.
Experimental data are compared with theoretical results given by a model proposed in the literature.
Comprehensive corrosion fatigue tests were developed in environments up to 10 bar of CO2 and the results have shown a fatigue life reduction.
The peak profile analysis data and the self-consistent predictions of texture evolution showed a good agreement with the experimental deformation texture evolution.
The results show that the model is in close agreement with experimental data.