Search results

Find Fk2-Fk+Jkdk2 is Non-negative, butFk2-Fk+Jk(dk+dk)2 is not necessarily non-negative.Actual reduction in value of the defined function is: Aredk=Fk2-F(xk+dk+dk)2.
Simulation results and analysis Simulation comparison standard BP algorithm, LMBP algorithm, and this paper improved the LM algorithm's advantages: Take six groups of temperature values obtained by the simulated fire test, the value of the original temperature of the neural network sample acquisition, as shown in table 1: Table 1 Neural network to obtain the original temperature of the sample Time(s) 0 50 100 150 200 250 300 350 400 450 500 Temperature Group 1 20.2 40.2 54.7 66.4 75.4 81.2 87.6 91.4 97.8 101.5 102.3 Group 2 21.8 41.8 59.3 68.7 76.5 82.1 88.6 90.9 97.4 99.7 103.2 Group 3 20.4 38.4 49.0 61.3 71.8 79.3 85.3 90.1 96.5 97.4 102.4 Group 4 19.6 36.3 48.6 58.8 69.5 77.4 82.9 88.4 93.5 96.7 100.2 Group 5 17.6 32.2 44.6 55.3 68.2 76.9 83.6 86.3 91.8 95.3 98.1 Group 6 17.4 28.2 41.4 52.6 63.4 74.2 82.1 84.2 91.4 92.8 94.6 Simulation process, in order to better training effect, followed by data processing is more convenient and usually requires the original sample data to normalization
processing, in which we have adopted the following normalized formula: RX=0.1+xi-xminxmax-xmin+0.8 After normalization of the data in the range [0.1,0.9].
Simulation results show that: this article by temperature experimental data of a large number of specific fire three convergence improved LMBP algorithm convergence speed and accuracy are superior to standard LMBP algorithm and three convergence algorithm in the calculation of the computer's memory and CPU consumption smaller, more convenient, especially suitable for the online calculation of the amount of data in engineering practice.

In Paper [7] it had been proposed that detecting sharp changes in image brightness is to capture important events and apply an edge detector to an image that may significantly reduce the amount of data to be processed and may therefore filter out information that may be regarded as less relevant, while preserving the important structural properties of an image.
Transferring a 2-D pixel array into statistically uncorrelated data set enhances the removal of redundant data, as a result, reduction of the amount of data is required to represent a digital image.
-Molina, “The Evolution of the Web and Implications for an Incremental Crawler,” Proceedings of the 26th International Conference on Very Large Data Bases, pp. 200 – 209, 2000

Experimental procedure Commercially available AlN powder (produced by carbothermal reduction method, Sinocera Advanced Materials Company, Fujian, China) was used as the starting material, which has an average particle size of 1.0 µm.
According to the data supplied by the manufacturer, the powder's oxygen impurity content was 1.2% and the nitride content was 32.5%.
All Raman data of the AlN ceramics were taken using a ×25 objective lens and 120 s collection time for consistency.
The results of hexagonal lattice parameters (ah, ch and unit cell volume) based on X-ray data are calculated and presented in Table 2.
The calculated lattice parameters based on X-ray data of AlN ceramics sintered at high pressure are little smaller than those calculated from the X-ray data of AlN powder.

For each actuation mode, the SMAHC beam was brought to a desired temperature set point, in which dynamic data were captured.
One run consisted of capturing dynamic data at all the chosen set point through a thermal cycle from room temperature to about 90 ºC.
Power current supplier for SMA wires resistive heating SMAHC Shaker Thermocouples Accelerometer Control system Data acquisition Microcomputer with control software Figure 2 Different modes for electrical activation of the active composite beam.
Resistive heating of the Ni-Ti SMA wires is done by a programmable DC power supply current (Agilent, E3633A model) (6) controlled by a specific software (7), while temperatures of the beam and electrical resistance of the Ni-Ti SMA wires are stored in a data acquisition system (Agilent, 34970A model) (8).
Time histories with a total length of 700 seconds were captured with a sampling rate of 1024 Hz to allow 50-frame averages with a bandwidth of 0 - 512 Hz and a frequency resolution of 0.25 Hz by Cattman of HBM data acquisition Spider 8 (9).

Direct polarization (DP) data were collected using a 30 s recycle delay, and cross polarization (CP) experiments were performed using a mixing time of 3ms, recycle delay of 4s, with proton decoupling.
Based on the TEM data the CdS NPs synthesized in PSS by ion-exchange and precipitation have diameters in the range of 3-7 nm and are well dispersed in the polymer matrix.
A HRTEM micrograph, Fig. 1, shows several particles near the film edge, and the selected area electron diffraction data reveals that the NP have adopted the cubic lattice type.
If the frequency/magnitude of the motion of TG has increased upon composite formation then one expects to see a narrowing of the peaks in the NMR data, especially in the 1H spectrum.
Using crystal structures located in the Cambridge Crystallagraphic Data Base, a model for the interface structure has been developed and is shown in Fig. 5.

The parameter N is determined by the statistical data of a specific channel locks.
The first example use the merging of three groups of data in [3], to applied the arranged algorithm to 22 waiting vessels, the utilization rate of effective area of the lock chamber(280 meters long and 32.8 meters wide) is 89.91%, the results are shown in Fig. 1 (the digital on figure represent the waiting vessel number).
Fig. 1 Lockage arrangement of three groups of data merging in [3] The second instance is the instance 2 given in [4].
Fig. 2 Lockage arrangement of data in [4] Conclusions Based on the analysis of the factors influencing of the ship lock navigation capacity, this paper from the angle of improving the once lockage tonnage to improve the navigation capacity of existing ship lock of the Three Gorges, establish ant colony optimization model for the ship lock arrangement, design and implement the corresponding algorithm.
[3] Sun Bo, Qi Huan, Zhang Xiaopan, etal: Dimensionality Reduction Quickly Arranging Algorithm of Lock Chambers in CoScheduling of Three Gorges Dam and Gezhouba Dam System(in Chinese), Computer Technology and Development, Vol. 16(2006), No. 12, pp. 207-212

PNN input and output signals can be time-varying function or procedural data compared with traditional neural networks.
By detecting the above parameters, the value of the effluent BOD can conduct online real-time estimates. 1) Soft-sensing Model of Swage Disposal Process Based on BFB BFB process is a complex, dynamic and biochemical reactions, since water quality and water level, as well as microbial growth conditions are always changing, the effluent BOD is also a corresponding change in water quality indicators, and difficult-line real-time measurement[8].Using the process of neural networks to predict, according to reactions characteristics of effluent biochemical to select DO, NH3-N, TOC three auxiliary variables ,which can basically reflect the variation of BOD, and are not related and in line with soft-measuring principle.According to a specific BFB process, a reaction period is about 3 hours，the sampling period should be based on the data trend of an increase to reduce.
Sampled data obtained in the form of: (6) The input data is synthesized in the 5-order polynomial form in the sampling period , thus train sample set, so that can not only meet the accuracy requirements, but also simplify the calculations. 2) Pretreatment of Date and Selection of Orthogonal Function In order to simplify the complexity of the training process, select the triangle basis function, the number of basis functions is determined by the accuracy of function approximating and the generalization ability of the network.
Table 1 Part of the process neural network training data Group 1 2 3 …… 50 t(min) TOC DO NH3-N TOC DO NH3-N TOC DO NH3-N …… TOC DO NH3-N 0 524.6 12.3 8 456 10.3 7 521 5.6 6 531 6.3 4 30 456.6 10.3 8 386 9.5 6 462 4.5 5 469 4.9 3.5 60 400 8.5 6.5 324 6.4 5 400 4.2 3.9 394 3.5 2.7 90 365 6.8 6 300 5.1 4 320 3.5 3 300 3 2.2 120 300 5.1 4 265 3.8 3.2 200 2.5 1.6 242 2.6 1.6 150 210 3.1 2.5 168 2.5 2.1 153 2.5 1 158 2.1 1.2 180 125 2.1 1.5 75 1.9 1.5 75 2 1 75 1.5 1 BOD 32 33 34 …… 35 Table 2 Predicted results and the relative error Number Detection value of BOD Calculate value of BOD Relative error 1 46 47.5 3.2% 2 44 42 4.5% 3 36 36.8 2.2% 2 Fuzzy control design and simulation of BFB Fuzzy control is a kind of knowledge model based on fuzzy reasoning and regard control experience as control, which is widely used in intelligent control.
In view of this situation, put forward fuzzy control strategy, which take Dissolved Oxygen (DO) as control variable, and take error(e), between DO measurement values and given value, error rate(ec), as two input variables of a fuzzy controller, design the rules of the fuzzy control, realized the BFB optimization control, BFB fuzzy control realization diagram as shown in Fig.2: Fig.2 BFB fuzzy control realization diagram BFB sewage treatment fuzzy control rendering which take DO as control variables in Fig.3: Fig.3 Fuzzy control effect of BFB The control effect show that taking DO for control variable of fuzzy control method, with the optimization target for energy conservation and emission reduction, the effluent water quality continuously become stable and the system operating normally. 3 Conclusion In order to solve the nonlinear and time-varying problem in BFB sewage treatment process, designed the outlet water quality BOD soft measurement model based on PNN; At the

Other works numerically and experimentally, utilizing a neural network to determine the approach that applyed on a variety of materials and process conditions, a neural network data were compiled through trial and error.
Analysis of DOE data: a.
Data with seven trials are collected randomly at each level of temperature.
ANOVA is a basic step in the DOE that is a formidable tool for decision-making based on data analysis.
Schmockel D., and Beth, M., “Spring back reduction in draw-bending process of sheet metals” .

The volume of a transferred electrode metal droplet which has the shape of a spherical segment with the base equal to the electrode cross section can be determined by the formula [23]: (5) The active surface area of the molten electrode metal droplet of can be found by the formula [22]: (6) Using the experimental data (Table 1) of the droplet transfer parameters [4], we verify the obtained formulas 4-6.
Table 1 - Surfacing parameters Power source - rectifier Electrode type Average parameters values (oscillograph) AKIP-4122/1V Number of short circuits during surfacing Short circuit duration τk.z., ms diode LB-52U Current 89+2.7 А Voltage 20.8+0.6 В Estimated rate of welding 0.25 m/min 17 6.7 ± 1.85 inverter 22 5.36 ± 1.34 diode LEP UONI 13/55 Current 88+2.7 А Voltage 21.5+0.6 В Estimated rate of welding 0.29 m/min 17 6.5±2.1 inverter 22 6 ± 1.9 diode CL-11 Current 86 А Voltageе 24.5+0.6 В Estimated rate of welding 0.27 m/min 12 12 ± 3.8 inverter 24 8.1 ± 2.3 Data in Table 1 show that the droplet transfer time decreases and the number of short-circuits increases when using an inverter rectifier which means that the transferred electrode metal droplets are smaller.
Table 2 - Average estimated data on the mass and radius of transferred electrode metal droplets Power source - rectifier Electrode type τk.z., 10–3 s Droplet mass m, g Droplet radius, R, mm Droplet volume V, mm3 diode LB-52U 6.7 ± 1.85 0.099 ± 0.002 1.39 ± 0.026 6.89 ± 1.9 inverter 5.36 ± 1.34 0.052 ± 0/015 1.05 ± 0.01 4.36 ± 1.38 diode LEP UONI 13/55 6.5±2.1 0.091 ± 0.004 1.3 ± 0.03 6.5 ± 1.99 inverter 6 ± 1.9 0.071 ± 0.002 1.23 ± 0.02 5.66 ± 1.8 diode CL-11 12 ± 3.8 0.57 ± 0.04 2.5 ± 0.05 15.48 ± 4.9 inverter 8.1 ± 2.3 0.175 ± 0.05 1.8 ± 0.04 10.28 ± 2.9 Analysis of the data in Table 2 shows that the use of an inverter rectifier makes it possible to reduce the volume of a transferred electrode metal droplet by 9-37% which provides a more stable fine-droplet transfer, especially when using high-alloy electrodes.
To prove the calculation results presented in Table 2 we have analyzed images of high-speed filming (KOMPAS and VEGAS programs) (Figure 1) which certify the calculated data presented in Table 2.
Figure 1 - Kinogram frame of electrode metal transfer in MMA (inverter rectifier, CL-11 electrodes) Decrease in time of the droplet at the end of the electrode (Table 1) and reduction of the geometric dimensions of the transferred droplets (Table 2) reduces heat content of the electrode metal droplets.

In further studies, reduction in partial discharge inception voltage and poorer resistance to electric tree aging in cryogenic temperature has been found for some dielectric materials [4-6].
DC heater G-M cooler heating cooling test platform transmission of heat by contact temperature sensor LabVIEW platform data acquisition unit(DAQ-2204) signal amplifier signal transfer into voltage temperature data storage display of temperature data acquisition unit(DAQ-2204) N Y PID temperature control cryogenic electrical characteristic test reach the preset temperature Fig.1.The schematic diagram of the temperature control system According to Fig.1, temperature signal, which is sampled and transferred into voltage signal, is sampled by data acquisition unit through amplifier and filter; After being A/D conversion, voltage data are collected by upper computer and displayed as temperature value.