lojmikak
Dołączył: 30 Sie 2010
Posty: 1597
Przeczytał: 0 tematów
Ostrzeżeń: 0/5
Skąd: olvovw
|
 Wysłany: Czw 15:28, 09 Gru 2010 Temat postu: ugg italia Integration of environmental parameters |
 |
|
,[link widoczny dla zalogowanych]
Integration of environmental parameters on Intelligent Gas Sensor
Degree of conversion. When the gas sensor after a period of time or environmental changes need to modify the model, they can re-use of gas sensor model in Figure 1 for the adaptive correction. Can be seen, the gas sensor neural network model and implementation of learning mainly by two steps: The first step is to provide a certain number of standard data model for the evolution of neural network learning, the identification data is input by the temperature, humidity and environmental parameters are identified Standard gas concentrations of gas composition; second step is learning from the neural network model, based on the completed model of gas sensor output characteristics to achieve the output voltage and gas concentration sampling conversion. At this point the model has become an independent online utility model, the gas concentration can be used for direct digital output, and allows concentration measurements independent of the environmental parameters. 4 Summary This paper discusses the use of artificial neural network inverse model of gas sensor fusion method of nonlinear, the neural network model can easily change according to the gas sensor to modify the model, which has more extensive adaptability and flexibility. Japanese companies used in the experiment of gas sensor FIGARO (TGs823) as subjects 25 No. 4 Zhuangzhe Min: Based on the integration of environmental parameters and the design of smart gas sensor for the sensor model identification 383,[link widoczny dla zalogowanych], 128 set to provide data about learning. When the learning rate r / is set to 0.43, momentum factor. 0.64, following a return of about 13,000 study, the objective function e can be stable. It can be seen that the neural network method has good convergence and stability, appropriate adjustments to the learning rate, 7, also can change the convergence rate of neural network algorithm. The actual detection of the fusion model is used, under different environmental parameters, the actual concentration value and the estimated value of the difference of the concentration of error of ± 0.8% full scale within the range shown in Figure 3. A deleted essays 0500l000l5002000250030003500 concentration (× 10) Figure 3 under different environmental parameters and estimated the concentration of the actual concentration of the error map \%; temperature is 30 ℃, humidity 50%; temperature is 50 ℃, humidity of 100% three cases, the actual gas concentration with the fusion model to estimate the concentration of error. Actual use show that the artificial neural network modeling of gas sensor, the sensor output to achieve non-linear correction conversion, the gas sensor output directly measured gas concentration to achieve the intelligent sensors. This method is also applicable to other nonlinear sensors. [References [1] MadniAM, YunWeijie, waJlLA. Mieromaehiningandartificialneuralnetworks: thefutureofsmartsensing [A]. IEEEAerospaceApplicationsConferenceProceedings [C],[link widoczny dla zalogowanych], Aspen,[link widoczny dla zalogowanych], USA: IEEE, l995, 2: ll7 a l3O. 【2] RathSK, PartraJC. KotAC. Intelligentpressuresensorwithself-calibrationcapabilityusingartificialneuralnetworks [A]. IEEEIntemafionalConferenceProceedingsonSystems, ManandCybernetics [C], Nashville, Tennessee,[link widoczny dla zalogowanych], USA: IEEE, 2000. 4:2563-2568.13 JPatraJC, VanDenBosA. ModelingofaJlintelligentpressuresensorusingfunctionallinkartificialneuralnetworks [J]. ISATransactions, 2000,39 (1): l5-27. [4] Zhang Zhengyong, Zhang Yaoxian, coke is. Semiconductor oxide gas sensor to test new principles and methods [J]. Sensors and Actuators, 2000, (1): 107 A llO. [5] Zhang Liming. Artificial neural network model and its application [M]. Shanghai: Fudan University Press. l994.
相关的主题文章：
[link widoczny dla zalogowanych]
[link widoczny dla zalogowanych]
[link widoczny dla zalogowanych]
Post został pochwalony 0 razy
|
|
