| ²é¿´: 434 | »Ø¸´: 1 | ||
lijunxue08½ð³æ (СÓÐÃûÆø)
|
[ÇóÖú]
ÇóÖú·ÒëÒ»¶Î
|
|
An adaptive filtering technique has been used wider and wider since the 1960s [10], [18], [23]. When the priori knowledge of input signal is unknown, the AF can adjust the weight coefficient after N iterations, so as to achieve the best filtering. The minimum mean square error as the standard of traditional AF has the drawback of excessive computing [23]. The neural network (NN)¡¯s whole connection among layers is a one-way connection. The learning process is composed of the input¡¯s forwardpropagation and errors¡¯ back propagation [28]. The hidden layer neuron uses sigmoid function. Although the increase in layers can improve the accuracy, but the network also needs more training time. Therefore, this article used the network topology that contains a hidden layer structure. Errors in hidden layer are decreased by increasing the number of neurons [18], [23], [28], [27]. The NN and AF are used separately or combined with other filters in the process and identified GPS and monitoring signals before, refer to [18], [29], [26], [9].However, the focus of this research is: (1) to examine the NN with AF in errors elimination of GPS survey observations, (2) to use the linear NN with AF to process the continuous RTK GPS observations, (3) to demonstrate the possibilities that the GPS data processing technique can be used in SHM, and (4) to analyse the deck bridge deformation based on RTK GPS monitoring system. |
» ²ÂÄãϲ»¶
085701Çóµ÷¼Á
ÒѾÓÐ8È˻ظ´
321Çóµ÷¼Á
ÒѾÓÐ11È˻ظ´
26¿¼Ñе÷¼Á0710 0860
ÒѾÓÐ12È˻ظ´
270»¯¹¤Çóµ÷¼Á
ÒѾÓÐ9È˻ظ´
Ò»Ö¾Ô¸0817»¯Ñ§¹¤³ÌÓë¼¼Êõ£¬Çóµ÷¼Á
ÒѾÓÐ23È˻ظ´
309Çóµ÷¼Á
ÒѾÓÐ7È˻ظ´
Ò»Ö¾Ô¸±±¾©»¯¹¤´óѧ£¬³õÊԳɼ¨350Çóµ÷¼Á
ÒѾÓÐ7È˻ظ´
081700»¯Ñ§¹¤³ÌÓë¼¼Êõ Ò»Ö¾Ô¸Öк£Ñó 323 Çóµ÷¼ÁѧУ
ÒѾÓÐ13È˻ظ´
Çóµ÷¼Á£º085600²ÄÁÏÓ뻯¹¤£¬¿¼²Ä¿Æ»ù£¬×Ü·Ö319
ÒѾÓÐ24È˻ظ´
ÍÁľ304Çóµ÷¼Á
ÒѾÓÐ4È˻ظ´
|
2Â¥2014-03-17 23:17:23














»Ø¸´´ËÂ¥