To explore a suitable multi-parameter model for predicting ranges of completely fluidic sprinkler, the ranges of a fluidic sprinkler were measured by varying four parameters such as installation height, diameter, nozzle tilt angle and working pressure. Based on the basic principles and algorithms of BP neural network and generalized radial basis(RBF)neural network, BP and RBF neural network models for range prediction of the fluidic sprinkler were established and their performance in the prediction was analyzed. The results show that the range is related nonlinearly to the working pressure and nozzle diameter. For the nozzle installed at 1.2 m height, 27° tilt angle and with a diameter ranged in 4-10 mm, when the working pressure is increased to 0.4 MPa, the range tends to reach a limit, but also the installation height is correlated positively with the range. Both BP and RBF neural net-works can better express the nonlinear relationship between the range and the main control parameters of the fluidic sprinkler. In terms of training time, the RBF network is 8.05 s longer than the BP network. In the prediction process, the BP network program can produce the same result unnecessarily in each run, however, this problem doesn′t exist in the RBF network. Further, the average prediction error of the RBF network against the measurements is 3.55% smaller than that of the BP network. Based on the overall effectiveness of both networks in prediction, the nozzle range predicted by RBF neural network model has an even better applicability.
HU Guang,ZHU Xing-Ye-*,YUAN Shou-Qi et al. Comparison of ranges of fluidic sprinkler predicted with BP and RBF neural network models[J]. Journal of Drainage and Irrigation Machinery Engin, 2019, 37(3): 263-269.
LYU Mingli, ZHANG Zhonghua, LIU Junping. Experimental study on hydraulic performance and combined uniformity of micro sprinklers[J]. Journal of drainage and irrigation machinery engineering, 2017,35(7):641-644.(in Chinese)
LOU Wengao, LIU Suiqing. On assessment of sustainable development level of regional water resource using artificial neural networks[J]. System sciences and comprehensive studies in agriculture, 2004,20(2):113-119.(in Chinese)
MOODY J,DARKEN C J. Fast learning in networks of locally-tuned processing units[J]. Neural computation, 2014, 1(2):281-294.
COUREY A J, HOLTZMAN D A, JACKSON S P, et al. Syner-gistic activation by the glutamine-rich domains of human transcription factor Sp1[J]. Cell, 1989, 59(5):827-836.
LIU Junping, YUAN Shouqi, LI Hong, et al. Analysis and experiment on influencing factors of range and spraying uniformity for complete fluidic sprinkler[J]. Transactions of the CSAM, 2008,39(11):51-54.(in Chinese)
任自中. 柴油机相继增压系统的理论与试验研究[J]. 内燃机工程,2001(1):32-37.
REN Zizhong. Theoretical and experimental study on the sequential turbocharging system of diesel engines[J]. Chinese internal combustion engine engineering, 2001(1):32-37.(in Chinese)