Water Pollution Pr ~ - diction with Evolutionary Neural Trees *
نویسندگان
چکیده
An evolutionary learning method for modeling and prediction of complex systems is described and applied to an environmental system. The method is based on tree-structured neural networks whose node type, weight, size and topology are dynamically adapted by genetic algorithms. Since the genetic algorithm used for training does not require error derivatives, a wide range of neural models can be identified. The application of this method to the prediction of water pollution shows comparable results to those achieved by well-engineered, conventional system-identification methods. ·rn Pmc. Int. Joint Conf on Artificial Intelligence (IJCAI) W01·kshop on AI and the Envimmnent, Montreal, Canada, August 20-25, 1995. This research was supported in part by the Real-World Computing Program under the project SIFOGA.
منابع مشابه
Southern Forest Experiment Station Do Different Young Plantation-grown Species Require Different Biomass Models?
--Sweetgum and water oak trees samp~ed from a plantation over 7 years were used to test whether primary tree component (bole wood, bole bark, limb wood, timb bark, and leaves) predictions could be summed to estima~e total bole. total limb. and total tree values. Estimattons by summing primary component predictions were not significantly different from predictions for the totals, but pr~diction ...
متن کاملDetermining water quality along the river with using evolutionary artificial neural networks (Case Study, Karoon River , Shahid Abbaspur-Arab Asad reach)
Rivers are important as the main source of supply for drinking, agriculture and industry.However, drinking water quality in terms of qualitative parameters, is the most important variable. Studias and predicting changes in quality parameters along a river, are one of the goals of water resources planners and managers. In this regard, many water quality models in order to maintain better water ...
متن کاملWater pollutant fingerprinting tracks recent industrial transfer from coastal to inland China: A case study
In recent years, China's developed regions have transferred industries to undeveloped regions. Large numbers of unlicensed or unregistered enterprises are widespread in these undeveloped regions and they are subject to minimal regulation. Current methods for tracing industrial transfers in these areas, based on enterprise registration information or economic surveys, do not work. We have develo...
متن کاملEvolving Neural Networks for Chlorophyll a Prediction
This paper studies the application of evolutionary arti cial neural networks to chlorophyll a pre diction in Lake Kasumigaura Unlike previous applications of arti cial neural networks in this eld the architecture of the arti cial neural network is evolved automatically rather than designed man ually The evolutionary system is able to nd a near optimal architecture of the arti cial neural networ...
متن کاملThe Response of Coniferous Trees to Industrial Pollution in North- Western Russia
The response of two dominant coniferous species (Picea obovata Ledeb. and Pinus sylvestris L.) of eastEuropean taiga to industrial pollution in the impact zone of the Severonickel metallurgical combine (Murmansk Region, Kola Peninsula) was investigated. The water deficit of the needles and shoots described by shoot water potential ( ) was used as indicator of the pollution level in the area. Th...
متن کامل