پیش بینی پارامترهای کیفی پساب خروجی تصفیه خانه فاضلاب مبتنی بر اصول و روش های داده کاوی (مطالعه موردی: تصفیه خانه فاضلاب بروجن)

ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﻣﺤﺪود ﺑﻮدن ﻣﻨﺎﺑﻊ آب در ﺟﻬﺎن، ﻣﺪلﺳﺎزی و ﭘﯿﺶﺑﯿﻨﯽ ﭘﺎراﻣﺘﺮﻫﺎی ﮐﯿﻔﯽ ﭘﺴﺎب ﺧﺮوﺟﯽ ﺗﺼﻔﯿﻪﺧﺎﻧﻪ ﻓﺎﺿﻼب، ﻧﻘﺶ ﻣﻬﻤﯽ را در ﺣﻮزه ﻣﺪﯾﺮﯾﺖ ﻣﻨﺎﺑﻊ آب در ﻣﻨﺎﻃﻖ ﻣﺨﺘﻠﻒ ﺟﻬﺎن اﯾﻔﺎ ﻣﯽﮐﻨﺪ. ﻣﺤﺪود ﺑﻮدن ﻣﻨﺎﺑﻊ آب ﺷﯿﺮﯾﻦ در ﺟﻬﺎن و ﻫﻤﭽﻨﯿﻦ رﺷﺪ روزاﻓﺰون ﺟﻤﻌﯿﺖ و ﺗﻮﺳﻌﻪ ﺷﻬﺮﻧﺸﯿﻨﯽ، روﯾﮑﺮد اﺳﺘﻔﺎده ﻣﺠﺪد از ﭘﺴﺎبﻫﺎی ﺷﻬﺮی را اﺟﺘﻨﺎبﻧﺎﭘﺬﯾ ﺮ ﺳﺎﺧﺘﻪ اﺳﺖ. در ﭼﻨﯿﻦ ﺷﺮاﯾﻄﯽ، اﺳﺘﻔﺎده از آب ﺑﺎزﯾﺎﻓﺘ ﯽ ﻣﯽﺗﻮاﻧﺪ ﺑﻪ ﻋﻨﻮان ﯾﮑﯽ از راهﻫﺎی ﻏﻠﺒﻪ ﺑﺮ ﮐﻢآﺑﯽ و ﺟﻠﻮﮔﯿﺮی از ﻫﺪر دادن ﻣﻨﺎﺑﻊ آب ﺗﻠﻘﯽ ﺷﻮد. در اﯾﻦ ﻣﻄﺎﻟﻌﻪ، از ﻣﺪل ﻫﺎی ﺷﺒﮑﻪ ﻋﺼﺒﯽ ﭼﻨﺪﻻﯾﻪmlp ، ﺷﺒﮑﻪ ﻋﺼﺒﯽ ﺷﻌﺎﻋﯽ ﭘﺎﯾﻪ rabf و ﻫﻤﭽﻨﯿﻦ ادﻏﺎم اﯾﻦ ﻣﺪلﻫﺎ ﺑﺎ ﭼﻨﺪﯾﻦ اﻟﮕﻮرﯾﺘﻢ دﯾﮕﺮ از ﺟﻤﻠﻪ، اﻟﮕﻮرﯾﺘﻢ ژﻧﺘﯿﮏga ، اﻟﮕﻮرﯾﺘﻢ ازدﺣﺎم ذرات psoو اﻟﮕﻮرﯾﺘﻢ ﺳﯿﻨﻮسﮐﺴﯿﻨﻮسsca ، ﺑﻪ ﻣﻨﻈﻮر ﭘ ﯿﺶ ﺑﯿﻨﯽ ﭘﺎراﻣﺘﺮﻫﺎی ﮐﯿﻘﯽ ﭘﺴﺎب ﺧﺮوﺟﯽ ﺗﺼﻔﯿﻪﺧﺎﻧﻪ ﻓﺎﺿﻼب از ﺟﻤﻠﻪ codeff ،bodeff و tsseffﺑﻬﺮه ﮔﺮﻓﺘﻪ ﺷﺪ. در واﻗﻊ ﻣﺰﯾﺖ ﺑﮑﺎرﮔﯿﺮی اﻟﮕﻮرﯾﺘﻢﻫﺎی ﺑﻬﯿﻨﻪﺳﺎزیpso ،ga وsca ، ﯾﺎﻓﺘﻦ ﻣﺪل ﻋﺼﺒﯽ ﺑﻬﯿﻨﻪ ﺑﻮده اﺳﺖ.4 ﭘﺎراﻣﺘﺮ ﻣﻮﺛﺮ ﺷﺎﻣﻞ اﮐﺴﯿﮋنﺧﻮاﻫﯽ ﺑﯿﻮﺷﯿﻤﯿﺎﯾﯽ 5روزهbod5 (، اﮐﺴ ﯿﮋنﺧﻮاﻫﯽ ﺷﯿﻤﯿﺎﯾﯽ )cod(، ﮐﻞ ﻣﻮاد ﺟﺎﻣﺪ ﻣﻌﻠﻖ در ﻓﺎﺿﻼب )tss( و ﺧﺎﺻﯿﺖ اﺳﯿﺪی ﯾﺎ ﻗﻠﯿﺎﯾﯽ ﻓﺎﺿﻼب )ph(، ﻃ ﯽ دوره آﻣﺎری 3 ﺳﺎﻟﻪ ) 1397ﺗﺎ1399 ( ﺑﻪ ﻋﻨﻮان ورودی ﺑﻪ ﻣﺪل اﻧﺘﺨﺎب ﺷﺪﻧﺪ. ﻫﻤﭽﻨﯿﻦ ﺳﺎﯾﺰﻫﺎی 70 %و 30% ﺑﻪ ﻋﻨﻮان ﺑﻬﺘﺮﯾﻦ ﺳﺎﯾﺰﻫﺎ ﺑﺮ ای ﻣﺮاﺣﻞ آﻣﻮزش و آزﻣﺎﯾ ﺶ ﺑﻪ ﻣﻨﻈﻮر ﻣﺪلﺳﺎزی ﭘﺎراﻣﺘﺮﻫﺎی bodeff وcodeff ﺗﻌﯿﯿﻦ ﺷﺪﻧﺪ. درآﺧﺮ ﻧﯿﺰ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺷﺎﺧﺺﻫﺎی آﻣﺎری، ﻧﺘﺎﯾﺞ ﺑﻪ دﺳﺖ آﻣﺪه، ﺣﺎﮐ ﯽ از ﻋﻤﻠﮑﺮد ﺑﻬﺘﺮ ﻣﺪلﻫﺎی ادﻏﺎم ﺷﺪه ﻧﺴﺒﺖ ﺑﻪ ﻣﺪل ﻫﺎی mlpو rabfﺑﻮد. از ﻟﺤﺎظ اوﻟﻮﯾﺖ ﻧﯿﺰ ﻣﺪل mlp-scaو rabf-scaﺑﻪ ﺗﺮﺗﯿﺐ، ﺑﯿﺶﺗﺮﯾﻦ دﻗﺖ )nse( و ﮐﻤﺘﺮﯾﻦ ﺧﻄﺎ )rmse( را دارا ﺑﻮدﻧﺪ.

prediction of effluent quality parameters of wastewater treatment plant using data mining methods (case study: borujen wastewater treatment plant)

prediction of effluent quality parameters of wastewater treatment plant is essential in managing water resources, limitation of fresh water resources in the world, furthermore the ever increasing population growth and the development of urbanization, have made the approach of urban wastewater reuse inevitable. in such a situation, the use of recycled water can be considered as one of the ways to overcome water shortage and prevent wastage of water resources. this research aimed to investigate the performance of prediction of effluent quality parameters of wastewater treatment plant in using recycled water. on the other hand, due to the health hazards caused by the discharge of wastewater from wastewater treatment plants to water sources, achieving a precise design and correct management of wastewater treatment plants (wwtp: wastewater treatment plant) is one of the important challenges of sustainable water resources management. 4 effective parameters including (bod), (cod), (tss) and (ph) of wastewater, were selected as input to the model, during a statistical period of 3 years (1397 to 1399). also, 70% and 30% sizes were determined as the best sizes for training and testing stages in order to model bodeff and codeff parameters. in this study, multi layer perceptron models (mlp), basic radial neural network models (rabf), as well as the integration of these models with several other algorithms, such as genetic algorithm (ga), particle swarm optimization algorithm (pso) and sine cosine algorithm (sca), were used in order to predict the quality parameters of wastewater treatment plant effluent.