[1]劉圣譽(yù),李彭,何義亮,等.基于人工神經(jīng)網(wǎng)絡(luò)的反硝化濾池外碳源投加控制[J].中國給水排水,2020,36(7):19-25.
點(diǎn)擊復(fù)制
基于人工神經(jīng)網(wǎng)絡(luò)的反硝化濾池外碳源投加控制
中國給水排水[ISSN:1000-4062/CN:12-1073/TU] 卷: 第36卷 期數(shù): 2020年第7期 頁碼: 19-25 欄目: 出版日期: 2020-04-01
- Title:
- External Carbon Source Dosage Control in Denitrification Biofilter Based on Artificial Neural Network
關(guān)鍵詞:
反硝化濾池; 深度脫氮; 外碳源; 人工神經(jīng)網(wǎng)絡(luò)
Keywords:- denitrification biofilter; advanced nitrogen removal; external carbon source; artificial neural network
摘要:- 針對(duì)反硝化濾池外碳源過量投加導(dǎo)致的出水總碳超標(biāo)與碳源浪費(fèi)問題，利用實(shí)際污水與小試裝置研究了最適外碳源投加量的影響因素，并應(yīng)用人工神經(jīng)網(wǎng)絡(luò)建立了外碳源投加模型與脫氮效果預(yù)測模型。結(jié)果表明，基于進(jìn)水總氮負(fù)荷與碳氮生化反應(yīng)計(jì)量守恒而進(jìn)行的外碳源投加可緩解碳源浪費(fèi)與污染問題，但脫氮效果缺乏穩(wěn)定性，可考慮通過進(jìn)水ORP、pH值、DO與溫度的綜合影響來進(jìn)行改進(jìn)。應(yīng)用自適應(yīng)學(xué)習(xí)速率動(dòng)量梯度下降法建立了輸入為5項(xiàng)進(jìn)水指標(biāo)、輸出為最適投加量的外碳源投加模型，相關(guān)系數(shù)為0.964 8，表明模型中進(jìn)水參數(shù)與最適投加量具有很好的相關(guān)性，外碳源投加模型的改進(jìn)具有可行性。應(yīng)用貝葉斯正則化法建立了輸入為5項(xiàng)進(jìn)水指標(biāo)、輸出為NO3- -N與NO2- -N濃度的脫氮效果預(yù)測模型，相關(guān)系數(shù)為0.908 5，表明預(yù)測反硝化濾池的脫氮效果具有一定可行性。外碳源投加模型可配合脫氮效果預(yù)測模型構(gòu)建反硝化濾池外碳源投加控制系統(tǒng)，完善污水廠的自動(dòng)化控制。
Abstract:- Excessive dosage of carbon source in the denitrification biofilter will result in total carbon over set standard in the effluent and waste of carbon source.Therefore, factors influencing the optimal dosage of external carbon source were explored in the laboratory test device feeding actual sewage, and the models of external carbon source dosage and denitrification performance prediction were built by applying artificial neural network.The problem of waste and pollution of carbon source could be alleviated by adding external carbon sources based on the total nitrogen load of influent and the conservation of carbon nitrogen biochemical reaction. However, the denitrification performance was not stable, and it could be improved by the combined effects of ORP, pH, DO and temperature. The adaptive learning rate momentum gradient descent algorithm was used to establish a carbon source dosage model with input of five influent indexes and output of an optimal dosage of external carbon source. The correlation coefficient was 0.964 8,indicating that there was a good correlation between the influent parameters and the optimal carbon dosage and the improvement of the model was feasible. The Bayesian-regularization algorithm was used to establish the denitrification performance prediction model with input of five influent indexes and output of NO3- -N and NO2- -N concentration.The correlation coefficient was 0.908 5, indicating that it was feasible to predict the performance of the denitrification biofilter. The external carbon source dosage control system of denitrification biofilter could be established by cooperation of the carbon source dosage model and the denitrification performance prediction model, in order to improve the automatic control of the sewage treatment plant.
關(guān)鍵詞：包括碳源投加的文章如下：
楊敏，郭興芳，孫永利，鄭興燦，李勱，熊會(huì)斌，申世峰，吳凡松黃煒琉，黃麒瑋，廖東，徐章亮，黃麗娜張玲玲1，曹洋2,顧淼1，單旺盛2,尚巍1，陳軼1，游佳1，張維1，李鵬峰1，呂小佳1，李家駒1劉圣譽(yù)1,2，李彭1,2，何義亮1,2，邵嘉慧2，任龍飛2白玉華1，張欣宇2，黃政鑫3，劉百倉4，陳艾2，朱芳琳1李宏斌，劉保成，李昌兵，張維，孫懷谷，韓麗 ?1
