王文成 邱勝朋 姚金峰 張寒

摘? 要： 根據(jù)廣西某地區(qū)的城鎮(zhèn)污水處理廠的工藝要求和特點(diǎn)，設(shè)計(jì)了一套基于S7?1200PLC的自動控制系統(tǒng)。針對處理過程中溶氧量控制環(huán)節(jié)大滯后、大慣性特征且難以建立精確模型導(dǎo)致的溶氧量不穩(wěn)定問題，提出使用神經(jīng)網(wǎng)絡(luò)預(yù)測模型PID算法對控制系統(tǒng)進(jìn)行改進(jìn)。由具有自學(xué)習(xí)和自適應(yīng)能力的單神經(jīng)元構(gòu)成基本的自適應(yīng)控制器NNC，同時(shí)使用基于預(yù)測輸出誤差的最小均方函數(shù)（LMS）進(jìn)行權(quán)值和偏置調(diào)整，并采用動態(tài)RBF神經(jīng)網(wǎng)絡(luò)對模型進(jìn)行在線辨識。算法驗(yàn)證結(jié)果表明，模型可以達(dá)到較快的收斂速度和精度，對比于傳統(tǒng)的PID算法，該算法對于溶氧量的控制可取得更好的穩(wěn)定性和動態(tài)響應(yīng)。使用VB語言進(jìn)行工程實(shí)現(xiàn)，對同類過程控制算法的設(shè)計(jì)應(yīng)用有一定的參考價(jià)值。

關(guān)鍵詞： 污水處理; 溶氧量; 神經(jīng)網(wǎng)絡(luò); 預(yù)測控制; PLC; Wincc

中圖分類號： TN876?34; TP183; TP273? ? ? ? ? ? ? ? 文獻(xiàn)標(biāo)識碼： A? ? ? ? ? ? ? ?文章編號： 1004?373X（2020）03?0104?05

Improvement of RBF model in prediction of PID in two stage AO sewage treatment

WANG Wencheng， QIU Shengpeng， YAO Jinfeng， ZHANG Han

（School of Mechanical and Control Engineering， Guilin University of Technology， Guilin 541006， China）

Abstract： An automatic control system based on S7?1200PLC is designed according to the technological requirements and characteristics of an urban wastewater treatment plant in Guangxi. To deal with the fact of instability of dissolved oxygen caused by large lag and great inertia of dissolved oxygen control and the difficulty in creating accurate model in the process of treatment， a neural network prediction model PID （Proportion Integral Differential） algorithm is proposed to improve the control system. The basic adaptive controller NNC （neural network controller） is composed of a single neuron with self?learning and adaptive ability. At the same time， the weight and offset are adjusted by least mean square （LMS） function which predicts the output error， and the dynamic RBF （Radial Basis Function） neural network is used to identify the model online. The verification results of the algorithm show that the model can achieve faster convergence speed and higher accuracy. Compared with the traditional PID algorithm， the algorithm can achieve better stability and dynamic response for dissolved oxygen control. The engineering implementation is performed with the VB （Visual Basic） language， which has certain reference value for the design and application of similar process control algorithms.

Keywords： sewage treatment; dissolved oxygen; neural network; prediction control; PLC; Wincc