Soft-computing Model For Sludge Volume Index Based On Recurrent Self-organizing RBF Neural Network

Sludge bulking is always one of the problem in the safe operation of the wastewater treatment system since the activated sludge process was first used, it not only occurs frequently, but also be universal. Since sludge bulking has the features of complicated mechanism, and many influential factors, it is hard to build a precise mathematical model to predict it; meanwhile, the sludge volume index(SVI), one of key parameters to quantity sludge bulking, is difficult to measure online. The existing methods for sludge bulking recognition have a few disadvantages, such as low precision, time-delay and so on. Hence, a soft-computing model basd on recurrent self-organizing RBF(RS-RBF) neural network is proposed to predict sludge bulking, the intelligent measuring system for sludge bulking can be complished and the sludge bulking recognition can be realized effectively.The main research works of this paper are as follows:1. The secondary variables selection for SVI soft-computing model. Analyse the main influencing factors of the sludge setting, based on the requirements of the sludge setting property, biochemical reaction mechanism of wastewater treatment and sludge bulking index-related research, extract the parameters associated with SVI. Analyse the importance of correlation parameters and dig through the important parameter information, correlation parameters are expressed as secondary variables, and finally five parameters(MLSS、COD、DO、pH、TN) are selected as secondary variables by PLS method.2. The design of the RS-RBF neural network. In order to improve the accuracy of SVI soft-computing model, though needs analysis about the structure of recurrent RBF neural network and the change of task, a design method of self-organizing for recurrent RBF neural network is proposed. The structure growing-pruning mechanism is based on the information processing ability and competitiveness, judgment is whether the hidden neurons of the recurrent neural RBF neural network should be added or deleted, which realized the recurrent RBF neural networs’ structure adjustment and improved the recurrent RBF neural network’s performance. The nonlinear system modeling simulation results show that the proposed RS-RBF neural network compared with other self-organizing neural network, a more compact structure and accurate prediction values have been got.3. The SVI soft-computing model study based on RS-RBF neural network. In view of the SVI on-line measurement, the RS-RBF neural network is proposed in this paper is applied to SVI soft-computing model design, and an adaptive second-order algorithm(ASOA) has been proposed to train RS-RBF, which realized the SVI on-line measurement. The design of SVI soft-computing model is applied to actual wastewater treatment process preparation platform, the simulation results show that the SVI soft-computing model compared with the traditional measurement methods needs less prior knowledge, and avoids the structure identification problem of complex system models, which can measure of SVI effectively.4. The design of the sludge bulking intelligent measuring system. In view of the sludge bulking intelligent measuring system, a SVI soft-computing platform is developed in this paper. The platform mainly includes the user management module, login module, data samples management module, model training module, model prediction module and model warning module. For the software design, firstly, using Visual Studio 2010 software designed the interface, and it provides the users for network model training, model prediction, and model early warning and so on. Then, the Matlab and Mysql softwares are used to write the background program, and soft-computing model based on RS-RBF neural network has been embedded to the platform, which realized the soft-computing model calculation. Finally, through the information transmission among users’ information management module, data processing module, and so on, realized the SVI prediction and display, achieved the purpose of sludge bulking recognition visualization.