Muhammad Sani Gaya

Feed-Forward Neural Network Approximation Applied to Activated Sludge System

The dynamic behavior of an activated sludge system is highly complex and uncertain. To efficiently control and operate the system, a reliable model capable of accurately describing the several time-varying parallel processes of the system is needed. Most of the existing models are too complex to use for design or control purposes. This paper presents a feed-forward neural network model for the system. The model validation was achieved through the use of appropriate international accepted data of the benchmark simulation model no. 1 (BSM1). Simulation studies revealed that the neural network model exhibited an outstanding performance in predicting the effluent quality, root mean square error (RMSE) of 0.0464, mean absolute deviation (MAD) of 0.0347, correlation coefficient (R) of 0.979 for chemical oxygen demand (COD) and RMSE of 0.1103, MAD of 0.0794, R of 0.841 for the total nitrogen (TN) could be acheived. The model is quite effective and suitable tool for the activated sludge system.

ANFIS inverse control of dissolved oxygen in an activated sludge process

The activated sludge process is the main versatile technology currently in use for wastewater treatment system. Dissolved oxygen (DO) is the key element in the process due to its significance influence upon the bacteria responsible for decomposing the organic pollutants in the wastewater. However, the non-linear nature of DO couple with the time-varying oxygen transfer rate makes the DO control quite complex. This paper presents an adaptive neuro fuzzy inference system (ANFIS) inverse control of dissolved oxygen in an activated sludge process. The performance of the proposed technique is illustrated with tracking of dissolved oxygen reference trajectory and for comparison PI controller is used. The simulation results revealed the effectiveness and the accuracy of the proposed controller in tracking the DO set point. The controller is valuable for an activated sludge process.

Application of adaptive decentralized PI controller for activated sludge process

This paper presents the application of adaptive decentralized PI controller to nonlinear activated sludge wastewater treatment plant (WWTP). Tuning of WWTP is a challenging task due to the variation and the high uncertainty of the parameters. Thus, a simple tuning method is applied in satisfying straighten effluent quality and hence optimizing the nitrogen removal of the plant. The PI controller parameters are adjusted directly by updating algorithm developed based on adaptive interaction algorithm theory. It was observed that the decentralized PI with approximate Frechet tuning algorithm offers an attractive tuning task for multivariable WWTP with improvement in energy saving and effluent violations of Benchmark Simulation Model No. 1.

Data-Driven Subspace Predictive Control for a MIMO System

Subspace-based Model Predictive Control (SMPC) is a combination of a result in subspace system identification with Model Predictive Control (MPC) method. Particularly, it uses the subspace linear predictor equation to predict the future value of the system in the MPC implementation, instead of the usual state-space representation. The recursive subspace identification which updates the estimation of the extended observability matrix online is presented here for a Multi Input-Multi Output (MIMO) system specifically for a nonlinear Biological Waste Water Treatment Process. Givens rotation is applied for recursive updating of QR decomposition of a matrix in this SMPC. In SMPC, the need to have an explicit state-space representation of the system is abolished, resulting in a control algorithm that performs system identification and controller design in a single simultaneous step. Additionally, SMPC algorithm will inherit the numerical robustness typical of subspace-based methods thus giving us an easily deployable control implementation in adaptive framework.

ANFIS Based Effluent pH Quality Prediction Model for an Activated Sludge Process

Activated sludge process is the most efficient technique used for municipal wastewater treatment plants. However, a pH value outside the limit of 6-9 could inhibit the activities of microorganisms responsible for treating the wastewater, and low pH value may cause damage to the treatment system. Therefore, prediction of pH value is essential for smooth and trouble-free operation of the process. This paper presents an adaptive neuro-fuzzy inference system (ANFIS) model for effluent pH quality prediction in the process. For comparison, artificial neural network is used. The model validation is achieved through use of full-scale data from the domestic wastewater treatment plant in Kuala Lumpur, Malaysia. Simulation results indicate that the ANFIS model predictions were highly accurate having the root mean square error (RMSE) of 0.18250, mean absolute percentage deviation (MAPD) of 9.482% and the correlation coefficient (R) of 0.72706. The proposed model is efficient and valuable tool for the activated sludge wastewater treatment process.

Anfis modelling of carbon removal in domestic wastewater treatment plant

Modelling of an ill-defined system such as the wastewater treatment plant is quite tedious and difficult. However, successful and optimal operation of the system relied upon a suitable model. Most of the available developed models were applied to industrial wastewater treatment plants. This paper presents adaptive neuro fuzzy inference system (ANFIS) model for carbon removal in the Bunu domestic wastewater treatment plant in Kuala Lumpur, Malaysia. For comparison feed-forward neural network (FFNN) was used. Simulation results revealed that ANFIS model is slightly better than the FFNN model, thus proving that the model is a reliable and valuable tool for the wastewater treatment plant.

Comparison of ANFIS and Neural Network Direct Inverse Control Applied to Wastewater Treatment System

Large disturbances and highly nonlinear nature of the wastewater treatment system makes its control very difficult and challenging. The control of the system using conventional techniques becomes hard and often impossible. This paper presents a comparison of an adaptive neuro-fuzzy inference system (ANFIS) and neural network (NN) inverse control applied to the system. The performances of the controllers were evaluated based on the rise time; percent overshot and the mean error. Simulation results revealed that the ANFIS controller performance was slightly better compared to the neural network controller. The proposed ANFIS controller is effective and useful to the process.

Neuro-fuzzy modelling of wastewater treatment system

Wastewater treatment system is highly uncertain and intricate system. Suitable model is a key to smooth and optimal operation of the system. The available wastewater treatment system models are too difficult to use and costly to experiment. This paper presents neuro-fuzzy modelling of wastewater treatment system. Adaptability, smoothness, effectiveness, reliability, less computational and empirical experimentation costs are some of the advantages of neuro-fuzzy approach. Simulation studies show that the proposed neuro-fuzzy technique yielded outstanding results. Thus, proven the technique is an efficient and valuable tool for modelling wastewater treatment system.

A LabVIEW-Based Simulator for the Activated Sludge Process

The activated sludge process is the main process in most urban wastewater treatment systems. It is considered complex in nature, and building its mathematical model in practice becomes difficult. A simple and easy way to build the simulation models for the system is needed. In this paper, a LabVIEW based simulator for the system is presented. LabVIEW offers a highly efficient, simple and flexible platform for simulation and control. Building applications in LabVIEW require less coding, and debugging is easy and fast. The proposed simulator utilizes the Benchmark Simulation Model no.1 (BSM 1) for the biochemical reactor and clarification processes. The operation of the simulator is via a graphical user interface (GUI) built in the LabVIEW environment. The simulation results can be displayed in digital and graphical forms. Simulation results obtained were compared with the results from other software simulation packages in which the COST/IWA Simulation Benchmark has been implemented. The developed simulator is very useful due to its efficiency and accuracy in simulating the wastewater process model. The simulator can serve as a training tool for plant operators/students to provide them with better knowledge and understanding of the process. (Less)