A recurrent neuro-fuzzy system and its application in inferential sensing

ReNFuzz-LF: A Recurrent Neurofuzzy System for Short-Term Load Forecasting

A neurofuzzy system is proposed for short-term electric load forecasting. The fuzzy rule base of ReNFuzz-LF consists of rules with dynamic consequent parts that are small-scale recurrent neural networks with one hidden layer, whose neurons have local output feedback. The particular representation maintains the local learning nature of the typical static fuzzy model, since the dynamic consequent parts of the fuzzy rules can be considered as subsystems operating at the subspaces defined by the fuzzy premise parts, and they are interconnected through the defuzzification part. The Greek power system is examined, and hourly based predictions are extracted for the whole year. The recurrent nature of the forecaster leads to the use of a minimal set of inputs, since the temporal relations of the electric load time-series are identified without any prior knowledge of the appropriate past load values being necessary. An extensive simulation analysis is conducted, and the forecaster’s performance is evaluated using appropriate metrics (APE, RMSE, forecast error duration curve). ReNFuzz-LF performs efficiently, attaining an average percentage error of 1.35% and an average yearly absolute error of 86.3 MW. Finally, the performance of the proposed forecaster is compared to a series of Computational Intelligence based models, such that the learning characteristics of ReNFuzz-LF are highlighted.

Development of soft sensor for neural network based control of distillation column

The present work is aimed at the design of Levenberg-Marquardt (LM) and adaptive linear network (ADALINE) based soft sensors and their application in inferential control of a multicomponent distillation process. Further the ADALINE sensor is trained online using past measurements, to adapt the changes in the inputs and is termed as dynamic ADALINE (D-ADALINE) sensor. The soft sensors are then used in the control loop to obtain LM based inferential controller (LMIC), ADALINE based inferential controller (ADIC) and D-ADALINE based inferential controller (DADIC) for the process. The performance of dynamic controller is also analyzed for different inputs and sampling intervals. The comparison of results shows the efficient and robust prediction capability of D-ADALINE sensor and hence DADIC proves to be the best controller.

A concurrent neuro-fuzzy inference system for screening the ecological risk in rivers

PURPOSE

A conceptual model to assess water quality in river basins was developed here. The model was based on ecological risk assessment principles, and incorporated a novel ranking and scoring system, based on self-organizing maps, to account for the likely ecological hazards posed by the presence of chemical substances in freshwater. This approach was used to study the chemical pollution in the Ebro River basin (Spain), whose currently applied environmental indices must be revised in terms of scientific accuracy.

METHODS

Ecological hazard indexes for chemical substances were calculated by pattern recognition of persistence, bioaccumulation, and toxicity properties. A fuzzy inference system was proposed to compute ecological risk points (ERP), which are a combination of the ecological hazard to aquatic sensitive organisms and environmental concentrations. By aggregating ERP, changes in water quality over time were estimated.

RESULTS

The proposed concurrent neuro-fuzzy model was applied to a comprehensive dataset of the network controlling the levels of dangerous substances, such as metals, pesticides, and polycyclic aromatic hydrocarbons, in the Ebro river basin. The approach was verified by comparison versus biological monitoring. The results showed that water quality in the Ebro river basin is affected by presence of micro-pollutants.

CONCLUSIONS

The ERP approach is suitable to analyze overall trends of potential threats to freshwater ecosystems by anticipating the likely impacts from multiple substances, although it does not account for synergies among pollutants. Anyhow, the model produces a convenient indicator to search for pollutant levels of concern.