My research interests are focused on methods of computational intelligence, pattern recognition and machine learning in application to:

• analysis and forecasting of time series with multiple seasonal cycles, 
• data classification,
• feature selection,
• combinatorial and continuous optimization.

Analysis and forecasting of time series with multiple seasonal cycles

I proposed many forecasting models based on neural networks, artificial immune systems, fuzzy logic, clustering methods, regression trees, kernel and nearest neighbor estimators. A key element of these models is appropriate data preprocessing, which helps to simplify the forecasting problem. I tested these models in the short-term electrical load forecasting. I summarized some of my achievements in this field in the book: Similarity-based Machine Learning Methods for Short-Term Load Forecasting. Academic Publishing House EXIT, Warsaw 2012 (habilitation monograph; in Polish). 

I collaborate with the Polish power system operator (PSE-Operator) in developing forecasting models.   

Data classification  

I proposed a new multi-class classifier based on immune system principles. The unique feature of this classifier is the embedded property of local feature selection. This method of feature selection was inspired by the binding of an antibody to an antigen, which occurs between amino acid residues forming an epitope and a paratope. 

Feature selection

I proposed a new search method to the feature selection problem – the tournament searching. The tournament feature selection method is a simple stochastic searching method with only one parameter controlling the global-local searching properties of the algorithm. It is less complicated and easier to use than other stochastic methods, e.g. the simulated annealing or genetic algorithm. 

Combinatorial and continuous optimization

In my PhD thesis I solved the problem of unit commitment and economic dispatch using evolutionary algorithms and simulated annealing. I proposed several methods of data representations, heuristic genetic operators, definitions of the objective function for the feasible and infeasible solutions, adaptive annealing schedule and hybrid methods combining simulated annealing and evolutionary algorithms. I summarized my research in this field in the book: New Optimization Methods for the Electric Power Unit Commitment Problem. Czestochowa University of Technology Publishing House, Czestochowa 2011 (based on the PhD thesis; in Polish). 

I collaborated with the Polish power system operator (PSE-Operator) in developing methods of long-term optimal dispatch of generating units.

I proposed directed mutation operators (tournament mutation, roulette wheel mutation) to improve the speeding up the convergence of genetic algorithms and  tournament searching.

I used evolutionary computation and tournament searching in other optimization problems such as: tuning of the parameters of the forecasting models and hysteresis model, localization of the power substations.