The problem of energy load forecasting has emerged as an essential area of research for electrical distributors seeking to minimize costs. This problem has a high degree of complexity; therefore, this paper solves the problem of short-term load forecasting for a day ahead using an adaptive fuzzy model, defined across the entire input space in order to share information between different areas. The proposed solution first separates the forecasting of daily load profiles into smaller, simpler subproblems, which are solved separably using a Takagi-Sugeno fuzzy model. This is done in order to solve smaller subproblems better, which brings improved forecasting accuracy after combining the subproblem results. The identification of the model is based on a recursive Gustafson-Kessel clustering and recursive weighted least mean squares, to which a combined membership function is proposed in order to improve domain partitioning. The model was tested on the real data obtained from a large Slovenian energy distribution company, at which the developed model forecast outperformed other methods, especially in the start of the week and the winter.
18 September 2017 in IEEE Transactions on Fuzzy Systems
Gradual drifts in data streams are usually hard to detect and often do not necessarily trigger the evolution of new fuzzy rules during model adaptation steps in order to represent the new, drifted data distribution(s) appropriately in the fuzzy model. Over time, they thus lead to oversized rules with untypically large local errors (typically also worsening the global model error), as representing joint local data distributions before and after a drift happened likewise. We therefore propose an incremental rule splitting concept for generalized fuzzy rules in order to autonomously compensate these negative effects of gradual drifts. Our splitting condition is based 1.) on the local error of rules measured in terms of a weighted contribution to the whole model error and 2.) on the size of the rules measured in terms of the volume of the associated clusters. We use the concept of statistical process control in order to omit an extra threshold parameter in our splitting condition. The splitting technique relies on the eigen-decomposition of the rule covariance matrix to adequately manipulate the largest eigenvector and eigenvalues in order to retrieve the new centers and contours of the two split rules. Furthermore, we guarantee sufficient flexibility in adapting the shapes and consequents of the split rules to the new drifted situation in the stream by integrating a specific dynamic and smooth forgetting concept of older samples, which formed the original (non-split) rules. Robustness against outliers is guaranteed by the realization of a two-layer model building process, where one layer represents the cluster partition and the other layer the rule partition: only clusters becoming significant over time are accepted as rules in the fuzzy model. The splitting concepts are integrated in the generalized smart evolving learning engine for fuzzy systems (termed as Gen-Smart-EFS) and successfully tested on two real-world application scenarios, engine test benches and rolling mills, the latter including a real-occurring gradual drift (whose position in the data is known). Results show clearly improved error trend lines over time when splitting is applied, compared to the case when it is not applied: reduction of the mean absolute model error by about one third (rolling mills) and about one half (engine test benches).
Missing values are common in real-world data stream applications. This paper proposes a modified evolving granular fuzzy rule-based model for function approximation and time series prediction in an online context where values may be missing. The fuzzy model is equipped with an incremental learning algorithm that simultaneously imputes missing data and adapts model parameters and structure over time. The evolving Fuzzy Granular Predictor (eFGP) handles single and multiple missing values on data samples by developing reduced-term consequent polynomials and utilizing time-varying granules. Missing at random (MAR) and missing completely at random (MCAR) values in nonstationary data streams are approached. Experiments to predict monthly weather conditions, the number of bikes hired on a daily basis, and the sound pressure on an airfoil from incomplete data streams show the usefulness of eFGP models. Results were compared with those of state-of-the-art fuzzy and neuro-fuzzy evolving modeling methods. A statistical hypothesis test shows that eFGP outperforms other evolving intelligent methods in online MAR and MCAR settings, regardless of the application.
3 December 2018 in IEEE Transactions on Industrial Electronics
The paper presents an electric arc furnace (EAF) optimization framework intended to define optimal control profiles for the EAF, in order to increase its efficiency and thus reduce the energy consumption. The framework aims to minimize controllable losses and to maximize energy transfer to the bath, and consequently minimize the operational costs. This is achieved through improved actuation of the EAF inputs, i.e. transformer power, oxygen lancing and carbon addition. To achieve maximal energy transfer to the bath and to reduce the heat losses from the arcs, proper properties of the slag, such as foaminess and basicity, are a subject of considerable attention. The framework is designed as an model-based optimization, intended to be executed online in parallel to the actual EAF process. In order to achieve sufficiently low computational complexity and to allow process optimization by arbitrary time intervals, the framework uses path constraints instead of end-point constraints. A combination of several optimization algorithms is used to solve the optimization problem. The validation of the framework was performed by comparing the predicted and the measured operational variables. Simulation results show that optimized operation profiles lead to significant decrease in operational costs and production times.