Thorsten Schnier

Classification of Energy Consumption in Buildings With Outlier Detection

In this paper, we propose an intelligent data-analysis method for modeling and prediction of daily electricity consumption in buildings. The objective is to enable a building-management system to be used for forecasting and detection of abnormal energy use. First, an outlier-detection method is proposed to identify abnormally high or low energy use in a building. Then a canonical variate analysis is employed to describe latent variables of daily electricity-consumption profiles, which can be used to group the data sets into different clusters. Finally, a simple classifier is used to predict the daily electricity-consumption profiles. A case study, based on a mixed-use environment, was studied. The results demonstrate that the method proposed in this paper can be used in conjunction with a building-management system to identify abnormal utility consumption and notify building operators in real time.

A software architecture for Twitter collection, search and geolocation services

The substantial increase of social networks and their combination with mobile devices make rigorous analysis of the outcomes of such system of paramount importance for intelligence gathering and decision making purposes. Since the introduction of Twitter system in 2006, tweeting emerged as an efficient open social network that attracted interest from various research/commercial and military communities. This paper investigates the current software architecture of Twitter system and put forward a new architecture dedicated for semantic and spatial analysis of Twitter data. Especially, Twitter Streaming API was used as a basis for tweet collection data stored in MySQL like database. While Lucene system together with WordNet lexical database linked to advanced natural language processing and PostGIS platform were used to ensure semantic and spatial analysis of the collected data. A functional diversity approach was implemented to enforce fault tolerance for the data collection part where its performances were evaluated through comparison with alternative approaches. The proposal enables the discovery of spatial patterns within geo-located Twitter and can provide the user or operator with useful unforeseen elements.

Digital filter design using multiple Pareto fronts

Evolutionary approaches have been used in a large variety of design domains, from aircraft engineering to the designs of analog filters. Many of these approaches use measures to improve the variety of solutions in the population. One such measure is clustering. In this paper, clustering and Pareto optimisation are combined into a single evolutionary design algorithm. The population is split into a number of clusters, and parent and offspring selection, as well as fitness calculation, are performed on a per-cluster basis. The objective of this is to prevent the system from converging prematurely to a local minimum and to encourage a number of different designs that fulfil the design criteria. Our approach is demonstrated in the domain of digital filter design. Using a polar coordinate based pole-zero representation, two different lowpass filter design problems are explored. The results are compared to designs created by a human expert. They demonstrate that the evolutionary process is able to create designs that are competitive with those created using a conventional design process by a human expert. They also demonstrate that each evolutionary run can produce a number of different designs with similar fitness values, but very different characteristics.

Using negative correlation to evolve fault-tolerant circuits

In this paper, we show how artificial evolution can be used to improve the fault-tolerance of electronic circuits. We show that evolution is able to improve the fault tolerance of a digital circuit, given a known fault model. Evolution is also able to create sets of different circuits that, when combined into an ensemble of circuits, have reduced correlation in their fault pattern, and therefore improved fault tolerance. An important part of the algorithm used to create the circuits is a measure of the correlation between the fault patterns of different circuits. Using this measure in the fitness, the circuits evolve towards different, highly fault-tolerant circuits. The measure also proves very useful for fitness sharing purposes. We have evolved a number of circuits for a simple 2×3 multiplier problem, and use these to demonstrate the performance under different simulated fault models.

Using multiple representations in evolutionary algorithms

Although evolutionary algorithms are very different from other artificial intelligence search algorithms, they face similar fundamental issues-representation and searching. There has been a large amount of work done in evolutionary computation on searching, such as recombination operators, mutation operators, selection schemes and various specialised operators. In comparison, research on different representations has not been as active. Most such research has been focused on a single representation, e.g. bit strings, real-valued vectors using Cartesian coordinates, etc. This paper proposes and studies multiple representations in an evolutionary algorithm and shows empirically how multiple representations can benefit searches as much as a good search operator could.

Target shape design optimization by evolving splines

Target shape design optimization problem (TS-DOP) is a miniature model for real world design optimization problems. It is proposed as a test bed to design and analyze optimization approaches for design optimization with tremendously reducing the running period of optimization process, while, the merit can be only achieved by correctly approximating the real design situation and satisfying the causality of design and evaluation. The representation of the designed object is mostly described by parameterization techniques. To realize the design optimization, is to vary the parameterized object by means of operating the relevant parameters. The solution of design optimization often involved the choice of suitable description for the designed object, which can be obtained by expanding the design freedom. When changing the description length, the original parameters of the designed object will then varied. This bring about the requirements for optimization algorithms to self-adapt their strategy parameters and related variables to perform consistently searching. We first put forwards a revised fitness evaluation mechanism for the TSDOP in order to more reasonably check the designed shape and direct optimization procedures. Based on the revised TSDOP framework, we further discuss the parameter setting problem for algorithms, especially evolution strategies, to adapt and initial their search strategy parameters. A solution method is proposed with solving a linear equations by a recursive way with linear time complexity. All discussions are limited with the B-spline parameterization framework, but may generally suit other parameterization techniques. Experiments are used to verify the causality of the revised fitness evaluation mechanism and to study the significance of the proposed method for suitable parameter settings of optimization algorithms during the adaptation of the description length for design optimization.

An Efficient Evolutionary Approach to Parameter Identification in a Building Thermal Model

Thermal models of buildings are often used to identify energy savings within a building. Given that a significant proportion of that energy is typically used to maintain building temperature, establishing the optimal control of the buildings thermal system is important. This requires an understanding of the thermal dynamics of the building, which is often obtained from physical thermal models. However, these models require detailed building parameters to be specified and these can often be difficult to determine. In this paper, we propose an evolutionary approach to parameter identification for thermal models that are formulated as an optimization task. A state-of-the-art evolutionary algorithm, i.e., SaNSDE+, has been developed. A fitness function is defined, which quantifies the difference between the energy-consumption time-series data that are derived from the identified parameters and that given by simulation with a set of predetermined target model parameters. In comparison with a conventional genetic algorithm, fast evolutionary programming, and two state-of-the-art evolutionary algorithms, our experimental results show that the proposed SaNSDE+ has significantly improved both the solution quality and the convergence speed, suggesting this is an effective tool for parameter identification for simulated building thermal models.

A software system for data mining with twitter

The rise of social network usage through mobile devices makes the rigorous analysis of these systems of paramount importance for intelligence gathering and decision making. This paper describes the outcome of a multidisciplinary project carried out for the purpose of Twitter data collection and analysis. In particular, a proposal for enabling the discovery of spatial patterns within geo-located Twitter content has been investigated and implemented.

Evolutionary Design Calibration

Evolutionary methods are nowb eginning to be used routinely in design applications. However, even with computing speeds growing continuously, for many complex design problems evolutionary computing times are so long that their use is not practical. Divide and conquer based methods sometimes improve the situation, but in most cases the biggest speed improvement can be gained by adding domain knowledge. Combining evolutionary methods with conventional design methods is one way of doing this. This paper shows how evolutionary computation can be used to improve designs created by conventional design methods. A digital filter design problem is used to illustrate howa conventionally derived design can be further improved by evolutionary calibration. Our experimental results showthat the evolutionary calibration algorithm is able to consistently improve the original designs by a considerable margin.

A Real-Time Control Oriented HCCI Combustion Model in 4-Stroke HCCI/SI GDI Engine and Model-Based Fast Calibration Development

or Homogeneous Charge Compression Ignition (HCCI) combustion, the auto-ignition process is very sensitive to in-cylinder conditions. This includes the change in in-cylinder temperature, the composition of chemical components and their concentrations. This sensitivity presents a major challenge for the accurate control of reliable and efficient HCCI combustion. This paper outlines our recent work: 1. a real-time control oriented gasoline-fueled HCCI combustion model and its implementation in Simulink with fixed step for the conversion into dSPACE Hardware-in-the-Loop (HIL) simulation purpose. 2. The development of model-based fast calibration for the best fuel efficiency and hydrocarbon emissions via evolutionary algorithm (EA).