Jiaming Li

Automatic liver parenchyma segmentation from abdominal CT images using support vector machines

This paper presents an automatic liver parenchyma segmentation algorithm that can segment liver in abdominal CT images. There are three major steps in the proposed approach. Firstly, a texture analysis is applied to input abdominal CT images to extract pixel level features. In this step, wavelet coefficients are used as texture descriptors. Secondly, support vector machines (SVMs) are implemented to classify the data into pixel-wised liver area or non-liver area. Finally, integrated morphological operations are designed to remove noise and finally delineate the liver. Our unique contributions to liver segmentation are twofold: one is that it has been proved through experiments that wavelet features present good classification result when SVMs are used; the other is that the combination of morphological operations with the pixel-wised SVM classifier can delineate volumetric liver accurately. The algorithm can be used in an advanced computer-aided liver disease diagnosis and liver surgical planning system. Examples of applying the proposed algorithm on real CT data are presented with performance validation based on the comparison between the automatically segmented results and manually segmented ones.

Evolutionary optimisation of distributed energy resources

Genetic optimisation is used to minimise operational costs across a system of electrical loads and generators controlled by local intelligent agents and connected to the electricity grid at market rates. Experimental results in a simulated environment show that coordinated market-sensitive behaviours are achieved. A large network of 500 loads and generators, each characterised by different randomly selected parameters, was optimised using a two-stage genetic algorithm to achieve scalability.

Unsupervised feature selection using swarm intelligence and consensus clustering for automatic fault detection and diagnosis in Heating Ventilation and Air Conditioning systems

Graphical abstractDisplay Omitted HighlightsOur algorithm aims to improve the feature quality in general fault diagnosis system.The algorithm filters out redundant features using consensus evolutionary clustering.The algorithm was tested on the ASHRAE-1312-RP experimental fault data.Sensitivity & specificity were >95%, with considerably less false positives up to as low as 1.6%. Various sensory and control signals in a Heating Ventilation and Air Conditioning (HVAC) system are closely interrelated which give rise to severe redundancies between original signals. These redundancies may cripple the generalization capability of an automatic fault detection and diagnosis (AFDD) algorithm. This paper proposes an unsupervised feature selection approach and its application to AFDD in a HVAC system. Using Ensemble Rapid Centroid Estimation (ERCE), the important features are automatically selected from original measurements based on the relative entropy between the low- and high-frequency features. The materials used is the experimental HVAC fault data from the ASHRAE-1312-RP datasets containing a total of 49 days of various types of faults and corresponding severity. The features selected using ERCE (Median normalized mutual information (NMI)=0.019) achieved the least redundancies compared to those selected using manual selection (Median NMI=0.0199) Complete Linkage (Median NMI=0.1305), Evidence Accumulation K-means (Median NMI=0.04) and Weighted Evidence Accumulation K-means (Median NMI=0.048). The effectiveness of the feature selection method is further investigated using two well-established time-sequence classification algorithms: (a) Nonlinear Auto-Regressive Neural Network with eXogenous inputs and distributed time delays (NARX-TDNN); and (b) Hidden Markov Models (HMM); where weighted average sensitivity and specificity of: (a) higher than 99% and 96% for NARX-TDNN; and (b) higher than 98% and 86% for HMM is observed. The proposed feature selection algorithm could potentially be applied to other model-based systems to improve the fault detection performance.

An Approach of Household Power Appliance Monitoring Based on Machine Learning

Monitoring household electrical consumption by employing appropriate techniques is of great significance to sustainable development of human society. This paper proposes one approach of nonintrusive appliance load monitoring (NIALM) for electrical consumption managing. This approach can automatically monitor the house power consumption of individual devices. It employs multiple-class support vector machine (M-SVM) to recognize different appliances. The approach is consisted of two stages. In stage one, harmonic feature analysis is applied on current signal. In stage two, a trained classifier based on M-SVM is applied to identify different appliances. This paper presents the principle of this approach, the experiment results on real data, and discussions on performance comparison with other study of supervised classification for household power appliance monitoring.

Set-Points Based Optimal Multi-Agent Coordination for Controlling Distributed Energy Loads

The management of a very large number of distributed energy resources, energy loads and generators, to create aggregated quantity of power is a hot research topic. We consider a multi-agent system comprising multiple energy loads, each with a dedicated controller. This paper introduces our latest research in self-organization of coordinated behavior of multiple agents. Energy resource agents coordinate with each other to achieve a balance between the overall consumption by the multi-agent collective and the stress on the community. In order to reduce the overall communication load while permitting efficient coordinated responses, information exchange is through indirect communications between resource agents and a broker agent. It gives a decentralized coordination approach that does not rely on intensive computation by a central processor. The algorithm presented here can coordinate different types of loads by controlling their set-points. The coordination strategy is optimized by a genetic algorithm. A fast coordination convergence has been achieved.

Literature review of power disaggregation

There are mainly two classes of approaches in power disaggregation, including Intrusive Load Monitoring (ILM) and Nonintrusive Load Monitoring (NILM). This paper presents the literature review on the NILM approaches. NILM is a process for detecting changes in the voltage and current going through a house, deducing what appliances are used in the house, as well as their individual energy consumption, with a single set of sensors. Different strategies and approaches for NILM systems have been developed over the past thirty years. This paper reviews the current state of the algorithms and systems of NILM. The paper points out that NILM can be utilised presently on available commercial devices and provides meaningful feedback. Our vision on the future of NILM is also summarized.

Automatic power load event detection and appliance classification based on power harmonic features in nonintrusive appliance load monitoring

Home electrical power monitoring plays an important role in reducing energy usage, and non-intrusive appliance load monitoring (NIALM) techniques are the most effective approach for estimating the electrical power consumption of individual appliances. Power load events detection is one of the most important steps in these techniques. This paper presents an automatic power load event detection method: edge symbol detector (ESD) for NIALM. The new transient detection approach can help the system locate all the load events (switch on and switch off) precisely. A modified power appliance classification technique based on power harmonic features and support vector machine (SVM), with higher recognition accuracy and faster computational speed, is also discussed. The experimental results of the new load events detection and classification technique are presented with promising results.

Sensor data fusion for accurate cloud presence prediction using Dempster-Shafer evidence theory.

Sensor data fusion technology can be used to best extract useful information from multiple sensor observations. It has been widely applied in various applications such as target tracking, surveillance, robot navigation, signal and image processing. This paper introduces a novel data fusion approach in a multiple radiation sensor environment using Dempster-Shafer evidence theory. The methodology is used to predict cloud presence based on the inputs of radiation sensors. Different radiation data have been used for the cloud prediction. The potential application areas of the algorithm include renewable power for virtual power station where the prediction of cloud presence is the most challenging issue for its photovoltaic output. The algorithm is validated by comparing the predicted cloud presence with the corresponding sunshine occurrence data that were recorded as the benchmark. Our experiments have indicated that comparing to the approaches using individual sensors, the proposed data fusion approach can increase correct rate of cloud prediction by ten percent, and decrease unknown rate of cloud prediction by twenty three percent.

Dynamic zone modelling for HVAC system control

This paper presents the development and validation of a dynamic zone model used for improved control of a heating, ventilation and air conditioning (HVAC) system to reduce energy consumption and improve the quality of the indoor environment. In particular, the paper focuses on a zone modelling technique that uses physical-principles based real-time model fitting and prediction methodology, taking advantage of genetic algorithm based problem solving. An air-conditioning zone model is deduced from an energy and mass balance and then expressed in terms of electric circuit theory, where the electric circuit is used to represent functions of the building elements. Experimental results for real-time zone model fitting and prediction are given. The results show that our model is capable of accurately predicting the indoor temperature of a dynamic zone. This dynamic model is useful for control strategies that require knowledge of the dynamic characteristics of HVAC systems.

Directed self-assembly of 2-dimensional mesoblocks using top-down/bottom-up design

In this paper we present a general design methodology suitable for a class of complex multi-agent systems which are capable of self-assembly. Our methodology is based on a top-down, bottom-up approach, which has the potential to achieve a range of global design goals whilst retaining emergent behaviour somewhere in the system, and thereby allowing access to a richer solution space. Our experimental environment is a software system to model 2-dimensional self-assembly of groups of autonomous agents, where agents are defined as square smart blocks. The general design goal for such systems is to direct the self-assembly process to produce a specified structure. The potential of this design methodology has been realised by demonstrating its application to a toy problem - the self-assembly of rectangles of different sizes and shapes in a two-dimensional mesoblock environment. The design procedure shows different choices available for decomposing a system goal into subsidiary goals, as well as the steps needed to ensure a match to what is achievable from the bottom-up process. Encouraging results have been obtained, which allows mesoblock rectangles of specified size to be assembled in a directed fashion. Two different approaches to the same problem were presented, showing the flexibility of the method.