I certify that this final work meets all the requirements as a Master's Thesis for the degree of Master of Science in Energy Engineering. This thesis, written by Futoon Eissa Alsharif under the guidance of his/her thesis supervisor and approved by his/her thesis committee, was submitted to and accepted by the Dean of Postgraduate Studies and Research on ………., in partial fulfillment of the requirements for the degree MAGISTER SCIENCE in Energy Engineering. Then, the new adaptive rate techniques are used for data conditioning, segmentation and feature extraction.
I would also like to express my gratitude to the Chairman of the Department of Electrical and Computer Engineering, Dr. I also certify that the proposed dissertation has not been previously or concurrently submitted for the award of a degree, to Effat University, any other university or Institution.
Problem Statement
Therefore, it results in the collection, transmission and processing of a large amount of unnecessary data. This work uses the event-driven sensing and analysis approaches to achieve real-time data compression and processing efficiency while ensuring high classification accuracy.
Objective
Importance and Motivation
Improved accuracy in terms of energy readings compared to the previous electromechanical meters where the measurement errors were quite sensitive and dependent on human operators. Occupancy detection: Information about the occupation of a place can be detected by the monitoring unit to control the appliances to apply energy saving measures. User-device interaction: Knowledge of how many times the user interacts with the devices is also useful in designing an appropriate load monitoring system.
These are potential applications to be integrated in the upcoming smart cities in Saudi Arabia, like NEOM.
- Smart Metering
- Smart Meter Data Acquisition
- Features Extraction
- Pattern Recognition
It provides accurate and up-to-date consumption data to connected persons and their systems and energy users. It presents a demand decomposition technique in a smart meter system with only a few customers who have .. the ability to monitor every device they own as a relatively practical situation for a potential smart distribution network. The definition of pattern recognition is supposed to be the categorization of data according to the information obtained or the statistical data obtained from the patterns [30].
The potential application of pattern recognition is considered as one of the significant features of pattern recognition [30]. In [36], using intrusive load monitoring (ILM), the authors discussed the use of Hidden Markov Models (HMMs) for device recognition.
![Figure 2: Applications and advantages of smart meter [15]](https://thumb-ap.123doks.com/thumbv2/azpdfco/10320150.0/23.918.191.724.102.384/figure-applications-advantages-smart-meter.webp)
- The Hypothesis
- The Smart Meter Database
- The Signal Reconstruction
- The Event–Driven Sensing (EDS)
- The Event–Driven Segmentation
- The Features Extraction
- The Classification Techniques
- The Cross–Validation
- The Evaluation Measures
The signal reconstruction process is the opposite of the sampling process, also known as interpolation [41]. In order to reconstruct a signal, the frequency of the analog signal must be lower than the Nyquist frequency, 𝑓𝑠. Thus, the design parameters of the traditional ADCs are selected for the worst case [43].
Event-driven sensitivity is a part of the class of sampling that introduces non-uniformity indirectly into the sampling process [46], [47]. The EDADC receives only the relevant information while the rest of the signal is ignored. In order to reduce the processing activity of the system and therefore reduce the energy consumption, the choice of the activity is very important [45].
The following mathematical expression describes the calculation of the maximum number of samples, 𝑁𝑚𝑎𝑥, that can be located within a selected 𝐿𝑟𝑒𝑓. One of these features is the ability to extract the value of the sampling frequency for each selected window. It can also compare the chosen length of the window with the activity of the signal it contains.
Of all major machine learning techniques, SVM is one of the most accurate and robust algorithms. The main steps of the ANN classification technique are clearly described in the following figure 17. The main steps of the Naïve Bayes classification techniques are described in the following figure.
Finally, the value of the average of the errors assigned in each fold is the k-cross-validation error estimate [54]. For event-driven sensing, the sampling frequency is not unique and adapts as a function of the temporal variations of the input signal [46], [47].
- Case (1)
- Case (2)
- Case (3)
- Case (4)
- Extended and Generalized Study for 6–Classes of Appliances
It confirms a drastic reduction of the proposed solution circuit complexity while comparing with the classical solution. In this case 𝐿𝑟𝑒𝑓 = 5 minutes results in a fixed number of 30 samples per segment regardless of the signal variations. Firstly, however, it can lead to a reduction in ADC performance in terms of signal-to-noise ratio (SNR), since the SNR output is inversely related to the quantum value used in the system [45], [56] .
In this paper, Qaisar and Futoon studied the examples of the considered actual energy consumption cases obtained by the EDS mechanism, which are shown in Figure 24. It shows that the proposed solution achieves a total of 2.1 times, 6.6 times, 1.6- times and 3.2 times the compression boost for the following devices; coffee. It provides a noticeable reduction in the arithmetic complexity and energy consumption of the proposed solution compared to the conventional approach.
The obtained percentage of recognition accuracy is summarized for the K-NN and SVM classification techniques respectively in Table 4 and Table 5. It results in 90 multidimensional time series of electrical energy-related characteristics of the considered devices [55]. The classification technique used in this case is the Naïve Bayes classifier [55]. Examples of considered cases of real energy consumption obtained with the EDS mechanism are shown in Figure 25 [55].
It shows that the proposed solution yields a total of 3.1 times, 6.6 times and 3.4 times more compression gains for kettles, refrigerators and freezers and fans respectively [55]. It shows that the proposed solution yields a total of 3.1 times, 6.6 times, 1.6 times and 3.4 times compression gains for the case of kettles, fans, refrigerators and freezers and microwave ovens, respectively. It shows that the proposed solution yields a total of 3.1 times, 6.6 times, 2.2 times and 6.0 times compression gains for the case of kettles, refrigerators and freezers, monitors and televisions, respectively.
It shows that the proposed solution achieves a total of 3.1x, 6.6x, 1.6x, 3.4x, 2.2x, and 6.0x compression gains for water heaters, chillers, and freezers, microwave ovens, fans, monitors and TVs. In short, the proposed solution results in a significant reduction in complexity and energy consumption, a reduction in data transfer performance compared to the classical method, and an increase in system performance.
An electricity bill with more data on consumption, that is now an unknown measurement in the meter at home. Load shedding optimization, understanding which device turns ON or OFF based on local or global electricity demand. The classic analog-to-digital converters sample and process the data based on the Nyquist principle.
In the following step, the essential adaptive data processing, segmentation and extraction techniques are proposed. Compared with classical counterparts, it confirms a significant compression and computation efficiency of the proposed method. They work based on event-driven sampling, and they can adjust their sampling rate according to the variations of the incoming signal.
Based on the results presented in the previous results section, it can be seen that different compression increments for the different implemented studied cases provide a noticeable reduction in the arithmetic complexity and energy consumption of the proposed solution compared to the conventional approach. In Europe, air conditioning mainly deals with warm radiators that work on the basis of circulating hot water in the home, and this hot water is supplied from centrally heated devices that heat the water and then supply it through good insolent pipes. to various buildings and facilities. The comparison between the proposed technique and state-of-the-art techniques is not obvious, as they are based on classical sampling and processing techniques.
The main advantage of the proposed system over the inverse equivalents is the significant compression gain. Moreover, it guarantees a significant efficiency in processing and energy consumption of the proposed solution over the equivalents. On the other hand, the disadvantage of this method is that it is still a new method, it has not been tested on a variety of devices or extended databases of devices.
Conclusion
Future Work
Laska et al., “Random Sampling for Analog-to-Information Conversion of Wideband Signals,” in 2006 IEEE Dallas/CAS Workshop on Design, Applications, Integration, and Software, Univ. Gruteser, “Neighborhood Watch: Security and Privacy Analysis of Automated Meter Reading Systems,” in Proceedings of the 2012 ACM Conference on Computer and Communications Security - CCS '12, Raleigh, North Carolina, USA, 2012, p. Reinhardt et al., "On the accuracy of device identification based on distributed load measurement data," in 2012 Sustainable Internet and ICT for Sustainability (SustainIT), 2012, pp.
Staake, "Leveraging slim meter data to identifiing home appliances," in 2012 IEEE International Conference on Pervasive Computing and Communications, 2012, pp. Hennebert, "ACS-F2—A new database of appliance consumer signatures," in 2014 6th International Conference of Soft Computing and Pattern Recognition (SoCPaR), 2014, pp. Ahmed, "Towards Efficient Energy Monitoring Using IoT," in 2018 IEEE 21st International Multi-Topic Conference (INMIC), 2018, pp.
Qaisar, "A Computationally Efficient EEG Signals Segmentation and De-noising Based on an Adaptive Rate Acquisition and Processing," in 2018 IEEE 3rd International Conference on Signal and Image Processing (ICSIP), 2018, pp. Yahiaoui, "Power-Efficient Analog to Digital Conversion for Li-ion battery voltage monitoring and measurement," in 2013 IEEE International Instrumentation and Measurement Technology Conference (I2MTC), 2013, pp. Dominique, "A smart power management system monitoring and measurement approach based on a signal-driven data acquisition," in 2015 Saudi Arabia Smart Grid (SASG), 2015, p.
Klauer, "Model Testing and Selection, Theory of," i International Encyclopedia of the Social & Behavioral Sciences, N. Gharbi, "En effektiv signalindsamling med en adaptiv hastighed A/D-konvertering," i 2013 IEEE International Conference on Circuits and Systems (ICCAS), 2013, s. Hennebert, "Appliance and state recognition using Hidden Markov Models," i 2014 International Conference on Data Science and Advanced Analytics (DSAA), okt.
![Figure 5: The system of appliance identification [27]](https://thumb-ap.123doks.com/thumbv2/azpdfco/10320150.0/31.918.212.727.111.355/figure-appliance-identification.webp)
