See below for the (brief) description of the smart grid project *Motivation*: By 2020, the number of installed smart electricity meters in Europe will reach 240 million, North America 150 million China 400 million Japan 60 million. Data analytics stakeholders are now competing to bring IT tools and capabilities that are new to the power grid. A wide variety of techniques and technologies to harvest, analyze and visualize power grid data are developed. They involve a number of disciplines, including statistics, clustering algorithms, data mining, machine learning, signal processing, pattern recognition, optimization and visualization methods.
The *objective* of our project is to design, implement and perform cases studies that exhibit the effectiveness, the robustness and the efficiency of the WSO2 platform in general and DAS/ML in particular in the wide area of the next generation energy systems. Our case studies should enjoy several desirable characteristics: They should be as realistic as possible, large scale but also detailed, of practical importance and requiring real-time strategy development. *Specifically* our efforts are divided into three phases that will presumably lead us to the conclusion that WSO2 is capable of playing the role of the soft part of the power grid. Initial stage - Work on simulation data for energy markets: 1. We start with a simple machine learning experiment that resamples the study found in the first 5 sections [1] on the WSO2 platform using DAS/ML. That is, harvest the required data and train the power producing (wind turbines and solar panels) and consuming devices (HVAC, water heaters, …) to bid on 15’ energy auctions in a way that utilise, in addition to historical data, weather prediction data too. 2. We next add game theoretical components. In particular we add the game proposed and utilized in the last sections of [1] which can be classified as non-cooperative, since players choose their strategy without communicating with each other. 3. Next we add a cooperative game component that is based on data analytics on selected configuration scenarios which are of significant importance for the emerging energy markets. Intermediate stage - real time data: Extend the above to real time simulation data and perhaps on virtual-physical scenarios where part of the data are coming from real meters. Extended stage: We investigate the full capabilities of WSO2 platform to 1. Extend the results found in some of the recent research papers , 2. Elucidate, through a mix of data analytics, machine learning and practical game, several unclear issues, 3. Validate/invalidate important conjectures (e.g. phase variation is related to market clearing price, Breass’ paradox [2] for power grid), 4. Explore the role batteries and electric cars on the overall power and economic stability of the power grids through attentive analytics and ML. [1] Intelligent Bidding in Smart Electricity Markets <http://www.igi-global.com/article/intelligent-bidding-in-smart-electricity-markets/146155> [2] https://arxiv.org/abs/1504.04319 On Mon, Jul 25, 2016 at 8:17 PM, Nirmal Fernando <nir...@wso2.com> wrote: > Also, you could get the Mean Square Error from the model summary page > which should be a good measurement about the model. > > On Mon, Jul 25, 2016 at 4:48 PM, Manolis Vavalis <m.vava...@gmail.com> > wrote: > >> >> On Jul 25, 2016, at 4:35 PM, Nirmal Fernando <nir...@wso2.com> wrote: >> >> You can use Random Forest Regression too. It should be more accurate than >> linear regression. >> >> >> Good idea. We will try it out right away. >> >> >> Also, can you please explain the use-case? >> >> >> I will ask the interns to describe the use-case on their own words and >> comment if required. >> >> Cheers >> >> Manolis >> >> >> On Mon, Jul 25, 2016 at 3:40 PM, Nihla Akram <ni...@wso2.com> wrote: >> >>> Hello Srinath, >>> >>> >>> Yes, we used WSO2 ML. >>> >>> We received some csv files containing the weather and other related data >>> from Magda in order to predict the clearing price using WSO2 ML. >>> >>> Initially we used Linear Regression with the default configurations. >>> >>> Below is the prediction obtained by changing the Parameter >>> configurations. These results are quite close to the initial predicted >>> values obtained from Magda. >>> >>> >>> Thanks, >>> Nihla >>> >>> On Mon, Jul 25, 2016 at 2:52 PM, Srinath Perera <srin...@wso2.com> >>> wrote: >>> >>>> adding archtecture@ >>>> >>>> What tool did you used to train the regression? Is it WSO2 ML. Can you >>>> share details about the process? >>>> >>>> --Srinath >>>> >>>> On Mon, Jul 25, 2016 at 2:04 PM, Sanjaya De Silva <sanja...@wso2.com> >>>> wrote: >>>> >>>>> Hi all, >>>>> >>>>> Following are the files that I used to train and test. >>>>> >>>>> On Mon, Jul 25, 2016 at 1:59 PM, Nihla Akram <ni...@wso2.com> wrote: >>>>> >>>>>> Hello All, >>>>>> >>>>>> >>>>>> The following are few attachments which was used to train and test >>>>>> the ML values obtained for clearing price. Please note that the predicted >>>>>> values weren't accurate. >>>>>> *trainerData.csv* is the file used to train the model. >>>>>> *resultData.csv* is the result file produced for predictions on* >>>>>> testData.csv* file. >>>>>> >>>>>> >>>>>> The configurations of the Model were as follows. >>>>>> Algorithm : Linear Regression >>>>>> Response variable : clearingprice >>>>>> Train data fraction : 0.7 >>>>>> >>>>>> >>>>>> >>>>>> Thanks, >>>>>> Nihla >>>>>> >>>>>> >>>>>> -- >>>>>> *Nihla Akram* >>>>>> Software Engineering Intern >>>>>> >>>>>> +94 72 667 9482 <%2B94%2072%6679482> >>>>>> >>>>>> >>>>> >>>>> >>>>> -- >>>>> Thank you >>>>> Best Regards >>>>> >>>>> Sanjaya De Silva >>>>> Trainee Software Engineer >>>>> WSO2 >>>>> +94774181056 >>>>> >>>> >>>> >>>> >>>> -- >>>> ============================ >>>> Srinath Perera, Ph.D. >>>> http://people.apache.org/~hemapani/ >>>> http://srinathsview.blogspot.com/ >>>> >>> >>> >>> >>> -- >>> *Nihla Akram* >>> Software Engineering Intern >>> >>> +94 72 667 9482 <%2B94%2072%6679482> >>> >>> >>> _______________________________________________ >>> Architecture mailing list >>> Architecture@wso2.org >>> https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture >>> >>> >> >> >> -- >> >> Thanks & regards, >> Nirmal >> >> Team Lead - WSO2 Machine Learner >> Associate Technical Lead - Data Technologies Team, WSO2 Inc. >> Mobile: +94715779733 >> Blog: http://nirmalfdo.blogspot.com/ >> >> >> >> > > > -- > > Thanks & regards, > Nirmal > > Team Lead - WSO2 Machine Learner > Associate Technical Lead - Data Technologies Team, WSO2 Inc. > Mobile: +94715779733 > Blog: http://nirmalfdo.blogspot.com/ > > > > _______________________________________________ > Architecture mailing list > Architecture@wso2.org > https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture > > -- Malith Jayasinghe WSO2, Inc. (http://wso2.com) Email : mali...@wso2.com Mobile : 0770704040 Lean . Enterprise . Middleware
_______________________________________________ Architecture mailing list Architecture@wso2.org https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture
