CITY, PROMAR and NET participated in the development of the necessary ICT architecture for the operation of the overall system. As it was mentioned in the description of the Task 3.1 the first initiative is the development of the Smart Meters (SM) and related ICT architecture in order to establish a reliable communication between SM and Central Monitoring and Control web Platform (CMCP). The information from the SM is transferred to the server/database segment thus all necessary compatibility of the different components should be checked and verified. The VPS (Virtual Private Server) used for hosting the website uses Ubuntu as its operating system which is a Linux based operating system. Moreover, this server is comprised of 1 CPU, 1GB of RAM and 10GB of disk space which can be upgraded anytime .
The overall ICT architecture is depicted in Figure 5.1 which comprises the following components
1. Development of advanced multi-modal data fusion/processing and computational intelligence
In this work some of the most popular classification algorithms and techniques were implemented in Matlab. In particular, various machine learning techniques were implemented regarding classification of Support Vector Machines (with the Linear (LIN), Radial Basis Function (RBF), Polynomial (POLY) kernels), Decision Trees, Random Forest, Naive-Bayes, Logistic Regression, Multi-Layer Perceptron and Adaptive Boosting (AdaBoost).
2. Distributed management and control
distributed control structures were analysed where the local regulators are designed with Model Predictive Control (MPC) strategies and it is assumed that there is some information exchange, transmitted among the local regulators so that they can predict the interaction effects (over the prediction horizon). Two cases of distributed MPC algorithms were considered, depending on the topology of the communication network:
- fully connected MPC algorithms: where information is transmitted (and received) from any local regulator to all the others;
- partially connected algorithms: where information is transmitted (and received) from any local regulator to a given subset of the others.
Whereas, as far as concerned the protocols for exchanging information among local regulators, we examine the following cases:
- Non-iterative Algorithms: where information is transmitted (and received) by the local regulators only once within each sampling time;
- Iterative Algorithms: where information can be transmitted (and received) by the local regulators many times within the sampling time.
Hierarchical multi-layer control structures are suitable for control and monitoring for complex large-scale systems, such as the DHS of InDeal. To comply with the project requirements, CITY has investigated hierarchical multi-layer architectures which are characterized by clearly separable slow and fast dynamics (i.e. working in different time scales, bandwidth). A conceptual scheme of such an architecture for a generic 2-layer system is shown below.
At the second stage of the project, CITY is being progressing to the development of the control algorithms that will constitute the Decision and Support System (DSS) of InDeal.
Here we summarize all the parameters, decision variables and forecasts applied in our mathematical formulation reported below.
Decision Variables to be used are:
Information by forecast units:
Based on the those the following will be considered:
- Boiler Operating Constraints
- Start-up/Shut-down costs
- Thermal Loads
- Physical plant layout Constraints
- Thermal Energy Storage
- Reliability constraints
- Objective (Cost) Function
- Overall Optimal Control Problem