This paper proposes a new methodology to optimize the deployment and management of the power grid smart monitoring infrastructure in order to gradually achieve better representations of the micro-generation (μg) activity as the volume of measurement increases.

A combination of simulated measurements and predictions with an optimization scheme is analyzed in order to test the efficiency of a population of potential monitoring configurations at every measure step. These solutions are formulated as a mapping between a set of available devices and a measurement mesh. Furthermore, smart meters sampling rates are included into the formulation in order to allow their dynamical adjustment. This way it is possible to prioritize the measurement of those locations where μg activity is higher.

The resulting solutions maximize measurement quality while minimize costs. In this regard, measurement quality accounts for a balance between the matching of proposed solutions with model predictions and sample diversity. Costs are optimized for both devices and their associated data management tariffs. This way the investment for devices and computing resources for both data storage and processing can be optimized while the deployment process evolves.

In order to highlight the proposed framework capabilities, a simulation through an Agent Based Modeling (AMB) scheme is implemented. In this case, a minimalistic Markovian scheme is used as μg forecasting model and μg activity is simulated by a probability distribution.