Demand Side Management in Smart Grid using Particle Swarm Optimization

Problems we are addressing

High peak load demand
High distribution cost at peak load
Load scheduling with consumers’ consent

Solving these problems can be beneficial for both Distribution Companies(DISCOMs) and customers

How Demand Side Management Solves Problem

Utilizing electricity tariffs, monetary incentives, and government regulations to :-

Mitigate the peak load demand
Reduce overall distribution cost
Maximize the use of renewable energy and battery storage

Proposed Algorithm for Demand Side Management(DSM) using Particle Swarm Optimization(PSO)

It is a population based stochastic optimization technique
Each particle contains records of their best fitness value, P_best, and the best fitness value of the entire swarm, G_best
At each iteration we calculate fitness value of a bird and then compare it with Global P_best and G_best to update them if required.
The swarm will move towards best solutions by minimizing objective function

Algorithm

Implementation of PSO:

Inputs:

Day ahead load price

Forecasted Load data

Optimisation Function : We take it to be inverse of Day a head load price

PSO Parameters : velocity_clamping_factor, individual learning rate, social parameter, etc.
User load preference data:

Outputs:

Results :

For Residential load data

For Commercial load data

For Industrial load data

Analysis :

Above results clearly show the reduction in peak load demand, total cost incurred
The proﬁt obtained in industrial load DSM was more than that of commercial and residential load. This is due to large ratings devices used in industries. So, shift of even a single controllable device from peak, can result in high load getting shifted and more eﬀective and percent cost reduction is maximum in Industrial Loads
When the data of end time was included along with start time. Hence, increasing the constraints on scheduling time. Due to these restrictions and enforcement’s on device to be scheduled, some sharp peaks were observed in the load proﬁle after DSM. Less cost reduction was observed in this case