Monitoring and Prediction of Household Power Consumption using Internet of Things and ARIMA

Electricity bills are significant household expenses. The cost of electricity increases as consumption increases. Customers in the Philippines receive a monthly reading from the conventional electric meters provided by the electricity supplier. Consumers cannot observe or monitor the electricity consumption of each appliance. Although numerous power monitoring systems have been developed nowadays, most do not have features such as maximum bill limit, standby power detection, or daily electricity bill forecasting. The paper introduces an energy management system for households that optimizes power consumption by monitoring and limiting usage. It sends notifications and allows for individual appliance control. It utilizes the AutoRegressive Integrated Moving Average (ARIMA) model for forecasting consumption and electricity bills, notifying owners when thresholds are exceeded, and sending bills to the power provider. The experiment results show the average hourly power consumption recorded by the system and referenced meter is 0.10151 kWh and 0.112917 kWh, respectively. This is with a positive correlation of 0.8459. With a strong correlation of 0.9418, the calculated mean percentage error for daily power consumption is 5.96%. ARIMA model with parameters (1, 0, 1) with an average MAPE of 14%, and the model with parameters (3, 0, 0) displayed a lower average MAPE of 8.68%. The delay in sending an SMS ranges from 5.87 seconds to 14.45 seconds, while the delay in sending an email ranges from 5.62 seconds to 14.82 seconds.