A Research On Intelligent Decision Making Non-Intrusive Load Monitoring

Non-intrusive load monitoring method uses aggregated data from a single signal source to estimate the usage and working state of specific loads.The results of non-intrusive load monitoring can enhance the awareness of energy consumption details and help save energy.Current approaches of non-intrusive load monitoring are not suitable to identify the simultaneous operation of multiple appliances and lack the generalizing ability to tackle the problem in different situations.To solve the problems above,a non-intrusive load monitoring method based on intelligent decision support system was proposed.The assistant decision ability and support decision ability were enhanced by combining the quantitative analysis of decision support system and qualitative analysis of artificial intelligence.A hierarchical decision tree model with pruning method was used to improve the accuracy and efficiency of load disaggregation.Cepstrum coefficients were added to the database of load signatures to improve the ability of identifying multiple appliance simultaneous operation.Deep neural network and its several variations were applied with external information to build a complex mapping between the aggregated signals and the performance of a specific application.User interface was added to the assembly component of decision support system to make it more flexible and scalableCurrent waveform data of real applications were used to estimate the proposed intelligent decision support system of non-intrusive load monitoring.Several decision trees and deep neural networks were compared to optimize the structure of intelligent decision model.Assembly component was applied to identify and log the features of new applications.The results indicated that the proposed approach can identify simultaneous operation of multiple appliances and gain more scalability.