Non-invasive load monitoring method based on combined hidden Markov model and approximate reasoning method

Abstract

The invention discloses a non-invasive load monitoring method based on a combined hidden Markov model and an approximate reasoning method, and the method comprises the steps: enabling a model of the method to be the combination of two types of extension factor hidden Markov models, and considering the characteristics of a total observation value and the sharp change of power in a short time; adding and introducing a noise hybrid model to simulate signal interference of an actual environment, converting two model optimization problems into a hybrid approximate reasoning problem after integer constraints and necessary constraints are added, and decomposing a total load curve to obtain working curves of all individual loads. According to the non-intrusive monitoring method adopted by the invention, the household load can be modeled and decomposed only by inputting the active power, the obtained result better conforms to the real power utilization situation, portability and privacy protection are achieved, the method can be quickly arranged on a system-on-chip on a household bus to work. And the load information does not need to be uploaded to a data processing center of the cloud, sothat privacy information such as key load use conditions and load operation conditions of users is protected.

Description

technical field [0001] The invention belongs to the field of information processing, in particular, relates to a non-invasive load monitoring method based on combined hidden Markov model and approximate reasoning method [0002] technical background [0003] The purpose of non-intrusive electrical load monitoring is to develop a tool that minimizes the impact on the monitoring object. It can provide households with electrical energy consumption and more specific information, and provide industrial and domestic electricity consumption behavior patterns and related data for the electricity market. With the deepening of experimental research, it has developed into today's non-invasive load monitoring system. [0004] But the cost of implementing computationally complex NILM algorithms may outweigh the benefits in terms of savings in consumption. For example, a simple NILM algorithm that detects appliances in use can save a lot of money if the operation of large appliances can b...

AI Technical Summary

Problems solved by technology

[0007] In view of the above deficiencies, the present invention provides a non-intrusive load monitoring method based on combined hidden Markov model and approximate reasoning method, adopts the combined structure of additive HMMs and subtractive HMMS, adds noisy mixed items and classical constraints, and transforms the optimization problem into For a pooling problem, a method combining hidden Markov models and approximate inference methods is used to obtain independent load power curves, which is computationally efficient, has no local optimum problems, and performs better than existing methods in practice

Images