Unlock instant, AI-driven research and patent intelligence for your innovation.

A system and method for judging abnormal power consumption based on reinforcement learning

A technology for abnormal power consumption and judgment, which is applied in the field of abnormal power consumption judgment system based on reinforcement learning, can solve the problems such as the difficulty of determining the judgment threshold and ratio, and achieve the effect of improving generalization ability and high flexibility

Active Publication Date: 2020-11-20
武汉格蓝若智能技术股份有限公司
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0014] Aiming at the technical problems existing in the prior art, the present invention provides a judgment method for abnormal power consumption based on reinforcement learning, which solves the problem in the prior art that the judgment threshold and ratio of the traditional classifier method are difficult to determine

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A system and method for judging abnormal power consumption based on reinforcement learning
  • A system and method for judging abnormal power consumption based on reinforcement learning
  • A system and method for judging abnormal power consumption based on reinforcement learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0067] Embodiment 1 provided by the present invention is an embodiment of an abnormal power consumption judgment system based on reinforcement learning provided by the present invention, such as figure 1 Shown is an interactive schematic diagram of an abnormal power consumption judgment system based on reinforcement learning provided by an embodiment of the present invention, which is composed of figure 1 It can be seen that, in an embodiment of an abnormal power consumption judgment system based on reinforcement learning provided by the present invention, the judgment system is a constructed DRQN ​​model, including: a memory bank, a Q network model and a target Q network model.

[0068] The memory bank is used to store the current state, the currently selected action, the state of the next step, and the reward and punishment value of the current round.

[0069] Specifically, the memory bank stores quadruples , Indicates the current state, Indicates the currently selecte...

Embodiment 2

[0091] Embodiment 2 provided by the present invention is an embodiment of an abnormal electricity consumption judgment method based on reinforcement learning provided by the present invention. The abnormal electricity consumption judgment method is based on an abnormal electricity consumption judgment system provided by the embodiment of the present invention. The judgment Methods include:

[0092] Step 1. Obtain the user abnormal power consumption probability sequence output by the classifier and the sample data of the corresponding original label, and divide the sample data into a training set and a test set.

[0093] Step 2, use the training set to iterate the DRQN ​​module, and complete the training of the Q network model during the iteration process of the DRQN ​​module.

[0094] The iterative process of the DRQN ​​module includes: according to the input n power consumption probability sequences, the reward and punishment value of the current round is determined dynamical...

Embodiment 3

[0119] Embodiment 3 provided by the present invention is a specific application embodiment of an abnormal power consumption judgment system based on reinforcement learning provided by the present invention, such as Figure 5 Shown is the flow chart of the abnormal user detection method provided by the embodiment of the present invention, consisting of Figure 5 It can be seen that in this specific application example, the user data sampled from the power companies of Province G and Province J are used to train the DRQN ​​model. This data set contains more than 300 users' electricity consumption data, and a single user has recorded up to 311 days electricity consumption records. The sampling frequency of this data set is 0.5h / time, and a single user has 48 electricity consumption records a day. We first downsample the electricity consumption data of these users to 1h / time, and then uniformly cut the user data to 300 days of electricity consumption records.

[0120] Establish ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The present invention relates to a system and method for judging abnormal power consumption based on reinforcement learning. The judgment system is a DRQN ​​(Deep Recurrent Q Network, deep recurrent Q network model) model for judging abnormal power consumption. The Q network model is based on the current state and the currently selected action as the input and output, and the state as the judgment index to determine the reward and punishment value of the current round; when the Q network model training reaches the set number of times, the network parameters of the target Q network model are synchronized to the network parameters of the Q network model; Input the power consumption probability sequence to be tested into the trained DRQN ​​model, use the state as the dynamic threshold of the power consumption probability sequence to be tested, judge whether the power consumption is abnormal according to the dynamic threshold, and use the current state as the judgment index to determine the reward and punishment value , and use the current state as a dynamic threshold, so the system can update the threshold according to real-time user power data, thereby effectively improving the generalization ability across user scenarios.

Description

technical field [0001] The invention relates to the field of power system load forecasting, in particular to a system and method for judging abnormal power consumption based on reinforcement learning. Background technique [0002] With the widespread application of smart meters, abnormal power consumption detection has become an important means to study customers' abnormal consumption behavior and discover unexpected power consumption events in time. In the operation of the power grid, whether it is the failure of the metering device or the theft of electricity by the user, it will not be possible to collect the real electricity consumption data of the user. These false electricity consumption data are called abnormal electricity consumption data. If these situations cannot be discovered and dealt with in time, it will seriously interfere and affect the normal electricity consumption of users and the power supply order of the power supply company, which will not only bring e...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62G06Q50/06
CPCG06Q50/06G06F18/217G06F18/214
Inventor 陈应林陈勉舟
Owner 武汉格蓝若智能技术股份有限公司