阅读说明：本技术 一种利用music分类的负荷开关事件检测方法和系统 (Load switch event detection method and system by using MUSIC classification ) 是由 翟明岳 于 2019-10-03 设计创作，主要内容包括：本发明的实施例公开一种利用MUSIC分类的负荷开关事件检测方法和系统,所述方法包括：步骤1,输入实测的功率信号序列S；步骤2,根据MUSIC分类原理检测负荷开关事件。具体为：如果第K个归一化窗口分数矢量H<Sub>K</Sub>满足判断条件|H<Sub>K</Sub>|≥e<Sub>0</Sub>,则在所述信号序列S的第K点处,检测到负荷开关事件；否则,未检测到负荷开关事件。其中,e<Sub>0</Sub>为负荷开关事件检测阈值。(The embodiment of the invention discloses a load switch event detection method and a system classified by MUSIC, wherein the method comprises the following steps: step 1, inputting an actually measured power signal sequence S; and 2, detecting a load switch event according to the MUSIC classification principle. The method specifically comprises the following steps: if the Kth normalized window score vector H K Satisfies the judgment condition | H K |≥e 0 Detecting a load switch event at the Kth point of the signal sequence S; otherwise, no load switch event is detected. Wherein e is 0 A threshold is detected for a load switch event.)

1. A method for load switch event detection using MUSIC classification, comprising:

step 1, inputting an actually measured power signal sequence S;

and 2, detecting a load switch event according to the MUSIC classification principle. The method specifically comprises the following steps: if the Kth normalized window score vector H_KSatisfies the judgment condition | H_K|≥e₀Detecting a load switch event at the Kth point of the signal sequence S; otherwise, no load switch event is detected. Wherein e is₀A threshold is detected for a load switch event.

2. The method of claim 1, wherein prior to step 2, the method further comprises:

step 3, calculating the K normalized window fraction vector H_KAnd said load switch event detection threshold e₀。

3. The method of claim 2, wherein step 3 comprises:

step 301, generating the nth signal first order difference sequence

wherein:

S_n: the nth element in the signal sequence S

S＝[S₁，S₂，…，S_N]The length of the signal sequence is N

If the element S_jSubscript j of>N, then S_j＝0。

Step 302, generating the nth signal second order difference sequence

wherein:

If the element S_jSubscript j of>N, then S_j＝0。

Step 303, calculating the n-th signal feature vector E_nThe method specifically comprises the following steps:

wherein:

[E_n]_i: the nth signal feature vector E_n1,2, …, n]

The nth signal second order difference sequence

The nth signal first order difference sequence

Step 304, calculating the K window fraction sequence h_KThe method specifically comprises the following steps:

wherein:

matrix [ E ]_K]^TE_KThe m-th feature vector of (2),

arranged according to the sequence of the characteristic values from big to small

[h_K]_j: the Kth window fraction vector h_KThe jth element of (1)

K＝1,2，...，N

j＝1,2，…，K

Step 305, obtaining the K-th normalized window fraction vector H_KTool for measuringThe body is as follows:

the first step is as follows: the score adjustment specifically comprises the following steps:

wherein:

σ_h: the Kth window fraction vector h_KThe mean square error of (c).

The second step is that: rearranging the K window fraction sequence h in descending order_KObtaining the K descending window fraction sequence

Thirdly, calculating an AU parameter sequence a, specifically:

wherein:

[a]_i: the ith element of the AU parameter sequence a

Fourthly, calculating a weight sequence W, which specifically comprises the following steps:

wherein:

[W]_i: the ith element of the weight sequence W

Fifthly, calculating the Kth normalized window fraction vector H_KThe method specifically comprises the following steps:

[H_K]_i＝[W]_i[h_K]_i

wherein:

[H_K]_i: the Kth normalized window score vector H_KThe ith element of

Step 306, calculating the load switch event detection threshold e₀The method specifically comprises the following steps:

wherein:

The nth signal secondary difference sequence

Sequence of mean values

4. A load switch event detection system using MUSIC classification, comprising:

the acquisition module inputs an actually measured power signal sequence S;

and the detection module detects the load switch event according to the MUSIC classification principle. The method specifically comprises the following steps: if the Kth normalized window score vector H_KSatisfies the judgment condition | H_K|≥e₀Detecting a load switch event at the Kth point of the signal sequence S; otherwise, no load switch event is detected. Wherein e is₀A threshold is detected for a load switch event.

5. The system of claim 4, further comprising:

a calculation module to calculate the Kth normalized window score vector H_KAnd said load switch event detection threshold e₀。

Technical Field

The invention relates to the field of electric power, in particular to a load switch event detection method and system.

Background

With the development of smart grids, the analysis of household electrical loads becomes more and more important. Through the analysis of the power load, a family user can obtain the power consumption information of each electric appliance and a refined list of the power charge in time; the power department can obtain more detailed user power utilization information, can improve the accuracy of power utilization load prediction, and provides a basis for overall planning for the power department. Meanwhile, the power utilization behavior of the user can be obtained by utilizing the power utilization information of each electric appliance, so that the method has guiding significance for the study of household energy consumption evaluation and energy-saving strategies.

The current electric load decomposition is mainly divided into an invasive load decomposition method and a non-invasive load decomposition method. The non-invasive load decomposition method does not need to install monitoring equipment on internal electric equipment of the load, and can obtain the load information of each electric equipment only according to the total information of the electric load. The non-invasive load decomposition method has the characteristics of less investment, convenience in use and the like, so that the method is suitable for decomposing household load electricity.

In the non-invasive load decomposition algorithm, the detection of the switching event of the electrical equipment is the most important link. The initial event detection takes the change value of the active power P as the judgment basis of the event detection, and is convenient and intuitive. This is because the power consumed by any one of the electric devices changes, and the change is reflected in the total power consumed by all the electric devices. Besides the need to set a reasonable threshold for the power variation value, this method also needs to solve the problem of the event detection method in practical application: a large peak (for example, a motor starting current is much larger than a rated current) appears in an instantaneous power value at the starting time of some electric appliances, so that an electric appliance steady-state power change value is inaccurate, and the judgment of a switching event is influenced, and the peak is actually pulse noise; moreover, the transient process of different household appliances is long or short (the duration and the occurrence frequency of impulse noise are different greatly), so that the determination of the power change value becomes difficult; due to the fact that the active power changes suddenly when the quality of the electric energy changes (such as voltage drop), misjudgment is likely to happen. The intensity of (impulse) noise is large and background noise has a large impact on the correct detection of switching events.

Load switching events that are now commonly used are often determined using changes in power data: when the power change value exceeds a preset threshold value, a load switch event is considered to occur. This approach, while simple and easy to implement, results in a significant drop in the accuracy of the switching event detection due to the impulse noise and the common use of non-linear loads.

Therefore, in the switching event detection process, how to improve the switching event detection accuracy is very important. Load switch event detection is the most important step in energy decomposition, and can detect the occurrence of an event and determine the occurrence time of the event. However, the accuracy of the detection of the switching event is greatly affected by noise in the power signal (power sequence), and particularly, impulse noise generally exists in the power signal, which further affects the detection accuracy. Therefore, it is currently a very important task to effectively improve the detection accuracy of the load switch event.

Load switching events that are now commonly used are often determined using changes in power data: when the power change value exceeds a preset threshold value, a load switch event is considered to occur. This approach, while simple and easy to implement, results in a significant drop in the accuracy of the switching event detection due to the impulse noise and the common use of non-linear loads.

Disclosure of Invention

The invention aims to provide a load switch event detection method and system by using MUSIC classification. The method has good switching event detection performance and is very simple in calculation.

In order to achieve the purpose, the invention provides the following scheme:

a method of load switch event detection using MUSIC classification, comprising:

step 1, inputting an actually measured power signal sequence S;

and 2, detecting a load switch event according to the MUSIC classification principle. The method specifically comprises the following steps: if the Kth normalized window score vector H_KSatisfies the judgment condition H_K≥e₀Detecting a load switch event at the Kth point of the signal sequence S; otherwise, no load switch event is detected. Wherein e is₀A threshold is detected for a load switch event.

A load switch event detection system using MUSIC classification, comprising:

the acquisition module inputs an actually measured power signal sequence S;

and the detection module detects the load switch event according to the MUSIC classification principle. The method specifically comprises the following steps: if the Kth normalized window score vector H_KSatisfies the judgment condition H_K≥e₀Detecting a load switch event at the Kth point of the signal sequence S; otherwise, no load switch event is detected. Wherein e is₀A threshold is detected for a load switch event.

According to the specific embodiment provided by the invention, the invention discloses the following technical effects:

although the transformer vibration and sound detection method is widely applied to load switch event monitoring and has a relatively mature technology, the vibration and sound detection method utilizes a vibration signal emitted by a transformer and is easily influenced by environmental noise, so that the method often cannot obtain a satisfactory result when applied in an actual working environment.

The invention aims to provide a load switch event detection method and system by using MUSIC classification. The method has good switching event detection performance and is very simple in calculation.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.

FIG. 1 is a schematic flow diagram of the process of the present invention;

FIG. 2 is a schematic diagram of the system of the present invention;

FIG. 3 is a flow chart illustrating an embodiment of the present invention.

Detailed Description

The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.

FIG. 1 is a schematic flow chart of a method for detecting a load switch event by MUSIC classification

Fig. 1 is a schematic flow chart of a load switch event detection method using MUSIC classification according to the present invention. As shown in fig. 1, the method for detecting a load switch event by using MUSIC classification specifically includes the following steps:

step 1, inputting an actually measured power signal sequence S;

and 2, detecting a load switch event according to the MUSIC classification principle. The method specifically comprises the following steps: if the Kth normalized window score vector H_KSatisfies the judgment condition H_K≥e₀Detecting a load switch event at the Kth point of the signal sequence S; otherwise, no load switch event is detected. Wherein e is₀A threshold is detected for a load switch event.

Before the step 2, the method further comprises:

step 3, calculating the fraction vector H of the Kth normalization window_KAnd said load switch event detection threshold e₀。

The step 3 comprises the following steps:

step 301, generating the nth signal first order difference sequence

The method specifically comprises the following steps:

wherein:

the nth signal first-order difference sequence [ N ═ 1,2]

S_n: the nth element in the signal sequence S

S＝[S₁，S₂，...，S_N]The length of the signal sequence is N

If the element S_jSubscript j of>N, then S_j＝0。

Step 302, generating the nth signal second order difference sequence

The method specifically comprises the following steps:

wherein:

the nth signal second order difference sequence [ N ═ 1,2]

If the element S_jSubscript j of>N, then S_j＝0。

Step 303, calculating the n-th signal feature vector E_nThe method specifically comprises the following steps:

wherein:

[E_n]_i: the nth signal feature vector E_n1,2, n, i]

The nth signal first order difference sequence

The ith element of

The nth signal second order difference sequence

The ith element of

The nth signal first order difference sequence

Maximum value of all elements in

The nth signal second order difference sequence

Maximum value of all elements in

Step 304, calculating the K window fraction sequence h_KThe method specifically comprises the following steps:

wherein:

matrix [ E ]_K]^TE_KThe m-th feature vector of (2),

arranged according to the sequence of the characteristic values from big to small

[h_K]_j: the Kth window fraction vector h_KThe jth element of (1)

K＝1,2，...，N

j＝1,2，...，K

Step 305, obtaining the K-th normalized window fraction vector H_KThe method specifically comprises the following steps:

the first step is as follows: the score adjustment specifically comprises the following steps:

wherein:

σ_h: the Kth window fraction vector h_KThe mean square error of (c).

The second step is that: rearranging the K window fraction sequence h in descending order_KObtaining the K descending window fraction sequence

Thirdly, calculating an AU parameter sequence a, specifically:

wherein:

[a]_i: the ith element of the AU parameter sequence a

The Kth descending window fraction sequence

The ith element of

The Kth descending window fraction sequence

Maximum value of

The Kth descending window fraction sequence

Minimum value of (2)

Fourthly, calculating a weight sequence W, which specifically comprises the following steps:

wherein:

[W]_i: the ith element of the weight sequence W

Fifthly, calculating the Kth normalized window fraction vector H_KThe method specifically comprises the following steps:

[H_K]_i＝[W]_i[h_K]_i

wherein:

[H_K]_i: the Kth normalized window score vector H_KThe ith element of

Step 306, calculating the load switch event detection threshold e₀The method specifically comprises the following steps:

wherein:

the nth signal primary differential sequence

Mean value of

The nth signal secondary difference sequence

Mean value of

Sequence of mean values

N is 1,2, mean value of N

Sequence of mean values

N is 1,2, mean value of N

Sequence of mean values

N1, 2, mean square error of N

Sequence of mean values

N1, 2, mean square error of N

Sequence of mean values

N is the maximum value of 1,2

Sequence of mean values

N is the maximum value of 1, 2.

FIG. 2 structural intent of a load switch event detection system using MUSIC classification

Fig. 2 is a schematic structural diagram of a load switch event detection system using MUSIC classification according to the present invention. As shown in fig. 2, the load switch event detection system using MUSIC classification includes the following structures:

the acquisition module 401 inputs an actually measured power signal sequence S;

the detection module 402 detects a load switch event according to the MUSIC classification principle. The method specifically comprises the following steps: if the Kth normalized window score vector H_KSatisfies the judgment condition | H_K|≥e₀Detecting a load switch event at the Kth point of the signal sequence S; otherwise, no load switch event is detected. Wherein e is₀A threshold is detected for a load switch event.

The system further comprises:

a calculation module 403 for calculating the Kth normalized window score vector H_KAnd said load switch event detection threshold e₀。

The following provides an embodiment for further illustrating the invention

FIG. 3 is a flow chart illustrating an embodiment of the present invention. As shown in fig. 3, the method specifically includes the following steps:

1. inputting a sequence of measured power signals

S＝[s₁,s₂,...,s_N-1,s_N]

Wherein:

s: a data sequence of measured power signal of length N

s_iN is the measured power signal with serial number i

2. Generating a first order difference sequence of signals

Wherein:

the nth signal first-order difference sequence [ N ═ 1,2]

S_n: the nth element in the signal sequence S

S＝[S₁，S₂，...，S_N]The length of the signal sequence is N

If the element S_jSubscript j of>N, then S_j＝0。

3. Generating a second order difference sequence of signals

Wherein:

the nth signal second order difference sequence [ N ═ 1,2]

If the element S_jSubscript j of>N, then S_j＝0。

4. Determining a signal feature vector

Wherein:

[E_n]_i: the nth signal feature vector E_n1,2, n, i]

The nth signal first order difference sequenceThe ith element of

The nth signal second order difference sequence

The ith element of

The nth signal first order difference sequence

Maximum value of all elements in

The nth signal second order difference sequence

Maximum value of all elements in

5. Finding a sequence of window scores

Wherein:

matrix [ E ]_K]^TE_KThe m-th feature vector of (2),

arranged according to the sequence of the characteristic values from big to small

[h_K]_j: the Kth window fraction vector h_KThe jth element of (1)

K＝1,2，...，N

j＝1,2，...，K

6. Calculating a normalized window score vector

The first step is as follows: the score adjustment specifically comprises the following steps:

wherein:

σ_h: the Kth window fraction vector h_KThe mean square error of (c).

The second step is that: rearranging the K window fraction sequence h in descending order_KObtaining the K descending window fraction sequence

Thirdly, calculating an AU parameter sequence a, specifically:

wherein:

[a]_i: the ith element of the AU parameter sequence a

The Kth descending window fraction sequence

The ith element of

The Kth descending window fraction sequence

Maximum value of

The Kth descending window fraction sequence

Minimum value of (2)

Fourthly, calculating a weight sequence W, which specifically comprises the following steps:

wherein:

[W]_i: the ith element of the weight sequence W

Fifthly, calculating the Kth normalized window fraction vector H_KThe method specifically comprises the following steps:

[H_K]_i＝[W]_i[h_K]_i

wherein:

[H_K]_i: the Kth normalized window score vector H_KThe ith element of

7. Determining load switch event detection threshold

Wherein:

the nth signal primary differential sequence

Mean value of

The nth signal secondary difference sequence

Mean value of

Sequence of mean values

N is 1,2, mean value of N

Sequence of mean values

N is 1,2, mean value of N

Sequence of mean valuesN1, 2, mean square error of N

Sequence of mean values

N1, 2, mean square error of N

Sequence of mean valuesN is the maximum value of 1,2

Sequence of mean values

N is the maximum value of 1, 2.

8. Detecting a switching event

Load switch events are detected according to the MUSIC classification principle. The method specifically comprises the following steps: if the Kth normalized window score vector H_KSatisfies the judgment condition | H_K|≥e₀Detecting a load switch event at the Kth point of the signal sequence S; otherwise, no load switch event is detected. Wherein e is₀A threshold is detected for a load switch event.

The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is simple because the system corresponds to the method disclosed by the embodiment, and the relevant part can be referred to the method part for description.

The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.