Support tensor machine based neutral point grounding mode decision method and system

• US 10,380,223 B2
• Filed: 03/05/2018
• Issued: 08/13/2019
• Est. Priority Date: 10/23/2017
• Status: Active Grant

• Current Assignee: State Grid Ningbo Electric Power Supply Company Limited, State Grid Zhejiang Economic Research Institute, State Grid Zhejiang Electric Power Company Limited, Zhejiang Huayun Electric Power Engineering Design Consulting Co., Ltd.
• Original Assignee: State Grid Zhejiang Electric Power Company Limited
• Inventors: He, Yingjing, Dan, Yangqing, Liu, Weijun, Zheng, Weimin, Zhang, Xiaodi, Zheng, Chaoming, Wang, Xiran, Shen, Shuyi, Yao, Yan, Zhu, Yanwei, Li, Fan, Zhou, Lin, Fang, Jiandi, Yu, Dan, Tang, Ren
• Primary Examiner(s): Mai, Tan V
• Application Number:
  US15/911,887
• Publication Number:
  US 20190121836A1
• Time in Patent Office: 526 Days
• Field of Search: 708520
• CPC Class Codes: G06F 17/16 Matrix or vector computation , e.g. matrix-matrix or matrix-vector multiplication, matrix factorization matrix transposition G06F7/78
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06Q 50/06 Energy or water supply
• CPC Class Codes: Y02E 40/70 Smart grids as climate change mitigation technology in the energy generation sector
• CPC Class Codes: Y04S 10/50 Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Abstract

A support tensor machine based neutral point grounding mode decision method and system adopts a support tensor machine method Based on three indexes, i.e., the power supply reliability index, safety index, and economical efficiency index, influences of different neutral point grounding modes are analyzed by employing the support tensor machine method to finally obtain a neutral point grounding mode capable of maximizing power supply reliability of a distribution network.

Claims

1. A support tensor machine based neutral point grounding mode decision method, comprising:

Step A, constructing a power supply reliability evaluation system according to a topological structure of a medium voltage distribution network and calculating a related index;

Step B, calculating a safety-related index of the distribution network according to related factors that affect the safety;

Step C, constructing an economic evaluation index system of the distribution network and calculating a related index;

Step D, constructing the power supply reliability index, the safety index of the distribution network and the economic evaluation index of the distribution network into a tensor form and decomposing the tensor form; and

Step E, constructing a support tensor machine kernel function, determining an optimization function and obtaining an optimal neutral point grounding mode via analysis.

2. The neutral point grounding mode decision method according to claim 1, wherein Step A specifically comprises the following steps:

A1, constructing a topology G=(V, E) of a network structure according to the structure of the distribution network, wherein V represents a set of all nodes, and E represents a set of edges connecting all the nodes;

A2, calculating a weight value r_ij of a branch, r_ij=α

_md_ij, i, j=1,2,3 . . . ,n;

m=1,2, . . . ,k;

wherein d_ij is a distance between nodes, and α

_m represents weighting coefficients in different scenarios;

A3, calculating a node current fluctuation value,

3. The neutral point grounding mode decision method according to claim 2, wherein in Step A2, according to an environment in which the line is located, the weighting coefficient has a value range of 0<

α

_m≤

1.

4. The neutral point grounding mode decision method according to claim 3, wherein Step B specifically comprises the following steps:

B1, performing statistics of related parameters;

a large-area failure probability τ

of a distribution network communication system, a failure probability M of a mobile cellular network, a failure probability W of a wireless sensor network, a failure probability F of optical fiber communication, and a defeat trip probability J of an automation system; and

B2, calculating a safety index of a distribution network as S=1−

τ

·

M·

W·

F·

J.

5. The neutral point grounding mode decision method according to claim 3, wherein Step C specifically comprises the following steps:

C1, calculating an equipment configuration cost,

6. The neutral point grounding mode decision method according to claim 5, wherein when power outage takes place in a period of power consumption peak, the value of μ

is 2, or otherwise, the value of μ

is 1.

7. The neutral point grounding mode decision method according to claim 3, wherein Step D specifically comprises the following steps:

D1, according to the above three types of indexes, constructing a tensor data form with its tensor machine represented in a form of p;

U×

V×

C₁×

C₂×

L×

C_n→

D;

wherein data is defined as P∈

p<sup>n×

m×

k</sup>, and U, V and C represent k indexes for economical efficiency, m indexes for safety of a distribution network, and n indexes for power supply reliability of the distribution network, respectively; and

D2, according to the nature of the decomposition uniqueness of a three-dimensional matrix, decomposing a data matrix P∈

P<sup>n×

m×

k</sup> into {circumflex over (P)}=G^(m)×

₁U⁽¹⁾×

₂V⁽²⁾×

₃{circumflex over (P)}_i3⁽³⁾, wherein {circumflex over (P)} represents a decomposition matrix of data, G^(m), U⁽¹⁾, V⁽²⁾ and {circumflex over (P)}_i3⁽³⁾ represent a neutral point grounding mode after decomposition, k indexes for the economical efficiency, m indexes for the safety of the distribution network, and n indexes for the power supply reliability of the distribution network, respectively.

8. The neutral point grounding mode decision method according to claim 2, wherein Step B specifically comprises the following steps:

B1, performing statistics of related parameters;

a large-area failure probability τ

of a distribution network communication system, a failure probability M of a mobile cellular network, a failure probability W of a wireless sensor network, a failure probability F of optical fiber communication, and a defeat trip probability J of an automation system; and

B2, calculating a safety index of a distribution network as S=1−

τ

·

M·

W·

F·

J.

9. The neutral point grounding mode decision method according to claim 2, wherein Step C specifically comprises the following steps:

C1 , calculating an equipment configuration cost,

10. The neutral point grounding mode decision method according to claim 9, wherein when power outage takes place in a period of power consumption peak, the value of μ

is 2, or otherwise, the value of μ

is 1.

11. The neutral point grounding mode decision method according to claim 2, wherein Step D specifically comprises the following steps:

D1, according to the above three types of indexes, constructing a tensor data form with its tensor machine represented in a form of p;

U×

V×

C₁×

C₂×

L×

C_n→

D;

wherein data is defined as P∈

p<sup>n×

m×

k</sup>, and U, V and C represent k indexes for economical efficiency, m indexes for safety of a distribution network, and n indexes for power supply reliability of the distribution network, respectively; and

D2, according to the nature of the decomposition uniqueness of a three-dimensional matrix, decomposing a data matrix P∈

p<sup>n×

m×

k</sup> into {circumflex over (P)}=G^(m)×

₁U⁽¹⁾×

₂V⁽²⁾×

₃{circumflex over (P)}_i3⁽³⁾, wherein {circumflex over (P)} represents a decomposition matrix of data, G^(m), U⁽¹⁾, V⁽²⁾ and {circumflex over (P)}_i3⁽³⁾ represent a neutral point grounding mode after decomposition, k indexes for the economical efficiency, m indexes for the safety of the distribution network, and n indexes for the power supply reliability of the distribution network, respectively.

12. The neutral point grounding mode decision method according to claim 3, wherein Step E specifically comprises the following steps:

E1, defining the distance between an estimation matrix and an observation matrix,

13. The neutral point grounding mode decision method according to claim 1, wherein Step B specifically comprises the following steps:

B1, performing statistics of related parameters;

a large-area failure probability τ

of a distribution network communication system, a failure probability M of a mobile cellular network, a failure probability W of a wireless sensor network, a failure probability F of optical fiber communication, and a defeat trip probability J of an automation system; and

B2, calculating a safety index of a distribution network as S=1−

τ

·

M·

W·

F·

J.

14. The neutral point grounding mode decision method according to claim 1, wherein Step C specifically comprises the following steps:

C1, calculating an equipment configuration cost,

15. The neutral point grounding mode decision method according to claim 14, wherein when power outage takes place in a period of power consumption peak, the value of μ

is 2, or otherwise, the value of μ

is 1.

16. The neutral point grounding mode decision method according to claim 1, wherein Step D specifically comprises the following steps:

D1, according to the above three types of indexes, constructing a tensor data form with its tensor machine represented in a form of p;

U×

V×

C₁×

C₂×

L×

C_n→

D;

wherein data is defined as P∈

p<sup>n×

m×

k</sup>, and U, V and C represent k indexes for economical efficiency, m indexes for safety of a distribution network, and n indexes for power supply reliability of the distribution network, respectively; and

D2, according to the nature of the decomposition uniqueness of a three-dimensional matrix, decomposing a data matrix P∈

p<sup>n×

m×

k</sup> into {circumflex over (P)}=G^(m)×

₁U⁽¹⁾×

₂V⁽²⁾×

₃{circumflex over (P)}_i3⁽³⁾, wherein {circumflex over (P)} represents a decomposition matrix of data, G^(m), U⁽¹⁾, V⁽²⁾ and {circumflex over (P)}_i3⁽³⁾ represent a neutral point grounding mode after decomposition, k indexes for the economical efficiency, m indexes for the safety of the distribution network, and n indexes for the power supply reliability of the distribution network, respectively.

17. The neutral point grounding mode decision method according to claim 1, wherein Step E specifically comprises the following steps:

E1, defining the distance between an estimation matrix and an observation matrix,

18. The neutral point grounding mode decision method according to claim 1, wherein Step E specifically comprises the following steps:

E1, defining the distance between an estimation matrix and an observation matrix,

19. A support tensor machine based neutral point grounding mode decision system, comprising:

a power supply reliability related index calculation unit to construct a power supply reliability evaluation system according to a topological structure of a medium voltage distribution network and calculate a related index;

a safety-related index calculation unit to calculate a safety-related index of the distribution network according to related factors that affect the safety;

an economical efficiency-related index calculation unit to construct an economic evaluation index system of the distribution network and calculate a related index;

a multi-index tensor data constructing and decomposing unit to construct the power supply reliability index, the safety index of the distribution network and the economic evaluation index of the distribution network into a tensor form and decompose the tensor form; and

an optimal neutral point grounding mode determining unit to construct a support tensor machine kernel function, determine an optimization function and obtain an optimal neutral point grounding mode via analysis.

20. The neutral point grounding mode decision system according to claim 19, wherein the multi-index tensor data constructing and decomposing unit comprises:

A distance calculation module for an estimation matrix and an observation matrix to calculate the distance between the estimation matrix and the observation matrix;

a prior probability calculation module to calculate a prior probability;

a joint probability calculation module to calculate a joint probability; and

a decomposed matrix gradient calculation module to calculate a gradient of a decomposed matrix.