Method and system for fast search of cascading failures in hybrid AC/DC power systems

Patent metadata

Field	Value
Publication number	US-11333698-B2
Application number	US-201816640786-A
Country	US
Kind code	B2
Filing date	Oct 23, 2018
Priority date	Aug 30, 2018
Publication date	May 17, 2022
Grant date	May 17, 2022

Abstract

A fast search method and system for cascading failures in hybrid AC/DC power systems. The fast search method for cascading failures in hybrid AC/DC power systems includes: Determining the initial failures and forming a failure set of the first-stage failure, setting the DC blocking or the failure level reaching the threshold as the search stop condition; Judge the cascading failures in the failure set of the current stage can trigger DC blocking, if a cascading failure can trigger DC blocking, store the DC blocking failure as the next-stage failure. Otherwise, cascading failures are searched on the AC side and the failure set of the next stage is determined based on the pruning search method and the outage risk value; If the search stop condition is satisfied, stop the search, otherwise, the number of failure stages is increased by one and return to judge the cascading failures to continue the search.

First claim

The invention claimed is:

1. A fast search method for cascading failures in hybrid AC/DC power systems, the method comprising steps of:
step 1: determining, using a processor, an initial failure and forming a failure set of first-stage failures; setting a DC blocking or a counted number of searched failure stages reaching a threshold as a search stop condition;
step 2: judging, using the processor, whether cascading failures in the current stage failure set can trigger a DC blocking; if the cascading failures cause the DC blocking, the DC blocking is stored, in a memory, in a next stage of the cascading failures; otherwise, using an outage risk value as a pruning basis to search the cascading failures on an AC side to determine a failure set of next stage failures; the search of the cascading failures on the AC side to determine the failure set of the next stage failures including:
calculating, using the processor, the outage risk value of a line without failure of each failure chain in turn;
the outage risk value of the failure chain is:
R i m =p i m ×I i m

where p i m a failure probability of a line i, and I i m is a consequence caused by a failure of the line i; and
taking, using the processor, the outage risk values as the pruning basis, sequencing corresponding failure chains by the calculated outage risk values, storing, using the processor, a first preset number of the failure chains in the memory based on the outage risk values, and the failure set of the next stage failures is searched on the AC side; and

step 3: if the search stop condition is satisfied, stopping, using the processor, the search; if the search stop condition is not satisfied, increasing, using the processor, the counted number of searched failure stages by one and returning to Step 2 to continue the search;
step 4: outputting, using the processor, the failure chains with the highest outage risk values based on the search results.

2. The method according to claim 1 , wherein, in the Step 1, according to an actual situation of a hybrid AC/DC power system, AC lines near a vicinity of a DC terminal location are selected as initial failures, and the initial failures are simulated in time-domain to check whether the initial failures will trigger the DC blocking.

3. The method according to claim 1 , wherein,
in the Step 2, when judging whether the cascading failures of the current stage failure set can trigger the DC blocking, extreme failures are identified by comparing a multi-infeed short circuit ratio (MISCR) with a preset threshold of MISCR; then, the extreme failures are judged whether the extreme failures are causing the DC blocking by time-domain simulations, and
the MISCR is:

M
i

=

S
aci

P
deqi

=

1



Z
eqii



⁢

P
di

+

∑

j
=
1

,

j
≠
1

n

⁢

⁢



Z
eqij



⁢

P
dj

where S aci , is a short-circuit capacity of a commutation bus i; P deqi is DC power of an i-th DC system, MW; Z eqii is a self-impedance of the commutation bus i; Z eqij is a mutual impedance of the commutation bus i and a communication bus j; P di is a power of DC system i; P dj is a power of DC system j.

4. The method according to claim 1 , wherein, in the Step 2, when judging whether the cascading failures of the current stage failure set can trigger the DC blocking, a Levenberg-Marquarelt Neural Network (LMNN) is used to judge whether non-extreme failures of the current stage failure set can trigger the DC blocking.

5. The method according to claim 4 , wherein, for the non-extreme failures, the using of the LMNN to judge whether the current stage failure set can trigger the DC blocking includes:
partitioning the AC/DC power system to be analyzed; a partition principle is that each DC system terminal location is as one area, cascading failure chains of different stages are randomly selected near a vicinity of the DC terminal location and as a training sample set;
for the training sample set, removing extreme cascading failure chains, and using the remaining non-extreme cascading failure chains for training of the LMNN;
utilizing the trained LMNN to quickly estimate a voltage dip degree of a commutation bus after failures and measure a severity of each of a plurality of AC failures; and
for the cascading failures of the current stage failure set having a severity that exceeds a preset threshold, using a time-domain simulation to verify whether the cascading failures of the current stage failure set having the severity that exceeds the preset threshold can cause the DC blocking.

6. The method according to claim 5 , wherein, in the process of using the LMNN to quickly estimate whether the non-extreme failures of the current stage can trigger the DC blocking, failure chains are sorted according to the severity of each of the AC failures, and the most severe failure of the AC failures is simulated by the time-domain simulation; if the simulated failure will trigger the DC blocking, a next failure is simulated continually; otherwise, if the simulated failure will not trigger the DC blocking, not simulating the remaining failures.

7. The method according to claim 1 , wherein
in the process of calculating the remaining line outage risk for each failure chain, the calculation is based on two aspects: the line outage probability and the line outage consequence,
the line outage probability of the line i is:
p i m =1−(1− p i hf )(1 −p i ol ),

where p i hf is a hidden failure probability of the line i after a m-stage failure, p i ol is a failure probability of the line i after failures, and
the consequence of the line i failing is:

I
i
m

=

Assignees

• Univ Shandong

Inventors

• Liu Yutian
• Zhu Yuanzhen
• Li Wenbo

Classifications

• H02J13/12

  Monitoring network conditions, e.g. electrical magnitudes or operational status · CPC title

• G01R31/086 Primary

  in power transmission or distribution networks, i.e. with interconnected conductors · CPC title

• H02J4/00

  Circuit arrangements for mains or distribution networks not specified as AC or DC; Circuit arrangements for mains or distribution networks combining AC and DC sections or sub-networks (arrangements using intermediate DC-AC-DC conversion H02J1/002; arrangements using high-voltage DC [HVDC] links H02J3/36) · CPC title

• Y04S10/52

  Outage or fault management, e.g. fault detection or location · CPC title

• H02J3/00 Primary

  Circuit arrangements for AC mains or AC distribution networks · CPC title