CN-122020586-A - Risk identification rule generation method, apparatus, device, medium and program product

CN122020586ACN 122020586 ACN122020586 ACN 122020586ACN-122020586-A

Abstract

The application discloses a risk identification rule generation method, device, equipment, medium and program product, relates to the technical field of oil and gas pipelines, and is used for efficiently and comprehensively identifying oil and gas pipeline faults and effectively avoiding oil and gas pipeline disaster accidents. The method comprises the steps of obtaining a plurality of risk event chains and a plurality of non-occurrence suppression events in the operation process of an oil-gas pipe network, wherein the risk event chains are used for representing the plurality of risk events arranged according to time sequence, the non-occurrence suppression events are used for representing the non-occurrence suppression events with associated risk events, constructing an initial frequent pattern tree based on the plurality of risk event chains, inserting each non-occurrence suppression event as an independent branch into the initial frequent pattern tree to obtain a target frequent pattern tree, and generating a key risk identification rule and a silencing risk identification rule based on the target frequent pattern tree.

Inventors

CHENG KEXIN
CHEN PENGCHAO
YANG YUFENG
LI RUI
SHI JIANCHENG
ZHANG QIANG
ZHANG XIXIANG
LI SHOUBAO

Assignees

国家石油天然气管网集团有限公司
国家石油天然气管网集团有限公司科学技术研究总院分公司

Dates

Publication Date: 20260512
Application Date: 20260121

Claims (11)

1. A risk identification rule generation method, comprising: acquiring a plurality of risk event chains and a plurality of non-occurrence inhibition events in the operation process of an oil-gas pipe network, wherein the risk event chains are used for representing a plurality of risk events arranged according to time sequence; Constructing an initial frequent pattern tree based on the multiple risk event chains, and inserting each non-occurrence suppression event into the initial frequent pattern tree as an independent branch to obtain a target frequent pattern tree; Generating a key risk identification rule and a silent risk identification rule based on the target frequent pattern tree, wherein the key risk identification rule is used for identifying abnormal events associated with risk events in the oil-gas pipe network, and the silent risk identification rule is used for identifying inhibition events associated with the risk events in the oil-gas pipe network.
2. The risk recognition rule generation method of claim 1, wherein the constructing an initial frequent pattern tree based on the plurality of risk event chains, and inserting each non-occurrence suppression event as an independent branch into the initial frequent pattern tree, before obtaining a target frequent pattern tree, further comprises: determining an occurrence time interval between a plurality of risk events in the risk event chain; And combining the risk events with the occurrence time interval smaller than or equal to the preset time interval into the same event to obtain a compressed risk event chain.
3. The risk identification rule generation method of claim 1, wherein the constructing an initial frequent pattern tree based on the plurality of risk event chains comprises: dividing a plurality of item sets based on the plurality of risk event chains, and setting the consequence severity weight of each risk event in each item set according to the disaster consequence severity level of each risk event in each item set; determining the weighted support degree of each item set according to the result severity weight of each risk event in each item set, wherein the result severity weight of each risk event and the weighted support degree meet a first formula, and the first formula is as follows: ; Wherein, the Is the item set; Count (I) is the number of times item set I appears in the plurality of item sets, N is the total number of the plurality of item sets; Weighting the consequence severity of the ith risk event in item set I; determining a term set with weighted support degree greater than or equal to a support degree threshold as a candidate term set; the initial frequent pattern tree is constructed based on a plurality of candidate sets.
4. A risk identification rule generation method according to claim 3, wherein the method further comprises: Setting causal position weights of all risk events in each candidate set according to event attributes of all risk events in each candidate set, and generating ordering weights of all risk events in each candidate set, wherein the causal position weights of the risk events and the ordering weights of the risk events meet a second formula, and the second formula is as follows: ; Wherein, the Ranking weights for the ith risk event in candidate set X; weighted support degree for candidate item set X; for weighting the weight coefficient of the support, For causal location weighting of the ith risk event in item set X, Is a weight coefficient of the causal position weight.
5. The risk identification rule generation method of claim 4, wherein the generating key risk identification rules and silent risk identification rules based on the target frequent pattern tree comprises: Identifying a plurality of frequent item sets in the target frequent pattern tree, and generating candidate rules based on every two non-empty subsets in all non-empty subsets of each frequent item set to obtain a plurality of candidate rules; determining the time attenuation confidence and the time sequence adjustment lifting degree of each candidate rule; determining candidate rules with time attenuation confidence degrees larger than or equal to a first confidence degree threshold and time sequence adjustment lifting degrees larger than or equal to a first lifting degree threshold as first target rules, sequencing the first target rules according to the sequence from large to small of the time sequence adjustment lifting degrees, merging the sequenced continuous first target rules, and generating the key risk identification rules; And determining candidate rules with time attenuation confidence degrees larger than or equal to a second confidence degree threshold and time sequence adjustment lifting degrees smaller than or equal to a second lifting degree threshold as second target rules, merging the second target rules, and generating the silencing risk identification rules.
6. The risk identification rule generation method of claim 5, wherein the time-decay confidence satisfies a third formula, the third formula being: ; the time sequence adjustment lifting degree meets a fourth formula, wherein the fourth formula is as follows: ; Wherein, the In order to decay the confidence level over time, The weighted support degree of intersection sets of two non-empty subsets A and B is provided, wherein A is the non-empty subset with the front occurrence time of the included risk event, and B is the non-empty subset with the rear occurrence time of the included risk event; Weighted support for non-null subset a; The average time difference from the risk event with the latest occurrence time in A to the risk event with the earliest occurrence time in B is shown as lambda is the attenuation coefficient; The lifting degree is adjusted for the time sequence; weighted support for non-null subset B; The time difference from the risk event with the earliest occurrence time in A to the risk event with the latest occurrence time in B.
7. The risk identification rule generation method of claim 6, further comprising: According to a fifth formula, calculating risk indexes of all first target rules in the key risk identification rules, and sequencing all first target rules in the key risk identification rules according to the order of the risk indexes from high to low, wherein the fifth formula is as follows: ; Wherein, the Risk index for the first target rule; 、、 are all weighted and ; For the weighted support of the first target rule, The maximum value of the weighted support degree in each first target rule in the key risk identification rule is added; the confidence level is attenuated for the time of the first target rule, A maximum value of the time attenuation confidence coefficient in each first target rule in the key risk identification rule is identified; adjusting a logarithmic transformation of the degree of lifting for the time sequence of the first target rule; According to a sixth formula, calculating risk indexes of all second target rules in the silent risk identification rules, and sequencing all second target rules in the silent risk identification rules according to the order of the risk indexes from high to low, wherein the sixth formula is as follows: ; Wherein, the Risk index for the second target rule; 、、 are all weighted and ; For the weighted support of the second target rule, The maximum value of the weight support degree in each second target rule in the silencing risk identification rule is added; the confidence level is attenuated for the time of the second target rule, A maximum value of time attenuation confidence in each second target rule in the silencing risk identification rule is identified; And adjusting reverse mapping of the lifting degree for the time sequence of the second target rule.
8. The risk identification rule generating device is characterized by comprising an acquisition unit and a processing unit; the acquisition unit is used for acquiring a plurality of risk event chains and a plurality of non-occurrence inhibition events in the operation process of the oil-gas pipe network, wherein the risk event chains are used for representing the plurality of risk events arranged according to time sequence, and the non-occurrence inhibition events are used for representing the inhibition events which do not occur and have associated risk events; The processing unit is used for constructing an initial frequent pattern tree based on the multiple risk event chains, and inserting each non-occurrence suppression event into the initial frequent pattern tree as an independent branch to obtain a target frequent pattern tree; the processing unit is further configured to generate a key risk identification rule and a silent risk identification rule based on the target frequent pattern tree, where the key risk identification rule is used to identify an abnormal event associated with a risk event in the oil-gas pipe network, and the silent risk identification rule is used to identify a suppression event associated with the risk event in the oil-gas pipe network.
9. A computer device comprising a processor, the processor being coupled to a memory, the memory for storing computer-executable instructions, the processor executing the computer-executable instructions stored in the memory to cause the computer device to implement the risk identification rule generation method of any one of claims 1-7.
10. A computer readable storage medium storing computer executable instructions which, when run on a computer device, cause the computer device to implement the risk identification rule generation method of any one of claims 1 to 7.
11. A computer program product comprising computer-executable instructions which, when run on a computer device, cause the computer device to implement the risk identification rule generation method of any one of claims 1 to 7.

Description

Risk identification rule generation method, apparatus, device, medium and program product Technical Field The present application relates to the field of oil and gas pipeline technologies, and in particular, to a risk identification rule generating method, apparatus, device, medium, and program product. Background The failure of a certain part of the oil and gas pipeline often easily causes chain reaction, thereby causing disaster accidents of the oil and gas pipeline and causing larger loss. However, the related technology often cannot support efficient and comprehensive recognition of oil and gas pipeline faults, and occurrence of disaster accidents of the oil and gas pipeline is difficult to effectively avoid. Disclosure of Invention The application aims to provide a risk identification rule generation method, device, equipment, medium and program product, which are used for efficiently and comprehensively identifying oil and gas pipeline faults and effectively avoiding oil and gas pipeline disaster accidents. In order to achieve the above purpose, the application adopts the following technical scheme: The embodiment of the application provides a risk identification rule generation method, which comprises the steps of obtaining a plurality of risk event chains and a plurality of non-occurrence inhibition events in the operation process of an oil-gas pipe network, wherein the risk event chains are used for representing the plurality of risk events which are arranged according to time sequence, the non-occurrence inhibition events are used for representing the non-occurrence inhibition events which are associated with the risk events, constructing an initial frequent pattern tree based on the plurality of risk event chains, inserting each non-occurrence inhibition event as an independent branch into the initial frequent pattern tree to obtain a target frequent pattern tree, generating a key risk identification rule and a silencing risk identification rule based on the target frequent pattern tree, wherein the key risk identification rule is used for identifying abnormal events which are associated with the risk events in the oil-gas pipe network, and the silencing risk identification rule is used for identifying the inhibition events which are associated with the risk events in the oil-gas pipe network. Based on the method, the risk event chain and the non-occurrence inhibition event can be combined to generate the target frequent pattern tree, and the key risk identification rule and the silencing risk identification rule are further generated, so that the key risk and the silencing risk are simultaneously mined, and disaster chain risk mining analysis is more efficient and comprehensive, thereby being capable of being used for efficiently and comprehensively identifying oil and gas pipeline faults and effectively avoiding oil and gas pipeline disaster accidents. In some embodiments, an initial frequent pattern tree is built based on a plurality of risk event chains, each non-occurrence suppression event is inserted into the initial frequent pattern tree as an independent branch, and before the target frequent pattern tree is obtained, the method further comprises the steps of determining occurrence time intervals among a plurality of risk events in the risk event chains, merging the risk events with occurrence time intervals smaller than or equal to a preset time interval into the same event, and obtaining a compressed risk event chain. In some embodiments, constructing an initial frequent pattern tree based on a plurality of risk event chains includes dividing a plurality of item sets based on the plurality of risk event chains, setting a consequence severity weight of each risk event in each item set according to a disaster result severity level of each risk event in each item set, determining a weighted support of each item set according to the consequence severity weights of each risk event in each item set, determining an item set with the weighted support being greater than or equal to a support threshold as a candidate item set, and constructing the initial frequent pattern tree based on the plurality of candidate item sets. The outcome severity weight and weighted support for each risk event satisfies a first formula: ; Wherein, the Is a set of items; count (I) is the number of times item set I appears in multiple item sets, N is the total number of multiple item sets; The outcome severity weight for the ith risk event in item set I. In some embodiments, the method further comprises setting causal position weights of the respective risk events in each candidate set according to event attributes of the respective risk events in each candidate set, and generating ranking weights of the respective risk events in each candidate set, wherein the causal position weights of the risk events and the ranking weights of the risk events satisfy a second formula: ; Wherein, the Ranking weights for the ith