CN105843210A - Power transformer defect information data mining method - Google Patents

Abstract

The invention discloses a power transformer defect data mining method, comprising: screening defect attributes on a historical defect data set _D0 of a power transformer to form _a defect data set D1 _; filling or deleting defect attributes in D1 to reduce data Noise; construct new attributes based _on the existing attributes of D1, discretize continuous attributes and rationally stratify subtype attributes to form defect data set D2 _; calculate the correlation between input attributes and target attributes, and delete irrelevant Attributes and the remaining attributes constitute the defect data set D ₃ ; use the Apriori algorithm to calculate the association relationship between the attributes of the defect data set; extract effective association rules, analyze the defect factors of power transformers, and form an association rule knowledge base. The invention has the following advantages: multi-dimensional and multi-layer excavation of defects in power transformers, convenient and quick extraction of correlations among defect attributes, providing basis for state evaluation of power transformers, and improving the accuracy of state evaluation.

Description

Data Mining Method for Power Transformer Defect Information

technical field

The invention relates to the technical field of data mining, in particular to a data mining method for power transformer defect information.

Background technique

The reliable and stable operation of the power system is the premise and basis for ensuring the power needed for economic development, social progress and improvement of people's living standards. As an important equipment in the power system, the power transformer undertakes functions such as power transmission and distribution, voltage conversion, etc., and its operating status and health level directly affect the safety, stability and reliability of the power system. The condition-based maintenance technology based on condition evaluation carries out active maintenance according to the condition evaluation results, and reasonably arranges the maintenance time and maintenance items, so as to reduce the failure rate of equipment and ensure the reliable operation of equipment.

Defect information, as an important data basis for power transformer condition evaluation, has the characteristics of numerous sources, rich attributes, large amount of data, low accuracy and high redundancy. In the past, the analysis of power transformer defect information mainly relied on statistical analysis, which could neither quickly obtain high-value information nor detect the potential relationship between defect information attributes, and lacked sufficient support for the evaluation of power transformer operating status.

Contents of the invention

The present invention aims to solve at least one of the above-mentioned technical problems.

Therefore, the object of the present invention is to propose a data mining method for power transformer defects.

In order to achieve the above object, an embodiment of the present invention discloses a method for mining defect data of power transformers, including the following steps: S1: Screen defect attributes on the historical defect data set D ₀ of power transformers, and retain those that may potentially be related to the mining target Relevant data to form a defect data set D ₁ ; S2: For the defect attributes in the defect data set D ₁ , at least one of filling in missing, correcting errors, directly deleting, deleting redundancy and eliminating inconsistency is used to reduce data noise; S3: Construct new attributes through data integration and data transformation for the redundant attributes _of the defect data set D1, discretize the continuous attributes and stratify the classification attributes to form the defect data set D2 _; S4: based on the defect data set D _2. Calculate the correlation between the input attribute and the target attribute, and delete irrelevant attributes to form the defect data set D ₃ ; S5: Based on the defect data set D ₃ , set the minimum support and minimum confidence, and use the Apriori algorithm to calculate the defect data set attributes S6: Extract effective association rules, analyze defect factors of power transformers, and form association rule knowledge base.

According to the power transformer defect data mining method of the embodiment of the present invention, through the associated mining method for power transformer defect information, a suitable defect data set is established, and problems such as omission, inconsistency and redundancy of multi-source heterogeneous defect data are eliminated, Reasonable screening of data attributes, using the Apriori algorithm to realize multi-dimensional and multi-layer mining of power transformer defect data, mining the relationship between defect attributes, providing a basis for state evaluation, improving the accuracy of power transformer state evaluation, and ensuring a more reasonable power transformer maintenance strategy efficient.

In addition, the power transformer defect data mining method according to the above-mentioned embodiments of the present invention may also have the following additional technical features:

Further, in step S1, the mining dimensions _of defect data set D1 include but are not limited to continuous type and type historical data.

Further, step S2 further includes: redefining the defect type of the power transformer based on the mining target, deleting repeated defects occurring in the same equipment, and retaining the record of the first defect.

Further, in step S3, the dimensions of the defect data set _D2 include running time, operating mechanism type, defect handling method, defect occurrence time, manufacturer qualification, equipment model, cause of defect occurrence, equipment operating environment, equipment operating location and At least one of substation names.

Further, step S4 further includes: performing feature selection _on the attributes of the defect data set D2, calculating the importance of each attribute based on the chi-square check, sorting the attributes according to the importance value, and retaining defect attributes whose importance is higher than the preset threshold .

Further, using the Apriori algorithm to calculate the association relationship between the attributes of the defect data set further includes: using the Apriori algorithm to mine the association rules between the defect-related factors of the power transformer, wherein the defect-related factors of the power transformer include manufacturers, Years of operation and type of defect.

Additional aspects and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

Description of drawings

The above and/or additional aspects and advantages of the present invention will become apparent and comprehensible from the description of the embodiments in conjunction with the following drawings, wherein:

Fig. 1 is a flowchart of a power transformer defect data mining method according to an embodiment of the present invention.

detailed description

Embodiments of the present invention are described in detail below, examples of which are shown in the drawings, wherein the same or similar reference numerals designate the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the figures are exemplary only for explaining the present invention and should not be construed as limiting the present invention.

In describing the present invention, it should be understood that the terms "center", "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", " The orientations or positional relationships indicated by "vertical", "horizontal", "top", "bottom", "inner" and "outer" are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and Simplified descriptions, rather than indicating or implying that the device or element referred to must have a particular orientation, be constructed and operate in a particular orientation, and thus should not be construed as limiting the invention. In addition, the terms "first" and "second" are used for descriptive purposes only, and should not be understood as indicating or implying relative importance.

In the description of the present invention, it should be noted that unless otherwise specified and limited, the terms "installation", "connection" and "connection" should be understood in a broad sense, for example, it can be a fixed connection or a detachable connection. Connected, or integrally connected; it may be mechanically connected or electrically connected; it may be directly connected or indirectly connected through an intermediary, and it may be the internal communication of two components. Those of ordinary skill in the art can understand the specific meanings of the above terms in the present invention in specific situations.

These and other aspects of embodiments of the invention will become apparent with reference to the following description and drawings. In these descriptions and drawings, some specific implementations of the embodiments of the present invention are specifically disclosed to represent some ways of implementing the principles of the embodiments of the present invention, but it should be understood that the scope of the embodiments of the present invention is not limited by This restriction. On the contrary, the embodiments of the present invention include all changes, modifications and equivalents coming within the spirit and scope of the appended claims.

First, let me introduce the basic concepts involved in using the Apriori algorithm: association rules and basic concepts.

Association rules represent rules that have certain association relationships between different domains in the database that meet the specified requirements. Let I={i ₁ , i ₂ ,...i _n } be a set of items. Given a transaction database D, where each transaction T is a collection of items, satisfying if itemsets and is shaped like The implication of is called an association rule, and X and Y are the premise and conclusion of the association rule;

The basic parameters for measuring association rules include Support, Confidence and Lift.

Support (support): Indicates the support of itemset X∪Y, that is, the ratio of both itemsets X and Y in the transaction database D, recorded as:

In the formula: |T(X∨Y)| represents the number of transactions including both X and Y; |T| represents the total number of transactions.

Confidence (confidence): Indicates the proportion of transactions where X appears in the transaction database D and also contains Y, recorded as:

Lift (lift): The lift ratio is the ratio of the confidence of the transaction database D to the confidence of the subsequent item, which is recorded as:

$\mathrm{<span\; class="notranslate">\; l</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; f</span>}\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; \&DoubleRightArrow;</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; \&DoubleRightArrow;</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>}$

Lift ratio Lift means that under the condition of X occurrence, the conditional probability of Y occurrence is the ratio of the prior probability of Y occurrence. When the boost ratio is greater than 1, it indicates that It is a directional relationship, that is, the appearance of X promotes the appearance of Y; when lift<1, it indicates that the appearance of X reduces the possibility of Y appearing.

A power transformer defect data mining method according to an embodiment of the present invention will be described below with reference to the accompanying drawings.

Fig. 1 is a flowchart of a power transformer defect data mining method according to an embodiment of the present invention. Please refer to Fig. 1, the power transformer defect data mining method of the embodiment of the present invention comprises the following steps:

S1: Screen defect attributes on the historical defect data set D ₀ of power transformers, retain relevant data that may be potentially related to the mining target, and form a defect data set D ₁ .

Specifically, the irrelevant attributes of the data set _D0 are deleted based on expert knowledge, including non-associated attributes such as "defect discoverer", "defect elimination person", "responsible unit" and "time of entering the maintenance department". Afterwards, 23 items of defect attributes are reserved to form a defect data set D ₁ .

S2: For the defect attributes in the defect data set D1, at least _one of filling in missing, correcting errors, directly deleting, deleting redundancy and eliminating inconsistency is used to reduce data noise.

Specifically, there are attribute value missing errors, _outliers , redundancy and inconsistencies in the defect data set D1. According to the existing problems, according to the types and characteristics of mining targets and missing attributes, the processing methods are as follows:

S201: Due to the need to measure the reliability of equipment from different manufacturers, it is necessary to compare the first occurrence time of each equipment defect. Repeated defects on the same equipment will seriously affect the data distribution of the equipment, making the correlation calculation results unreliable. Therefore, according to "functional location", " Considering factors such as "substation", "equipment number" and "defect occurrence time", only the first defect is retained and the remaining redundant defects are deleted.

S202: For the classification attribute, for example, there are missing values or outliers in the attribute "power transformer model", the missing values can be filled or the wrong values can be corrected through joint analysis of factors such as "substation name", "voltage level" and "manufacturer" . When the missing data cannot be compensated by joint analysis of other attributes, the record will be deleted.

S3: Construct new attributes for the redundant attributes of defect data set D ₁ through data integration and data transformation, discretize continuous attributes and stratify categorical attributes to form defect data set D ₂ .

Specifically, some attributes in the defect dataset _D1 are redundant and have low value density. New attributes are constructed through data integration and data transformation, which not only reduces the attribute dimension, but also improves the expressive ability of the defect dataset. The specific method includes the following steps:

S301: Based on the two attribute items of "defect handling measures" and "defect handling results", construct the "defect handling method" defect, divide the defect handling measures into simple methods, replacement methods, comprehensive methods and other methods, and divide various The division of treatments into these four methods makes the data easier to understand.

S302: Construct the attribute of "equipment operating life" through "defect discovery time" and "equipment commissioning time", and quantify the continuous attribute based on expert knowledge, and divide it into "operating life N<1 year", "1 year < Operational period N<5 years", "5 years <operational period N<10 years", "10 years <operational period N<15 years", "15 years <operational period N<20 years" and "operational period N>20 years " and other 6 attribute values.

S303: According to the attributes of "device type" and "manufacturer", construct the attribute of "manufacturer qualification" and divide it into three attribute values of "foreign capital", "joint venture" and "domestic".

S304: Quantify and layer the data in the defect data set D ₁ to establish a power transformer defect data set D ₂ .

S4: Based on the defect data set D ₂ , calculate the correlation between the input attribute and the target attribute, and delete irrelevant attributes to form the defect data set D ₃ . It should be noted that for different mining targets, the target attributes are different.

Specifically, the power transformer defect data set D ₂ still contains many attributes, and the purpose of further streamlining the data set is achieved by examining the importance of attributes. The importance of attributes can be jointly investigated from two aspects: first, from the perspective of the attributes themselves; second, from the perspective of the correlation between input attributes and target attributes. From the attribute itself, the important attribute should carry more information, that is, the variance is larger. According to the actual situation, some standards for measuring the variance are formulated. When the attribute variance is less than the specified standard, it is considered unimportant. From the point of view of the correlation between input attributes and target attributes, important attributes should have significant significance in the classification prediction of target attributes. For different types of input attributes and target attributes, the measurement methods adopted are also different. The specific situation is shown in Table 1, which is a table of different variable test methods.

Table 1 Measurement methods of different types of variables

Since the defect attributes of power transformers are concentrated into classification attributes, the chi-square check method is firstly used to measure the correlation between attributes. Chi-square verification belongs to the category of hypothesis testing in statistics, and mainly involves the following four steps: proposing a null hypothesis, selecting and calculating test statistics, determining the level of significance, conclusion and decision-making. Among them, the test statistic of the chi method test is the Peason chi square statistic, and its data is defined as:

${\mathrm{<span\; class="notranslate">\; \&chi;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; r</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; c</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}\frac{{<span\; class="notranslate">\; (</span>{{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}}^{\mathrm{<span\; class="notranslate">\; o</span>}}<span\; class="notranslate">\; -</span>{{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}}^{\mathrm{<span\; class="notranslate">\; e</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{{{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}}^{\mathrm{<span\; class="notranslate">\; e</span>}}}$

In the formula: r is the number of rows of the contingency table, c is the number of columns of the contingency table; f ^o is the observed frequency, f ^e is the expected frequency.

The importance of measuring attributes is measured by "importance". Importance (Importance) is not the size of the correlation coefficient. This value is calculated by calculating the probability p of the chi-square statistic at a specific significance level, and by comparing the (1-p) values between variables to measure its importance; usually the Larger values indicate that the variable is more important.

Set the importance I>0.95, when the importance value is greater than 0.95, keep it, and if the importance is less than 0.9, it will be deleted directly; for the attributes with high importance, delete the attributes whose importance is lower than the standard, and form the power transformer defect data set D ₃ .

S5: Based on the defect data set D ₃ , set the minimum support and minimum confidence, and use the Apriori algorithm to calculate the relationship between the attributes of the defect data set.

Specifically, the main flow of the Apriori algorithm is as follows:

Input: defect database D ₃ ; minimum support minsup

Output: set R of all strong association rules in _D3

algorithm:

F ₁ =find_frequent_1-itemset(D ₃ )

for(k=2; k++)

{C _k ＝ appriori_gen(F _k-1 ,minsup);

foreachtransactiont∈D

{C _t = subset(C _k ,t);.

foreach candidate c ∈ C _t

c.count++; }

${\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; C</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; |</span>\frac{\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; c</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; t</span>}}{<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; |</span>}<span\; class="notranslate">\; \&Greater\; Equal;</span>\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}\mathrm{<span\; class="notranslate">\; the\; s</span>}\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ;</span><span\; class="notranslate">\; \}</span>$

return F = ∪ _k F _k ;

R = generate_rule(F);

Return(R);

procedure apriori_gen(F _k-1 :frequent(k-1)-itemsets);

minsup: minimum support threshold)

foreach item set f ₁ ∈ F _k-1

foreach item set f ₂ ∈ F _k-1

if((f ₁ [1]=f ₂ [1]∧f ₁ [2]=f ₂ [2])∧∧f ₁ [k-2]=f ₂ [k-2]∧f ₁ [k- 1]<f ₂ [k-1]))

then{c=f ₁ [1], f ₁ [2],, f ₁ [k-1], f ₂ [k-1]};

ifhas_infrequent_subset(c,F _k-1 )then

deleteC;

else addcto C _k ; }

return C _k ;

procedure has_infrequent_subset(c:candidatek-itemset; F _k-1 :frequent(k-1)-itemset)

foreach(k-1)-subsetsofc

returnTRUE;

else return FALSE;

S6: Extract effective association rules, analyze defect factors of power transformers, and form association rule knowledge base.

In an example of the present invention, taking the power transformer defect type as the latter item, the association rules extracted based on the apriori algorithm are shown in Table 2, and Table 2 is a table of strong association rules for power transformers.

Table 2 Strong association rules for power transformers

From the above table, it can be seen that the probability of defects in the cooling system of the equipment of manufacturer A is nearly 90% during the operation period of 5-10 years. The manufacturer's operation and maintenance strategy for power transformer equipment. By changing the antecedent and posterior attributes of association rules, the defect factors of power transformers are analyzed from multi-angle, multi-dimensional, and multi-level associations.

The power transformer defect data mining method of the embodiment of the present invention combines the particularity of the power industry, applies the association rules to the selection and analysis of the association rules of the power transformer defect information, and proposes to use the association rules in the data mining technology to analyze the power transformer defect data. Analysis of the basic ideas and specific solutions. Through the extraction and analysis of strong association rules, a reference basis is provided for the state evaluation of power transformers. The accuracy of state evaluation is higher, and the maintenance strategy of power transformers is more reasonable and more targeted.

In addition, other configurations and functions of the power transformer defect data mining method of the embodiment of the present invention are known to those skilled in the art, and will not be repeated in order to reduce redundancy.

In the description of this specification, descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific examples", or "some examples" mean that specific features described in connection with the embodiment or example , structure, material or feature is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Although the embodiments of the present invention have been shown and described, those skilled in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the invention is defined by the claims and their equivalents.

Claims (6)

1. a power transformer defective data method for digging, it is characterised in that comprise the following steps:

S1: the historical defect data collection D to power transformer₀Screening defect attribute, retains and excavates target There may be the related data of potential association, form defective data collection D₁；

S2: to defective data collection D₁In defect attribute by fill up disappearance, right the wrong, directly deletion, In deletion redundancy and elimination discordance, at least one is to reduce noise data；

S3: to defective data collection D₁Redundant attributes by data integration and the new attribute of data transition structure, Discretization carried out for continuous attribute and categorical attribute is layered, forming defective data collection D₂；

S4: based on defective data collection D₂, calculate the dependency between input attribute and objective attribute target attribute, delete not Association attributes constitutes defective data collection D₃；

S5: based on defective data collection D₃, minimum support and min confidence are set, use Apriori to calculate Method calculates the incidence relation between defective data set attribute；

S6: extract efficient association rule, analyze the defect factors of power transformer, form correlation rule knowledge Storehouse.

Power transformer defective data method for digging the most according to claim 1, it is characterised in that In step sl, defective data collection D₁Excavation dimension including but not limited to electric pressure, manufacturer, Unit type, time of putting into operation, disfigurement discovery time, defect type, defect processing measure and transformer station's title In interior continuous, classifying type historical data.

Power transformer defective data method for digging the most according to claim 1, it is characterised in that Step S2 farther includes: redefines power transformer defect type based on excavating target, deletes same The repeated defects that equipment occurs, retains defect record first.

Power transformer defective data method for digging the most according to claim 1, it is characterised in that In step s3, defective data collection D₂Dimension include operation time, operating mechanism type, defect processing Mode, defect time of origin, manufacturer's qualification, unit type, defect occurrence cause, equipment run ring In border, equipment operational site and transformer station's title at least one.

Power transformer defective data method for digging the most according to claim 1, it is characterised in that Step S4 farther includes: for defective data collection D₂Attribute carry out feature selection, based on card side verify Calculate each Attribute Significance, carry out attribute sequence according to importance value, retain importance degree higher than predetermined threshold value Defect attribute.

Power transformer defective data method for digging the most according to claim 1, it is characterised in that The incidence relation using Apriori algorithm to calculate between defective data set attribute farther includes: use Apriori Algorithm carries out the association rule mining between the defect correlative factor of described power transformer, wherein, described electric power The defect correlative factor of transformator includes manufacturer, runs the time limit and defect type.