CN107451402A - A kind of equipment health degree appraisal procedure and device based on alarm data analysis - Google Patents

Abstract

The invention mainly belongs to the field of equipment health degree analysis, and in particular relates to a method and device for evaluating equipment health degree based on alarm data analysis. The present invention regards the alarm event as an evaluation index, determines the index score and index weight by the alarm feature, evaluates the health degree of the equipment objectively, and provides a basis for guiding subsequent equipment failure management and maintenance work.

Description

A method and device for evaluating equipment health based on alarm data analysis

technical field

The invention mainly belongs to the field of equipment health degree analysis, and in particular relates to a method and device for evaluating equipment health degree based on alarm data analysis.

Background technique

Equipment health refers to the goodness of the overall operation of the equipment, and is a comprehensive evaluation of the operation status of the entire equipment. The commonly used health assessment method is based on the health assessment of equipment operating status parameters, that is, by monitoring the equipment in operation and obtaining equipment status parameters to evaluate the status. However, in practical applications, the equipment is packaged in the cabinet, and it is difficult to obtain the parameters of the equipment in operation, and most of them are tested by manual discharge test and conductance test, which is not only complicated, but also requires strong professional requirements.

It has been reported in the literature that some methods based on alarm data analysis are used to evaluate the health of communication base stations/systems and/or networks in the field of communication and network. Alarms reflect equipment failures, and alarm analysis can effectively evaluate equipment status. As an intuitive representation of equipment status, alarms are easier to obtain than data. Using alarm data as a data set, mining features related to equipment operating status, and establishing a health evaluation model can well avoid time-consuming issues based on equipment status parameters. Existing problems. However, the method in the report is only applied to the fields of communication and network, and the selection and scoring of evaluation indicators are different for different evaluation objects, so its application has limitations.

The main idea of evaluating the health degree is to evaluate and score each parameter reflecting the performance status of the evaluation object, and then weight and fuse each evaluation result to obtain the final evaluation result. The selection and scoring of evaluation indicators are different for different evaluation objects. Compared with the evaluation score, it is more difficult to determine the weight of each parameter in the equipment performance evaluation. The currently applied methods mainly include: fuzzy analytic hierarchy process, linear weighting method, principal component analysis method, fuzzy comprehensive evaluation method, entropy weight method, BP neural network method and support vector machine method. Among them, the existing work using fuzzy analytic hierarchy process and linear weighting method all determine weights in a subjective way, and human subjective factors may cause bias. The principal component analysis method is determined by the user's attention, which still has a subjective component. The shortcoming of the fuzzy comprehensive evaluation method is that it cannot solve the evaluation problem related to fuzziness and randomness. The entropy weight method uses the entropy of each parameter as the weight. The neural network method and the support vector machine method use the fuzzy comprehensive evaluation method to obtain the evaluation value and establish the model through training. The quality of the model is subject to the fuzzy comprehensive evaluation method. In short, there are human factors in the weight determination methods of the existing evaluation methods, which will inevitably affect the objectivity of the evaluation results.

Contents of the invention

In view of the above problems, the present invention proposes a method and device for evaluating equipment health based on alarm data analysis. The present invention regards the alarm event as an evaluation index, determines the index score and index weight by the alarm feature, evaluates the health degree of the equipment objectively, and provides a basis for guiding subsequent equipment failure management and maintenance work.

The present invention is achieved through the following technical solutions:

A method for evaluating equipment health based on alarm data analysis, the method comprising the following steps:

Data collection: collect the alarm data of the equipment and write it into the database according to the alarm occurrence time;

Evaluation data selection: select the alarm event name, alarm level, alarm occurrence time, alarm end time, and alarm repetition times from the database as evaluation data;

Selection of evaluation indicators: select alarm events with different alarm event names as evaluation indicators to form an evaluation indicator set;

Calculation of health degree: Score each alarm event by calculating the number of alarm occurrences and the average alarm duration of each evaluation index within the time to be evaluated, and assign different weights to the alarm events according to the alarm level. The scores of each evaluation index are combined with the weight to calculate the health of the device .

Further, the scoring of each alarm event by calculating the number of alarm occurrences and the average alarm duration of each evaluation index within a certain period of time is specifically:

21) Divide the time to be evaluated into q scoring moments according to the time axis, {t ₁ ,t ₂ ,…,t _i ,…,t _q } (t ₀ is the starting point of time), and the scoring interval is {Δ ₁ =t ₁ -t ₀ ,Δ ₂ ＝t ₂ -t ₁ ,...,Δ _j ＝t _j -t _j-1 ,...,Δ _q ＝t _q -t _q-1 }; for the evaluation index in the evaluation index set, that is, alarm The event a _i is scored at the scoring time t _j (j=1,2,…,q), and the score of the alarm event a _i at the time point t _j is

in:

is the number of occurrences of the evaluation index a _i within the interval Δ _j ;

To evaluate the average alarm duration of the indicator a _i in the interval Δ _j ,

with are the end time and occurrence time of the alarm when the evaluation index (alarm event) a _i occurs for the kth time within the duration Δ _j ;

23) Repeat the above scoring at other scoring moments to obtain the scoring sequence of the evaluation index a _i within the time to be evaluated

24) Score all the evaluation indicators in the evaluation index set to obtain the evaluation index score set {X ₁ ,X ₂ ,…,X _i ,…X _m }, the set element X _i is a row or column vector, including q scoring moments The obtained indicator score, m is the number of evaluation indicators.

Further, the assigning different weights to alarm events according to the alarm level is specifically:

31) Construct a judgment matrix:

The judgment matrix was constructed using the 1-9 ratio scale (Table 1) commonly used in the AHP. Comparing all the evaluation indicators with each other, according to the alarm level according to the 1-9 ratio scale method, the relative importance of each indicator is given, and the judgment matrix A is constructed;

Table 1 1-9 ratio scale method

In the present invention, Table 1 can also be described by the following mathematical formula;

b _ij is the element in row i and column j in the judgment matrix A, representing the relative importance of the i-th evaluation index relative to the j-th evaluation index, b _ij =1/b _ji ;

L is the difference between the alarm levels of the i-th evaluation index and the j-th evaluation index;

32) Calculate the weight:

w _i is the weight of the i-th evaluation index;

in

M _i =b _i1 ×b _i2 ×... b _im ;

m is the number of indicators.

Further, eliminate the correlation between various evaluation indicators;

41) Index score normalization processing:

Evaluation index normalized score set {Y ₁ ,Y ₂ ,…,Y _i ,…Y _m },

in,

is the score of a _i and the maximum score in Xi _, is the minimum _score in Xi;

m is the number of evaluation indicators;

42) Use the Gram-Schmidt orthogonal method to eliminate the correlation between indicators, and obtain a group of orthogonal sets {β _i ,i=1,2,...,m} that have no correlation with each other;

β ₁ = Y ₁ ;

β ₂ ＝Y ₂ -[Y ₂ ,β ₁ ]/[β ₁ ,β ₁ ]×β ₁ , [Y ₂ ,β ₁ ] is the inner product of Y ₂ and β ₁ , [β ₁ ,β ₁ ] is Inner product of β ₁ and β ₁ ;

The orthogonal set {β _i ,i=1,2,...,m} obtained after eliminating the correlation is used as the score of the evaluation index combined with the weight to calculate the equipment health degree. β _i is a row or column vector, including k scoring moments The score of the evaluation index.

Further, calculating the health degree of the equipment by combining the scores of the evaluation indicators after the correlation elimination is combined with the weight is as follows: the scores and weights of the evaluation indicators are weighted and summed to obtain the health score of the equipment, and the calculation formula is:

is the equipment health score, H is a row or column vector, which contains the health score at all scoring moments, β _i is the score sequence of the i-th evaluation index after eliminating correlation, and contains the scores of the evaluation index at q scoring moments , ω _i is the weight of the score, m is the number of evaluation indicators. Health score at each scoring moment It is equal to the sum of the products of the evaluation indicator score and the score weight at this moment.

Further, the consistency check is carried out on the weight distribution, so that the consistency ratio CR=CI/RI<0.10, otherwise, the element values of the judgment matrix should be adjusted (using the 1-9 ratio scaling method shown in Table 1), Redistribute the values of the weight coefficients until the consistency ratio is less than 0.1;

Among them, the consistency index Where λmax is the largest characteristic root of the judgment matrix, A is the judgment matrix, W is the weight matrix, AW is the product of two matrices, (AW) _i is the i-th element of the product of the matrix, and w _i is the weight of the i-th indicator;

RI is the average random consistency index value.

Furthermore, before the evaluation data is selected, the alarm data is preprocessed to remove bad data such as flash memory and duplication.

A device health evaluation device based on alarm data analysis, the device uses the above evaluation method, and the device includes a data acquisition module, an evaluation data selection module, an evaluation index selection module, and a health degree calculation module;

The data acquisition module collects the alarm data of the equipment, and writes it into the database according to the alarm occurrence time;

The evaluation data selection module selects the alarm event name, alarm level, alarm occurrence time, alarm end time, and alarm repetition times from the database as evaluation data;

The evaluation indicator selection module selects various alarm events according to the name of the alarm event, and uses them as evaluation indicators to form an evaluation indicator set;

The health degree calculation module scores each alarm event by calculating the number of alarm occurrences and the average alarm duration of each alarm event within the time to be evaluated to obtain the alarm event score value, and eliminates the correlation to obtain the alarm event score value, and according to the alarm level Different weights are assigned to the alarm events to obtain the weight value of the alarm event, and the weighted sum of the alarm event score value and the alarm event weight value is obtained to obtain the device health degree.

Further, the health calculation module includes an alarm event scoring module, a weight calculation module, and a weighted sum module;

The alarm event scoring module multiplies the number of times the alarm event occurs within the time to be evaluated by the average alarm occurrence time to obtain the alarm event score value; and eliminates the correlation between the score values of different alarm events to obtain the alarm event score value ;

The weight calculation module uses the alarm level of the alarm event to construct a judgment matrix, and uses the judgment matrix to obtain the weight value of the alarm event;

The weighted sum module multiplies the alarm event score value and the alarm event weight value correspondingly to obtain the contribution component of the alarm event to the health degree; adds the contribution components of each alarm event in the evaluation index set to obtain the Health.

Further, the device is a switching power supply device, and the alarm events include battery power supply alarm, DC output voltage too high alarm, AC input phase loss alarm, rectifier module failure alarm, AC input frequency is too high alarm, AC input voltage is too low Alarm, AC input voltage is too high alarm, AC input frequency is too low alarm.

Further, the device is a storage battery device, and the alarm event includes the total voltage is too low, the total voltage is too high, the voltage of a single cell is too high, the voltage of a single cell is too low, and the voltage of the middle point of the battery pack is unbalanced.

Beneficial technical effect of the present invention:

1) The present invention evaluates the health of equipment based entirely on alarm data, has almost no technical requirements, and is very convenient to implement;

2) The present invention uses the alarm event as an evaluation index, and the index score and index weight are determined by the alarm characteristics, and the evaluation result is very intuitive and easy to understand;

3) According to the device health evaluation model proposed by the present invention, not only the real-time health score of the device can be obtained, but also the change curve of the device health degree can be drawn over a period of time.

Description of drawings

FIG. 1 is a schematic diagram of the daily change trend of the health degree of the switching power supply within a month calculated in Embodiment 1 of the present invention.

detailed description

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

On the contrary, the invention covers any alternatives, modifications, equivalent methods and schemes within the spirit and scope of the invention as defined by the claims. Further, in order to make the public have a better understanding of the present invention, some specific details are described in detail in the detailed description of the present invention below. The present invention can be fully understood by those skilled in the art without the description of these detailed parts.

Example 1

The method of the invention is used to evaluate the health of a switching power supply in a communication base station. The goal is to evaluate the change trend of the health of the switching power supply within a month, and score the health of the switching power supply every other day, that is, the scoring interval is 1 day.

S1. Collect the alarm data of the switching power supply of the communication base station, span a month, and write it into the database according to the time sequence of the alarm occurrence.

S2. Eliminate invalid data such as missing, repeated, and transient data in the alarm data, and extract data attributes such as alarm event name, alarm level, alarm occurrence time, alarm end time, and alarm repetition times.

Missing alarm means that the alarm data information is incomplete, and some field values are empty. Alarm transient means that the time from alarm generation to elimination is very short; alarm repetition means that the attribute values of several alarm records are completely consistent. Alarm data with these characteristics either cannot provide complete data attributes, are missing data, or do not reflect correct alarm events, so they are eliminated through preprocessing before evaluation.

S3. Selecting health evaluation indicators. There are a total of eight different alarm events in the alarm data, which are selected as the health evaluation indicators of the switching power supply, as shown in Table 2.

S4. Calculate the health degree of the power supply equipment. The observation period is one month (31 days), and the health evaluation is performed once a day, that is, the scoring interval Δ is one day.

Scoring intervals can be uniform or non-uniform. The size of the scoring interval can be the time granularity that the maintainer is interested in; at the same time, it is necessary to ensure that the scoring interval contains a certain amount of alarm data. For example, if the time to be evaluated is 1 month, and you want to examine the daily trend of equipment health, then the scoring interval is selected as 1 day. The scoring interval can also choose a smaller time granularity, such as 6 hours, 1 hour, etc. The applicant found that the scoring intervals were 1 day, 6 hours, and 1 hour, and the change trend of equipment health was roughly the same.

On each day, count the number of alarms and the average alarm duration of each indicator in the past day, and calculate the scores of each indicator in 31 time windows, as shown in Table 3. Take the evaluation indicator a1 at the first scoring moment, that is, the first day, as an example to illustrate the calculation process of the indicator score. In the first day, a1 occurred 5 times, and the duration of each time was 29.05, 66.45, 10.38, 60.85 and 87.12 minutes, so

minute, This value is the element in row 1, column 1 in Table 3.

Table 2 Evaluation indicators of switching power supply

Table 3 Switching Power Supply Index Scores

The index scores are normalized so that the index scores are all between 0 and 1. The results are shown in Table 4. Taking the value of X ₁ at time t ₁ as The normalized value calculation of is taken as an example. the largest value in the X ₁ sequence is 253.85, the minimum is 0, so the normalized value at time t ₁ This value is the value of the element in row 1 and column 1 of Table 4.

Table 4 Normalized indicator scores

Then, use the Schmidt orthogonal method to process the index scores, and the index scores after eliminating the correlation are shown in Table 5.

Table 5 Scores of each index after eliminating index correlation

According to the alarm level of each index of the switching power supply in Table 2, the weight of each index is determined by the analytic hierarchy process. The judgment matrix and weight results are shown in Table 6. There are eight kinds of alarm events, so m=8. There are three different alarm levels, so the judgment matrix takes the value 1-3-5. Take the weight calculation of the evaluation index a1 as an example. The corresponding evaluation value in the judgment matrix, that is, the values in the first row in Table 6 are 1, 1, 3, 3, 5, 5, 5, 5, so the product M ₁ of the elements in this row = 1×1×3×3 ×5×5×5×5＝5625, M ₁ to the 8th power, its 8th root In the same way, the product of the elements of the remaining seven rows can be obtained Normalize to get the weight:

Other weights can be calculated in the same way.

Table 6. Calculation of weights using AHP

Consistency metrics are computed on the weights. consistency index Where λ _max is the largest characteristic root of the judgment matrix, namely A is the judgment matrix, W is the weight matrix, AW is the product of two matrices, (AW) _i is its i-th element, and w _i is the weight of the i-th indicator.

Calculate the consistency ratio CR=CI/RI, where RI is the average random consistency index value, as shown in Table 7, where m is the index number in the judgment matrix.

Table 7 Average random consistency index RI value

When the consistency ratio CR=CI/RI<0.10, it is considered that the judgment matrix has relatively good consistency, and the weight distribution is considered reasonable.

In Example 1, the judgment matrix

weight matrix

Since the number of evaluation indicators is 8, RI=1.41 is known from Table 7, so CR=CI/RI=0.012395÷1.41=0.008791<0.1 passed the consistency test.

The health of the switching power supply on the first day In the same way, the health of other days can be calculated. Taking the health degree as the vertical axis and the corresponding observation days as the horizontal axis, draw the health degree change curve of the switching power supply every day within a month, as shown in Figure 1. The curve reflects the daily change trend of the switching power supply health degree within a month.

At the same time, this embodiment also includes a device health evaluation device based on alarm data analysis, the device uses the above evaluation method, and the device includes a data acquisition module, an evaluation data selection module, an evaluation index selection module, and a health degree calculation module;

After the monitored power supply equipment generates an alarm, it sends it to the data acquisition module through the RS485, RS232 or Ethernet interface. The format of the alarm event name, alarm event type, alarm level, alarm occurrence time, alarm end time, alarm repetition times, alarm reason, alarm summary, etc. is written into the database.

The data acquisition module collects the alarm data of the equipment, and writes it into the database according to the alarm occurrence time;

The evaluation data selection module selects the alarm event name, alarm level, alarm occurrence time, alarm end time, and alarm repetition times from the database as evaluation data;

According to the name of the alarm event, the evaluation index selection module selects various alarm events as health evaluation indicators to form an evaluation index set (the alarm events in the index set have no order requirements);

The alarm events include battery power supply alarm, DC output voltage too high alarm, AC input phase loss alarm, rectifier module failure alarm, AC input frequency too high alarm, AC input voltage too low alarm, AC input voltage Low frequency warning.

The health degree calculation module scores each alarm event by calculating the number of alarm occurrences and the average alarm duration of each alarm event within the time to be evaluated to obtain the alarm event score value, and eliminates the correlation to obtain the alarm event score value, and according to the alarm level Different weights are assigned to the alarm events to obtain the weight value of the alarm event, and the weighted sum of the alarm event score value and the alarm event weight value is obtained to obtain the device health degree.

Further, the health calculation module includes an alarm event scoring module, a weight calculation module, and a weighted sum module;

The alarm event scoring module multiplies the number of times the alarm event occurs within the time to be evaluated by the average alarm occurrence time to obtain the alarm event score value, and eliminates the correlation to obtain the alarm event score value;

The weight calculation module uses the alarm level of the alarm event to construct a judgment matrix, and uses the judgment matrix to obtain the weight value of the alarm event;

The weighted sum module multiplies the alarm event score value and the alarm event weight value correspondingly to obtain the contribution component of the alarm event to the health degree; adds the contribution components of each alarm event in the evaluation index set to obtain the Health.

Claims (10)

1. An equipment health degree evaluation method based on alarm data analysis is characterized by comprising the following steps:

data acquisition: collecting alarm data of equipment, and writing the alarm data into a database according to alarm occurrence time;

selecting evaluation data: selecting the alarm event name, the alarm level, the alarm occurrence time, the alarm ending time and the alarm repetition times from a database as evaluation data;

selecting evaluation indexes: selecting different alarm events as evaluation indexes to form an evaluation index set;

and (3) calculating the health degree: the alarm event score value is obtained by calculating the alarm occurrence frequency and the average alarm duration of each alarm event in the time to be evaluated, different weights are given to the alarm events according to the alarm levels to obtain alarm event weight values, and the equipment health degree is obtained by combining the score value of each alarm event with the weight values.

2. The evaluation method according to claim 1, wherein the scoring of each alarm event by calculating the number of alarm occurrences and the average alarm duration for each evaluation index within a certain period of time to obtain the alarm event score value is specifically:

21) dividing the time to be evaluated into q scoring moments on the time axis, { t₁，t₂，...，t_i，...，t_qA scoring interval of { Δ }₁＝t₁-t₀，Δ₂＝t₂-t₁，…，Δ_j＝t_j-t_j-1，…，Δ_q＝t_q-t_q-1}；t₀Is the starting point of time;

the scoring intervals are uniform intervals or non-uniform intervals, and are selected according to the evaluation time;

22) for the evaluation index a in the evaluation index set_iAt the scoring time t_jScore, alarm event a_iAt a point in time t_jIs scored asm is the number of evaluation indexes, j is 1, 2.., q, q is the number of scoring time;

wherein:

to evaluate the index a_iAt intervals of time delta_jThe number of internal occurrences;

to evaluate the index a_iAt intervals of time delta_jThe average alarm duration within which the alarm occurred,

andrespectively is an evaluation index a_iAt a time duration of_jThe alarm ending time and the occurrence time when the kth time occurs;

23) repeating the scoring at other scoring moments to obtain an evaluation index a in the time to be evaluated_iScore sequence of (3)

24) All alarm events in the evaluation index set are scored to obtain the evaluation index set { X }₁，X₂，…，X_i，…X_m}, set element X_iThe row or column vector includes index scores obtained at q scoring times, and m is the number of evaluation indexes.

3. The evaluation method according to claim 1, wherein said assigning different weights to alarm events according to alarm level is specifically:

31) constructing a judgment matrix:

comparing all the evaluation indexes with each other pairwise, obtaining according to the alarm level, and constructing a judgment matrix A;

<mrow> <msub> <mi>b</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>2</mn> <mo>|</mo> <mi>L</mi> <mo>|</mo> <mo>-</mo> <mn>1</mn> <mo>,</mo> <mi>L</mi> <mo>&gt;</mo> <mn>0</mn> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mn>1</mn> <mo>,</mo> <mi>L</mi> <mo>=</mo> <mn>0</mn> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mfrac> <mn>1</mn> <mrow> <mn>2</mn> <mo>|</mo> <mi>L</mi> <mo>|</mo> <mo>-</mo> <mn>1</mn> </mrow> </mfrac> <mo>,</mo> <mi>L</mi> <mo>&lt;</mo> <mn>0</mn> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow>

b_ijto determine the element in the ith row and the jth column of the matrix A, which represents the relative importance of the ith evaluation index with respect to the jth evaluation index, b_ij＝1/b_ji；

L is the difference value of the alarm levels of the ith evaluation index and the jth evaluation index;

32) calculating the weight:

<mrow> <msub> <mi>w</mi> <mi>i</mi> </msub> <mo>=</mo> <mover> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>&amp;OverBar;</mo> </mover> <mo>/</mo> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </msubsup> <mover> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>&amp;OverBar;</mo> </mover> <mo>;</mo> </mrow>

w_ithe weight of the ith evaluation index;

wherein

<mrow> <mover> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>&amp;OverBar;</mo> </mover> <mo>=</mo> <mroot> <msub> <mi>M</mi> <mi>i</mi> </msub> <mi>m</mi> </mroot> </mrow>

M_i＝b_i1×b_i2×.....b_im；

m is the index number.

4. The evaluation method according to claim 1, wherein the correlation between the respective evaluation indexes is eliminated;

41) index score normalization processing:

evaluation index normalization to diversity { Y }₁，Y₂，…，Y_i，…Y_m}，

Wherein,

is a_iScore sequence X of (2)_iThe maximum score of (a) of (b),is X_iMinimum of (1)Scoring;

m is the number of evaluation indexes;

42) utilizing the gram-Schmidt orthogonal method to eliminate the correlation between the indexes to obtain a group of orthogonal sets { β) without correlation between the indexes_i，i＝1，2，…，m}；

β₁＝Y₁；

β₂＝Y₂-[Y₂，β₁]/[β₁，β₁]×β₁，[Y₂，β₁]Is Y₂And β₁Inner product of [ β ]₁，β₁]Is β₁And β₁Inner product of (d);

<mrow> <msub> <mi>&amp;beta;</mi> <mi>i</mi> </msub> <mo>=</mo> <msub> <mi>Y</mi> <mi>i</mi> </msub> <mo>-</mo> <mfrac> <mrow> <mo>&amp;lsqb;</mo> <msub> <mi>Y</mi> <mn>2</mn> </msub> <mo>,</mo> <msub> <mi>&amp;beta;</mi> <mn>1</mn> </msub> <mo>&amp;rsqb;</mo> </mrow> <mrow> <mo>&amp;lsqb;</mo> <msub> <mi>&amp;beta;</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mi>&amp;beta;</mi> <mn>1</mn> </msub> <mo>&amp;rsqb;</mo> </mrow> </mfrac> <mo>&amp;times;</mo> <msub> <mi>&amp;beta;</mi> <mn>1</mn> </msub> <mo>-</mo> <mo>...</mo> <mo>-</mo> <mfrac> <mrow> <mo>&amp;lsqb;</mo> <msub> <mi>Y</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>&amp;beta;</mi> <mrow> <mi>i</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>&amp;rsqb;</mo> </mrow> <mrow> <mo>&amp;lsqb;</mo> <msub> <mi>&amp;beta;</mi> <mrow> <mi>i</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>,</mo> <msub> <mi>&amp;beta;</mi> <mrow> <mi>i</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>&amp;rsqb;</mo> </mrow> </mfrac> <mo>&amp;times;</mo> <msub> <mi>&amp;beta;</mi> <mrow> <mi>i</mi> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>,</mo> <mi>i</mi> <mo>&gt;</mo> <mn>2</mn> <mo>;</mo> </mrow>

to eliminate the orthogonal set β obtained after correlation_iβ degree of health of the equipment was calculated by combining score of 1,2, …, m as evaluation index and weight_iThe row or column vector contains the scores of the evaluation index at q scoring times.

5. The evaluation method according to claim 1, wherein the score obtained after the elimination of the correlation of each evaluation index is combined with the weight calculation to obtain the health degree of the equipment, specifically: the scores and the weights of all the evaluation indexes are weighted and summed to obtain the health degree score of the equipment, and the calculation formula is

<mrow> <mi>H</mi> <mo>=</mo> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </msubsup> <msub> <mi>w</mi> <mi>i</mi> </msub> <msub> <mi>&amp;beta;</mi> <mi>i</mi> </msub> <mo>;</mo> </mrow>

Is the equipment health score, H is a row or column vector containing the health scores at all scored moments, β_iIs the ith evaluation index elimination correlation score sequence which comprises q beatsScore, ω, of evaluation index at time_iIs the weight of the score, m is the number of evaluation indexes; health score per scored timeEqual to the sum of the products of the evaluation index score and the score weight at that time.

6. The evaluation method according to claim 3, wherein the weight distribution is subjected to a consistency check such that the consistency ratio CR ═ CI/RI <0.10, otherwise, the values of the elements of the decision matrix are adjusted and the values of the weight coefficients are redistributed until the consistency ratio is less than 0.1;

wherein the consistency indexWherein λ_maxTo determine the largest root of the features of the matrix,a is the decision matrix, W is the weight matrix, AW is the product of the two matrices, (AW)_iIs the i-th element of the product of the matrices, w_iIs the weight of the ith index;

RI is the average random consistency index value.

7. The assessment method of claim 1 wherein the alarm data is pre-processed to remove missing, snap, duplicate bad data prior to assessment data selection.

8. An equipment health degree evaluation device based on alarm data analysis is characterized by comprising a data acquisition module, an evaluation data selection module, an evaluation index selection module and a health degree calculation module;

the data acquisition module acquires alarm data of the equipment and writes the alarm data into a database according to alarm occurrence time;

the evaluation data selection module selects the alarm event name, the alarm level, the alarm occurrence time, the alarm ending time and the alarm repetition times from the database as evaluation data;

the evaluation index selection module selects different alarm events as health degree evaluation indexes to form an evaluation index set;

the health degree calculation module scores alarm events by calculating the alarm occurrence times and the average alarm duration of each evaluation index in the time to be evaluated to obtain alarm event score values, gives different weights to the alarm events according to the alarm levels to obtain alarm event weight values, and weights the alarm event score values and the alarm event weight values to obtain the equipment health degree.

9. The apparatus of claim 8, wherein the health calculation module comprises an alarm event scoring module, a weight calculation module, a weighted sum module;

the alarm event scoring module multiplies the occurrence frequency of the alarm event in the time to be evaluated by the average alarm occurrence time length to obtain an alarm event score value; eliminating the correlation among the alarm events to obtain the score of the alarm event;

the weight calculation module utilizes the alarm level of the alarm event to construct a judgment matrix, and utilizes the judgment matrix to obtain the weight value of the alarm event;

the weighting sum module multiplies the alarm event score and the alarm event weight value correspondingly to obtain the contribution component of the alarm event to the health degree; and adding the contribution components of each alarm event in the evaluation index set to obtain the health degree of the equipment to be evaluated.

10. The apparatus of claim 8, wherein the device is a switching power supply device or a battery device;

when the equipment is switching power supply equipment, the alarm event comprises a battery power supply alarm, a direct current output voltage overhigh alarm, an alternating current input open-phase alarm, a rectifier module fault alarm, an alternating current input frequency overhigh alarm, an alternating current input voltage overlow alarm, an alternating current input voltage overhigh alarm and an alternating current input frequency overlow alarm;

when the equipment is storage battery equipment, the alarm events comprise that the total voltage is too low, the total voltage is too high, the voltage of a certain single battery is too low, and the voltage of the middle point of the battery pack is unbalanced.