CN114444738B - Electrical equipment maintenance cycle generation method - Google Patents
Electrical equipment maintenance cycle generation method Download PDFInfo
- Publication number
- CN114444738B CN114444738B CN202210362556.XA CN202210362556A CN114444738B CN 114444738 B CN114444738 B CN 114444738B CN 202210362556 A CN202210362556 A CN 202210362556A CN 114444738 B CN114444738 B CN 114444738B
- Authority
- CN
- China
- Prior art keywords
- electrical equipment
- training
- attribute
- maintenance
- decision tree
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The invention discloses a method for generating a maintenance cycle of electrical equipment, which comprises the following steps: constructing a training sample set consisting of electrical equipment data, generating a plurality of training subsets from the training sample set by using a self-service method, and then respectively carrying out electrical equipment characteristic screening on the electrical equipment data in each training subset to obtain a training characteristic corresponding to each training subset; selecting a decision tree generation algorithm matched with the nth training subset from an ID3 algorithm and a CART algorithm according to the training feature types and the number of the nth training subset, and processing the training features by using the decision tree generation algorithm to generate an electrical equipment maintenance period decision tree corresponding to the nth training subset; generating a plurality of decision trees of the maintenance period of the electrical equipment so as to obtain a decision forest of the maintenance period of the electrical equipment; and inputting the characteristic value of the electrical equipment with the maintenance period to be obtained into the electrical equipment maintenance period decision forest, and outputting the electrical equipment maintenance period by the electrical equipment maintenance period decision forest.
Description
Technical Field
The invention relates to a method for generating a maintenance cycle of electrical equipment, and belongs to the field of maintenance cycle prediction.
Background
The electrical equipment is high in price, the maintenance cost is high, the requirements on the quality and the speciality of maintenance personnel are high, correspondingly, more labor cost is consumed in each maintenance, and therefore the manpower resource capable of maintaining the electrical equipment is limited. On the other hand, if the electrical equipment cannot be maintained in time, additional unnecessary wear may be generated, and even a safety accident may be caused.
In order to maintain each electrical device as efficiently and in time as possible under the condition of limited human maintenance resources, it is necessary to obtain maintenance cycle data of each electrical device as much as possible. Similarly, the maintenance cycle of the electrical equipment basically needs to be judged by professional electrical equipment maintenance personnel, and in order to reduce or avoid the consumption of the part of human maintenance resources, how to accurately and efficiently acquire the maintenance cycle of each electrical equipment as far as possible without depending on the human judgment of the professional maintenance personnel is an urgent problem to be solved.
Disclosure of Invention
The present invention provides a method for generating a maintenance cycle of an electrical device, which overcomes the disadvantages of the prior art.
To solve the technical problem, the invention adopts the following technical scheme:
a method for generating an electrical equipment maintenance cycle includes the steps of:
step S1: constructing a training sample set consisting of electrical equipment data, generating a plurality of training subsets from the training sample set by using a self-service method, and then respectively carrying out electrical equipment characteristic screening on the electrical equipment data in each training subset to obtain a training characteristic corresponding to each training subset;
step S2: selecting a decision tree generation algorithm matched with the nth training subset from an ID3 algorithm and a CART algorithm according to the training feature type and the number of the nth training subset, and processing the training features by using the decision tree generation algorithm to generate an electrical equipment maintenance period decision tree corresponding to the nth training subset;
step S3: repeating the step S2 for multiple times, wherein the value of n is changed in each repeated process to generate a plurality of decision trees of the maintenance period of the electrical equipment, and further obtain a decision forest of the maintenance period of the electrical equipment;
step S4: and inputting the characteristic value of the electrical equipment with the maintenance period to be obtained into the electrical equipment maintenance period decision forest, and outputting the electrical equipment maintenance period by the electrical equipment maintenance period decision forest.
The beneficial effects of the invention are as follows:
the training characteristics of the electrical equipment are usually associated with each other, so that the number of the types of the training characteristics used for generating the decision tree of the maintenance period of the electrical equipment is increased, the accuracy of the decision tree of the maintenance period of the electrical equipment is unlikely to be obviously improved, and moreover, the generation time of the decision tree of the maintenance period of the electrical equipment and the judgment time are greatly increased due to the increase of the number of the types of the training characteristics used, based on the reality, tests aiming at the ID3 algorithm and the CART algorithm can find that the generation of the decision tree of the maintenance period of the electrical equipment by adopting the ID3 algorithm has high accuracy and high generation speed under the condition that the number of the types of the training characteristics is not large, the number of the types of the training characteristics used is further increased on the basis of the ID3 algorithm, so that the accuracy of the decision tree of the maintenance period of the electrical equipment is not obviously increased, in contrast, in the case of the CART algorithm, if the number of training feature types is small, the accuracy of the generated electrical equipment maintenance period decision tree is not high, and the generation time is relatively short, but if the number of training feature types to be used is further increased, both the accuracy and the generation time of the generated electrical equipment maintenance period decision tree are increased. On the basis, a proper decision tree generation algorithm is adopted to generate the electrical equipment maintenance period decision tree according to the training feature types and the number of the training subsets, so that the judgment accuracy is improved, the judgment time is reduced, the accuracy of the electrical equipment maintenance period decision forest is balanced with the judgment time, and the accurate maintenance period of the electrical equipment is obtained without professional intervention of electrical equipment maintainers.
In step S1 of the present invention, the electrical device data is first preprocessed before forming the training sample set, and the preprocessing process is as follows: the method comprises the steps of obtaining a maximum value Tmax of an electrical equipment maintenance time interval and a minimum value Tmin of the electrical equipment maintenance time interval of all electrical equipment data, setting maintenance batch number p, Tmax-Tmin = p T according to electrical equipment maintenance capacity, wherein p is an integer not less than 2, sequentially setting maintenance periods [ Tmin, Tmin + T ], [ Tmin + T, Tmin +2T ], [ Tmin + (p-2) T, Tmin + (p-1) T ], [ Tmin + (p-1) T, Tmax ], and associating each electrical equipment data with a corresponding maintenance period through a respective electrical equipment maintenance time interval.
In the preprocessing process, the characteristics of the electrical equipment, which are irrelevant to the maintenance period, in the data of the electrical equipment are removed.
In the step S2 of the present invention,
if the training features of the nth training subset contain the environmental state attribute, the decision tree generation algorithm matched with the nth training subset is an ID3 algorithm;
if the training features of the nth training subset do not contain the environmental state attribute and the number of the training features of the nth training subset is more than 5, the decision tree generation algorithm matched with the nth training subset is a CART algorithm;
if the training features of the nth training subset do not include the environmental status attribute and the number of training features of the nth training subset is not greater than 5, the decision tree generation algorithm matched with the nth training subset is the ID3 algorithm.
After the preprocessing is finished, the electrical equipment data only comprise an environment state attribute, a unit water temperature attribute, a starting battery voltage attribute, an output line voltage attribute, an output phase current attribute, an active power attribute, a reactive power attribute, a power factor attribute, a generator set number attribute, a combustion engine rotating speed attribute and a combustion engine temperature attribute, and the number of training characteristics of the nth training subset is not more than 8.
The invention adjusts the value of p so that each maintenance period has corresponding electrical equipment data.
In step S3, each electrical equipment maintenance cycle decision tree is tested by using a test sample, a threshold is set, electrical equipment maintenance cycle decision trees with accuracy lower than the threshold are rejected, and an electrical equipment maintenance cycle decision forest is formed by using electrical equipment maintenance cycle decision trees with accuracy higher than the threshold.
In step S3, a part of the electrical equipment maintenance period decision tree with accuracy higher than the threshold is selected, so that the number ratio of the electrical equipment maintenance period decision tree generated by the CART algorithm to the electrical equipment maintenance period decision tree generated by the ID3 algorithm in the electrical equipment maintenance period decision forest is 0.9-1.1.
Other features and advantages of the present invention will be disclosed in more detail in the following detailed description of the invention and the accompanying drawings.
Drawings
The invention is further described below with reference to the accompanying drawings:
fig. 1 is a flowchart of a method for generating a maintenance cycle of an electrical device according to an embodiment of the present invention.
Detailed Description
The technical solutions of the embodiments of the present invention are explained and illustrated below with reference to the drawings of the embodiments of the present invention, but the following embodiments are only preferred embodiments of the present invention, and not all embodiments. Based on the embodiments in the implementation, other embodiments obtained by those skilled in the art without any creative effort belong to the protection scope of the present invention.
In the following description, the appearances of the indicating orientation or positional relationship such as the terms "inner", "outer", "upper", "lower", "left", "right", etc. are only for convenience in describing the embodiments and for simplicity in description, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and are not to be construed as limiting the present invention.
Example (b):
the embodiment provides a method for generating a maintenance cycle of electrical equipment, which comprises the following steps:
step S0: obtaining a large amount of different electrical equipment data, wherein in this embodiment, 3773 pieces of electrical equipment data are obtained in total, where the electrical equipment data is historical electrical equipment data, each piece of electrical equipment data corresponds to a different piece of electrical equipment, the electrical equipment data includes a large amount of feature data of corresponding electrical equipment, one piece of feature data is an electrical equipment maintenance time interval determined by a professional electrical equipment maintenance worker for the electrical equipment in a historical maintenance process, and the electrical equipment maintenance time interval is an ideal theoretical time interval between two adjacent times of maintenance of the electrical equipment, rather than an actual maintenance time interval;
if the actual maintenance time interval for the electrical equipment is less than the electrical equipment maintenance time interval, the electrical equipment is subjected to excessive maintenance, and the human resources of the electrical equipment maintenance personnel are wasted;
usually, the electrical equipment characteristic data included in the electrical equipment data includes 103 attributes, such as an equipment brand attribute, an environmental state attribute, a unit water temperature attribute, a starting battery voltage attribute, an output line voltage attribute, an output phase current attribute, an active power attribute, a reactive power attribute, a power factor attribute, a generator set quantity attribute, a combustion engine rotation speed attribute, and a combustion engine temperature attribute, besides an electrical equipment maintenance time interval;
the method aims to predict and judge the electrical equipment maintenance time interval through electrical equipment characteristics, which are not the electrical equipment maintenance time interval, in the electrical equipment data, so that the electrical equipment characteristics which are obviously irrelevant to the electrical equipment maintenance time interval or have low relevance are eliminated to avoid adverse effects on the prediction speed and the prediction accuracy;
the electrical equipment maintenance time interval is usually a point value, and is limited by the amount of the historical electrical equipment data, and the value of the electrical equipment maintenance time interval in the historical electrical equipment data may only be 1 day, 2 days, 12 days, 10 days, or 15 days, etc., so when the electrical equipment maintenance time interval prediction is directly performed on an electrical equipment for which no electrical equipment maintenance time interval is obtained based on the original historical electrical equipment data, the obtained electrical equipment maintenance time interval result may only be the electrical equipment maintenance time interval value existing in the historical electrical equipment data for 1 day, 2 days, 12 days, 10 days, or 15 days, etc., however, according to actual experience, the actual electrical equipment maintenance time interval result of the electrical equipment for which no electrical equipment maintenance time interval is obtained may also be the result that the electrical equipment maintenance time interval result does not exist in the historical electrical equipment data for 1.2 days, 13 days, etc., therefore, the accuracy of predicting the electrical equipment maintenance time interval of unknown electrical equipment by directly adopting original historical electrical equipment data is very low, and in the actual electrical equipment maintenance process, even if the electrical equipment maintenance time interval is known, the actual time interval between two times of maintenance is basically unlikely to just reach the electrical equipment maintenance time interval, and based on the practical conditions and situation consideration, the maintenance requirement on the electrical equipment under the condition of not wasting human resources can be basically ensured as long as the actual maintenance time interval of the electrical equipment by maintenance personnel is not greatly different from the electrical equipment maintenance time interval;
the electrical equipment with the similar numerical values of the electrical equipment maintenance time intervals is classified into one maintenance batch, maintenance personnel mainly carry out unified maintenance on the electrical equipment of the same maintenance batch each time the maintenance personnel start working, if the human resources for maintaining the electrical equipment are sufficient, the number of the maintenance batches can be more, the difference between the electrical equipment maintenance time intervals of the electrical equipment corresponding to the same maintenance batch is smaller, the number of the electrical equipment is also smaller, and therefore the occurrence of the situation that the electrical equipment is overhauled or not maintained in time is reduced;
factors affecting the number p of the maintenance batches, such as human resources, maintenance tools and the like, which are collectively called the maintenance capability of the electrical equipment;
for the above reasons, the electrical device data preprocessing process further includes the following steps:
obtaining a maximum value Tmax and a minimum value Tmin of electrical equipment maintenance time intervals in all historical electrical equipment data, setting p according to electrical equipment maintenance capacity, wherein p is an integer not less than 2, Tmax-Tmin = p T, then obtaining maintenance periods [ Tmin, Tmin + T ], [ Tmin + T, Tmin +2T ], [ Tmin + (p-2) T, Tmin + (p-1) T ], [ Tmin + (p-1) T, Tmax ] in turn, finding a maintenance period corresponding to the electrical equipment maintenance time interval in each historical electrical equipment data, thereby enabling the historical electrical equipment data to be associated with the corresponding maintenance period, for example, if p =3, obtaining the maintenance periods [ Tmin, Tmin + T ], [ Tmin +2T, Tmax ] in turn, wherein if the electrical equipment maintenance period of one piece of electrical equipment data is Tmin + 5.1, the maintenance cycle of the historical electrical equipment data is [ Tmin + T, Tmin + 2T);
predicting the maintenance cycles of other electrical equipment according to the maintenance cycles of historical electrical equipment data, classifying the electrical equipment with the same maintenance cycle into the same maintenance batch, and further meeting the requirement of making an actual maintenance plan;
theoretically, the maintenance time interval of the electrical equipment of one electrical equipment should be any value between Tmax and Tmin, that is, it is required to ensure that each maintenance period has corresponding historical electrical equipment data, so as to ensure that each maintenance period is possible to be a prediction result when the maintenance period prediction is performed on the electrical equipment subsequently, avoid omission of the prediction result, and improve the prediction accuracy;
however, if the amount of the historical electrical equipment data is small, there is a possibility that a part of the maintenance period does not have corresponding historical electrical equipment data, and at this time, the number of the maintenance period is reduced by appropriately reducing the value of p, so that the time width of a single maintenance period is increased, and the problem is solved;
it should be noted that the device brand attribute in the historical electrical device data is not a technical attribute, and there is a certain correlation between the device brand attribute and the electrical device maintenance time interval as a point value, but the correlation is weak, and on this basis, the correlation between the device brand attribute and the maintenance period as a range value is further sharply weakened, so the device brand attribute in the historical electrical device data needs to be removed together in the preprocessing process;
finally, the electrical equipment data after the preprocessing in the embodiment only retains the environmental state attribute, the unit water temperature attribute, the starting battery voltage attribute, the output line voltage attribute, the output phase current attribute, the active power attribute, the reactive power attribute, the power factor attribute, the generator set number attribute, the combustion engine rotating speed attribute and the combustion engine temperature attribute, and meanwhile, the maintenance period associated with the electrical equipment data is also obtained.
Step S1: forming a training sample set by using a part of preprocessed electrical equipment data, wherein the rest of each electrical equipment data can be used as an independent test sample, generating a plurality of training subsets from the training sample set by using a self-help method, so that each training subset has the same number of electrical equipment data as the training sample set, and therefore each training subset in the embodiment also has 3773 pieces of electrical equipment data (a part of electrical equipment data in each training subset is repeated), and screening the electrical equipment data in the n-th training subset by using a random number and a random variety of electrical equipment features so as to obtain corresponding training features;
for example, for the 6 th training subset, the electrical equipment characteristics of the internal electrical equipment data are uniformly screened to obtain 4 training characteristics which are respectively a starting battery voltage attribute, an output line voltage attribute, an output phase voltage attribute and an output phase current attribute, for the 1002 th training subset, the electrical equipment characteristics of the internal electrical equipment data are uniformly screened to obtain 2 training characteristics which are respectively a combustion engine rotation speed attribute and a combustion engine temperature attribute, the screening process is repeated to finally obtain the training characteristics corresponding to each training subset, and the number and the type of the training characteristics corresponding to each training subset are correspondingly determined;
step S2: selecting a decision tree generation algorithm matched with the nth training subset from an ID3 algorithm and a CART algorithm according to the training feature type and the number of the nth training subset, and processing the training features of the nth training subset by using the decision tree generation algorithm to generate an electrical equipment maintenance period decision tree corresponding to the nth training subset;
each training subset can generate a decision tree of the maintenance period of the electrical equipment, each training feature serves as a classification node, the sequence of the classification nodes is determined by an ID3 algorithm or a CART algorithm, the method for generating the decision tree by the ID3 algorithm or the CART algorithm aiming at the training features is the prior art, and the embodiment is not repeated;
the test sample is input into the maintenance period decision tree of the electrical equipment, so that a maintenance period judgment result corresponding to the test sample can be obtained, and whether the judgment result of the maintenance period decision tree is accurate or not can be obtained by comparing the maintenance period corresponding to the test sample with the maintenance period judgment result. Similarly, for example, a case of p =3 is taken, if one of the test samples is input to the electrical equipment maintenance period decision tree, the obtained maintenance period determination result is [ Tmin, Tmin + T ], and the actual maintenance period corresponding to the test sample is [ Tmin + T, Tmin +2T "), it is determined that the electrical equipment maintenance period decision tree has determined the test sample incorrectly, but if the maintenance period determination result obtained by the electrical equipment maintenance period decision tree and the maintenance period corresponding to the test sample are both [ Tmin + T, Tmin + 2T"), it is determined that the electrical equipment maintenance period decision tree has determined the test sample correctly;
inputting a large number of test samples into the electrical equipment maintenance period decision tree in sequence, and further knowing whether the maintenance period judgment result of the electrical equipment maintenance period decision tree for each test sample is correct or not, so that the accuracy of the electrical equipment maintenance period decision tree can be finally obtained;
the method is characterized in that a certain correlation exists between most attributes of an environment state attribute, a unit water temperature attribute, a starting battery voltage attribute, an output line voltage attribute, an output phase current attribute, an active power attribute, a reactive power attribute, a power factor attribute, a generator set quantity attribute, a combustion engine rotating speed attribute and a combustion engine temperature attribute, so that the quantity of training feature types used for generating an electrical equipment maintenance period decision tree is increased, the accuracy of the electrical equipment maintenance period decision tree is unlikely to be obviously improved, the generation time of the electrical equipment maintenance period decision tree and the judgment time are greatly increased due to the increase of the quantity of the training feature types, and based on the practical situation, tests aiming at an ID3 algorithm and a CART algorithm can find that under the condition that the quantity of the training feature types is not large, the method for generating the electrical equipment maintenance period decision tree by adopting the ID3 algorithm is high in accuracy and high in generation speed, the number of the used training feature types is further increased on the basis of the ID3 algorithm, the accuracy of the electrical equipment maintenance period decision tree is not obviously increased, the generation time is greatly increased, and on the contrary, under the condition of adopting the CART algorithm, if the number of the training feature types is not large, the accuracy of the generated electrical equipment maintenance period decision tree is not high, the generation time is relatively short, but if the number of the used training feature types is further increased, the accuracy and the generation time of the generated electrical equipment maintenance period decision tree are increased. Particularly, the environmental state attribute is not a numerical value, but is a state value such as high temperature and high humidity, low temperature and high humidity, high temperature and low humidity, low temperature and low humidity, and the environmental state attribute is adopted as a training feature, the environmental state attribute in the electrical equipment maintenance period decision tree generated by the ID3 algorithm is used for judging that the environmental state attribute is closer to the root node position, even if the number of types of the training features used in the generation process is large, a part of the training features cannot play a classification effect, but the judgment accuracy of the final maintenance period is usually not very low;
based on the analysis and test results, in this step, when the training feature type and number of the nth training subset are just obtained, the training feature type and number are firstly analyzed, and if the training features of the nth training subset include the environmental state attribute, the decision tree generation algorithm matched with the nth training subset is the ID3 algorithm; if the training features of the nth training subset do not contain the environmental state attribute and the number of the training features of the nth training subset is more than 5, the decision tree generation algorithm matched with the nth training subset is a CART algorithm; if the training features of the nth training subset do not contain the environmental state attribute and the number of the training features of the nth training subset is not greater than 5, the decision tree generation algorithm matched with the nth training subset is the ID3 algorithm.
If the number of the types of the adopted training features exceeds a certain range, the judgment accuracy of the maintenance period cannot be further increased, and the judgment time is also prolonged, so that in the step S1, the number of the training features of each training subset is required to be ensured to be not more than 8;
step S3: repeating the step S2 for multiple times, wherein the value of n is changed in each repeated process to generate a plurality of decision trees of the maintenance period of the electrical equipment, and further obtain a decision forest of the maintenance period of the electrical equipment;
inputting the same test sample into each electrical equipment maintenance period decision tree in the electrical equipment maintenance period decision forest, wherein each electrical equipment maintenance period decision tree can obtain a judgment result of a maintenance period for the test sample, and the judgment result with the largest occurrence frequency is the judgment result of the electrical equipment maintenance period decision forest;
the accuracy of the forest decision of the maintenance period of the electrical equipment is improved by enlarging the scale, namely increasing the number of decision trees of the maintenance period of the electrical equipment, but the scale of the forest decision of the maintenance period of the electrical equipment is overlarge, and the judgment time is overlong when the maintenance period is judged;
overall, the ID3 algorithm generates a less training feature number for the electrical equipment maintenance cycle decision tree, the CART algorithm generates a more training feature number for the electrical equipment maintenance cycle decision tree, the ID3 algorithm generates an electrical equipment maintenance cycle decision tree with a slightly lower accuracy than the CART algorithm, but the judgment time is short, in order to balance the number of the decision trees of the maintenance period of the electrical equipment generated by the two algorithms and ensure that the decision forest of the maintenance period of the electrical equipment has high accuracy and high judgment speed, a part of the decision trees of the maintenance period of the electrical equipment with accuracy higher than a threshold value are removed in the step, thereby ensuring that the quantity ratio of the electric equipment maintenance period decision tree generated by the CART algorithm to the electric equipment maintenance period decision tree generated by the ID3 algorithm in the final electric equipment maintenance period decision forest is between 0.9 and 1.1.
Finally, the accuracy of the forest decision of the maintenance period of the electrical equipment in the embodiment is 70.06%, the generation time is only 9 minutes and 21 seconds, and in contrast, the accuracy of the forest decision of the maintenance period of the electrical equipment generated only by using the CART algorithm is only 64.69%, the generation time is 20 minutes and 23 seconds, the accuracy of the forest decision of the maintenance period of the electrical equipment generated only by using the ID3 algorithm is only 57.09%, and the generation time is 4 minutes and 14 seconds.
Step S4: and after the electrical equipment maintenance period decision forest is generated, inputting the characteristic value of the electrical equipment to be subjected to maintenance period into the electrical equipment maintenance period decision forest, and outputting the electrical equipment maintenance period by the electrical equipment maintenance period decision forest.
While the invention has been described with reference to specific embodiments, it will be understood by those skilled in the art that the present invention may be practiced without limitation to such specific embodiments. Any modification which does not depart from the functional and structural principles of the present invention is intended to be included within the scope of the claims.
Claims (5)
1. A method for generating a maintenance cycle of an electrical device, comprising the steps of:
step S1: constructing a training sample set consisting of electrical equipment data, generating a plurality of training subsets from the training sample set by using a self-service method, and then respectively carrying out electrical equipment characteristic screening on the electrical equipment data in each training subset to obtain a training characteristic corresponding to each training subset; in step S1, the electrical device data is first preprocessed before forming the training sample set, and the preprocessing process is as follows: obtaining a maximum value Tmax of an electrical equipment maintenance time interval and a minimum value Tmin of the electrical equipment maintenance time interval of all electrical equipment data, setting a maintenance batch number p, Tmax-Tmin = p T according to the electrical equipment maintenance capacity, wherein p is an integer not less than 2, sequentially setting maintenance periods [ Tmin, Tmin + T ], [ Tmin + T, Tmin +2T ], [ Tmin + (p-2). T, Tmin + (p-1). T ], [ Tmin + (p-1). T, Tmax ], and associating each electrical equipment data with a corresponding maintenance period through a respective electrical equipment maintenance time interval;
step S2: selecting a decision tree generation algorithm matched with the nth training subset from an ID3 algorithm and a CART algorithm according to the training feature types and the number of the nth training subset, and processing the training features by using the decision tree generation algorithm to generate an electrical equipment maintenance period decision tree corresponding to the nth training subset; in the step S2, in the step S,
if the training features of the nth training subset contain the environmental state attribute, the decision tree generation algorithm matched with the nth training subset is an ID3 algorithm;
if the training features of the nth training subset do not contain the environmental state attribute and the number of the training features of the nth training subset is more than 5, the decision tree generation algorithm matched with the nth training subset is a CART algorithm;
if the training features of the nth training subset do not contain the environmental state attribute and the number of the training features of the nth training subset is not more than 5, the decision tree generation algorithm matched with the nth training subset is an ID3 algorithm; step S3: repeating the step S2 for multiple times, wherein the value of n is changed in each repeated process to generate a plurality of decision trees of the maintenance period of the electrical equipment, and further obtain a decision forest of the maintenance period of the electrical equipment; in the step S3, each electrical equipment maintenance cycle decision tree is tested by using the test sample, a threshold is set, the electrical equipment maintenance cycle decision trees with the accuracy rate lower than the threshold are rejected, and the electrical equipment maintenance cycle decision forest is formed by using the electrical equipment maintenance cycle decision tree with the accuracy rate higher than the threshold;
step S4: and inputting the characteristic value of the electrical equipment with the maintenance period to be obtained into the electrical equipment maintenance period decision forest, and outputting the electrical equipment maintenance period by the electrical equipment maintenance period decision forest.
2. The electrical equipment maintenance cycle generation method of claim 1, wherein electrical equipment features that are not related to the maintenance cycle within the electrical equipment data are removed during the preprocessing.
3. The electrical equipment maintenance cycle generation method according to claim 1, wherein after the preprocessing is finished, the electrical equipment data only includes an environmental state attribute, a unit water temperature attribute, a starting battery voltage attribute, an output line voltage attribute, an output phase current attribute, an active power attribute, a reactive power attribute, a power factor attribute, a generator set number attribute, a combustion engine rotation speed attribute, and a combustion engine temperature attribute, and the number of the training features of the n-th training subset is not more than 8.
4. The electrical equipment maintenance cycle generation method of claim 1, wherein the value of p is adjusted so that each maintenance cycle has corresponding electrical equipment data.
5. The electrical equipment maintenance cycle generation method of claim 1, wherein in step S3, a part of the electrical equipment maintenance cycle decision tree with accuracy higher than the threshold is selected, so that the ratio of the number of the electrical equipment maintenance cycle decision tree generated by the CART algorithm to the number of the electrical equipment maintenance cycle decision tree generated by the ID3 algorithm in the electrical equipment maintenance cycle decision forest is between 0.9 and 1.1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210362556.XA CN114444738B (en) | 2022-04-08 | 2022-04-08 | Electrical equipment maintenance cycle generation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210362556.XA CN114444738B (en) | 2022-04-08 | 2022-04-08 | Electrical equipment maintenance cycle generation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114444738A CN114444738A (en) | 2022-05-06 |
| CN114444738B true CN114444738B (en) | 2022-09-09 |
Family
ID=81359750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210362556.XA Active CN114444738B (en) | 2022-04-08 | 2022-04-08 | Electrical equipment maintenance cycle generation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114444738B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114662997B (en) * | 2022-05-20 | 2022-08-09 | 成都运荔枝科技有限公司 | Cold chain transport capacity demand prediction method and cold chain transport capacity distribution method |
| CN116070787A (en) * | 2023-03-08 | 2023-05-05 | 中环洁集团股份有限公司 | Equipment maintenance period prediction method, system, equipment and readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107742160A (en) * | 2017-10-12 | 2018-02-27 | 深圳达实智能股份有限公司 | A kind of electromechanical equipment maintenance periodic time self-adapting Forecasting Methodology and device |
| CN109635962A (en) * | 2018-12-17 | 2019-04-16 | 广州甘来信息科技有限公司 | Based on repair time prediction technique, device, equipment and the storage medium from the machine of dealer |
| CN111105050A (en) * | 2019-12-23 | 2020-05-05 | 远景智能国际私人投资有限公司 | Method, device and equipment for generating fan maintenance plan and storage medium |
| CN112633617A (en) * | 2019-09-24 | 2021-04-09 | 北京国双科技有限公司 | Maintenance strategy generation method and device, computer equipment and readable storage medium |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2019101160A4 (en) * | 2019-09-30 | 2019-10-31 | Chen, Jingxuan MISS | Application of decision tree and random forest in cash loan |
| CN111340125A (en) * | 2020-03-03 | 2020-06-26 | 温州大学 | Method for training rupture data of endometriotic cyst based on random forest algorithm |
| CN112052875A (en) * | 2020-07-30 | 2020-12-08 | 华控清交信息科技(北京)有限公司 | Method and device for training tree model |
-
2022
- 2022-04-08 CN CN202210362556.XA patent/CN114444738B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107742160A (en) * | 2017-10-12 | 2018-02-27 | 深圳达实智能股份有限公司 | A kind of electromechanical equipment maintenance periodic time self-adapting Forecasting Methodology and device |
| CN109635962A (en) * | 2018-12-17 | 2019-04-16 | 广州甘来信息科技有限公司 | Based on repair time prediction technique, device, equipment and the storage medium from the machine of dealer |
| CN112633617A (en) * | 2019-09-24 | 2021-04-09 | 北京国双科技有限公司 | Maintenance strategy generation method and device, computer equipment and readable storage medium |
| CN111105050A (en) * | 2019-12-23 | 2020-05-05 | 远景智能国际私人投资有限公司 | Method, device and equipment for generating fan maintenance plan and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 决策树算法的比较与应用研究;张海燕等;《华北电力技术》;20170630;第2017年卷(第06期);第2节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114444738A (en) | 2022-05-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114444738B (en) | Electrical equipment maintenance cycle generation method | |
| CN112731159B (en) | Method for pre-judging and positioning battery faults of battery compartment of energy storage power station | |
| US20220239122A1 (en) | Server-side characterisation of rechargeable batteries | |
| CN107478455B (en) | Timing truncation reliability test method suitable for Weibull distribution type product | |
| KR20200140093A (en) | Prediction Method and Prediction System for predicting Capacity Change according to Charging / Discharging Cycle of Battery | |
| CN110221222A (en) | Cell safety blanking voltage prediction technique, device and battery management system | |
| CN110059845B (en) | Metering device clock error trend prediction method based on time sequence evolution gene model | |
| CN112147530B (en) | Battery state evaluation method and device | |
| CN109613440B (en) | Battery grading method, device, equipment and storage medium | |
| CN109800446B (en) | Method and device for estimating voltage inconsistency in discharging process of lithium ion battery | |
| CN109886328B (en) | Electric vehicle charging facility fault prediction method and system | |
| CN112904219A (en) | Big data-based power battery health state prediction method | |
| CN111861023A (en) | Statistical-based hybrid wind power prediction method and device | |
| CN114114039A (en) | Method and device for evaluating consistency of single battery cells of battery system | |
| CN112836720A (en) | Building operation and maintenance equipment abnormity diagnosis method and system and computer readable storage medium | |
| CN115201692A (en) | Method for analyzing and predicting residual life of lithium battery | |
| CN113379005A (en) | Intelligent energy management system and method for power grid power equipment | |
| CN117154263A (en) | Lithium battery cascade utilization charging and discharging system and control method | |
| CN116128690B (en) | Carbon emission cost value calculation method, device, equipment and medium | |
| CN108647838B (en) | Power grid situation sensing method based on random matrix theory and online sequence extreme learning machine | |
| CN116249186A (en) | Data processing method and device of wireless network equipment, storage medium and electronic equipment | |
| CN113671287B (en) | Intelligent detection method, system and readable storage medium for power grid automation terminal | |
| CN115270983A (en) | Switch cabinet fault prediction method based on AdaBoost-RBF algorithm | |
| CN115980607A (en) | Battery state of health (SOH) estimation method based on incremental capacity curve characteristic points | |
| CN111476392A (en) | Metering device clock error trend prediction method based on social perception |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |